From 3ac635367eb589bee8edcc722f812a89970e14b7 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Thu, 16 Oct 2025 13:55:27 -0500 Subject: drm/amd: Check that VPE has reached DPM0 in idle handler [Why] Newer VPE microcode has functionality that will decrease DPM level only when a workload has run for 2 or more seconds. If VPE is turned off before this DPM decrease and the PMFW doesn't reset it when power gating VPE, the SOC can get stuck with a higher DPM level. This can happen from amdgpu's ring buffer test because it's a short quick workload for VPE and VPE is turned off after 1s. [How] In idle handler besides checking fences are drained check PMFW version to determine if it will reset DPM when power gating VPE. If PMFW will not do this, then check VPE DPM level. If it is not DPM0 reschedule delayed work again until it is. v2: squash in return fix (Alex) Cc: Peyton.Lee@amd.com Reported-by: Sultan Alsawaf Reviewed-by: Sultan Alsawaf Tested-by: Sultan Alsawaf Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4615 Reviewed-by: Lijo Lazar Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c | 34 +++++++++++++++++++++++++++++---- 1 file changed, 30 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c index 474bfe36c0c2..aa78c2ee9e21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c @@ -322,6 +322,26 @@ static int vpe_early_init(struct amdgpu_ip_block *ip_block) return 0; } +static bool vpe_need_dpm0_at_power_down(struct amdgpu_device *adev) +{ + switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) { + case IP_VERSION(6, 1, 1): + return adev->pm.fw_version < 0x0a640500; + default: + return false; + } +} + +static int vpe_get_dpm_level(struct amdgpu_device *adev) +{ + struct amdgpu_vpe *vpe = &adev->vpe; + + if (!adev->pm.dpm_enabled) + return 0; + + return RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_request_lv)); +} + static void vpe_idle_work_handler(struct work_struct *work) { struct amdgpu_device *adev = @@ -329,11 +349,17 @@ static void vpe_idle_work_handler(struct work_struct *work) unsigned int fences = 0; fences += amdgpu_fence_count_emitted(&adev->vpe.ring); + if (fences) + goto reschedule; - if (fences == 0) - amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE); - else - schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT); + if (vpe_need_dpm0_at_power_down(adev) && vpe_get_dpm_level(adev) != 0) + goto reschedule; + + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE); + return; + +reschedule: + schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT); } static int vpe_common_init(struct amdgpu_vpe *vpe) -- cgit v1.2.3 From d0da43def215543d7b62a32af67e181619b3293a Mon Sep 17 00:00:00 2001 From: Daniel Palmer Date: Sat, 18 Oct 2025 14:44:49 +0900 Subject: drm/radeon: Clean up pdev->dev instances in probe Get a struct device pointer from the start and use it. Signed-off-by: Daniel Palmer Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/radeon_drv.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 88e821d67af7..eb5138da8d5b 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -262,6 +262,7 @@ static int radeon_pci_probe(struct pci_dev *pdev, unsigned long flags = 0; struct drm_device *ddev; struct radeon_device *rdev; + struct device *dev = &pdev->dev; const struct drm_format_info *format; int ret; @@ -277,7 +278,7 @@ static int radeon_pci_probe(struct pci_dev *pdev, case CHIP_VERDE: case CHIP_OLAND: case CHIP_HAINAN: - dev_info(&pdev->dev, + dev_info(dev, "SI support disabled by module param\n"); return -ENODEV; } @@ -289,7 +290,7 @@ static int radeon_pci_probe(struct pci_dev *pdev, case CHIP_HAWAII: case CHIP_KABINI: case CHIP_MULLINS: - dev_info(&pdev->dev, + dev_info(dev, "CIK support disabled by module param\n"); return -ENODEV; } @@ -303,11 +304,11 @@ static int radeon_pci_probe(struct pci_dev *pdev, if (ret) return ret; - rdev = devm_drm_dev_alloc(&pdev->dev, &kms_driver, typeof(*rdev), ddev); + rdev = devm_drm_dev_alloc(dev, &kms_driver, typeof(*rdev), ddev); if (IS_ERR(rdev)) return PTR_ERR(rdev); - rdev->dev = &pdev->dev; + rdev->dev = dev; rdev->pdev = pdev; ddev = rdev_to_drm(rdev); ddev->dev_private = rdev; -- cgit v1.2.3 From 16c0681617b8a045773d4d87b6140002fa75b03b Mon Sep 17 00:00:00 2001 From: Daniel Palmer Date: Sat, 18 Oct 2025 14:44:50 +0900 Subject: drm/radeon: Do not kfree() devres managed rdev Since the allocation of the drivers main structure was changed to devm_drm_dev_alloc() rdev is managed by devres and we shouldn't be calling kfree() on it. This fixes things exploding if the driver probe fails and devres cleans up the rdev after we already free'd it. Fixes: a9ed2f052c5c ("drm/radeon: change drm_dev_alloc to devm_drm_dev_alloc") Signed-off-by: Daniel Palmer Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/radeon_kms.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index 645e33bf7947..ba1446acd703 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -84,7 +84,6 @@ void radeon_driver_unload_kms(struct drm_device *dev) rdev->agp = NULL; done_free: - kfree(rdev); dev->dev_private = NULL; } -- cgit v1.2.3 From 3eb8c0b4c091da0a623ade0d3ee7aa4a93df1ea4 Mon Sep 17 00:00:00 2001 From: Daniel Palmer Date: Sat, 18 Oct 2025 14:44:51 +0900 Subject: drm/radeon: Remove calls to drm_put_dev() Since the allocation of the drivers main structure was changed to devm_drm_dev_alloc() drm_put_dev()'ing to trigger it to be free'd should be done by devres. However, drm_put_dev() is still in the probe error and device remove paths. When the driver fails to probe warnings like the following are shown because devres is trying to drm_put_dev() after the driver already did it. [ 5.642230] radeon 0000:01:05.0: probe with driver radeon failed with error -22 [ 5.649605] ------------[ cut here ]------------ [ 5.649607] refcount_t: underflow; use-after-free. [ 5.649620] WARNING: CPU: 0 PID: 357 at lib/refcount.c:28 refcount_warn_saturate+0xbe/0x110 Fixes: a9ed2f052c5c ("drm/radeon: change drm_dev_alloc to devm_drm_dev_alloc") Signed-off-by: Daniel Palmer Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/radeon_drv.c | 25 ++++--------------------- 1 file changed, 4 insertions(+), 21 deletions(-) diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index eb5138da8d5b..350f88af888d 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -315,17 +315,17 @@ static int radeon_pci_probe(struct pci_dev *pdev, ret = pci_enable_device(pdev); if (ret) - goto err_free; + return ret; pci_set_drvdata(pdev, ddev); ret = radeon_driver_load_kms(ddev, flags); if (ret) - goto err_agp; + goto err; ret = drm_dev_register(ddev, flags); if (ret) - goto err_agp; + goto err; if (rdev->mc.real_vram_size <= (8 * 1024 * 1024)) format = drm_format_info(DRM_FORMAT_C8); @@ -338,30 +338,14 @@ static int radeon_pci_probe(struct pci_dev *pdev, return 0; -err_agp: +err: pci_disable_device(pdev); -err_free: - drm_dev_put(ddev); return ret; } -static void -radeon_pci_remove(struct pci_dev *pdev) -{ - struct drm_device *dev = pci_get_drvdata(pdev); - - drm_put_dev(dev); -} - static void radeon_pci_shutdown(struct pci_dev *pdev) { - /* if we are running in a VM, make sure the device - * torn down properly on reboot/shutdown - */ - if (radeon_device_is_virtual()) - radeon_pci_remove(pdev); - #if defined(CONFIG_PPC64) || defined(CONFIG_MACH_LOONGSON64) /* * Some adapters need to be suspended before a @@ -614,7 +598,6 @@ static struct pci_driver radeon_kms_pci_driver = { .name = DRIVER_NAME, .id_table = pciidlist, .probe = radeon_pci_probe, - .remove = radeon_pci_remove, .shutdown = radeon_pci_shutdown, .driver.pm = &radeon_pm_ops, }; -- cgit v1.2.3 From bc6d54ac7e7436721a19443265f971f890c13cc5 Mon Sep 17 00:00:00 2001 From: Kenneth Feng Date: Fri, 28 Mar 2025 10:34:57 +0800 Subject: drm/amd/display: pause the workload setting in dm v1: Pause the workload setting in dm when doinn idle optimization v2: Rebase patch to latest kernel code base (kernel 6.16) Reviewed-by: Alex Deucher Signed-off-by: Kenneth Feng Signed-off-by: Alex Deucher Signed-off-by: Yang Wang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index 1ec9d03ad747..f08121a2b838 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -248,6 +248,8 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) struct vblank_control_work *vblank_work = container_of(work, struct vblank_control_work, work); struct amdgpu_display_manager *dm = vblank_work->dm; + struct amdgpu_device *adev = drm_to_adev(dm->ddev); + int r; mutex_lock(&dm->dc_lock); @@ -277,7 +279,16 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) if (dm->active_vblank_irq_count == 0) { dc_post_update_surfaces_to_stream(dm->dc); + + r = amdgpu_dpm_pause_power_profile(adev, true); + if (r) + dev_warn(adev->dev, "failed to set default power profile mode\n"); + dc_allow_idle_optimizations(dm->dc, true); + + r = amdgpu_dpm_pause_power_profile(adev, false); + if (r) + dev_warn(adev->dev, "failed to restore the power profile mode\n"); } mutex_unlock(&dm->dc_lock); -- cgit v1.2.3 From edddaada9e1c5620c184da8c2cdefbf31a930376 Mon Sep 17 00:00:00 2001 From: Jinzhou Su Date: Wed, 15 Oct 2025 09:39:44 +0800 Subject: drm/amdgpu: clear bad page info of ras module Clear bad page info of ras module. Signed-off-by: Jinzhou Su Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index ff34e1c0d9e2..b22ce0a567f2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -612,6 +612,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, return size; } +static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev); + /** * DOC: AMDGPU RAS debugfs EEPROM table reset interface * @@ -636,6 +638,11 @@ static ssize_t amdgpu_ras_debugfs_eeprom_write(struct file *f, (struct amdgpu_device *)file_inode(f)->i_private; int ret; + if (amdgpu_uniras_enabled(adev)) { + ret = amdgpu_uniras_clear_badpages_info(adev); + return ret ? ret : size; + } + ret = amdgpu_ras_eeprom_reset_table( &(amdgpu_ras_get_context(adev)->eeprom_control)); @@ -1543,6 +1550,21 @@ out_fini_err_data: return ret; } +static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev) +{ + struct ras_cmd_dev_handle req = {0}; + int ret; + + ret = amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__CLEAR_BAD_PAGE_INFO, + &req, sizeof(req), NULL, 0); + if (ret) { + dev_err(adev->dev, "Failed to clear bad pages info, ret: %d\n", ret); + return ret; + } + + return 0; +} + static int amdgpu_uniras_query_block_ecc(struct amdgpu_device *adev, struct ras_query_if *info) { -- cgit v1.2.3 From 28d4de7ebf288d628cd3a03c0e57311b94a93ee7 Mon Sep 17 00:00:00 2001 From: Perry Yuan Date: Tue, 14 Oct 2025 15:05:19 +0800 Subject: drm/amdgpu: get rev_id from strap register or IP-discovery table Query the sub-revision field in the IP Discovery table for the VFs to obtain their revision ID. Meanwhile, read the revision ID from the strap register for the PF. Signed-off-by: Perry Yuan Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 4 ++++ drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c | 24 +++++++++++++----------- 2 files changed, 17 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 0d1dd587db5f..e716097dfde4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1843,6 +1843,10 @@ static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev) if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E; + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) && + adev->rev_id == 0x3) + adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E; + if (!(adev->flags & AMD_IS_APU) && !amdgpu_sriov_vf(adev)) { vram_info = RREG32(regBIF_BIOS_SCRATCH_4); adev->gmc.vram_vendor = vram_info & 0xF; diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index 1c22bc11c1f8..bdfd2917e3ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -41,19 +41,21 @@ static void nbio_v7_9_remap_hdp_registers(struct amdgpu_device *adev) static u32 nbio_v7_9_get_rev_id(struct amdgpu_device *adev) { - u32 tmp; - - tmp = IP_VERSION_SUBREV(amdgpu_ip_version_full(adev, NBIO_HWIP, 0)); - /* If it is VF or subrevision holds a non-zero value, that should be used */ - if (tmp || amdgpu_sriov_vf(adev)) - return tmp; + u32 rev_id; - /* If discovery subrev is not updated, use register version */ - tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); - tmp = REG_GET_FIELD(tmp, RCC_STRAP0_RCC_DEV0_EPF0_STRAP0, - STRAP_ATI_REV_ID_DEV0_F0); + /* + * fetch the sub-revision field from the IP-discovery table + * (returns zero if the table entry is not populated). + */ + if (amdgpu_sriov_vf(adev)) { + rev_id = IP_VERSION_SUBREV(amdgpu_ip_version_full(adev, NBIO_HWIP, 0)); + } else { + rev_id = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); + rev_id = REG_GET_FIELD(rev_id, RCC_STRAP0_RCC_DEV0_EPF0_STRAP0, + STRAP_ATI_REV_ID_DEV0_F0); + } - return tmp; + return rev_id; } static void nbio_v7_9_mc_access_enable(struct amdgpu_device *adev, bool enable) -- cgit v1.2.3 From 214eb7e83d222ed542d41e23f3296ee71eedf31b Mon Sep 17 00:00:00 2001 From: Jinzhou Su Date: Tue, 21 Oct 2025 09:31:11 +0800 Subject: drm/amdgpu: Add uniras version in sysfs Display uniras version in sysfs version interface when uniras enable. v2: display ras version detail info Signed-off-by: Jinzhou Su Reviewed-by: Tao Zhou Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 32 +++++++++++++++++++++++++++++++- 1 file changed, 31 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index b22ce0a567f2..7f899d3e1bfa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1950,12 +1950,42 @@ static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev, return sysfs_emit(buf, "feature mask: 0x%x\n", con->features); } +static bool amdgpu_ras_get_version_info(struct amdgpu_device *adev, u32 *major, + u32 *minor, u32 *rev) +{ + int i; + + if (!adev || !major || !minor || !rev || !amdgpu_uniras_enabled(adev)) + return false; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_RAS) { + *major = adev->ip_blocks[i].version->major; + *minor = adev->ip_blocks[i].version->minor; + *rev = adev->ip_blocks[i].version->rev; + return true; + } + } + + return false; +} + static ssize_t amdgpu_ras_sysfs_version_show(struct device *dev, struct device_attribute *attr, char *buf) { struct amdgpu_ras *con = container_of(attr, struct amdgpu_ras, version_attr); - return sysfs_emit(buf, "table version: 0x%x\n", con->eeprom_control.tbl_hdr.version); + u32 major, minor, rev; + ssize_t size = 0; + + size += sysfs_emit_at(buf, size, "table version: 0x%x\n", + con->eeprom_control.tbl_hdr.version); + + if (amdgpu_ras_get_version_info(con->adev, &major, &minor, &rev)) + size += sysfs_emit_at(buf, size, "ras version: %u.%u.%u\n", + major, minor, rev); + + return size; } static ssize_t amdgpu_ras_sysfs_schema_show(struct device *dev, -- cgit v1.2.3 From 13ff4f63fcddfc84ec8632f1443936b00aa26725 Mon Sep 17 00:00:00 2001 From: Matthew Schwartz Date: Mon, 20 Oct 2025 16:09:34 -0700 Subject: drm/amd/display: Don't program BLNDGAM_MEM_PWR_FORCE when CM low-power is disabled on DCN30 Before commit 33056a97ae5e ("drm/amd/display: Remove double checks for `debug.enable_mem_low_power.bits.cm`"), dpp3_program_blnd_lut(NULL) checked the low-power debug flag before calling dpp3_power_on_blnd_lut(false). After commit 33056a97ae5e ("drm/amd/display: Remove double checks for `debug.enable_mem_low_power.bits.cm`"), dpp3_program_blnd_lut(NULL) unconditionally calls dpp3_power_on_blnd_lut(false). The BLNDGAM power helper writes BLNDGAM_MEM_PWR_FORCE when CM low-power is disabled, causing immediate SRAM power toggles instead of deferring at vupdate. This can disrupt atomic color/LUT sequencing during transitions between direct scanout and composition within gamescope's DRM backend on Steam Deck OLED. To fix this, leave the BLNDGAM power state unchanged when low-power is disabled, matching dpp3_power_on_hdr3dlut and dpp3_power_on_shaper. Fixes: 33056a97ae5e ("drm/amd/display: Remove double checks for `debug.enable_mem_low_power.bits.cm`") Signed-off-by: Matthew Schwartz Reviewed-by: Harry Wentland Reviewed-by: Mario Limonciello Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c index 94d0dc3461d2..fe52df01caf7 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c @@ -582,9 +582,6 @@ static void dpp3_power_on_blnd_lut( dpp_base->ctx->dc->optimized_required = true; dpp_base->deferred_reg_writes.bits.disable_blnd_lut = true; } - } else { - REG_SET(CM_MEM_PWR_CTRL, 0, - BLNDGAM_MEM_PWR_FORCE, power_on == true ? 0 : 1); } } -- cgit v1.2.3 From fca0c66b22303de0d1d6313059baf4dc960a4753 Mon Sep 17 00:00:00 2001 From: Yang Wang Date: Wed, 22 Oct 2025 14:12:21 +0800 Subject: drm/amd/pm: fix smu table id bound check issue in smu_cmn_update_table() 'table_index' is a variable defined by the smu driver (kmd) 'table_id' is a variable defined by the hw smu (pmfw) This code should use table_index as a bounds check. Fixes: caad2613dc4bd ("drm/amd/powerplay: move table setting common code to smu_cmn.c") Signed-off-by: Yang Wang Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index f532f7c69259..a8961a8f5c42 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -969,7 +969,7 @@ int smu_cmn_update_table(struct smu_context *smu, table_index); uint32_t table_size; int ret = 0; - if (!table_data || table_id >= SMU_TABLE_COUNT || table_id < 0) + if (!table_data || table_index >= SMU_TABLE_COUNT || table_id < 0) return -EINVAL; table_size = smu_table->tables[table_index].size; -- cgit v1.2.3 From c52238c9fb414555c68340cd80e487d982c1921c Mon Sep 17 00:00:00 2001 From: John Smith Date: Tue, 21 Oct 2025 11:08:13 +0200 Subject: drm/amd/pm/powerplay/smumgr: Fix PCIeBootLinkLevel value on Fiji Previously this was initialized with zero which represented PCIe Gen 1.0 instead of using the maximum value from the speed table which is the behaviour of all other smumgr implementations. Fixes: 18edef19ea44 ("drm/amd/powerplay: implement fw image related smu interface for Fiji.") Signed-off-by: John Smith Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c index d2dbd90bb427..0a876c840c79 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c @@ -2024,7 +2024,7 @@ static int fiji_init_smc_table(struct pp_hwmgr *hwmgr) table->VoltageResponseTime = 0; table->PhaseResponseTime = 0; table->MemoryThermThrottleEnable = 1; - table->PCIeBootLinkLevel = 0; /* 0:Gen1 1:Gen2 2:Gen3*/ + table->PCIeBootLinkLevel = (uint8_t) (data->dpm_table.pcie_speed_table.count); table->PCIeGenInterval = 1; table->VRConfig = 0; -- cgit v1.2.3 From 92b0a6ae6672857ddeabf892223943d2f0e06c97 Mon Sep 17 00:00:00 2001 From: John Smith Date: Tue, 21 Oct 2025 11:09:09 +0200 Subject: drm/amd/pm/powerplay/smumgr: Fix PCIeBootLinkLevel value on Iceland Previously this was initialized with zero which represented PCIe Gen 1.0 instead of using the maximum value from the speed table which is the behaviour of all other smumgr implementations. Fixes: 18aafc59b106 ("drm/amd/powerplay: implement fw related smu interface for iceland.") Signed-off-by: John Smith Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c index 1f50f1e74c48..aa3ae9b115c4 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c @@ -2028,7 +2028,7 @@ static int iceland_init_smc_table(struct pp_hwmgr *hwmgr) table->VoltageResponseTime = 0; table->PhaseResponseTime = 0; table->MemoryThermThrottleEnable = 1; - table->PCIeBootLinkLevel = 0; + table->PCIeBootLinkLevel = (uint8_t) (data->dpm_table.pcie_speed_table.count); table->PCIeGenInterval = 1; result = iceland_populate_smc_svi2_config(hwmgr, table); -- cgit v1.2.3 From 6142aa066061919a436c34c8a182674b48636311 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 22 Oct 2025 14:03:24 +0300 Subject: drm/amdgpu/userqueue: Fix use after free in amdgpu_userq_buffer_vas_list_cleanup() The amdgpu_userq_buffer_va_list_del() function frees "va_cursor" but it is dereferenced on the next line when we print the debug message. Print the debug message first and then free it. Fixes: 2a28f9665dca ("drm/amdgpu: track the userq bo va for its obj management") Signed-off-by: Dan Carpenter Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index 9d4751a39c20..2200e0bbf040 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -159,9 +159,9 @@ static int amdgpu_userq_buffer_vas_list_cleanup(struct amdgpu_device *adev, r = -EINVAL; goto err; } - amdgpu_userq_buffer_va_list_del(mapping, va_cursor); dev_dbg(adev->dev, "delete the userq:%p va:%llx\n", queue, va_cursor->gpu_addr); + amdgpu_userq_buffer_va_list_del(mapping, va_cursor); } err: amdgpu_bo_unreserve(queue->vm->root.bo); -- cgit v1.2.3 From abd3f876404cafb107cb34bacb74706bfee11cbe Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 22 Oct 2025 09:12:54 -0400 Subject: drm/amdgpu: fix SPDX headers on amdgpu_cper.c/h These should be MIT. The driver in general is MIT and the license text at the top of the files is MIT so fix it. Fixes: 92d5d2a09de1 ("drm/amdgpu: Introduce funcs for populating CPER") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4654 Reviewed-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c index ef996493115f..425a3e564360 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c @@ -1,4 +1,4 @@ -// SPDX-License-Identifier: GPL-2.0 +// SPDX-License-Identifier: MIT /* * Copyright 2025 Advanced Micro Devices, Inc. * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h index bcb97d245673..353421807387 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2025 Advanced Micro Devices, Inc. * -- cgit v1.2.3 From 72c5482cb0f3d3c772c9de50e5a4265258a53f81 Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 22 Oct 2025 09:14:55 -0400 Subject: drm/amdgpu: fix SPDX header on amd_cper.h This should be MIT. The driver in general is MIT and the license text at the top of the file is MIT so fix it. Fixes: 523b69c65445 ("drm/amd/include: Add amd cper header") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4654 Reviewed-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/include/amd_cper.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/include/amd_cper.h b/drivers/gpu/drm/amd/include/amd_cper.h index 086869264425..a252ee4c7874 100644 --- a/drivers/gpu/drm/amd/include/amd_cper.h +++ b/drivers/gpu/drm/amd/include/amd_cper.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2025 Advanced Micro Devices, Inc. * -- cgit v1.2.3 From 68c20d7b1779f97d600e61b9e95726c0cd609e2a Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 22 Oct 2025 09:17:37 -0400 Subject: drm/amdgpu: fix SPDX header on irqsrcs_vcn_5_0.h This should be MIT. The driver in general is MIT and the license text at the top of the file is MIT so fix it. Fixes: d1bb64651095 ("drm/amdgpu: add irq source ids for VCN5_0/JPEG5_0") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4654 Reviewed-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h index 64b553e7de1a..e7fdcee22a71 100644 --- a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h +++ b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2024 Advanced Micro Devices, Inc. All rights reserved. -- cgit v1.2.3 From 102c4f7c554ac5a5ecf0023fa0612beb58e3b0bd Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Wed, 22 Oct 2025 09:19:55 -0400 Subject: drm/amdgpu: fix SPDX header on cyan_skillfish_reg_init.c This should be MIT. The driver in general is MIT and the license text at the top of the file is MIT so fix it. Fixes: e8529dbc75ca ("drm/amdgpu: add ip offset support for cyan skillfish") Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4654 Reviewed-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c index 96616a865aac..ed1e25661706 100644 --- a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c +++ b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c @@ -1,4 +1,4 @@ -// SPDX-License-Identifier: GPL-2.0 +// SPDX-License-Identifier: MIT /* * Copyright 2018 Advanced Micro Devices, Inc. * -- cgit v1.2.3 From 00dc2ff519781309f1f1ff344a638f61c33884c7 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Wed, 22 Oct 2025 17:46:26 +0530 Subject: drm/amdgpu: Make SR-IOV critical region checks overflow-safe MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The function amdgpu_virt_init_critical_region() contained an invalid check for a negative init_hdr_offset value: if (init_hdr_offset < 0) Since init_hdr_offset is an unsigned 32-bit integer, this condition can never be true and triggers a Smatch warning: warn: unsigned 'init_hdr_offset' is never less than zero In addition, the subsequent bounds check: if ((init_hdr_offset + init_hdr_size) > vram_size) was vulnerable to integer overflow when adding the two unsigned values. Thus, by promoting offset and size to 64-bit and using check_add_overflow() to safely validate the sum against VRAM size. Fixes: 07009df6494d ("drm/amdgpu: Introduce SRIOV critical regions v2 during VF init") Reported by: Dan Carpenter Cc: Ellen Pan Cc: Christian König Cc: Alex Deucher Signed-off-by: Srinivasan Shanmugam Reviewed-by: Ellen Pan Reviewed-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 66e9cd103597..45f2ad083338 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -937,9 +937,10 @@ static uint8_t amdgpu_virt_crit_region_calc_checksum(uint8_t *buf_start, uint8_t int amdgpu_virt_init_critical_region(struct amdgpu_device *adev) { struct amd_sriov_msg_init_data_header *init_data_hdr = NULL; - uint32_t init_hdr_offset = adev->virt.init_data_header.offset; - uint32_t init_hdr_size = adev->virt.init_data_header.size_kb << 10; - uint64_t vram_size; + u64 init_hdr_offset = adev->virt.init_data_header.offset; + u64 init_hdr_size = (u64)adev->virt.init_data_header.size_kb << 10; /* KB → bytes */ + u64 vram_size; + u64 end; int r = 0; uint8_t checksum = 0; @@ -957,7 +958,7 @@ int amdgpu_virt_init_critical_region(struct amdgpu_device *adev) return -EINVAL; vram_size <<= 20; - if ((init_hdr_offset + init_hdr_size) > vram_size) { + if (check_add_overflow(init_hdr_offset, init_hdr_size, &end) || end > vram_size) { dev_err(adev->dev, "init_data_header exceeds VRAM size, exiting\n"); return -EINVAL; } -- cgit v1.2.3 From 6f1ee58a5e56c4a7b5b3fe9323e4510df156449c Mon Sep 17 00:00:00 2001 From: "Jesse.Zhang" Date: Tue, 21 Oct 2025 10:01:46 +0800 Subject: drm/amd/pm: smu13: Enable VCN_RESET for pgm 7 with appropriate firmware version This patch extends the VCN_RESET capability check to include pgm 7 when the firmware version is 0x07551400 or newer. Signed-off-by: Jesse Zhang Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index cc7a0e061ba1..0a7d2cea7dc6 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -450,7 +450,8 @@ static void smu_v13_0_6_init_caps(struct smu_context *smu) ((pgm == 4) && (fw_ver >= 0x4557000))) smu_v13_0_6_cap_set(smu, SMU_CAP(SDMA_RESET)); - if ((pgm == 0) && (fw_ver >= 0x00558200)) + if ((pgm == 0 && fw_ver >= 0x00558200) || + (pgm == 7 && fw_ver >= 0x07551400)) smu_v13_0_6_cap_set(smu, SMU_CAP(VCN_RESET)); } -- cgit v1.2.3 From 84564d2920b8c858d96cb7471b45203d35f63b61 Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Tue, 21 Oct 2025 15:14:06 +0530 Subject: drm/amdgpu: null check for hmm_pfns ptr before freeing it Due to low memory or when num of pages is too big to be accomodated, allocation could fail for pfn's. Chekc hmm_pfns for NULL before calling the kvfree for the it. Signed-off-by: Sunil Khatri Acked-by: Arunpravin Paneer Selvam Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c index d6f903a2d573..7e5a09b0bc78 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c @@ -286,7 +286,9 @@ void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range) if (!range) return; - kvfree(range->hmm_range.hmm_pfns); + if (range->hmm_range.hmm_pfns) + kvfree(range->hmm_range.hmm_pfns); + amdgpu_bo_unref(&range->bo); kfree(range); } -- cgit v1.2.3 From 90ef1dcb1d2bb84ad998e845e26a2a297a7ddfd6 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Thu, 23 Oct 2025 10:52:21 +0530 Subject: drm/amdgpu: Fix pointer casts when reading dynamic region sizes MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The function amdgpu_virt_get_dynamic_data_info() writes a 64-bit size value. In two places (amdgpu_bios.c and amdgpu_discovery.c), the code passed the address of a smaller variable by casting it to u64 *, which is unsafe. This could make the function write more bytes than the smaller variable can hold, possibly overwriting nearby memory. Reported by static analysis tools. v2: Dynamic region size comes from the host (SR-IOV setup) and is always fixed to 5 MB. (Lijo/Ellen) 5 MB easily fits inside a 32-bit value, so using a 64-bit type is not needed. It also avoids extra type casts Fixes: b4a8fcc7826a ("drm/amdgpu: Add logic for VF ipd and VF bios to init from dynamic crit_region offsets") Reported by: Dan Carpenter Cc: Ellen Pan Cc: Christian König Cc: Alex Deucher Signed-off-by: Srinivasan Shanmugam Reviewed-by: Lijo Lazar Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c index db705bf723f1..35d04e69aec0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c @@ -103,7 +103,7 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev) { uint8_t __iomem *bios = NULL; resource_size_t vram_base; - resource_size_t size = 256 * 1024; /* ??? */ + u32 size = 256U * 1024U; /* ??? */ if (!(adev->flags & AMD_IS_APU)) if (amdgpu_device_need_post(adev)) @@ -126,7 +126,7 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev) */ if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) { if (amdgpu_virt_get_dynamic_data_info(adev, - AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, (uint64_t *)&size)) { + AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, &size)) { amdgpu_bios_release(adev); return false; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index a7cb4665f485..fa2a22dfa048 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -311,7 +311,7 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, */ if (amdgpu_virt_get_dynamic_data_info(adev, AMD_SRIOV_MSG_IPD_TABLE_ID, binary, - (uint64_t *)&adev->discovery.size)) { + &adev->discovery.size)) { dev_err(adev->dev, "failed to read discovery info from dynamic critical region."); ret = -EINVAL; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 45f2ad083338..f2ce8f506aa8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -1102,7 +1102,7 @@ out: } int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev, - int data_id, uint8_t *binary, uint64_t *size) + int data_id, uint8_t *binary, u32 *size) { uint32_t data_offset = 0; uint32_t data_size = 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 2a13cc892a13..14d864be5800 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -443,7 +443,7 @@ void amdgpu_virt_init(struct amdgpu_device *adev); int amdgpu_virt_init_critical_region(struct amdgpu_device *adev); int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev, - int data_id, uint8_t *binary, uint64_t *size); + int data_id, uint8_t *binary, u32 *size); bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev); int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev); -- cgit v1.2.3 From 1454642960b0a4fafd48b08919d39350ee2ac372 Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Fri, 18 Jul 2025 14:20:45 -0500 Subject: drm/amd: Re-introduce property to control adaptive backlight modulation commit 0887054d14ae ("drm/amd: Drop abm_level property") dropped the abm level property in favor of sysfs control. Since then there have been discussions that compositors showed an interest in modifying a vendor specific property instead. So re-introduce the abm level property, but with different semantics. Rather than being an integer it's now an enum. One of the enum options is 'sysfs', and that is because there is still a sysfs file for use by userspace when the compositor doesn't support this property. If usespace has not modified this property, the default value will be for sysfs to control it. Once userspace has set the property stop allowing sysfs control. The property is only attached to non-OLED eDP panels. Cc: Xaver Hugl Reviewed-by: Harry Wentland Signed-off-by: Mario Limonciello Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 60 ++++++++++++++++++++++- drivers/gpu/drm/amd/amdgpu/amdgpu_display.h | 7 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h | 2 + drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 39 ++++++++++++++- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h | 1 + 5 files changed, 106 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 51bab32fd8c6..c74b95bd41b5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -1365,6 +1365,64 @@ static const struct drm_prop_enum_list amdgpu_dither_enum_list[] = { { AMDGPU_FMT_DITHER_ENABLE, "on" }, }; +/** + * DOC: property for adaptive backlight modulation + * + * The 'adaptive backlight modulation' property is used for the compositor to + * directly control the adaptive backlight modulation power savings feature + * that is part of DCN hardware. + * + * The property will be attached specifically to eDP panels that support it. + * + * The property is by default set to 'sysfs' to allow the sysfs file 'panel_power_savings' + * to be able to control it. + * If set to 'off' the compositor will ensure it stays off. + * The other values 'min', 'bias min', 'bias max', and 'max' will control the + * intensity of the power savings. + * + * Modifying this value can have implications on color accuracy, so tread + * carefully. + */ +static int amdgpu_display_setup_abm_prop(struct amdgpu_device *adev) +{ + const struct drm_prop_enum_list props[] = { + { ABM_SYSFS_CONTROL, "sysfs" }, + { ABM_LEVEL_OFF, "off" }, + { ABM_LEVEL_MIN, "min" }, + { ABM_LEVEL_BIAS_MIN, "bias min" }, + { ABM_LEVEL_BIAS_MAX, "bias max" }, + { ABM_LEVEL_MAX, "max" }, + }; + struct drm_property *prop; + int i; + + if (!adev->dc_enabled) + return 0; + + prop = drm_property_create(adev_to_drm(adev), DRM_MODE_PROP_ENUM, + "adaptive backlight modulation", + 6); + if (!prop) + return -ENOMEM; + + for (i = 0; i < ARRAY_SIZE(props); i++) { + int ret; + + ret = drm_property_add_enum(prop, props[i].type, + props[i].name); + + if (ret) { + drm_property_destroy(adev_to_drm(adev), prop); + + return ret; + } + } + + adev->mode_info.abm_level_property = prop; + + return 0; +} + int amdgpu_display_modeset_create_props(struct amdgpu_device *adev) { int sz; @@ -1411,7 +1469,7 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev) "dither", amdgpu_dither_enum_list, sz); - return 0; + return amdgpu_display_setup_abm_prop(adev); } void amdgpu_display_update_priority(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h index 930c171473b4..49a29bf47a37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h @@ -55,4 +55,11 @@ int amdgpu_display_resume_helper(struct amdgpu_device *adev); int amdgpu_display_get_scanout_buffer(struct drm_plane *plane, struct drm_scanout_buffer *sb); +#define ABM_SYSFS_CONTROL -1 +#define ABM_LEVEL_OFF 0 +#define ABM_LEVEL_MIN 1 +#define ABM_LEVEL_BIAS_MIN 2 +#define ABM_LEVEL_BIAS_MAX 3 +#define ABM_LEVEL_MAX 4 + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 20460cfd09bc..dc8d2f52c7d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -326,6 +326,8 @@ struct amdgpu_mode_info { struct drm_property *audio_property; /* FMT dithering */ struct drm_property *dither_property; + /* Adaptive Backlight Modulation (power feature) */ + struct drm_property *abm_level_property; /* hardcoded DFP edid from BIOS */ const struct drm_edid *bios_hardcoded_edid; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index e3be2620f959..5b446e4000ef 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -5200,6 +5200,7 @@ static int initialize_plane(struct amdgpu_display_manager *dm, static void setup_backlight_device(struct amdgpu_display_manager *dm, struct amdgpu_dm_connector *aconnector) { + struct amdgpu_dm_backlight_caps *caps; struct dc_link *link = aconnector->dc_link; int bl_idx = dm->num_of_edps; @@ -5219,6 +5220,13 @@ static void setup_backlight_device(struct amdgpu_display_manager *dm, dm->num_of_edps++; update_connector_ext_caps(aconnector); + caps = &dm->backlight_caps[aconnector->bl_idx]; + + /* Only offer ABM property when non-OLED and user didn't turn off by module parameter */ + if (!caps->ext_caps->bits.oled && amdgpu_dm_abm_level < 0) + drm_object_attach_property(&aconnector->base.base, + dm->adev->mode_info.abm_level_property, + ABM_SYSFS_CONTROL); } static void amdgpu_set_panel_orientation(struct drm_connector *connector); @@ -7291,6 +7299,20 @@ int amdgpu_dm_connector_atomic_set_property(struct drm_connector *connector, } else if (property == adev->mode_info.underscan_property) { dm_new_state->underscan_enable = val; ret = 0; + } else if (property == adev->mode_info.abm_level_property) { + switch (val) { + case ABM_SYSFS_CONTROL: + dm_new_state->abm_sysfs_forbidden = false; + break; + case ABM_LEVEL_OFF: + dm_new_state->abm_sysfs_forbidden = true; + dm_new_state->abm_level = ABM_LEVEL_IMMEDIATE_DISABLE; + break; + default: + dm_new_state->abm_sysfs_forbidden = true; + dm_new_state->abm_level = val; + }; + ret = 0; } return ret; @@ -7333,6 +7355,13 @@ int amdgpu_dm_connector_atomic_get_property(struct drm_connector *connector, } else if (property == adev->mode_info.underscan_property) { *val = dm_state->underscan_enable; ret = 0; + } else if (property == adev->mode_info.abm_level_property) { + if (!dm_state->abm_sysfs_forbidden) + *val = ABM_SYSFS_CONTROL; + else + *val = (dm_state->abm_level != ABM_LEVEL_IMMEDIATE_DISABLE) ? + dm_state->abm_level : 0; + ret = 0; } return ret; @@ -7385,10 +7414,16 @@ static ssize_t panel_power_savings_store(struct device *device, return -EINVAL; drm_modeset_lock(&dev->mode_config.connection_mutex, NULL); - to_dm_connector_state(connector->state)->abm_level = val ?: - ABM_LEVEL_IMMEDIATE_DISABLE; + if (to_dm_connector_state(connector->state)->abm_sysfs_forbidden) + ret = -EBUSY; + else + to_dm_connector_state(connector->state)->abm_level = val ?: + ABM_LEVEL_IMMEDIATE_DISABLE; drm_modeset_unlock(&dev->mode_config.connection_mutex); + if (ret) + return ret; + drm_kms_helper_hotplug_event(dev); return count; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index db75e991ac7b..5a7aa903bd3c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -993,6 +993,7 @@ struct dm_connector_state { bool underscan_enable; bool freesync_capable; bool update_hdcp; + bool abm_sysfs_forbidden; uint8_t abm_level; int vcpi_slots; uint64_t pbn; -- cgit v1.2.3 From 9daa8f19e45b94cff1164082dd9d187f8753379c Mon Sep 17 00:00:00 2001 From: Mario Limonciello Date: Mon, 6 Oct 2025 10:09:09 -0500 Subject: drm/amd/display: Bump minimum for frame_warn_limit [Why] The bigger of CONFIG_FRAME_WARN and frame_warn_limit is used to trigger warnings about large stack frames. The dml_core_mode_support() stack frame has grown to 2056. [How] Update frame_warn_limit to 2056 so that CONFIG_FRAME_WARN of 2048 doesn't cause a failure. Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4609 Reviewed-by: Aurabindo Pillai Signed-off-by: Mario Limonciello Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dml2/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile index 4c21ce42054c..4ff0adef233a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml2/Makefile @@ -35,7 +35,7 @@ ifneq ($(CONFIG_FRAME_WARN),0) frame_warn_limit := 3072 endif else - frame_warn_limit := 2048 + frame_warn_limit := 2056 endif ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) -- cgit v1.2.3 From 423ef48d41cbcfd5ee2039085d59e1f29b020ef4 Mon Sep 17 00:00:00 2001 From: Dillon Varone Date: Tue, 30 Sep 2025 12:17:13 -0400 Subject: drm/amd/display: Fix DMUB reset sequence for DCN32 [WHY&HOW] Backport reset sequence fixes implemented on DCN401 to DCN32 to address stability issues when resetting the DMUB. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Dillon Varone Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c | 50 ++++++++++++++-------- drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c | 17 ++++++-- 2 files changed, 45 insertions(+), 22 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c index ce041f6239dc..7e9856289910 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c @@ -89,50 +89,58 @@ static inline void dmub_dcn32_translate_addr(const union dmub_addr *addr_in, void dmub_dcn32_reset(struct dmub_srv *dmub) { union dmub_gpint_data_register cmd; - const uint32_t timeout = 100000; - uint32_t in_reset, is_enabled, scratch, i, pwait_mode; + const uint32_t timeout_us = 1 * 1000 * 1000; //1s + const uint32_t poll_delay_us = 1; //1us + uint32_t i = 0; + uint32_t enabled, in_reset, scratch, pwait_mode; - REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); - REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_enabled); + REG_GET(DMCUB_CNTL, + DMCUB_ENABLE, &enabled); + REG_GET(DMCUB_CNTL2, + DMCUB_SOFT_RESET, &in_reset); - if (in_reset == 0 && is_enabled != 0) { + if (enabled && in_reset == 0) { cmd.bits.status = 1; cmd.bits.command_code = DMUB_GPINT__STOP_FW; cmd.bits.param = 0; dmub->hw_funcs.set_gpint(dmub, cmd); - for (i = 0; i < timeout; ++i) { - if (dmub->hw_funcs.is_gpint_acked(dmub, cmd)) - break; - - udelay(1); - } - - for (i = 0; i < timeout; ++i) { + for (; i < timeout_us; i++) { scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; - udelay(1); + udelay(poll_delay_us); } - for (i = 0; i < timeout; ++i) { + for (; i < timeout_us; i++) { REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &pwait_mode); if (pwait_mode & (1 << 0)) break; - udelay(1); + udelay(poll_delay_us); } - /* Force reset in case we timed out, DMCUB is likely hung. */ } - if (is_enabled) { + if (enabled) { REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); udelay(1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); } + if (i >= timeout_us) { + /* timeout should never occur */ + BREAK_TO_DEBUGGER(); + } + + REG_UPDATE(DMCUB_REGION3_CW2_TOP_ADDRESS, DMCUB_REGION3_CW2_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW3_TOP_ADDRESS, DMCUB_REGION3_CW3_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW4_TOP_ADDRESS, DMCUB_REGION3_CW4_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW5_TOP_ADDRESS, DMCUB_REGION3_CW5_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW7_TOP_ADDRESS, DMCUB_REGION3_CW7_ENABLE, 0); + REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); @@ -141,7 +149,7 @@ void dmub_dcn32_reset(struct dmub_srv *dmub) REG_WRITE(DMCUB_OUTBOX0_WPTR, 0); REG_WRITE(DMCUB_SCRATCH0, 0); - /* Clear the GPINT command manually so we don't send anything during boot. */ + /* Clear the GPINT command manually so we don't reset again. */ cmd.all = 0; dmub->hw_funcs.set_gpint(dmub, cmd); } @@ -163,7 +171,9 @@ void dmub_dcn32_backdoor_load(struct dmub_srv *dmub, dmub_dcn32_get_fb_base_offset(dmub, &fb_base, &fb_offset); + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); dmub_dcn32_translate_addr(&cw0->offset, fb_base, fb_offset, &offset); @@ -193,7 +203,9 @@ void dmub_dcn32_backdoor_load_zfb_mode(struct dmub_srv *dmub, { union dmub_addr offset; + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); offset = cw0->offset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c index b31adbd0d685..95542299e3b3 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c @@ -81,7 +81,7 @@ void dmub_dcn401_reset(struct dmub_srv *dmub) dmub->hw_funcs.set_gpint(dmub, cmd); for (; i < timeout_us; i++) { - scratch = dmub->hw_funcs.get_gpint_response(dmub); + scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; @@ -97,11 +97,24 @@ void dmub_dcn401_reset(struct dmub_srv *dmub) } } + if (enabled) { + REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); + udelay(1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); + } + if (i >= timeout_us) { /* timeout should never occur */ BREAK_TO_DEBUGGER(); } + REG_UPDATE(DMCUB_REGION3_CW2_TOP_ADDRESS, DMCUB_REGION3_CW2_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW3_TOP_ADDRESS, DMCUB_REGION3_CW3_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW4_TOP_ADDRESS, DMCUB_REGION3_CW4_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW5_TOP_ADDRESS, DMCUB_REGION3_CW5_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW7_TOP_ADDRESS, DMCUB_REGION3_CW7_ENABLE, 0); + REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); @@ -134,7 +147,6 @@ void dmub_dcn401_backdoor_load(struct dmub_srv *dmub, /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); dmub_dcn401_translate_addr(&cw0->offset, fb_base, fb_offset, &offset); @@ -168,7 +180,6 @@ void dmub_dcn401_backdoor_load_zfb_mode(struct dmub_srv *dmub, /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); offset = cw0->offset; -- cgit v1.2.3 From 610cf76e9453b71f19e8c18e49876a8c5952e170 Mon Sep 17 00:00:00 2001 From: Dmytro Laktyushkin Date: Tue, 14 Oct 2025 13:05:26 -0400 Subject: drm/amd/display: Add opp count validation to dml2.1 Newer asics can have mismatching dpp and opp counts and dml needs to account for this. Reviewed-by: Charlene Liu Signed-off-by: Dmytro Laktyushkin Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- .../dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c | 1 + .../dml21/src/dml2_core/dml2_core_dcn4_calcs.c | 27 +++++++++++++++++++--- .../dml21/src/dml2_core/dml2_core_shared_types.h | 3 +++ 3 files changed, 28 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c index 6ee37386f672..eba948e187c1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c @@ -28,6 +28,7 @@ struct dml2_core_ip_params core_dcn4_ip_caps_base = { .writeback_interface_buffer_size_kbytes = 90, //Number of pipes after DCN Pipe harvesting .max_num_dpp = 4, + .max_num_opp = 4, .max_num_otg = 4, .max_num_wb = 1, .max_dchub_pscl_bw_pix_per_clk = 4, diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index 4ccdb179b001..f16792f79bef 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -4047,7 +4047,9 @@ static bool ValidateODMMode(enum dml2_odm_mode ODMMode, bool UseDSC, unsigned int NumberOfDSCSlices, unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, unsigned int MaxNumDPP, + unsigned int MaxNumOPP, double DISPCLKRequired, unsigned int NumberOfDPPRequired, unsigned int MaxHActiveForDSC, @@ -4063,7 +4065,7 @@ static bool ValidateODMMode(enum dml2_odm_mode ODMMode, if (DISPCLKRequired > MaxDispclk) return false; - if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP) + if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP || (TotalNumberOfActiveOPP + NumberOfDPPRequired) > MaxNumOPP) return false; if (are_odm_segments_symmetrical) { if (HActive % (NumberOfDPPRequired * pixels_per_clock_cycle)) @@ -4109,7 +4111,9 @@ static noinline_for_stack void CalculateODMMode( double MaxDispclk, bool DSCEnable, unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, unsigned int MaxNumDPP, + unsigned int MaxNumOPP, double PixelClock, unsigned int NumberOfDSCSlices, @@ -4179,7 +4183,9 @@ static noinline_for_stack void CalculateODMMode( UseDSC, NumberOfDSCSlices, TotalNumberOfActiveDPP, + TotalNumberOfActiveOPP, MaxNumDPP, + MaxNumOPP, DISPCLKRequired, NumberOfDPPRequired, MaxHActiveForDSC, @@ -8358,6 +8364,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; mode_lib->ms.support.TotalAvailablePipesSupport = true; for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { @@ -8393,7 +8400,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.max_dispclk_freq_mhz, false, // DSCEnable mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), mode_lib->ms.support.NumberOfDSCSlices[k], @@ -8412,7 +8421,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.max_dispclk_freq_mhz, true, // DSCEnable mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), mode_lib->ms.support.NumberOfDSCSlices[k], @@ -8516,20 +8527,23 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 1; + mode_lib->ms.NoOfOPP[k] = 1; if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 4; + mode_lib->ms.NoOfOPP[k] = 4; } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 3; + mode_lib->ms.NoOfOPP[k] = 3; } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 2; + mode_lib->ms.NoOfOPP[k] = 2; } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 2) { mode_lib->ms.MPCCombine[k] = true; mode_lib->ms.NoOfDPP[k] = 2; - mode_lib->ms.TotalNumberOfActiveDPP++; } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 1; @@ -8540,7 +8554,6 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out if ((mode_lib->ms.MinDPPCLKUsingSingleDPP[k] > mode_lib->ms.max_dppclk_freq_mhz) || !mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { mode_lib->ms.MPCCombine[k] = true; mode_lib->ms.NoOfDPP[k] = 2; - mode_lib->ms.TotalNumberOfActiveDPP++; } } #if defined(__DML_VBA_DEBUG__) @@ -8548,8 +8561,16 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out #endif } + mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.TotalNumberOfActiveDPP += mode_lib->ms.NoOfDPP[k]; + mode_lib->ms.TotalNumberOfActiveOPP += mode_lib->ms.NoOfOPP[k]; + } if (mode_lib->ms.TotalNumberOfActiveDPP > (unsigned int)mode_lib->ip.max_num_dpp) mode_lib->ms.support.TotalAvailablePipesSupport = false; + if (mode_lib->ms.TotalNumberOfActiveOPP > (unsigned int)mode_lib->ip.max_num_opp) + mode_lib->ms.support.TotalAvailablePipesSupport = false; mode_lib->ms.TotalNumberOfSingleDPPSurfaces = 0; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h index ff1c47347610..051c31ec2f0e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h @@ -36,6 +36,7 @@ struct dml2_core_ip_params { unsigned int max_line_buffer_lines; unsigned int writeback_interface_buffer_size_kbytes; unsigned int max_num_dpp; + unsigned int max_num_opp; unsigned int max_num_otg; unsigned int TDLUT_33cube_count; unsigned int max_num_wb; @@ -570,6 +571,7 @@ struct dml2_core_internal_mode_support { enum dml2_odm_mode ODMMode[DML2_MAX_PLANES]; unsigned int SurfaceSizeInMALL[DML2_MAX_PLANES]; unsigned int NoOfDPP[DML2_MAX_PLANES]; + unsigned int NoOfOPP[DML2_MAX_PLANES]; bool MPCCombine[DML2_MAX_PLANES]; double dcfclk_deepsleep; double MinDPPCLKUsingSingleDPP[DML2_MAX_PLANES]; @@ -580,6 +582,7 @@ struct dml2_core_internal_mode_support { bool PTEBufferSizeNotExceeded[DML2_MAX_PLANES]; bool DCCMetaBufferSizeNotExceeded[DML2_MAX_PLANES]; unsigned int TotalNumberOfActiveDPP; + unsigned int TotalNumberOfActiveOPP; unsigned int TotalNumberOfSingleDPPSurfaces; unsigned int TotalNumberOfDCCActiveDPP; unsigned int Total3dlutActive; -- cgit v1.2.3 From 7c228b1aabce1ee9c4d050cb68a43638dc5b3900 Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Tue, 7 Oct 2025 10:40:26 -0400 Subject: drm/amd/display: Fix wrong index for DCN401 cursor offload [Why] Payloads are ignored because the wrong index is written as part of the pipe update implementation for DCN401. [How] Align it to the DCN35 implementation and ensure the + 1 is added. Reviewed-by: Alvin Lee Signed-off-by: Nicholas Kazlauskas Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 23ecab4bcbba..b9c357a40707 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -2978,7 +2978,7 @@ void dcn401_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pip return; stream_idx = top_pipe->pipe_idx; - write_idx = cs->offload_streams[stream_idx].write_idx; + write_idx = cs->offload_streams[stream_idx].write_idx + 1; /* new payload (+1) */ payload_idx = write_idx % ARRAY_SIZE(cs->offload_streams[stream_idx].payloads); p = &cs->offload_streams[stream_idx].payloads[payload_idx].pipe_data[pipe->pipe_idx].dcn401; -- cgit v1.2.3 From 6cf0552e03033da74eff5137a10b33030c8a450a Mon Sep 17 00:00:00 2001 From: Nicholas Kazlauskas Date: Tue, 7 Oct 2025 10:39:07 -0400 Subject: drm/amd/display: Fix notification of vtotal to DMU for cursor offload [Why] It was placed after the early return and the notification is never sent. [How] Place it after .set_drr and before the return. Reviewed-by: Alvin Lee Signed-off-by: Nicholas Kazlauskas Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 4c964cf28f68..c4ef63011f47 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -493,13 +493,14 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, 1, *adjust); stream->adjust.timing_adjust_pending = false; + + if (dc->hwss.notify_cursor_offload_drr_update) + dc->hwss.notify_cursor_offload_drr_update(dc, dc->current_state, stream); + return true; } } - if (dc->hwss.notify_cursor_offload_drr_update) - dc->hwss.notify_cursor_offload_drr_update(dc, dc->current_state, stream); - return false; } -- cgit v1.2.3 From e6a8a000cfe6a1106c17ab4a47eb6dd21596968c Mon Sep 17 00:00:00 2001 From: Austin Zheng Date: Tue, 21 Oct 2025 13:52:58 +0800 Subject: drm/amd/display: Rename dml2 to dml2_0 folder [Why] dml2 folder contains all logic for all versions of DML2 This is currently DML2.0 and DML2.1. Rename dml2 to dml2_0 folder to reflect this better (dml2_0 for DML2.0). [How] Rename dml2 to dml2_0 folder and update dml2 references to use dml2_0 folder. Reviewed-by: Dillon Varone Signed-off-by: Austin Zheng Signed-off-by: waynelin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/Makefile | 2 +- drivers/gpu/drm/amd/display/dc/core/dc.c | 2 +- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 2 +- drivers/gpu/drm/amd/display/dc/core/dc_state.c | 4 +- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- drivers/gpu/drm/amd/display/dc/dml2/Makefile | 141 - drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h | 94 - .../drm/amd/display/dc/dml2/display_mode_core.c | 10335 -------------- .../drm/amd/display/dc/dml2/display_mode_core.h | 204 - .../display/dc/dml2/display_mode_core_structs.h | 2033 --- .../amd/display/dc/dml2/display_mode_lib_defines.h | 79 - .../drm/amd/display/dc/dml2/display_mode_util.c | 798 -- .../drm/amd/display/dc/dml2/display_mode_util.h | 76 - .../dc/dml2/dml21/dml21_translation_helper.c | 929 -- .../dc/dml2/dml21/dml21_translation_helper.h | 28 - .../drm/amd/display/dc/dml2/dml21/dml21_utils.c | 516 - .../drm/amd/display/dc/dml2/dml21/dml21_utils.h | 50 - .../drm/amd/display/dc/dml2/dml21/dml21_wrapper.c | 468 - .../drm/amd/display/dc/dml2/dml21/dml21_wrapper.h | 135 - .../dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h | 373 - .../dc/dml2/dml21/inc/dml2_external_lib_deps.h | 10 - .../drm/amd/display/dc/dml2/dml21/inc/dml_top.h | 46 - .../dc/dml2/dml21/inc/dml_top_dchub_registers.h | 189 - .../dc/dml2/dml21/inc/dml_top_display_cfg_types.h | 520 - .../dc/dml2/dml21/inc/dml_top_policy_types.h | 13 - .../dml2/dml21/inc/dml_top_soc_parameter_types.h | 215 - .../amd/display/dc/dml2/dml21/inc/dml_top_types.h | 750 -- .../dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c | 661 - .../dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h | 12 - .../dml21/src/dml2_core/dml2_core_dcn4_calcs.c | 13339 ------------------ .../dml21/src/dml2_core/dml2_core_dcn4_calcs.h | 39 - .../dml2/dml21/src/dml2_core/dml2_core_factory.c | 39 - .../dml2/dml21/src/dml2_core/dml2_core_factory.h | 13 - .../dml21/src/dml2_core/dml2_core_shared_types.h | 2335 ---- .../dc/dml2/dml21/src/dml2_core/dml2_core_utils.c | 786 -- .../dc/dml2/dml21/src/dml2_core/dml2_core_utils.h | 43 - .../dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c | 785 -- .../dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h | 14 - .../dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c | 50 - .../dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h | 13 - .../dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c | 198 - .../dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h | 13 - .../dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c | 39 - .../dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h | 13 - .../dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c | 706 - .../dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h | 22 - .../dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c | 2394 ---- .../dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h | 33 - .../dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c | 83 - .../dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h | 13 - .../src/dml2_standalone_libraries/lib_float_math.c | 147 - .../src/dml2_standalone_libraries/lib_float_math.h | 25 - .../dml2/dml21/src/dml2_top/dml2_top_interfaces.c | 49 - .../dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c | 10 - .../dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h | 9 - .../dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c | 1170 -- .../dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h | 14 - .../amd/display/dc/dml2/dml21/src/inc/dml2_debug.h | 189 - .../dml21/src/inc/dml2_internal_shared_types.h | 1004 -- .../amd/display/dc/dml2/dml2_dc_resource_mgmt.c | 1174 -- .../amd/display/dc/dml2/dml2_dc_resource_mgmt.h | 52 - .../gpu/drm/amd/display/dc/dml2/dml2_dc_types.h | 43 - .../drm/amd/display/dc/dml2/dml2_internal_types.h | 157 - .../drm/amd/display/dc/dml2/dml2_mall_phantom.c | 910 -- .../drm/amd/display/dc/dml2/dml2_mall_phantom.h | 52 - drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c | 311 - drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h | 47 - .../amd/display/dc/dml2/dml2_translation_helper.c | 1525 --- .../amd/display/dc/dml2/dml2_translation_helper.h | 41 - drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c | 560 - drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h | 149 - drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c | 704 - drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h | 309 - drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h | 32 - .../gpu/drm/amd/display/dc/dml2/dml_depedencies.h | 33 - .../amd/display/dc/dml2/dml_display_rq_dlg_calc.c | 573 - .../amd/display/dc/dml2/dml_display_rq_dlg_calc.h | 63 - drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h | 31 - drivers/gpu/drm/amd/display/dc/dml2_0/Makefile | 140 + drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h | 94 + .../drm/amd/display/dc/dml2_0/display_mode_core.c | 10337 ++++++++++++++ .../drm/amd/display/dc/dml2_0/display_mode_core.h | 204 + .../display/dc/dml2_0/display_mode_core_structs.h | 2032 +++ .../display/dc/dml2_0/display_mode_lib_defines.h | 79 + .../drm/amd/display/dc/dml2_0/display_mode_util.c | 798 ++ .../drm/amd/display/dc/dml2_0/display_mode_util.h | 74 + .../dc/dml2_0/dml21/dml21_translation_helper.c | 929 ++ .../dc/dml2_0/dml21/dml21_translation_helper.h | 28 + .../drm/amd/display/dc/dml2_0/dml21/dml21_utils.c | 516 + .../drm/amd/display/dc/dml2_0/dml21/dml21_utils.h | 50 + .../amd/display/dc/dml2_0/dml21/dml21_wrapper.c | 466 + .../amd/display/dc/dml2_0/dml21/dml21_wrapper.h | 135 + .../dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h | 372 + .../dc/dml2_0/dml21/inc/dml2_external_lib_deps.h | 10 + .../drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h | 46 + .../dc/dml2_0/dml21/inc/dml_top_dchub_registers.h | 189 + .../dml2_0/dml21/inc/dml_top_display_cfg_types.h | 520 + .../dc/dml2_0/dml21/inc/dml_top_policy_types.h | 13 + .../dml2_0/dml21/inc/dml_top_soc_parameter_types.h | 215 + .../display/dc/dml2_0/dml21/inc/dml_top_types.h | 750 ++ .../dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c | 661 + .../dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h | 12 + .../dml21/src/dml2_core/dml2_core_dcn4_calcs.c | 13341 +++++++++++++++++++ .../dml21/src/dml2_core/dml2_core_dcn4_calcs.h | 39 + .../dml2_0/dml21/src/dml2_core/dml2_core_factory.c | 39 + .../dml2_0/dml21/src/dml2_core/dml2_core_factory.h | 13 + .../dml21/src/dml2_core/dml2_core_shared_types.h | 2335 ++++ .../dml2_0/dml21/src/dml2_core/dml2_core_utils.c | 786 ++ .../dml2_0/dml21/src/dml2_core/dml2_core_utils.h | 43 + .../dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c | 785 ++ .../dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h | 14 + .../dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c | 50 + .../dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h | 13 + .../dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c | 198 + .../dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h | 13 + .../dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c | 39 + .../dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h | 13 + .../dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c | 706 + .../dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h | 22 + .../dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c | 2394 ++++ .../dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h | 33 + .../dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c | 83 + .../dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h | 13 + .../src/dml2_standalone_libraries/lib_float_math.c | 147 + .../src/dml2_standalone_libraries/lib_float_math.h | 25 + .../dml21/src/dml2_top/dml2_top_interfaces.c | 49 + .../dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c | 10 + .../dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h | 9 + .../dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c | 1170 ++ .../dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h | 14 + .../display/dc/dml2_0/dml21/src/inc/dml2_debug.h | 189 + .../dml21/src/inc/dml2_internal_shared_types.h | 1004 ++ .../amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c | 1174 ++ .../amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h | 52 + .../gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h | 43 + .../amd/display/dc/dml2_0/dml2_internal_types.h | 157 + .../drm/amd/display/dc/dml2_0/dml2_mall_phantom.c | 911 ++ .../drm/amd/display/dc/dml2_0/dml2_mall_phantom.h | 52 + .../gpu/drm/amd/display/dc/dml2_0/dml2_policy.c | 311 + .../gpu/drm/amd/display/dc/dml2_0/dml2_policy.h | 47 + .../display/dc/dml2_0/dml2_translation_helper.c | 1528 +++ .../display/dc/dml2_0/dml2_translation_helper.h | 41 + drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c | 560 + drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h | 149 + .../gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c | 704 + .../gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h | 309 + drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h | 32 + .../drm/amd/display/dc/dml2_0/dml_depedencies.h | 34 + .../display/dc/dml2_0/dml_display_rq_dlg_calc.c | 573 + .../display/dc/dml2_0/dml_display_rq_dlg_calc.h | 63 + .../gpu/drm/amd/display/dc/dml2_0/dml_logging.h | 32 + .../drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h | 2 +- drivers/gpu/drm/amd/display/dc/inc/core_types.h | 4 +- drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h | 4 +- drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h | 2 +- .../amd/display/dc/resource/dcn32/dcn32_resource.c | 2 +- .../amd/display/dc/resource/dcn35/dcn35_resource.c | 2 +- .../display/dc/resource/dcn351/dcn351_resource.c | 2 +- .../amd/display/dc/resource/dcn36/dcn36_resource.c | 2 +- .../display/dc/resource/dcn401/dcn401_resource.c | 2 +- .../dcn401/dcn401_soc_and_ip_translator.h | 2 +- 161 files changed, 49049 insertions(+), 49046 deletions(-) delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/Makefile delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_lib_defines.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h delete mode 100644 drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/Makefile create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_lib_defines.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h create mode 100644 drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h diff --git a/drivers/gpu/drm/amd/display/dc/Makefile b/drivers/gpu/drm/amd/display/dc/Makefile index dc943abd6dba..7277ed21552f 100644 --- a/drivers/gpu/drm/amd/display/dc/Makefile +++ b/drivers/gpu/drm/amd/display/dc/Makefile @@ -36,7 +36,7 @@ DC_LIBS += dcn30 DC_LIBS += dcn301 DC_LIBS += dcn31 DC_LIBS += dml -DC_LIBS += dml2 +DC_LIBS += dml2_0 DC_LIBS += soc_and_ip_translator endif diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index c4ef63011f47..246ea9754810 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -83,7 +83,7 @@ #include "hw_sequencer_private.h" #if defined(CONFIG_DRM_AMD_DC_FP) -#include "dml2/dml2_internal_types.h" +#include "dml2_0/dml2_internal_types.h" #include "soc_and_ip_translator.h" #endif diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index bc5dedf5f60c..8d0cdbf14e17 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -95,7 +95,7 @@ #define DC_LOGGER \ dc->ctx->logger #define DC_LOGGER_INIT(logger) -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define UNABLE_TO_SPLIT -1 diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c index c61300a7cb1c..2de8ef4a58ec 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_state.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c @@ -35,8 +35,8 @@ #include "link_enc_cfg.h" #if defined(CONFIG_DRM_AMD_DC_FP) -#include "dml2/dml2_wrapper.h" -#include "dml2/dml2_internal_types.h" +#include "dml2_0/dml2_wrapper.h" +#include "dml2_0/dml2_internal_types.h" #endif #define DC_LOGGER \ diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 66d9da5426f1..182a35f6b439 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -42,7 +42,7 @@ #include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dmub/inc/dmub_cmd.h" diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile deleted file mode 100644 index 4ff0adef233a..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile +++ /dev/null @@ -1,141 +0,0 @@ -# SPDX-License-Identifier: MIT */ -# -# Copyright 2023 Advanced Micro Devices, Inc. -# -# Permission is hereby granted, free of charge, to any person obtaining a -# copy of this software and associated documentation files (the "Software"), -# to deal in the Software without restriction, including without limitation -# the rights to use, copy, modify, merge, publish, distribute, sublicense, -# and/or sell copies of the Software, and to permit persons to whom the -# Software is furnished to do so, subject to the following conditions: -# -# The above copyright notice and this permission notice shall be included in -# all copies or substantial portions of the Software. -# -# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL -# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR -# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, -# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR -# OTHER DEALINGS IN THE SOFTWARE. -# -# Authors: AMD -# -# Makefile for dml2. - -dml2_ccflags := $(CC_FLAGS_FPU) -dml2_rcflags := $(CC_FLAGS_NO_FPU) - -ifneq ($(CONFIG_FRAME_WARN),0) - ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) - ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) - frame_warn_limit := 4096 - else - frame_warn_limit := 3072 - endif - else - frame_warn_limit := 2056 - endif - - ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) - frame_warn_flag := -Wframe-larger-than=$(frame_warn_limit) - endif -endif - -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2 -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_core -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_mcg/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_dpmm/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_pmo/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_standalone_libraries/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/inc -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/inc -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/ - -CFLAGS_$(AMDDALPATH)/dc/dml2/display_mode_core.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/display_mode_util.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_wrapper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_utils.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_policy.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_translation_helper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_mall_phantom.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml_display_rq_dlg_calc.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_dc_resource_mgmt.o := $(dml2_ccflags) - -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/display_mode_core.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/display_mode_util.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_wrapper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_utils.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_policy.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_translation_helper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_mall_phantom.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml_display_rq_dlg_calc.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_dc_resource_mgmt.o := $(dml2_rcflags) - -DML2 = display_mode_core.o display_mode_util.o dml2_wrapper.o \ - dml2_utils.o dml2_policy.o dml2_translation_helper.o dml2_dc_resource_mgmt.o dml2_mall_phantom.o \ - dml_display_rq_dlg_calc.o - -AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2/,$(DML2)) - -AMD_DISPLAY_FILES += $(AMD_DAL_DML2) - -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_ccflags) - -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_rcflags) - -DML21 := src/dml2_top/dml2_top_interfaces.o -DML21 += src/dml2_top/dml2_top_soc15.o -DML21 += src/dml2_core/dml2_core_dcn4.o -DML21 += src/dml2_core/dml2_core_utils.o -DML21 += src/dml2_core/dml2_core_factory.o -DML21 += src/dml2_core/dml2_core_dcn4_calcs.o -DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o -DML21 += src/dml2_dpmm/dml2_dpmm_factory.o -DML21 += src/dml2_mcg/dml2_mcg_dcn4.o -DML21 += src/dml2_mcg/dml2_mcg_factory.o -DML21 += src/dml2_pmo/dml2_pmo_dcn3.o -DML21 += src/dml2_pmo/dml2_pmo_factory.o -DML21 += src/dml2_pmo/dml2_pmo_dcn4_fams2.o -DML21 += src/dml2_standalone_libraries/lib_float_math.o -DML21 += dml21_translation_helper.o -DML21 += dml21_wrapper.o -DML21 += dml21_utils.o - -AMD_DAL_DML21 = $(addprefix $(AMDDALPATH)/dc/dml2/dml21/,$(DML21)) - -AMD_DISPLAY_FILES += $(AMD_DAL_DML21) - diff --git a/drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h b/drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h deleted file mode 100644 index e450445bc05d..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h +++ /dev/null @@ -1,94 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __CMNTYPES_H__ -#define __CMNTYPES_H__ - -#ifdef __GNUC__ -#if __GNUC__ == 4 && __GNUC_MINOR__ > 7 -typedef unsigned int uint; -#endif -#endif - -typedef signed char int8, *pint8; -typedef signed short int16, *pint16; -typedef signed int int32, *pint32; -typedef signed int64, *pint64; - -typedef unsigned char uint8, *puint8; -typedef unsigned short uint16, *puint16; -typedef unsigned int uint32, *puint32; -typedef unsigned uint64, *puint64; - -typedef unsigned long int ulong; -typedef unsigned char uchar; -typedef unsigned int uint; - -typedef void *pvoid; -typedef char *pchar; -typedef const void *const_pvoid; -typedef const char *const_pchar; - -typedef struct rgba_struct { - uint8 a; - uint8 r; - uint8 g; - uint8 b; -} rgba_t; - -typedef struct { - uint8 blue; - uint8 green; - uint8 red; - uint8 alpha; -} gen_color_t; - -typedef union { - uint32 val; - gen_color_t f; -} gen_color_u; - -// -// Types to make it easy to get or set the bits of a float/double. -// Avoids automatic casting from int to float and back. -// -#if 0 -typedef union { - uint32 i; - float f; -} uintfloat32; - -typedef union { - uint64 i; - double f; -} uintfloat64; - -#ifndef UNREFERENCED_PARAMETER -#define UNREFERENCED_PARAMETER(x) x = x -#endif -#endif - -#endif //__CMNTYPES_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c deleted file mode 100644 index 4b9b2e84d381..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c +++ /dev/null @@ -1,10335 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "display_mode_core.h" -#include "display_mode_util.h" -#include "display_mode_lib_defines.h" - -#include "dml_assert.h" - -#define DML2_MAX_FMT_420_BUFFER_WIDTH 4096 -#define TB_BORROWED_MAX 400 -#define DML_MAX_VSTARTUP_START 1023 - -// --------------------------- -// Declaration Begins -// --------------------------- -static void CalculateBytePerPixelAndBlockSizes( - enum dml_source_format_class SourcePixelFormat, - enum dml_swizzle_mode SurfaceTiling, - // Output - dml_uint_t *BytePerPixelY, - dml_uint_t *BytePerPixelC, - dml_float_t *BytePerPixelDETY, - dml_float_t *BytePerPixelDETC, - dml_uint_t *BlockHeight256BytesY, - dml_uint_t *BlockHeight256BytesC, - dml_uint_t *BlockWidth256BytesY, - dml_uint_t *BlockWidth256BytesC, - dml_uint_t *MacroTileHeightY, - dml_uint_t *MacroTileHeightC, - dml_uint_t *MacroTileWidthY, - dml_uint_t *MacroTileWidthC); - -static dml_float_t CalculateWriteBackDISPCLK( - enum dml_source_format_class WritebackPixelFormat, - dml_float_t PixelClock, - dml_float_t WritebackHRatio, - dml_float_t WritebackVRatio, - dml_uint_t WritebackHTaps, - dml_uint_t WritebackVTaps, - dml_uint_t WritebackSourceWidth, - dml_uint_t WritebackDestinationWidth, - dml_uint_t HTotal, - dml_uint_t WritebackLineBufferSize, - dml_float_t DISPCLKDPPCLKVCOSpeed); - -static void CalculateVMRowAndSwath( - struct display_mode_lib_scratch_st *s, - struct CalculateVMRowAndSwath_params_st *p); - -static void CalculateOutputLink( - dml_float_t PHYCLKPerState, - dml_float_t PHYCLKD18PerState, - dml_float_t PHYCLKD32PerState, - dml_float_t Downspreading, - dml_bool_t IsMainSurfaceUsingTheIndicatedTiming, - enum dml_output_encoder_class Output, - enum dml_output_format_class OutputFormat, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_float_t PixelClockBackEnd, - dml_float_t ForcedOutputLinkBPP, - dml_uint_t DSCInputBitPerComponent, - dml_uint_t NumberOfDSCSlices, - dml_float_t AudioSampleRate, - dml_uint_t AudioSampleLayout, - enum dml_odm_mode ODMModeNoDSC, - enum dml_odm_mode ODMModeDSC, - enum dml_dsc_enable DSCEnable, - dml_uint_t OutputLinkDPLanes, - enum dml_output_link_dp_rate OutputLinkDPRate, - - // Output - dml_bool_t *RequiresDSC, - dml_bool_t *RequiresFEC, - dml_float_t *OutBpp, - enum dml_output_type_and_rate__type *OutputType, - enum dml_output_type_and_rate__rate *OutputRate, - dml_uint_t *RequiredSlots); - -static void CalculateODMMode( - dml_uint_t MaximumPixelsPerLinePerDSCUnit, - dml_uint_t HActive, - enum dml_output_encoder_class Output, - enum dml_output_format_class OutputFormat, - enum dml_odm_use_policy ODMUse, - dml_float_t StateDispclk, - dml_float_t MaxDispclk, - dml_bool_t DSCEnable, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t MaxNumDPP, - dml_float_t PixelClock, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKRampingMargin, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_uint_t NumberOfDSCSlices, - - // Output - dml_bool_t *TotalAvailablePipesSupport, - dml_uint_t *NumberOfDPP, - enum dml_odm_mode *ODMMode, - dml_float_t *RequiredDISPCLKPerSurface); - -static dml_float_t CalculateRequiredDispclk( - enum dml_odm_mode ODMMode, - dml_float_t PixelClock, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKRampingMargin, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_float_t MaxDispclkSingle); - -static void CalculateSinglePipeDPPCLKAndSCLThroughput( - dml_float_t HRatio, - dml_float_t HRatioChroma, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_float_t MaxDCHUBToPSCLThroughput, - dml_float_t MaxPSCLToLBThroughput, - dml_float_t PixelClock, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t HTaps, - dml_uint_t HTapsChroma, - dml_uint_t VTaps, - dml_uint_t VTapsChroma, - - // Output - dml_float_t *PSCL_THROUGHPUT, - dml_float_t *PSCL_THROUGHPUT_CHROMA, - dml_float_t *DPPCLKUsingSingleDPP); - -static void CalculateDPPCLK( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_float_t DPPCLKUsingSingleDPP[], - dml_uint_t DPPPerSurface[], - - // Output - dml_float_t *GlobalDPPCLK, - dml_float_t Dppclk[]); - -static void CalculateMALLUseForStaticScreen( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t MALLAllocatedForDCNFinal, - enum dml_use_mall_for_static_screen_mode *UseMALLForStaticScreen, - dml_uint_t SurfaceSizeInMALL[], - dml_bool_t one_row_per_frame_fits_in_buffer[], - - // Output - dml_bool_t UsesMALLForStaticScreen[]); - -static dml_uint_t dscceComputeDelay( - dml_uint_t bpc, - dml_float_t BPP, - dml_uint_t sliceWidth, - dml_uint_t numSlices, - enum dml_output_format_class pixelFormat, - enum dml_output_encoder_class Output); - -static dml_uint_t dscComputeDelay(enum dml_output_format_class pixelFormat, - enum dml_output_encoder_class Output); - -static dml_bool_t CalculatePrefetchSchedule(struct display_mode_lib_scratch_st *scratch, - struct CalculatePrefetchSchedule_params_st *p); - -static dml_float_t RoundToDFSGranularity(dml_float_t Clock, dml_bool_t round_up, dml_float_t VCOSpeed); - -static void CalculateDCCConfiguration( - dml_bool_t DCCEnabled, - dml_bool_t DCCProgrammingAssumesScanDirectionUnknown, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t SurfaceWidthLuma, - dml_uint_t SurfaceWidthChroma, - dml_uint_t SurfaceHeightLuma, - dml_uint_t SurfaceHeightChroma, - dml_uint_t nomDETInKByte, - dml_uint_t RequestHeight256ByteLuma, - dml_uint_t RequestHeight256ByteChroma, - enum dml_swizzle_mode TilingFormat, - dml_uint_t BytePerPixelY, - dml_uint_t BytePerPixelC, - dml_float_t BytePerPixelDETY, - dml_float_t BytePerPixelDETC, - enum dml_rotation_angle SourceScan, - // Output - dml_uint_t *MaxUncompressedBlockLuma, - dml_uint_t *MaxUncompressedBlockChroma, - dml_uint_t *MaxCompressedBlockLuma, - dml_uint_t *MaxCompressedBlockChroma, - dml_uint_t *IndependentBlockLuma, - dml_uint_t *IndependentBlockChroma); - -static dml_uint_t CalculatePrefetchSourceLines( - dml_float_t VRatio, - dml_uint_t VTaps, - dml_bool_t Interlace, - dml_bool_t ProgressiveToInterlaceUnitInOPP, - dml_uint_t SwathHeight, - enum dml_rotation_angle SourceScan, - dml_bool_t ViewportStationary, - dml_uint_t SwathWidth, - dml_uint_t ViewportHeight, - dml_uint_t ViewportXStart, - dml_uint_t ViewportYStart, - - // Output - dml_uint_t *VInitPreFill, - dml_uint_t *MaxNumSwath); - -static dml_uint_t CalculateVMAndRowBytes( - dml_bool_t ViewportStationary, - dml_bool_t DCCEnable, - dml_uint_t NumberOfDPPs, - dml_uint_t BlockHeight256Bytes, - dml_uint_t BlockWidth256Bytes, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t SurfaceTiling, - dml_uint_t BytePerPixel, - enum dml_rotation_angle SourceScan, - dml_uint_t SwathWidth, - dml_uint_t ViewportHeight, - dml_uint_t ViewportXStart, - dml_uint_t ViewportYStart, - dml_bool_t GPUVMEnable, - dml_uint_t GPUVMMaxPageTableLevels, - dml_uint_t GPUVMMinPageSizeKBytes, - dml_uint_t PTEBufferSizeInRequests, - dml_uint_t Pitch, - dml_uint_t DCCMetaPitch, - dml_uint_t MacroTileWidth, - dml_uint_t MacroTileHeight, - - // Output - dml_uint_t *MetaRowByte, - dml_uint_t *PixelPTEBytesPerRow, - dml_uint_t *PixelPTEBytesPerRowStorage, // for PTE buffer size check - dml_uint_t *dpte_row_width_ub, - dml_uint_t *dpte_row_height, - dml_uint_t *dpte_row_height_linear, - dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, - dml_uint_t *dpte_row_width_ub_one_row_per_frame, - dml_uint_t *dpte_row_height_one_row_per_frame, - dml_uint_t *MetaRequestWidth, - dml_uint_t *MetaRequestHeight, - dml_uint_t *meta_row_width, - dml_uint_t *meta_row_height, - dml_uint_t *PixelPTEReqWidth, - dml_uint_t *PixelPTEReqHeight, - dml_uint_t *PTERequestSize, - dml_uint_t *DPDE0BytesFrame, - dml_uint_t *MetaPTEBytesFrame); - -static dml_float_t CalculateTWait( - dml_uint_t PrefetchMode, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, - dml_bool_t SynchronizeDRRDisplaysForUCLKPStateChangeFinal, - dml_bool_t DRRDisplay, - dml_float_t DRAMClockChangeLatency, - dml_float_t FCLKChangeLatency, - dml_float_t UrgentLatency, - dml_float_t SREnterPlusExitTime); - -static void CalculatePrefetchMode( - enum dml_prefetch_modes AllowForPStateChangeOrStutterInVBlank, - dml_uint_t *MinPrefetchMode, - dml_uint_t *MaxPrefetchMode); - -static void CalculateRowBandwidth( - dml_bool_t GPUVMEnable, - enum dml_source_format_class SourcePixelFormat, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_bool_t DCCEnable, - dml_float_t LineTime, - dml_uint_t MetaRowByteLuma, - dml_uint_t MetaRowByteChroma, - dml_uint_t meta_row_height_luma, - dml_uint_t meta_row_height_chroma, - dml_uint_t PixelPTEBytesPerRowLuma, - dml_uint_t PixelPTEBytesPerRowChroma, - dml_uint_t dpte_row_height_luma, - dml_uint_t dpte_row_height_chroma, - // Output - dml_float_t *meta_row_bw, - dml_float_t *dpte_row_bw); - -static void CalculateFlipSchedule( - dml_float_t HostVMInefficiencyFactor, - dml_float_t UrgentExtraLatency, - dml_float_t UrgentLatency, - dml_uint_t GPUVMMaxPageTableLevels, - dml_bool_t HostVMEnable, - dml_uint_t HostVMMaxNonCachedPageTableLevels, - dml_bool_t GPUVMEnable, - dml_uint_t HostVMMinPageSize, - dml_float_t PDEAndMetaPTEBytesPerFrame, - dml_float_t MetaRowBytes, - dml_float_t DPTEBytesPerRow, - dml_float_t BandwidthAvailableForImmediateFlip, - dml_uint_t TotImmediateFlipBytes, - enum dml_source_format_class SourcePixelFormat, - dml_float_t LineTime, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_float_t Tno_bw, - dml_bool_t DCCEnable, - dml_uint_t dpte_row_height, - dml_uint_t meta_row_height, - dml_uint_t dpte_row_height_chroma, - dml_uint_t meta_row_height_chroma, - dml_bool_t use_one_row_for_frame_flip, - - // Output - dml_float_t *DestinationLinesToRequestVMInImmediateFlip, - dml_float_t *DestinationLinesToRequestRowInImmediateFlip, - dml_float_t *final_flip_bw, - dml_bool_t *ImmediateFlipSupportedForPipe); - -static dml_float_t CalculateWriteBackDelay( - enum dml_source_format_class WritebackPixelFormat, - dml_float_t WritebackHRatio, - dml_float_t WritebackVRatio, - dml_uint_t WritebackVTaps, - dml_uint_t WritebackDestinationWidth, - dml_uint_t WritebackDestinationHeight, - dml_uint_t WritebackSourceHeight, - dml_uint_t HTotal); - -static void CalculateVUpdateAndDynamicMetadataParameters( - dml_uint_t MaxInterDCNTileRepeaters, - dml_float_t Dppclk, - dml_float_t DISPCLK, - dml_float_t DCFClkDeepSleep, - dml_float_t PixelClock, - dml_uint_t HTotal, - dml_uint_t VBlank, - dml_uint_t DynamicMetadataTransmittedBytes, - dml_uint_t DynamicMetadataLinesBeforeActiveRequired, - dml_uint_t InterlaceEnable, - dml_bool_t ProgressiveToInterlaceUnitInOPP, - dml_float_t *TSetup, - dml_float_t *Tdmbf, - dml_float_t *Tdmec, - dml_float_t *Tdmsks, - dml_uint_t *VUpdateOffsetPix, - dml_uint_t *VUpdateWidthPix, - dml_uint_t *VReadyOffsetPix); - -static void PixelClockAdjustmentForProgressiveToInterlaceUnit(struct dml_display_cfg_st *display_cfg, dml_bool_t ptoi_supported); - -static dml_float_t TruncToValidBPP( - dml_float_t LinkBitRate, - dml_uint_t Lanes, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_float_t PixelClock, - dml_float_t DesiredBPP, - dml_bool_t DSCEnable, - enum dml_output_encoder_class Output, - enum dml_output_format_class Format, - dml_uint_t DSCInputBitPerComponent, - dml_uint_t DSCSlices, - dml_uint_t AudioRate, - dml_uint_t AudioLayout, - enum dml_odm_mode ODMModeNoDSC, - enum dml_odm_mode ODMModeDSC, - // Output - dml_uint_t *RequiredSlotsSingle); - -static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( - struct display_mode_lib_scratch_st *s, - struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *p); - -static void CalculateDCFCLKDeepSleep( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t BytePerPixelY[], - dml_uint_t BytePerPixelC[], - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_uint_t SwathWidthY[], - dml_uint_t SwathWidthC[], - dml_uint_t DPPPerSurface[], - dml_float_t HRatio[], - dml_float_t HRatioChroma[], - dml_float_t PixelClock[], - dml_float_t PSCL_THROUGHPUT[], - dml_float_t PSCL_THROUGHPUT_CHROMA[], - dml_float_t Dppclk[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_uint_t ReturnBusWidth, - - // Output - dml_float_t *DCFCLKDeepSleep); - -static void CalculateUrgentBurstFactor( - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, - dml_uint_t swath_width_luma_ub, - dml_uint_t swath_width_chroma_ub, - dml_uint_t SwathHeightY, - dml_uint_t SwathHeightC, - dml_float_t LineTime, - dml_float_t UrgentLatency, - dml_float_t CursorBufferSize, - dml_uint_t CursorWidth, - dml_uint_t CursorBPP, - dml_float_t VRatio, - dml_float_t VRatioC, - dml_float_t BytePerPixelInDETY, - dml_float_t BytePerPixelInDETC, - dml_uint_t DETBufferSizeY, - dml_uint_t DETBufferSizeC, - // Output - dml_float_t *UrgentBurstFactorCursor, - dml_float_t *UrgentBurstFactorLuma, - dml_float_t *UrgentBurstFactorChroma, - dml_bool_t *NotEnoughUrgentLatencyHiding); - -static dml_float_t RequiredDTBCLK( - dml_bool_t DSCEnable, - dml_float_t PixelClock, - enum dml_output_format_class OutputFormat, - dml_float_t OutputBpp, - dml_uint_t DSCSlices, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_uint_t AudioRate, - dml_uint_t AudioLayoutSingle); - -static void UseMinimumDCFCLK( - struct display_mode_lib_scratch_st *scratch, - struct UseMinimumDCFCLK_params_st *p); - -static void CalculatePixelDeliveryTimes( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_float_t VRatioPrefetchY[], - dml_float_t VRatioPrefetchC[], - dml_uint_t swath_width_luma_ub[], - dml_uint_t swath_width_chroma_ub[], - dml_uint_t DPPPerSurface[], - dml_float_t HRatio[], - dml_float_t HRatioChroma[], - dml_float_t PixelClock[], - dml_float_t PSCL_THROUGHPUT[], - dml_float_t PSCL_THROUGHPUT_CHROMA[], - dml_float_t Dppclk[], - dml_uint_t BytePerPixelC[], - enum dml_rotation_angle SourceScan[], - dml_uint_t NumberOfCursors[], - dml_uint_t CursorWidth[], - dml_uint_t CursorBPP[], - dml_uint_t BlockWidth256BytesY[], - dml_uint_t BlockHeight256BytesY[], - dml_uint_t BlockWidth256BytesC[], - dml_uint_t BlockHeight256BytesC[], - - // Output - dml_float_t DisplayPipeLineDeliveryTimeLuma[], - dml_float_t DisplayPipeLineDeliveryTimeChroma[], - dml_float_t DisplayPipeLineDeliveryTimeLumaPrefetch[], - dml_float_t DisplayPipeLineDeliveryTimeChromaPrefetch[], - dml_float_t DisplayPipeRequestDeliveryTimeLuma[], - dml_float_t DisplayPipeRequestDeliveryTimeChroma[], - dml_float_t DisplayPipeRequestDeliveryTimeLumaPrefetch[], - dml_float_t DisplayPipeRequestDeliveryTimeChromaPrefetch[], - dml_float_t CursorRequestDeliveryTime[], - dml_float_t CursorRequestDeliveryTimePrefetch[]); - -static void CalculateMetaAndPTETimes( - dml_bool_t use_one_row_for_frame[], - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t GPUVMEnable, - dml_uint_t MetaChunkSize, - dml_uint_t MinMetaChunkSizeBytes, - dml_uint_t HTotal[], - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_float_t DestinationLinesToRequestRowInVBlank[], - dml_float_t DestinationLinesToRequestRowInImmediateFlip[], - dml_bool_t DCCEnable[], - dml_float_t PixelClock[], - dml_uint_t BytePerPixelY[], - dml_uint_t BytePerPixelC[], - enum dml_rotation_angle SourceScan[], - dml_uint_t dpte_row_height[], - dml_uint_t dpte_row_height_chroma[], - dml_uint_t meta_row_width[], - dml_uint_t meta_row_width_chroma[], - dml_uint_t meta_row_height[], - dml_uint_t meta_row_height_chroma[], - dml_uint_t meta_req_width[], - dml_uint_t meta_req_width_chroma[], - dml_uint_t meta_req_height[], - dml_uint_t meta_req_height_chroma[], - dml_uint_t dpte_group_bytes[], - dml_uint_t PTERequestSizeY[], - dml_uint_t PTERequestSizeC[], - dml_uint_t PixelPTEReqWidthY[], - dml_uint_t PixelPTEReqHeightY[], - dml_uint_t PixelPTEReqWidthC[], - dml_uint_t PixelPTEReqHeightC[], - dml_uint_t dpte_row_width_luma_ub[], - dml_uint_t dpte_row_width_chroma_ub[], - - // Output - dml_float_t DST_Y_PER_PTE_ROW_NOM_L[], - dml_float_t DST_Y_PER_PTE_ROW_NOM_C[], - dml_float_t DST_Y_PER_META_ROW_NOM_L[], - dml_float_t DST_Y_PER_META_ROW_NOM_C[], - dml_float_t TimePerMetaChunkNominal[], - dml_float_t TimePerChromaMetaChunkNominal[], - dml_float_t TimePerMetaChunkVBlank[], - dml_float_t TimePerChromaMetaChunkVBlank[], - dml_float_t TimePerMetaChunkFlip[], - dml_float_t TimePerChromaMetaChunkFlip[], - dml_float_t time_per_pte_group_nom_luma[], - dml_float_t time_per_pte_group_vblank_luma[], - dml_float_t time_per_pte_group_flip_luma[], - dml_float_t time_per_pte_group_nom_chroma[], - dml_float_t time_per_pte_group_vblank_chroma[], - dml_float_t time_per_pte_group_flip_chroma[]); - -static void CalculateVMGroupAndRequestTimes( - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t GPUVMEnable, - dml_uint_t GPUVMMaxPageTableLevels, - dml_uint_t HTotal[], - dml_uint_t BytePerPixelC[], - dml_float_t DestinationLinesToRequestVMInVBlank[], - dml_float_t DestinationLinesToRequestVMInImmediateFlip[], - dml_bool_t DCCEnable[], - dml_float_t PixelClock[], - dml_uint_t dpte_row_width_luma_ub[], - dml_uint_t dpte_row_width_chroma_ub[], - dml_uint_t vm_group_bytes[], - dml_uint_t dpde0_bytes_per_frame_ub_l[], - dml_uint_t dpde0_bytes_per_frame_ub_c[], - dml_uint_t meta_pte_bytes_per_frame_ub_l[], - dml_uint_t meta_pte_bytes_per_frame_ub_c[], - - // Output - dml_float_t TimePerVMGroupVBlank[], - dml_float_t TimePerVMGroupFlip[], - dml_float_t TimePerVMRequestVBlank[], - dml_float_t TimePerVMRequestFlip[]); - -static void CalculateStutterEfficiency( - struct display_mode_lib_scratch_st *scratch, - struct CalculateStutterEfficiency_params_st *p); - -static void CalculateSwathAndDETConfiguration( - struct display_mode_lib_scratch_st *scratch, - struct CalculateSwathAndDETConfiguration_params_st *p); - -static void CalculateSwathWidth( - dml_bool_t ForceSingleDPP, - dml_uint_t NumberOfActiveSurfaces, - enum dml_source_format_class SourcePixelFormat[], - enum dml_rotation_angle SourceScan[], - dml_bool_t ViewportStationary[], - dml_uint_t ViewportWidth[], - dml_uint_t ViewportHeight[], - dml_uint_t ViewportXStart[], - dml_uint_t ViewportYStart[], - dml_uint_t ViewportXStartC[], - dml_uint_t ViewportYStartC[], - dml_uint_t SurfaceWidthY[], - dml_uint_t SurfaceWidthC[], - dml_uint_t SurfaceHeightY[], - dml_uint_t SurfaceHeightC[], - enum dml_odm_mode ODMMode[], - dml_uint_t BytePerPixY[], - dml_uint_t BytePerPixC[], - dml_uint_t Read256BytesBlockHeightY[], - dml_uint_t Read256BytesBlockHeightC[], - dml_uint_t Read256BytesBlockWidthY[], - dml_uint_t Read256BytesBlockWidthC[], - dml_uint_t BlendingAndTiming[], - dml_uint_t HActive[], - dml_float_t HRatio[], - dml_uint_t DPPPerSurface[], - - // Output - dml_uint_t SwathWidthSingleDPPY[], - dml_uint_t SwathWidthSingleDPPC[], - dml_uint_t SwathWidthY[], - dml_uint_t SwathWidthC[], - dml_uint_t MaximumSwathHeightY[], - dml_uint_t MaximumSwathHeightC[], - dml_uint_t swath_width_luma_ub[], - dml_uint_t swath_width_chroma_ub[]); - -static dml_float_t CalculateExtraLatency( - dml_uint_t RoundTripPingLatencyCycles, - dml_uint_t ReorderingBytes, - dml_float_t DCFCLK, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t PixelChunkSizeInKByte, - dml_uint_t TotalNumberOfDCCActiveDPP, - dml_uint_t MetaChunkSize, - dml_float_t ReturnBW, - dml_bool_t GPUVMEnable, - dml_bool_t HostVMEnable, - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t NumberOfDPP[], - dml_uint_t dpte_group_bytes[], - dml_float_t HostVMInefficiencyFactor, - dml_uint_t HostVMMinPageSize, - dml_uint_t HostVMMaxNonCachedPageTableLevels); - -static dml_uint_t CalculateExtraLatencyBytes( - dml_uint_t ReorderingBytes, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t PixelChunkSizeInKByte, - dml_uint_t TotalNumberOfDCCActiveDPP, - dml_uint_t MetaChunkSize, - dml_bool_t GPUVMEnable, - dml_bool_t HostVMEnable, - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t NumberOfDPP[], - dml_uint_t dpte_group_bytes[], - dml_float_t HostVMInefficiencyFactor, - dml_uint_t HostVMMinPageSize, - dml_uint_t HostVMMaxNonCachedPageTableLevels); - -static dml_float_t CalculateUrgentLatency( - dml_float_t UrgentLatencyPixelDataOnly, - dml_float_t UrgentLatencyPixelMixedWithVMData, - dml_float_t UrgentLatencyVMDataOnly, - dml_bool_t DoUrgentLatencyAdjustment, - dml_float_t UrgentLatencyAdjustmentFabricClockComponent, - dml_float_t UrgentLatencyAdjustmentFabricClockReference, - dml_float_t FabricClockSingle); - -static dml_bool_t UnboundedRequest( - enum dml_unbounded_requesting_policy UseUnboundedRequestingFinal, - dml_uint_t TotalNumberOfActiveDPP, - dml_bool_t NoChromaOrLinear, - enum dml_output_encoder_class Output); - -static void CalculateSurfaceSizeInMall( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t MALLAllocatedForDCN, - enum dml_use_mall_for_static_screen_mode UseMALLForStaticScreen[], - dml_bool_t DCCEnable[], - dml_bool_t ViewportStationary[], - dml_uint_t ViewportXStartY[], - dml_uint_t ViewportYStartY[], - dml_uint_t ViewportXStartC[], - dml_uint_t ViewportYStartC[], - dml_uint_t ViewportWidthY[], - dml_uint_t ViewportHeightY[], - dml_uint_t BytesPerPixelY[], - dml_uint_t ViewportWidthC[], - dml_uint_t ViewportHeightC[], - dml_uint_t BytesPerPixelC[], - dml_uint_t SurfaceWidthY[], - dml_uint_t SurfaceWidthC[], - dml_uint_t SurfaceHeightY[], - dml_uint_t SurfaceHeightC[], - dml_uint_t Read256BytesBlockWidthY[], - dml_uint_t Read256BytesBlockWidthC[], - dml_uint_t Read256BytesBlockHeightY[], - dml_uint_t Read256BytesBlockHeightC[], - dml_uint_t ReadBlockWidthY[], - dml_uint_t ReadBlockWidthC[], - dml_uint_t ReadBlockHeightY[], - dml_uint_t ReadBlockHeightC[], - - // Output - dml_uint_t SurfaceSizeInMALL[], - dml_bool_t *ExceededMALLSize); - -static void CalculateDETBufferSize( - dml_uint_t DETSizeOverride[], - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - dml_bool_t ForceSingleDPP, - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t UnboundedRequestEnabled, - dml_uint_t nomDETInKByte, - dml_uint_t MaxTotalDETInKByte, - dml_uint_t ConfigReturnBufferSizeInKByte, - dml_uint_t MinCompressedBufferSizeInKByte, - dml_uint_t ConfigReturnBufferSegmentSizeInkByte, - dml_uint_t CompressedBufferSegmentSizeInkByteFinal, - enum dml_source_format_class SourcePixelFormat[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_uint_t RotesY[], - dml_uint_t RoundedUpMaxSwathSizeBytesC[], - dml_uint_t DPPPerSurface[], - // Output - dml_uint_t DETBufferSizeInKByte[], - dml_uint_t *CompressedBufferSizeInkByte); - -static void CalculateMaxDETAndMinCompressedBufferSize( - dml_uint_t ConfigReturnBufferSizeInKByte, - dml_uint_t ConfigReturnBufferSegmentSizeInKByte, - dml_uint_t ROBBufferSizeInKByte, - dml_uint_t MaxNumDPP, - dml_bool_t nomDETInKByteOverrideEnable, - dml_uint_t nomDETInKByteOverrideValue, - - // Output - dml_uint_t *MaxTotalDETInKByte, - dml_uint_t *nomDETInKByte, - dml_uint_t *MinCompressedBufferSizeInKByte); - -static dml_uint_t DSCDelayRequirement( - dml_bool_t DSCEnabled, - enum dml_odm_mode ODMMode, - dml_uint_t DSCInputBitPerComponent, - dml_float_t OutputBpp, - dml_uint_t HActive, - dml_uint_t HTotal, - dml_uint_t NumberOfDSCSlices, - enum dml_output_format_class OutputFormat, - enum dml_output_encoder_class Output, - dml_float_t PixelClock, - dml_float_t PixelClockBackEnd); - -static dml_bool_t CalculateVActiveBandwithSupport( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - dml_bool_t NotUrgentLatencyHiding[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[]); - -static void CalculatePrefetchBandwithSupport( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - dml_bool_t NotUrgentLatencyHiding[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_float_t cursor_bw_pre[], - dml_float_t prefetch_vmrow_bw[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[], - - // Output - dml_float_t *PrefetchBandwidth, - dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch, - dml_float_t *FractionOfUrgentBandwidth, - dml_bool_t *PrefetchBandwidthSupport); - -static dml_float_t CalculateBandwidthAvailableForImmediateFlip( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t cursor_bw_pre[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[]); - -static void CalculateImmediateFlipBandwithSupport( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - enum dml_immediate_flip_requirement ImmediateFlipRequirement[], - dml_float_t final_flip_bw[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_float_t cursor_bw_pre[], - dml_float_t prefetch_vmrow_bw[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[], - - // Output - dml_float_t *TotalBandwidth, - dml_float_t *TotalBandwidthNotIncludingMALLPrefetch, - dml_float_t *FractionOfUrgentBandwidth, - dml_bool_t *ImmediateFlipBandwidthSupport); - -// --------------------------- -// Declaration Ends -// --------------------------- - -static dml_uint_t dscceComputeDelay( - dml_uint_t bpc, - dml_float_t BPP, - dml_uint_t sliceWidth, - dml_uint_t numSlices, - enum dml_output_format_class pixelFormat, - enum dml_output_encoder_class Output) -{ - // valid bpc = source bits per component in the set of {8, 10, 12} - // valid bpp = increments of 1/16 of a bit - // min = 6/7/8 in N420/N422/444, respectively - // max = such that compression is 1:1 - //valid sliceWidth = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode) - //valid numSlices = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4} - //valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420} - - // fixed value - dml_uint_t rcModelSize = 8192; - - // N422/N420 operate at 2 pixels per clock - dml_uint_t pixelsPerClock, lstall, D, initalXmitDelay, w, s, ix, wx, p, l0, a, ax, L, - Delay, pixels; - - if (pixelFormat == dml_420) - pixelsPerClock = 2; - // #all other modes operate at 1 pixel per clock - else if (pixelFormat == dml_444) - pixelsPerClock = 1; - else if (pixelFormat == dml_n422) - pixelsPerClock = 2; - else - pixelsPerClock = 1; - - //initial transmit delay as per PPS - initalXmitDelay = (dml_uint_t)(dml_round(rcModelSize / 2.0 / BPP / pixelsPerClock, 1)); - - //compute ssm delay - if (bpc == 8) - D = 81; - else if (bpc == 10) - D = 89; - else - D = 113; - - //divide by pixel per cycle to compute slice width as seen by DSC - w = sliceWidth / pixelsPerClock; - - //422 mode has an additional cycle of delay - if (pixelFormat == dml_420 || pixelFormat == dml_444 || pixelFormat == dml_n422) - s = 0; - else - s = 1; - - //main calculation for the dscce - ix = initalXmitDelay + 45; - wx = (w + 2) / 3; - p = 3 * wx - w; - l0 = ix / w; - a = ix + p * l0; - ax = (a + 2) / 3 + D + 6 + 1; - L = (ax + wx - 1) / wx; - if ((ix % w) == 0 && p != 0) - lstall = 1; - else - lstall = 0; - Delay = L * wx * (numSlices - 1) + ax + s + lstall + 22; - - //dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels - pixels = Delay * 3 * pixelsPerClock; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: bpc: %u\n", __func__, bpc); - dml_print("DML::%s: BPP: %f\n", __func__, BPP); - dml_print("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); - dml_print("DML::%s: numSlices: %u\n", __func__, numSlices); - dml_print("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); - dml_print("DML::%s: Output: %u\n", __func__, Output); - dml_print("DML::%s: pixels: %u\n", __func__, pixels); -#endif - return pixels; -} - -static dml_uint_t dscComputeDelay(enum dml_output_format_class pixelFormat, enum dml_output_encoder_class Output) -{ - dml_uint_t Delay = 0; - - if (pixelFormat == dml_420) { - // sfr - Delay = Delay + 2; - // dsccif - Delay = Delay + 0; - // dscc - input deserializer - Delay = Delay + 3; - // dscc gets pixels every other cycle - Delay = Delay + 2; - // dscc - input cdc fifo - Delay = Delay + 12; - // dscc gets pixels every other cycle - Delay = Delay + 13; - // dscc - cdc uncertainty - Delay = Delay + 2; - // dscc - output cdc fifo - Delay = Delay + 7; - // dscc gets pixels every other cycle - Delay = Delay + 3; - // dscc - cdc uncertainty - Delay = Delay + 2; - // dscc - output serializer - Delay = Delay + 1; - // sft - Delay = Delay + 1; - } else if (pixelFormat == dml_n422) { - // sfr - Delay = Delay + 2; - // dsccif - Delay = Delay + 1; - // dscc - input deserializer - Delay = Delay + 5; - // dscc - input cdc fifo - Delay = Delay + 25; - // dscc - cdc uncertainty - Delay = Delay + 2; - // dscc - output cdc fifo - Delay = Delay + 10; - // dscc - cdc uncertainty - Delay = Delay + 2; - // dscc - output serializer - Delay = Delay + 1; - // sft - Delay = Delay + 1; - } else { - // sfr - Delay = Delay + 2; - // dsccif - Delay = Delay + 0; - // dscc - input deserializer - Delay = Delay + 3; - // dscc - input cdc fifo - Delay = Delay + 12; - // dscc - cdc uncertainty - Delay = Delay + 2; - // dscc - output cdc fifo - Delay = Delay + 7; - // dscc - output serializer - Delay = Delay + 1; - // dscc - cdc uncertainty - Delay = Delay + 2; - // sft - Delay = Delay + 1; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); - dml_print("DML::%s: Delay = %u\n", __func__, Delay); -#endif - - return Delay; -} - -static dml_bool_t CalculatePrefetchSchedule(struct display_mode_lib_scratch_st *scratch, - struct CalculatePrefetchSchedule_params_st *p) -{ - struct CalculatePrefetchSchedule_locals_st *s = &scratch->CalculatePrefetchSchedule_locals; - - s->MyError = false; - s->DPPCycles = 0; - s->DISPCLKCycles = 0; - s->DSTTotalPixelsAfterScaler = 0.0; - s->LineTime = 0.0; - s->dst_y_prefetch_equ = 0.0; - s->prefetch_bw_oto = 0.0; - s->Tvm_oto = 0.0; - s->Tr0_oto = 0.0; - s->Tvm_oto_lines = 0.0; - s->Tr0_oto_lines = 0.0; - s->dst_y_prefetch_oto = 0.0; - s->TimeForFetchingMetaPTE = 0.0; - s->TimeForFetchingRowInVBlank = 0.0; - s->LinesToRequestPrefetchPixelData = 0.0; - s->HostVMDynamicLevelsTrips = 0; - s->trip_to_mem = 0.0; - s->Tvm_trips = 0.0; - s->Tr0_trips = 0.0; - s->Tvm_trips_rounded = 0.0; - s->Tr0_trips_rounded = 0.0; - s->max_Tsw = 0.0; - s->Lsw_oto = 0.0; - s->Tpre_rounded = 0.0; - s->prefetch_bw_equ = 0.0; - s->Tvm_equ = 0.0; - s->Tr0_equ = 0.0; - s->Tdmbf = 0.0; - s->Tdmec = 0.0; - s->Tdmsks = 0.0; - s->prefetch_sw_bytes = 0.0; - s->prefetch_bw_pr = 0.0; - s->bytes_pp = 0.0; - s->dep_bytes = 0.0; - s->min_Lsw_oto = 0.0; - s->Tsw_est1 = 0.0; - s->Tsw_est3 = 0.0; - - if (p->GPUVMEnable == true && p->HostVMEnable == true) { - s->HostVMDynamicLevelsTrips = p->HostVMMaxNonCachedPageTableLevels; - } else { - s->HostVMDynamicLevelsTrips = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); - dml_print("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->GPUVMPageTableLevels); - dml_print("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); - dml_print("DML::%s: VStartup = %u\n", __func__, p->VStartup); - dml_print("DML::%s: MaxVStartup = %u\n", __func__, p->MaxVStartup); - dml_print("DML::%s: HostVMEnable = %u\n", __func__, p->HostVMEnable); - dml_print("DML::%s: HostVMInefficiencyFactor= %f\n", __func__, p->HostVMInefficiencyFactor); - dml_print("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); -#endif - CalculateVUpdateAndDynamicMetadataParameters( - p->MaxInterDCNTileRepeaters, - p->myPipe->Dppclk, - p->myPipe->Dispclk, - p->myPipe->DCFClkDeepSleep, - p->myPipe->PixelClock, - p->myPipe->HTotal, - p->myPipe->VBlank, - p->DynamicMetadataTransmittedBytes, - p->DynamicMetadataLinesBeforeActiveRequired, - p->myPipe->InterlaceEnable, - p->myPipe->ProgressiveToInterlaceUnitInOPP, - p->TSetup, - - // Output - &s->Tdmbf, - &s->Tdmec, - &s->Tdmsks, - p->VUpdateOffsetPix, - p->VUpdateWidthPix, - p->VReadyOffsetPix); - - s->LineTime = p->myPipe->HTotal / p->myPipe->PixelClock; - s->trip_to_mem = p->UrgentLatency; - s->Tvm_trips = p->UrgentExtraLatency + s->trip_to_mem * (p->GPUVMPageTableLevels * (s->HostVMDynamicLevelsTrips + 1) - 1); - - if (p->DynamicMetadataVMEnabled == true) { - *p->Tdmdl = p->TWait + s->Tvm_trips + s->trip_to_mem; - } else { - *p->Tdmdl = p->TWait + p->UrgentExtraLatency; - } - -#ifdef __DML_VBA_ALLOW_DELTA__ - if (DynamicMetadataEnable == false) { - *Tdmdl = 0.0; - } -#endif - - if (p->DynamicMetadataEnable == true) { - if (p->VStartup * s->LineTime < *p->TSetup + *p->Tdmdl + s->Tdmbf + s->Tdmec + s->Tdmsks) { - *p->NotEnoughTimeForDynamicMetadata = true; - dml_print("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); - dml_print("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - dml_print("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - dml_print("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - dml_print("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); - } else { - *p->NotEnoughTimeForDynamicMetadata = false; - } - } else { - *p->NotEnoughTimeForDynamicMetadata = false; - } - - *p->Tdmdl_vm = (p->DynamicMetadataEnable == true && p->DynamicMetadataVMEnabled == true && p->GPUVMEnable == true ? p->TWait + s->Tvm_trips : 0); - - if (p->myPipe->ScalerEnabled) - s->DPPCycles = (dml_uint_t)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCL); - else - s->DPPCycles = (dml_uint_t)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCLLBOnly); - - s->DPPCycles = (dml_uint_t)(s->DPPCycles + p->myPipe->NumberOfCursors * p->DPPCLKDelayCNVCCursor); - - s->DISPCLKCycles = (dml_uint_t)p->DISPCLKDelaySubtotal; - - if (p->myPipe->Dppclk == 0.0 || p->myPipe->Dispclk == 0.0) - return true; - - *p->DSTXAfterScaler = (dml_uint_t) dml_round(s->DPPCycles * p->myPipe->PixelClock / p->myPipe->Dppclk + s->DISPCLKCycles * p->myPipe->PixelClock / p->myPipe->Dispclk + p->DSCDelay, 1.0); - *p->DSTXAfterScaler = (dml_uint_t) dml_round(*p->DSTXAfterScaler + (p->myPipe->ODMMode != dml_odm_mode_bypass ? 18 : 0) + (p->myPipe->DPPPerSurface - 1) * p->DPP_RECOUT_WIDTH + - ((p->myPipe->ODMMode == dml_odm_mode_split_1to2 || p->myPipe->ODMMode == dml_odm_mode_mso_1to2) ? (dml_float_t)p->myPipe->HActive / 2.0 : 0) + - ((p->myPipe->ODMMode == dml_odm_mode_mso_1to4) ? (dml_float_t)p->myPipe->HActive * 3.0 / 4.0 : 0), 1.0); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); - dml_print("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); - dml_print("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); - dml_print("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); - dml_print("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); - dml_print("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); - dml_print("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); - dml_print("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); - dml_print("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); -#endif - - if (p->OutputFormat == dml_420 || (p->myPipe->InterlaceEnable && p->myPipe->ProgressiveToInterlaceUnitInOPP)) - *p->DSTYAfterScaler = 1; - else - *p->DSTYAfterScaler = 0; - - s->DSTTotalPixelsAfterScaler = *p->DSTYAfterScaler * p->myPipe->HTotal + *p->DSTXAfterScaler; - *p->DSTYAfterScaler = (dml_uint_t)(dml_floor(s->DSTTotalPixelsAfterScaler / p->myPipe->HTotal, 1)); - *p->DSTXAfterScaler = (dml_uint_t)(s->DSTTotalPixelsAfterScaler - ((dml_float_t) (*p->DSTYAfterScaler * p->myPipe->HTotal))); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); - dml_print("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); -#endif - - s->MyError = false; - - s->Tr0_trips = s->trip_to_mem * (s->HostVMDynamicLevelsTrips + 1); - - if (p->GPUVMEnable == true) { - s->Tvm_trips_rounded = dml_ceil(4.0 * s->Tvm_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; - s->Tr0_trips_rounded = dml_ceil(4.0 * s->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; - if (p->GPUVMPageTableLevels >= 3) { - *p->Tno_bw = p->UrgentExtraLatency + s->trip_to_mem * (dml_float_t) ((p->GPUVMPageTableLevels - 2) * (s->HostVMDynamicLevelsTrips + 1) - 1); - } else if (p->GPUVMPageTableLevels == 1 && p->myPipe->DCCEnable != true) { - s->Tr0_trips_rounded = dml_ceil(4.0 * p->UrgentExtraLatency / s->LineTime, 1.0) / 4.0 * s->LineTime; - *p->Tno_bw = p->UrgentExtraLatency; - } else { - *p->Tno_bw = 0; - } - } else if (p->myPipe->DCCEnable == true) { - s->Tvm_trips_rounded = s->LineTime / 4.0; - s->Tr0_trips_rounded = dml_ceil(4.0 * s->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; - *p->Tno_bw = 0; - } else { - s->Tvm_trips_rounded = s->LineTime / 4.0; - s->Tr0_trips_rounded = s->LineTime / 2.0; - *p->Tno_bw = 0; - } - s->Tvm_trips_rounded = dml_max(s->Tvm_trips_rounded, s->LineTime / 4.0); - s->Tr0_trips_rounded = dml_max(s->Tr0_trips_rounded, s->LineTime / 4.0); - - if (p->myPipe->SourcePixelFormat == dml_420_8 || p->myPipe->SourcePixelFormat == dml_420_10 || p->myPipe->SourcePixelFormat == dml_420_12) { - s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC / 4; - } else { - s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC; - } - - s->prefetch_bw_pr = s->bytes_pp * p->myPipe->PixelClock / (dml_float_t)p->myPipe->DPPPerSurface; - if (p->myPipe->VRatio < 1.0) - s->prefetch_bw_pr = p->myPipe->VRatio * s->prefetch_bw_pr; - - s->max_Tsw = (dml_max(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) * s->LineTime); - - s->prefetch_sw_bytes = p->PrefetchSourceLinesY * p->swath_width_luma_ub * p->myPipe->BytePerPixelY + p->PrefetchSourceLinesC * p->swath_width_chroma_ub * p->myPipe->BytePerPixelC; - s->prefetch_bw_oto = dml_max(s->prefetch_bw_pr, s->prefetch_sw_bytes / s->max_Tsw); - - s->min_Lsw_oto = dml_max(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML_MAX_VRATIO_PRE_OTO__; - s->min_Lsw_oto = dml_max(s->min_Lsw_oto, 1.0); - s->Lsw_oto = dml_ceil(4.0 * dml_max(s->prefetch_sw_bytes / s->prefetch_bw_oto / s->LineTime, s->min_Lsw_oto), 1.0) / 4.0; - - if (p->GPUVMEnable == true) { - s->Tvm_oto = dml_max3( - s->Tvm_trips, - *p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->prefetch_bw_oto, - s->LineTime / 4.0); - } else - s->Tvm_oto = s->LineTime / 4.0; - - if ((p->GPUVMEnable == true || p->myPipe->DCCEnable == true)) { - s->Tr0_oto = dml_max4( - s->Tr0_trips, - (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_oto, - (s->LineTime - s->Tvm_oto)/2.0, - s->LineTime / 4.0); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Tr0_oto max0 = %f\n", __func__, (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_oto); - dml_print("DML::%s: Tr0_oto max1 = %f\n", __func__, s->Tr0_trips); - dml_print("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime - s->Tvm_oto); - dml_print("DML::%s: Tr0_oto max3 = %f\n", __func__, s->LineTime / 4); -#endif - } else - s->Tr0_oto = (s->LineTime - s->Tvm_oto) / 2.0; - - s->Tvm_oto_lines = dml_ceil(4.0 * s->Tvm_oto / s->LineTime, 1) / 4.0; - s->Tr0_oto_lines = dml_ceil(4.0 * s->Tr0_oto / s->LineTime, 1) / 4.0; - s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto; - - s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + dml_max(p->TWait + p->TCalc, *p->Tdmdl)) / s->LineTime - (*p->DSTYAfterScaler + (dml_float_t) *p->DSTXAfterScaler / (dml_float_t)p->myPipe->HTotal); - s->dst_y_prefetch_equ = dml_min(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); - dml_print("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); - dml_print("DML::%s: *Tno_bw = %f\n", __func__, *p->Tno_bw); - dml_print("DML::%s: UrgentExtraLatency = %f\n", __func__, p->UrgentExtraLatency); - dml_print("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); - dml_print("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml_print("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - dml_print("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); - dml_print("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - dml_print("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); - dml_print("DML::%s: prefetch_sw_bytes = %f\n", __func__, s->prefetch_sw_bytes); - dml_print("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); - dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, p->PDEAndMetaPTEBytesFrame); - dml_print("DML::%s: MetaRowByte = %u\n", __func__, p->MetaRowByte); - dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - dml_print("DML::%s: Tvm_trips = %f\n", __func__, s->Tvm_trips); - dml_print("DML::%s: Tr0_trips = %f\n", __func__, s->Tr0_trips); - dml_print("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); - dml_print("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); - dml_print("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); - dml_print("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); - dml_print("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); - dml_print("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); - dml_print("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); - dml_print("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); -#endif - - s->dst_y_prefetch_equ = dml_floor(4.0 * (s->dst_y_prefetch_equ + 0.125), 1) / 4.0; - s->Tpre_rounded = s->dst_y_prefetch_equ * s->LineTime; - - dml_print("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); - - dml_print("DML::%s: LineTime: %f\n", __func__, s->LineTime); - dml_print("DML::%s: VStartup: %u\n", __func__, p->VStartup); - dml_print("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); - dml_print("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); - dml_print("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); - dml_print("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); - dml_print("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - dml_print("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - dml_print("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - dml_print("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); - dml_print("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); - dml_print("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); - dml_print("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); - - s->dep_bytes = dml_max(p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor, p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor); - - if (s->prefetch_sw_bytes < s->dep_bytes) { - s->prefetch_sw_bytes = 2 * s->dep_bytes; - } - - *p->DestinationLinesToRequestVMInVBlank = 0; - *p->DestinationLinesToRequestRowInVBlank = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixDataBWLuma = 0; - if (s->dst_y_prefetch_equ > 1) { - - if (s->Tpre_rounded - *p->Tno_bw > 0) { - s->PrefetchBandwidth1 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + 2 * p->MetaRowByte - + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor - + s->prefetch_sw_bytes) - / (s->Tpre_rounded - *p->Tno_bw); - s->Tsw_est1 = s->prefetch_sw_bytes / s->PrefetchBandwidth1; - } else - s->PrefetchBandwidth1 = 0; - - if (p->VStartup == p->MaxVStartup && (s->Tsw_est1 / s->LineTime < s->min_Lsw_oto) && s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw > 0) { - s->PrefetchBandwidth1 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + 2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / - (s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw); - } - - if (s->Tpre_rounded - *p->Tno_bw - 2 * s->Tr0_trips_rounded > 0) - s->PrefetchBandwidth2 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + s->prefetch_sw_bytes) / - (s->Tpre_rounded - *p->Tno_bw - 2 * s->Tr0_trips_rounded); - else - s->PrefetchBandwidth2 = 0; - - if (s->Tpre_rounded - s->Tvm_trips_rounded > 0) { - s->PrefetchBandwidth3 = (2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + s->prefetch_sw_bytes) / - (s->Tpre_rounded - s->Tvm_trips_rounded); - s->Tsw_est3 = s->prefetch_sw_bytes / s->PrefetchBandwidth3; - } - else - s->PrefetchBandwidth3 = 0; - - - if (p->VStartup == p->MaxVStartup && (s->Tsw_est3 / s->LineTime < s->min_Lsw_oto) && s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded > 0) { - s->PrefetchBandwidth3 = (2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / (s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded); - } - - if (s->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded > 0) - s->PrefetchBandwidth4 = s->prefetch_sw_bytes / (s->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded); - else - s->PrefetchBandwidth4 = 0; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Tpre_rounded: %f\n", __func__, s->Tpre_rounded); - dml_print("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); - dml_print("DML::%s: Tvm_trips_rounded: %f\n", __func__, s->Tvm_trips_rounded); - dml_print("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); - dml_print("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); - dml_print("DML::%s: PrefetchBandwidth1: %f\n", __func__, s->PrefetchBandwidth1); - dml_print("DML::%s: PrefetchBandwidth2: %f\n", __func__, s->PrefetchBandwidth2); - dml_print("DML::%s: PrefetchBandwidth3: %f\n", __func__, s->PrefetchBandwidth3); - dml_print("DML::%s: PrefetchBandwidth4: %f\n", __func__, s->PrefetchBandwidth4); -#endif - { - dml_bool_t Case1OK; - dml_bool_t Case2OK; - dml_bool_t Case3OK; - - if (s->PrefetchBandwidth1 > 0) { - if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth1 >= s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth1 >= s->Tr0_trips_rounded) { - Case1OK = true; - } else { - Case1OK = false; - } - } else { - Case1OK = false; - } - - if (s->PrefetchBandwidth2 > 0) { - if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth2 >= s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth2 < s->Tr0_trips_rounded) { - Case2OK = true; - } else { - Case2OK = false; - } - } else { - Case2OK = false; - } - - if (s->PrefetchBandwidth3 > 0) { - if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth3 < s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth3 >= s->Tr0_trips_rounded) { - Case3OK = true; - } else { - Case3OK = false; - } - } else { - Case3OK = false; - } - - if (Case1OK) { - s->prefetch_bw_equ = s->PrefetchBandwidth1; - } else if (Case2OK) { - s->prefetch_bw_equ = s->PrefetchBandwidth2; - } else if (Case3OK) { - s->prefetch_bw_equ = s->PrefetchBandwidth3; - } else { - s->prefetch_bw_equ = s->PrefetchBandwidth4; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Case1OK: %u\n", __func__, Case1OK); - dml_print("DML::%s: Case2OK: %u\n", __func__, Case2OK); - dml_print("DML::%s: Case3OK: %u\n", __func__, Case3OK); - dml_print("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); -#endif - - if (s->prefetch_bw_equ > 0) { - if (p->GPUVMEnable == true) { - s->Tvm_equ = dml_max3(*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->prefetch_bw_equ, s->Tvm_trips, s->LineTime / 4); - } else { - s->Tvm_equ = s->LineTime / 4; - } - - if ((p->GPUVMEnable == true || p->myPipe->DCCEnable == true)) { - s->Tr0_equ = dml_max4((p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_equ, s->Tr0_trips, (s->LineTime - s->Tvm_equ) / 2, s->LineTime / 4); - } else { - s->Tr0_equ = (s->LineTime - s->Tvm_equ) / 2; - } - } else { - s->Tvm_equ = 0; - s->Tr0_equ = 0; - dml_print("DML::%s: prefetch_bw_equ equals 0!\n", __func__); - } - } - - - if (s->dst_y_prefetch_oto < s->dst_y_prefetch_equ) { - *p->DestinationLinesForPrefetch = s->dst_y_prefetch_oto; - s->TimeForFetchingMetaPTE = s->Tvm_oto; - s->TimeForFetchingRowInVBlank = s->Tr0_oto; - - *p->DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; - *p->DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; - } else { - *p->DestinationLinesForPrefetch = s->dst_y_prefetch_equ; - s->TimeForFetchingMetaPTE = s->Tvm_equ; - s->TimeForFetchingRowInVBlank = s->Tr0_equ; - - if (p->VStartup == p->MaxVStartup && p->EnhancedPrefetchScheduleAccelerationFinal != 0) { - *p->DestinationLinesToRequestVMInVBlank = dml_floor(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; - *p->DestinationLinesToRequestRowInVBlank = dml_floor(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; - } else { - *p->DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; - *p->DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; - } - } - - s->LinesToRequestPrefetchPixelData = *p->DestinationLinesForPrefetch - *p->DestinationLinesToRequestVMInVBlank - 2 * *p->DestinationLinesToRequestRowInVBlank; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DestinationLinesForPrefetch = %f\n", __func__, *p->DestinationLinesForPrefetch); - dml_print("DML::%s: DestinationLinesToRequestVMInVBlank = %f\n", __func__, *p->DestinationLinesToRequestVMInVBlank); - dml_print("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); - dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); - dml_print("DML::%s: DestinationLinesToRequestRowInVBlank = %f\n", __func__, *p->DestinationLinesToRequestRowInVBlank); - dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml_print("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); -#endif - - if (s->LinesToRequestPrefetchPixelData >= 1 && s->prefetch_bw_equ > 0) { - *p->VRatioPrefetchY = (dml_float_t)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData; - *p->VRatioPrefetchY = dml_max(*p->VRatioPrefetchY, 1.0); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - dml_print("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); - dml_print("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); -#endif - if ((p->SwathHeightY > 4) && (p->VInitPreFillY > 3)) { - if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillY - 3.0) / 2.0) { - *p->VRatioPrefetchY = dml_max(*p->VRatioPrefetchY, - (dml_float_t)p->MaxNumSwathY * p->SwathHeightY / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillY - 3.0) / 2.0)); - } else { - s->MyError = true; - dml_print("DML::%s: MyErr set. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); - *p->VRatioPrefetchY = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - dml_print("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); -#endif - } - - *p->VRatioPrefetchC = (dml_float_t)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData; - *p->VRatioPrefetchC = dml_max(*p->VRatioPrefetchC, 1.0); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - dml_print("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); - dml_print("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); -#endif - if ((p->SwathHeightC > 4) && (p->VInitPreFillC > 3)) { - if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillC - 3.0) / 2.0) { - *p->VRatioPrefetchC = dml_max(*p->VRatioPrefetchC, (dml_float_t)p->MaxNumSwathC * p->SwathHeightC / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillC - 3.0) / 2.0)); - } else { - s->MyError = true; - dml_print("DML::%s: MyErr set. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); - *p->VRatioPrefetchC = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - dml_print("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - dml_print("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); -#endif - } - - *p->RequiredPrefetchPixDataBWLuma = (dml_float_t)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData - * p->myPipe->BytePerPixelY - * p->swath_width_luma_ub / s->LineTime; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - dml_print("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); - dml_print("DML::%s: RequiredPrefetchPixDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixDataBWLuma); -#endif - *p->RequiredPrefetchPixDataBWChroma = (dml_float_t)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData - *p->myPipe->BytePerPixelC - *p->swath_width_chroma_ub / s->LineTime; - } else { - s->MyError = true; - dml_print("DML:%s: MyErr set. LinesToRequestPrefetchPixelData: %f, should be > 0\n", __func__, s->LinesToRequestPrefetchPixelData); - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixDataBWLuma = 0; - *p->RequiredPrefetchPixDataBWChroma = 0; - } - - dml_print("DML: Tpre: %fus - sum of time to request meta pte, 2 x data pte + meta data, swaths\n", (dml_float_t)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingMetaPTE); - dml_print("DML: Tvm: %fus - time to fetch page tables for meta surface\n", s->TimeForFetchingMetaPTE); - dml_print("DML: Tr0: %fus - time to fetch first row of data pagetables and first row of meta data (done in parallel)\n", s->TimeForFetchingRowInVBlank); - dml_print("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (dml_float_t)s->LinesToRequestPrefetchPixelData * s->LineTime); - dml_print("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((dml_float_t) (*p->DSTXAfterScaler) / (dml_float_t)p->myPipe->HTotal)) * s->LineTime); - dml_print("DML: Tvstartup - TSetup - Tcalc - Twait - Tpre - To > 0\n"); - dml_print("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingMetaPTE - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((dml_float_t) (*p->DSTXAfterScaler) / (dml_float_t)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); - dml_print("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); - - } else { - s->MyError = true; - dml_print("DML::%s: MyErr set, dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); - s->TimeForFetchingMetaPTE = 0; - s->TimeForFetchingRowInVBlank = 0; - *p->DestinationLinesToRequestVMInVBlank = 0; - *p->DestinationLinesToRequestRowInVBlank = 0; - s->LinesToRequestPrefetchPixelData = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixDataBWLuma = 0; - *p->RequiredPrefetchPixDataBWChroma = 0; - } - - { - dml_float_t prefetch_vm_bw; - dml_float_t prefetch_row_bw; - - if (p->PDEAndMetaPTEBytesFrame == 0) { - prefetch_vm_bw = 0; - } else if (*p->DestinationLinesToRequestVMInVBlank > 0) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, p->PDEAndMetaPTEBytesFrame); - dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - dml_print("DML::%s: DestinationLinesToRequestVMInVBlank = %f\n", __func__, *p->DestinationLinesToRequestVMInVBlank); - dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); -#endif - prefetch_vm_bw = p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / (*p->DestinationLinesToRequestVMInVBlank * s->LineTime); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); -#endif - } else { - prefetch_vm_bw = 0; - s->MyError = true; - dml_print("DML::%s: MyErr set. DestinationLinesToRequestVMInVBlank=%f (should be > 0)\n", __func__, *p->DestinationLinesToRequestVMInVBlank); - } - - if (p->MetaRowByte + p->PixelPTEBytesPerRow == 0) { - prefetch_row_bw = 0; - } else if (*p->DestinationLinesToRequestRowInVBlank > 0) { - prefetch_row_bw = (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / (*p->DestinationLinesToRequestRowInVBlank * s->LineTime); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: MetaRowByte = %u\n", __func__, p->MetaRowByte); - dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - dml_print("DML::%s: DestinationLinesToRequestRowInVBlank = %f\n", __func__, *p->DestinationLinesToRequestRowInVBlank); - dml_print("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); -#endif - } else { - prefetch_row_bw = 0; - s->MyError = true; - dml_print("DML::%s: MyErr set. DestinationLinesToRequestRowInVBlank=%f (should be > 0)\n", __func__, *p->DestinationLinesToRequestRowInVBlank); - } - - *p->prefetch_vmrow_bw = dml_max(prefetch_vm_bw, prefetch_row_bw); - } - - if (s->MyError) { - s->TimeForFetchingMetaPTE = 0; - s->TimeForFetchingRowInVBlank = 0; - *p->DestinationLinesToRequestVMInVBlank = 0; - *p->DestinationLinesToRequestRowInVBlank = 0; - *p->DestinationLinesForPrefetch = 0; - s->LinesToRequestPrefetchPixelData = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixDataBWLuma = 0; - *p->RequiredPrefetchPixDataBWChroma = 0; - } - - return s->MyError; -} // CalculatePrefetchSchedule - -static void CalculateBytePerPixelAndBlockSizes( - enum dml_source_format_class SourcePixelFormat, - enum dml_swizzle_mode SurfaceTiling, - - // Output - dml_uint_t *BytePerPixelY, - dml_uint_t *BytePerPixelC, - dml_float_t *BytePerPixelDETY, - dml_float_t *BytePerPixelDETC, - dml_uint_t *BlockHeight256BytesY, - dml_uint_t *BlockHeight256BytesC, - dml_uint_t *BlockWidth256BytesY, - dml_uint_t *BlockWidth256BytesC, - dml_uint_t *MacroTileHeightY, - dml_uint_t *MacroTileHeightC, - dml_uint_t *MacroTileWidthY, - dml_uint_t *MacroTileWidthC) -{ - if (SourcePixelFormat == dml_444_64) { - *BytePerPixelDETY = 8; - *BytePerPixelDETC = 0; - *BytePerPixelY = 8; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml_444_32 || SourcePixelFormat == dml_rgbe) { - *BytePerPixelDETY = 4; - *BytePerPixelDETC = 0; - *BytePerPixelY = 4; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml_444_16 || SourcePixelFormat == dml_mono_16) { - *BytePerPixelDETY = 2; - *BytePerPixelDETC = 0; - *BytePerPixelY = 2; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml_444_8 || SourcePixelFormat == dml_mono_8) { - *BytePerPixelDETY = 1; - *BytePerPixelDETC = 0; - *BytePerPixelY = 1; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml_rgbe_alpha) { - *BytePerPixelDETY = 4; - *BytePerPixelDETC = 1; - *BytePerPixelY = 4; - *BytePerPixelC = 1; - } else if (SourcePixelFormat == dml_420_8) { - *BytePerPixelDETY = 1; - *BytePerPixelDETC = 2; - *BytePerPixelY = 1; - *BytePerPixelC = 2; - } else if (SourcePixelFormat == dml_420_12) { - *BytePerPixelDETY = 2; - *BytePerPixelDETC = 4; - *BytePerPixelY = 2; - *BytePerPixelC = 4; - } else { - *BytePerPixelDETY = (dml_float_t) (4.0 / 3); - *BytePerPixelDETC = (dml_float_t) (8.0 / 3); - *BytePerPixelY = 2; - *BytePerPixelC = 4; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); - dml_print("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); - dml_print("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); - dml_print("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); - dml_print("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); -#endif - if ((SourcePixelFormat == dml_444_64 || SourcePixelFormat == dml_444_32 - || SourcePixelFormat == dml_444_16 - || SourcePixelFormat == dml_444_8 - || SourcePixelFormat == dml_mono_16 - || SourcePixelFormat == dml_mono_8 - || SourcePixelFormat == dml_rgbe)) { - if (SurfaceTiling == dml_sw_linear) { - *BlockHeight256BytesY = 1; - } else if (SourcePixelFormat == dml_444_64) { - *BlockHeight256BytesY = 4; - } else if (SourcePixelFormat == dml_444_8) { - *BlockHeight256BytesY = 16; - } else { - *BlockHeight256BytesY = 8; - } - *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; - *BlockHeight256BytesC = 0; - *BlockWidth256BytesC = 0; - } else { - if (SurfaceTiling == dml_sw_linear) { - *BlockHeight256BytesY = 1; - *BlockHeight256BytesC = 1; - } else if (SourcePixelFormat == dml_rgbe_alpha) { - *BlockHeight256BytesY = 8; - *BlockHeight256BytesC = 16; - } else if (SourcePixelFormat == dml_420_8) { - *BlockHeight256BytesY = 16; - *BlockHeight256BytesC = 8; - } else { - *BlockHeight256BytesY = 8; - *BlockHeight256BytesC = 8; - } - *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; - *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); - dml_print("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); - dml_print("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); - dml_print("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); -#endif - - if (SurfaceTiling == dml_sw_linear) { - *MacroTileHeightY = *BlockHeight256BytesY; - *MacroTileWidthY = 256 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 256 / *BytePerPixelC / *MacroTileHeightC; - } - } else if (SurfaceTiling == dml_sw_64kb_d || SurfaceTiling == dml_sw_64kb_d_t || SurfaceTiling == dml_sw_64kb_d_x || SurfaceTiling == dml_sw_64kb_r_x) { - *MacroTileHeightY = 16 * *BlockHeight256BytesY; - *MacroTileWidthY = 65536 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = 16 * *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 65536 / *BytePerPixelC / *MacroTileHeightC; - } - } else { - *MacroTileHeightY = 32 * *BlockHeight256BytesY; - *MacroTileWidthY = 65536 * 4 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = 32 * *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 65536 * 4 / *BytePerPixelC / *MacroTileHeightC; - } - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); - dml_print("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); - dml_print("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); - dml_print("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); -#endif -} // CalculateBytePerPixelAndBlockSizes - -static noinline_for_stack dml_float_t CalculateTWait( - dml_uint_t PrefetchMode, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, - dml_bool_t SynchronizeDRRDisplaysForUCLKPStateChangeFinal, - dml_bool_t DRRDisplay, - dml_float_t DRAMClockChangeLatency, - dml_float_t FCLKChangeLatency, - dml_float_t UrgentLatency, - dml_float_t SREnterPlusExitTime) -{ - dml_float_t TWait = 0.0; - - if (PrefetchMode == 0 && - !(UseMALLForPStateChange == dml_use_mall_pstate_change_full_frame) && !(UseMALLForPStateChange == dml_use_mall_pstate_change_sub_viewport) && - !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe) && !(SynchronizeDRRDisplaysForUCLKPStateChangeFinal && DRRDisplay)) { - TWait = dml_max3(DRAMClockChangeLatency + UrgentLatency, SREnterPlusExitTime, UrgentLatency); - } else if (PrefetchMode <= 1 && !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe)) { - TWait = dml_max3(FCLKChangeLatency + UrgentLatency, SREnterPlusExitTime, UrgentLatency); - } else if (PrefetchMode <= 2 && !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe)) { - TWait = dml_max(SREnterPlusExitTime, UrgentLatency); - } else { - TWait = UrgentLatency; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: PrefetchMode = %u\n", __func__, PrefetchMode); - dml_print("DML::%s: TWait = %f\n", __func__, TWait); -#endif - return TWait; -} // CalculateTWait - - -/// @brief Calculate the "starting point" for prefetch calculation -/// if AllowForPStateChangeOrStutterInVBlank is set as a particular requirement, then the mode evalulation -/// will only be done at the given mode. If no specific requirement (i.e. *_if_possible), then will just go from -/// try all the prefetch mode in decreasing order of "difficulty" (start from 0 which means all power saving -/// features). -static void CalculatePrefetchMode( - enum dml_prefetch_modes AllowForPStateChangeOrStutterInVBlank, - dml_uint_t *MinPrefetchMode, - dml_uint_t *MaxPrefetchMode) -{ - if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_uclk_fclk_and_stutter_if_possible) { - *MinPrefetchMode = 0; // consider all pwr saving features - *MaxPrefetchMode = 3; // consider just urgent latency - } else { - if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_none) { - *MinPrefetchMode = 3; - } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_stutter) { - *MinPrefetchMode = 2; - } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_fclk_and_stutter) { - *MinPrefetchMode = 1; - } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_uclk_fclk_and_stutter) { - *MinPrefetchMode = 0; - } else { - dml_print("ERROR: Invalid AllowForPStateChangeOrStutterInVBlank setting! val=%u\n", AllowForPStateChangeOrStutterInVBlank); - ASSERT(0); - } - *MaxPrefetchMode = *MinPrefetchMode; - } -} // CalculatePrefetchMode - -static dml_float_t CalculateWriteBackDISPCLK( - enum dml_source_format_class WritebackPixelFormat, - dml_float_t PixelClock, - dml_float_t WritebackHRatio, - dml_float_t WritebackVRatio, - dml_uint_t WritebackHTaps, - dml_uint_t WritebackVTaps, - dml_uint_t WritebackSourceWidth, - dml_uint_t WritebackDestinationWidth, - dml_uint_t HTotal, - dml_uint_t WritebackLineBufferSize, - dml_float_t DISPCLKDPPCLKVCOSpeed) -{ - dml_float_t DISPCLK_H, DISPCLK_V, DISPCLK_HB; - - DISPCLK_H = PixelClock * dml_ceil(WritebackHTaps / 8.0, 1) / WritebackHRatio; - DISPCLK_V = PixelClock * (WritebackVTaps * dml_ceil(WritebackDestinationWidth / 6.0, 1) + 8.0) / (dml_float_t) HTotal; - DISPCLK_HB = PixelClock * WritebackVTaps * (WritebackDestinationWidth * WritebackVTaps - WritebackLineBufferSize / 57.0) / 6.0 / (dml_float_t) WritebackSourceWidth; - return RoundToDFSGranularity(dml_max3(DISPCLK_H, DISPCLK_V, DISPCLK_HB), 1, DISPCLKDPPCLKVCOSpeed); -} - -static dml_float_t CalculateWriteBackDelay( - enum dml_source_format_class WritebackPixelFormat, - dml_float_t WritebackHRatio, - dml_float_t WritebackVRatio, - dml_uint_t WritebackVTaps, - dml_uint_t WritebackDestinationWidth, - dml_uint_t WritebackDestinationHeight, - dml_uint_t WritebackSourceHeight, - dml_uint_t HTotal) -{ - dml_float_t CalculateWriteBackDelay; - dml_float_t Line_length; - dml_float_t Output_lines_last_notclamped; - dml_float_t WritebackVInit; - - WritebackVInit = (WritebackVRatio + WritebackVTaps + 1) / 2; - Line_length = dml_max((dml_float_t) WritebackDestinationWidth, dml_ceil((dml_float_t)WritebackDestinationWidth / 6.0, 1.0) * WritebackVTaps); - Output_lines_last_notclamped = WritebackDestinationHeight - 1 - dml_ceil(((dml_float_t)WritebackSourceHeight - (dml_float_t) WritebackVInit) / (dml_float_t)WritebackVRatio, 1.0); - if (Output_lines_last_notclamped < 0) { - CalculateWriteBackDelay = 0; - } else { - CalculateWriteBackDelay = Output_lines_last_notclamped * Line_length + (HTotal - WritebackDestinationWidth) + 80; - } - return CalculateWriteBackDelay; -} - -static void CalculateVUpdateAndDynamicMetadataParameters( - dml_uint_t MaxInterDCNTileRepeaters, - dml_float_t Dppclk, - dml_float_t Dispclk, - dml_float_t DCFClkDeepSleep, - dml_float_t PixelClock, - dml_uint_t HTotal, - dml_uint_t VBlank, - dml_uint_t DynamicMetadataTransmittedBytes, - dml_uint_t DynamicMetadataLinesBeforeActiveRequired, - dml_uint_t InterlaceEnable, - dml_bool_t ProgressiveToInterlaceUnitInOPP, - - // Output - dml_float_t *TSetup, - dml_float_t *Tdmbf, - dml_float_t *Tdmec, - dml_float_t *Tdmsks, - dml_uint_t *VUpdateOffsetPix, - dml_uint_t *VUpdateWidthPix, - dml_uint_t *VReadyOffsetPix) -{ - dml_float_t TotalRepeaterDelayTime; - TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2 / Dppclk + 3 / Dispclk); - *VUpdateWidthPix = (dml_uint_t)(dml_ceil((14.0 / DCFClkDeepSleep + 12.0 / Dppclk + TotalRepeaterDelayTime) * PixelClock, 1.0)); - *VReadyOffsetPix = (dml_uint_t)(dml_ceil(dml_max(150.0 / Dppclk, TotalRepeaterDelayTime + 20.0 / DCFClkDeepSleep + 10.0 / Dppclk) * PixelClock, 1.0)); - *VUpdateOffsetPix = (dml_uint_t)(dml_ceil(HTotal / 4.0, 1.0)); - *TSetup = (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock; - *Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / Dispclk; - *Tdmec = HTotal / PixelClock; - - if (DynamicMetadataLinesBeforeActiveRequired == 0) { - *Tdmsks = VBlank * HTotal / PixelClock / 2.0; - } else { - *Tdmsks = DynamicMetadataLinesBeforeActiveRequired * HTotal / PixelClock; - } - if (InterlaceEnable == 1 && ProgressiveToInterlaceUnitInOPP == false) { - *Tdmsks = *Tdmsks / 2; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); - dml_print("DML::%s: VBlank = %u\n", __func__, VBlank); - dml_print("DML::%s: HTotal = %u\n", __func__, HTotal); - dml_print("DML::%s: PixelClock = %f\n", __func__, PixelClock); - dml_print("DML::%s: Dppclk = %f\n", __func__, Dppclk); - dml_print("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); - dml_print("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); - dml_print("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); - - dml_print("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); - dml_print("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); - dml_print("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); - - dml_print("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); -#endif -} - -static void CalculateRowBandwidth( - dml_bool_t GPUVMEnable, - enum dml_source_format_class SourcePixelFormat, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_bool_t DCCEnable, - dml_float_t LineTime, - dml_uint_t MetaRowByteLuma, - dml_uint_t MetaRowByteChroma, - dml_uint_t meta_row_height_luma, - dml_uint_t meta_row_height_chroma, - dml_uint_t PixelPTEBytesPerRowLuma, - dml_uint_t PixelPTEBytesPerRowChroma, - dml_uint_t dpte_row_height_luma, - dml_uint_t dpte_row_height_chroma, - // Output - dml_float_t *meta_row_bw, - dml_float_t *dpte_row_bw) -{ - if (DCCEnable != true) { - *meta_row_bw = 0; - } else if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12 || SourcePixelFormat == dml_rgbe_alpha) { - *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime) - + VRatioChroma * MetaRowByteChroma - / (meta_row_height_chroma * LineTime); - } else { - *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime); - } - - if (GPUVMEnable != true) { - *dpte_row_bw = 0; - } else if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12 || SourcePixelFormat == dml_rgbe_alpha) { - *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime) - + VRatioChroma * PixelPTEBytesPerRowChroma - / (dpte_row_height_chroma * LineTime); - } else { - *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime); - } -} - -/// @brief Determine immediate flip schedule given bw remaining after considering the prefetch schedule -/// @param BandwidthAvailableForImmediateFlip Bandwidth available for iflip for all planes -static void CalculateFlipSchedule( - dml_float_t HostVMInefficiencyFactor, - dml_float_t UrgentExtraLatency, - dml_float_t UrgentLatency, - dml_uint_t GPUVMMaxPageTableLevels, - dml_bool_t HostVMEnable, - dml_uint_t HostVMMaxNonCachedPageTableLevels, - dml_bool_t GPUVMEnable, - dml_uint_t HostVMMinPageSize, - dml_float_t PDEAndMetaPTEBytesPerFrame, - dml_float_t MetaRowBytes, - dml_float_t DPTEBytesPerRow, - dml_float_t BandwidthAvailableForImmediateFlip, - dml_uint_t TotImmediateFlipBytes, - enum dml_source_format_class SourcePixelFormat, - dml_float_t LineTime, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_float_t Tno_bw, - dml_bool_t DCCEnable, - dml_uint_t dpte_row_height, - dml_uint_t meta_row_height, - dml_uint_t dpte_row_height_chroma, - dml_uint_t meta_row_height_chroma, - dml_bool_t use_one_row_for_frame_flip, - - // Output - dml_float_t *DestinationLinesToRequestVMInImmediateFlip, - dml_float_t *DestinationLinesToRequestRowInImmediateFlip, - dml_float_t *final_flip_bw, - dml_bool_t *ImmediateFlipSupportedForPipe) -{ - dml_float_t min_row_time = 0.0; - dml_uint_t HostVMDynamicLevelsTrips = 0; - dml_float_t TimeForFetchingMetaPTEImmediateFlip = 0; - dml_float_t TimeForFetchingRowInVBlankImmediateFlip = 0; - dml_float_t ImmediateFlipBW = 0; // @brief The immediate flip bandwidth for this pipe - - if (GPUVMEnable == true && HostVMEnable == true) { - HostVMDynamicLevelsTrips = HostVMMaxNonCachedPageTableLevels; - } else { - HostVMDynamicLevelsTrips = 0; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); - dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); - dml_print("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); - dml_print("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); -#endif - - if (TotImmediateFlipBytes > 0) { - if (use_one_row_for_frame_flip) { - ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + 2.0 * DPTEBytesPerRow) * BandwidthAvailableForImmediateFlip / (dml_float_t) TotImmediateFlipBytes; - } else { - ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + DPTEBytesPerRow) * BandwidthAvailableForImmediateFlip / (dml_float_t) TotImmediateFlipBytes; - } - if (GPUVMEnable == true) { - TimeForFetchingMetaPTEImmediateFlip = dml_max3(Tno_bw + PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / ImmediateFlipBW, - UrgentExtraLatency + UrgentLatency * (GPUVMMaxPageTableLevels * (HostVMDynamicLevelsTrips + 1) - 1), - LineTime / 4.0); - } else { - TimeForFetchingMetaPTEImmediateFlip = 0; - } - if ((GPUVMEnable == true || DCCEnable == true)) { - TimeForFetchingRowInVBlankImmediateFlip = dml_max3((MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / ImmediateFlipBW, UrgentLatency * (HostVMDynamicLevelsTrips + 1), LineTime / 4.0); - } else { - TimeForFetchingRowInVBlankImmediateFlip = 0; - } - - *DestinationLinesToRequestVMInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime), 1.0) / 4.0; - *DestinationLinesToRequestRowInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime), 1.0) / 4.0; - - if (GPUVMEnable == true) { - *final_flip_bw = dml_max(PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInImmediateFlip * LineTime), - (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime)); - } else if ((GPUVMEnable == true || DCCEnable == true)) { - *final_flip_bw = (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime); - } else { - *final_flip_bw = 0; - } - } else { - TimeForFetchingMetaPTEImmediateFlip = 0; - TimeForFetchingRowInVBlankImmediateFlip = 0; - *DestinationLinesToRequestVMInImmediateFlip = 0; - *DestinationLinesToRequestRowInImmediateFlip = 0; - *final_flip_bw = 0; - } - - if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_rgbe_alpha) { - if (GPUVMEnable == true && DCCEnable != true) { - min_row_time = dml_min(dpte_row_height * LineTime / VRatio, dpte_row_height_chroma * LineTime / VRatioChroma); - } else if (GPUVMEnable != true && DCCEnable == true) { - min_row_time = dml_min(meta_row_height * LineTime / VRatio, meta_row_height_chroma * LineTime / VRatioChroma); - } else { - min_row_time = dml_min4(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio, dpte_row_height_chroma * LineTime / VRatioChroma, meta_row_height_chroma * LineTime / VRatioChroma); - } - } else { - if (GPUVMEnable == true && DCCEnable != true) { - min_row_time = dpte_row_height * LineTime / VRatio; - } else if (GPUVMEnable != true && DCCEnable == true) { - min_row_time = meta_row_height * LineTime / VRatio; - } else { - min_row_time = dml_min(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio); - } - } - - if (*DestinationLinesToRequestVMInImmediateFlip >= 32 || *DestinationLinesToRequestRowInImmediateFlip >= 16 || TimeForFetchingMetaPTEImmediateFlip + 2 * TimeForFetchingRowInVBlankImmediateFlip > min_row_time) { - *ImmediateFlipSupportedForPipe = false; - } else { - *ImmediateFlipSupportedForPipe = true; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); - dml_print("DML::%s: DCCEnable = %u\n", __func__, DCCEnable); - - dml_print("DML::%s: MetaRowBytes = %f\n", __func__, MetaRowBytes); - dml_print("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); - dml_print("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); - dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); - dml_print("DML::%s: ImmediateFlipBW = %f\n", __func__, ImmediateFlipBW); - dml_print("DML::%s: PDEAndMetaPTEBytesPerFrame = %f\n", __func__, PDEAndMetaPTEBytesPerFrame); - dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); - dml_print("DML::%s: LineTime = %f\n", __func__, LineTime); - dml_print("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); - - dml_print("DML::%s: DestinationLinesToRequestVMInImmediateFlip = %f\n", __func__, *DestinationLinesToRequestVMInImmediateFlip); - dml_print("DML::%s: DestinationLinesToRequestRowInImmediateFlip = %f\n", __func__, *DestinationLinesToRequestRowInImmediateFlip); - dml_print("DML::%s: TimeForFetchingMetaPTEImmediateFlip = %f\n", __func__, TimeForFetchingMetaPTEImmediateFlip); - dml_print("DML::%s: TimeForFetchingRowInVBlankImmediateFlip = %f\n", __func__, TimeForFetchingRowInVBlankImmediateFlip); - dml_print("DML::%s: min_row_time = %f\n", __func__, min_row_time); - dml_print("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); -#endif -} // CalculateFlipSchedule - -static dml_float_t RoundToDFSGranularity(dml_float_t Clock, dml_bool_t round_up, dml_float_t VCOSpeed) -{ - if (Clock <= 0.0) - return 0.0; - else { - if (round_up) - return VCOSpeed * 4.0 / dml_floor(VCOSpeed * 4.0 / Clock, 1.0); - else - return VCOSpeed * 4.0 / dml_ceil(VCOSpeed * 4.0 / Clock, 1.0); - } -} - -static void CalculateDCCConfiguration( - dml_bool_t DCCEnabled, - dml_bool_t DCCProgrammingAssumesScanDirectionUnknown, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t SurfaceWidthLuma, - dml_uint_t SurfaceWidthChroma, - dml_uint_t SurfaceHeightLuma, - dml_uint_t SurfaceHeightChroma, - dml_uint_t nomDETInKByte, - dml_uint_t RequestHeight256ByteLuma, - dml_uint_t RequestHeight256ByteChroma, - enum dml_swizzle_mode TilingFormat, - dml_uint_t BytePerPixelY, - dml_uint_t BytePerPixelC, - dml_float_t BytePerPixelDETY, - dml_float_t BytePerPixelDETC, - enum dml_rotation_angle SourceScan, - // Output - dml_uint_t *MaxUncompressedBlockLuma, - dml_uint_t *MaxUncompressedBlockChroma, - dml_uint_t *MaxCompressedBlockLuma, - dml_uint_t *MaxCompressedBlockChroma, - dml_uint_t *IndependentBlockLuma, - dml_uint_t *IndependentBlockChroma) -{ - dml_uint_t DETBufferSizeForDCC = nomDETInKByte * 1024; - - dml_uint_t yuv420; - dml_uint_t horz_div_l; - dml_uint_t horz_div_c; - dml_uint_t vert_div_l; - dml_uint_t vert_div_c; - - dml_uint_t swath_buf_size; - dml_float_t detile_buf_vp_horz_limit; - dml_float_t detile_buf_vp_vert_limit; - - dml_uint_t MAS_vp_horz_limit; - dml_uint_t MAS_vp_vert_limit; - dml_uint_t max_vp_horz_width; - dml_uint_t max_vp_vert_height; - dml_uint_t eff_surf_width_l; - dml_uint_t eff_surf_width_c; - dml_uint_t eff_surf_height_l; - dml_uint_t eff_surf_height_c; - - dml_uint_t full_swath_bytes_horz_wc_l; - dml_uint_t full_swath_bytes_horz_wc_c; - dml_uint_t full_swath_bytes_vert_wc_l; - dml_uint_t full_swath_bytes_vert_wc_c; - - dml_uint_t req128_horz_wc_l; - dml_uint_t req128_horz_wc_c; - dml_uint_t req128_vert_wc_l; - dml_uint_t req128_vert_wc_c; - - dml_uint_t segment_order_horz_contiguous_luma; - dml_uint_t segment_order_horz_contiguous_chroma; - dml_uint_t segment_order_vert_contiguous_luma; - dml_uint_t segment_order_vert_contiguous_chroma; - - typedef enum{ - REQ_256Bytes, - REQ_128BytesNonContiguous, - REQ_128BytesContiguous, - REQ_NA - } RequestType; - - RequestType RequestLuma; - RequestType RequestChroma; - - yuv420 = ((SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12) ? 1 : 0); - horz_div_l = 1; - horz_div_c = 1; - vert_div_l = 1; - vert_div_c = 1; - - if (BytePerPixelY == 1) - vert_div_l = 0; - if (BytePerPixelC == 1) - vert_div_c = 0; - - if (BytePerPixelC == 0) { - swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 256; - detile_buf_vp_horz_limit = (dml_float_t) swath_buf_size / ((dml_float_t) RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l)); - detile_buf_vp_vert_limit = (dml_float_t) swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l)); - } else { - swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 2 * 256; - detile_buf_vp_horz_limit = (dml_float_t) swath_buf_size / ((dml_float_t) RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l) + (dml_float_t) RequestHeight256ByteChroma * BytePerPixelC / (1 + horz_div_c) / (1 + yuv420)); - detile_buf_vp_vert_limit = (dml_float_t) swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l) + 256.0 / RequestHeight256ByteChroma / (1 + vert_div_c) / (1 + yuv420)); - } - - if (SourcePixelFormat == dml_420_10) { - detile_buf_vp_horz_limit = 1.5 * detile_buf_vp_horz_limit; - detile_buf_vp_vert_limit = 1.5 * detile_buf_vp_vert_limit; - } - - detile_buf_vp_horz_limit = dml_floor(detile_buf_vp_horz_limit - 1, 16); - detile_buf_vp_vert_limit = dml_floor(detile_buf_vp_vert_limit - 1, 16); - - MAS_vp_horz_limit = SourcePixelFormat == dml_rgbe_alpha ? 3840 : 6144; - MAS_vp_vert_limit = SourcePixelFormat == dml_rgbe_alpha ? 3840 : (BytePerPixelY == 8 ? 3072 : 6144); - max_vp_horz_width = (dml_uint_t)(dml_min((dml_float_t) MAS_vp_horz_limit, detile_buf_vp_horz_limit)); - max_vp_vert_height = (dml_uint_t)(dml_min((dml_float_t) MAS_vp_vert_limit, detile_buf_vp_vert_limit)); - eff_surf_width_l = (SurfaceWidthLuma > max_vp_horz_width ? max_vp_horz_width : SurfaceWidthLuma); - eff_surf_width_c = eff_surf_width_l / (1 + yuv420); - eff_surf_height_l = (SurfaceHeightLuma > max_vp_vert_height ? max_vp_vert_height : SurfaceHeightLuma); - eff_surf_height_c = eff_surf_height_l / (1 + yuv420); - - full_swath_bytes_horz_wc_l = eff_surf_width_l * RequestHeight256ByteLuma * BytePerPixelY; - full_swath_bytes_vert_wc_l = eff_surf_height_l * 256 / RequestHeight256ByteLuma; - if (BytePerPixelC > 0) { - full_swath_bytes_horz_wc_c = eff_surf_width_c * RequestHeight256ByteChroma * BytePerPixelC; - full_swath_bytes_vert_wc_c = eff_surf_height_c * 256 / RequestHeight256ByteChroma; - } else { - full_swath_bytes_horz_wc_c = 0; - full_swath_bytes_vert_wc_c = 0; - } - - if (SourcePixelFormat == dml_420_10) { - full_swath_bytes_horz_wc_l = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_horz_wc_l * 2.0 / 3.0, 256.0)); - full_swath_bytes_horz_wc_c = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_horz_wc_c * 2.0 / 3.0, 256.0)); - full_swath_bytes_vert_wc_l = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_vert_wc_l * 2.0 / 3.0, 256.0)); - full_swath_bytes_vert_wc_c = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_vert_wc_c * 2.0 / 3.0, 256.0)); - } - - if (2 * full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 0; - req128_horz_wc_c = 0; - } else if (full_swath_bytes_horz_wc_l < 1.5 * full_swath_bytes_horz_wc_c && 2 * full_swath_bytes_horz_wc_l + full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 0; - req128_horz_wc_c = 1; - } else if (full_swath_bytes_horz_wc_l >= 1.5 * full_swath_bytes_horz_wc_c && full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 1; - req128_horz_wc_c = 0; - } else { - req128_horz_wc_l = 1; - req128_horz_wc_c = 1; - } - - if (2 * full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 0; - req128_vert_wc_c = 0; - } else if (full_swath_bytes_vert_wc_l < 1.5 * full_swath_bytes_vert_wc_c && 2 * full_swath_bytes_vert_wc_l + full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 0; - req128_vert_wc_c = 1; - } else if (full_swath_bytes_vert_wc_l >= 1.5 * full_swath_bytes_vert_wc_c && full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 1; - req128_vert_wc_c = 0; - } else { - req128_vert_wc_l = 1; - req128_vert_wc_c = 1; - } - - if (BytePerPixelY == 2) { - segment_order_horz_contiguous_luma = 0; - segment_order_vert_contiguous_luma = 1; - } else { - segment_order_horz_contiguous_luma = 1; - segment_order_vert_contiguous_luma = 0; - } - - if (BytePerPixelC == 2) { - segment_order_horz_contiguous_chroma = 0; - segment_order_vert_contiguous_chroma = 1; - } else { - segment_order_horz_contiguous_chroma = 1; - segment_order_vert_contiguous_chroma = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); - dml_print("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - dml_print("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); - dml_print("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); - dml_print("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); - dml_print("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); - dml_print("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); - dml_print("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); - dml_print("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); -#endif - - if (DCCProgrammingAssumesScanDirectionUnknown == true) { - if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { - RequestLuma = REQ_256Bytes; - } else if ((req128_horz_wc_l == 1 && segment_order_horz_contiguous_luma == 0) || (req128_vert_wc_l == 1 && segment_order_vert_contiguous_luma == 0)) { - RequestLuma = REQ_128BytesNonContiguous; - } else { - RequestLuma = REQ_128BytesContiguous; - } - if (req128_horz_wc_c == 0 && req128_vert_wc_c == 0) { - RequestChroma = REQ_256Bytes; - } else if ((req128_horz_wc_c == 1 && segment_order_horz_contiguous_chroma == 0) || (req128_vert_wc_c == 1 && segment_order_vert_contiguous_chroma == 0)) { - RequestChroma = REQ_128BytesNonContiguous; - } else { - RequestChroma = REQ_128BytesContiguous; - } - } else if (!dml_is_vertical_rotation(SourceScan)) { - if (req128_horz_wc_l == 0) { - RequestLuma = REQ_256Bytes; - } else if (segment_order_horz_contiguous_luma == 0) { - RequestLuma = REQ_128BytesNonContiguous; - } else { - RequestLuma = REQ_128BytesContiguous; - } - if (req128_horz_wc_c == 0) { - RequestChroma = REQ_256Bytes; - } else if (segment_order_horz_contiguous_chroma == 0) { - RequestChroma = REQ_128BytesNonContiguous; - } else { - RequestChroma = REQ_128BytesContiguous; - } - } else { - if (req128_vert_wc_l == 0) { - RequestLuma = REQ_256Bytes; - } else if (segment_order_vert_contiguous_luma == 0) { - RequestLuma = REQ_128BytesNonContiguous; - } else { - RequestLuma = REQ_128BytesContiguous; - } - if (req128_vert_wc_c == 0) { - RequestChroma = REQ_256Bytes; - } else if (segment_order_vert_contiguous_chroma == 0) { - RequestChroma = REQ_128BytesNonContiguous; - } else { - RequestChroma = REQ_128BytesContiguous; - } - } - - if (RequestLuma == REQ_256Bytes) { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 256; - *IndependentBlockLuma = 0; - } else if (RequestLuma == REQ_128BytesContiguous) { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 128; - *IndependentBlockLuma = 128; - } else { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 64; - *IndependentBlockLuma = 64; - } - - if (RequestChroma == REQ_256Bytes) { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 256; - *IndependentBlockChroma = 0; - } else if (RequestChroma == REQ_128BytesContiguous) { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 128; - *IndependentBlockChroma = 128; - } else { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 64; - *IndependentBlockChroma = 64; - } - - if (DCCEnabled != true || BytePerPixelC == 0) { - *MaxUncompressedBlockChroma = 0; - *MaxCompressedBlockChroma = 0; - *IndependentBlockChroma = 0; - } - - if (DCCEnabled != true) { - *MaxUncompressedBlockLuma = 0; - *MaxCompressedBlockLuma = 0; - *IndependentBlockLuma = 0; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); - dml_print("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); - dml_print("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); - dml_print("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); - dml_print("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); - dml_print("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); -#endif - -} // CalculateDCCConfiguration - -static dml_uint_t CalculatePrefetchSourceLines( - dml_float_t VRatio, - dml_uint_t VTaps, - dml_bool_t Interlace, - dml_bool_t ProgressiveToInterlaceUnitInOPP, - dml_uint_t SwathHeight, - enum dml_rotation_angle SourceScan, - dml_bool_t ViewportStationary, - dml_uint_t SwathWidth, - dml_uint_t ViewportHeight, - dml_uint_t ViewportXStart, - dml_uint_t ViewportYStart, - - // Output - dml_uint_t *VInitPreFill, - dml_uint_t *MaxNumSwath) -{ - - dml_uint_t vp_start_rot = 0; - dml_uint_t sw0_tmp = 0; - dml_uint_t MaxPartialSwath = 0; - dml_float_t numLines = 0; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VRatio = %f\n", __func__, VRatio); - dml_print("DML::%s: VTaps = %u\n", __func__, VTaps); - dml_print("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); - dml_print("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); - dml_print("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); - dml_print("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); -#endif - if (ProgressiveToInterlaceUnitInOPP) - *VInitPreFill = (dml_uint_t)(dml_floor((VRatio + (dml_float_t) VTaps + 1) / 2.0, 1)); - else - *VInitPreFill = (dml_uint_t)(dml_floor((VRatio + (dml_float_t) VTaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1)); - - if (ViewportStationary) { - if (SourceScan == dml_rotation_180 || SourceScan == dml_rotation_180m) { - vp_start_rot = SwathHeight - (((dml_uint_t) (ViewportYStart + ViewportHeight - 1) % SwathHeight) + 1); - } else if (SourceScan == dml_rotation_270 || SourceScan == dml_rotation_90m) { - vp_start_rot = ViewportXStart; - } else if (SourceScan == dml_rotation_90 || SourceScan == dml_rotation_270m) { - vp_start_rot = SwathHeight - (((dml_uint_t)(ViewportYStart + SwathWidth - 1) % SwathHeight) + 1); - } else { - vp_start_rot = ViewportYStart; - } - sw0_tmp = SwathHeight - (vp_start_rot % SwathHeight); - if (sw0_tmp < *VInitPreFill) { - *MaxNumSwath = (dml_uint_t)(dml_ceil((*VInitPreFill - sw0_tmp) / (dml_float_t) SwathHeight, 1) + 1); - } else { - *MaxNumSwath = 1; - } - MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (vp_start_rot + *VInitPreFill - 1) % SwathHeight)); - } else { - *MaxNumSwath = (dml_uint_t)(dml_ceil((*VInitPreFill - 1.0) / (dml_float_t) SwathHeight, 1) + 1); - if (*VInitPreFill > 1) { - MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (*VInitPreFill - 2) % SwathHeight)); - } else { - MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (*VInitPreFill + SwathHeight - 2) % SwathHeight)); - } - } - numLines = *MaxNumSwath * SwathHeight + MaxPartialSwath; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); - dml_print("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); - dml_print("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); - dml_print("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); - dml_print("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); -#endif - return (dml_uint_t)(numLines); - -} // CalculatePrefetchSourceLines - -static dml_uint_t CalculateVMAndRowBytes( - dml_bool_t ViewportStationary, - dml_bool_t DCCEnable, - dml_uint_t NumberOfDPPs, - dml_uint_t BlockHeight256Bytes, - dml_uint_t BlockWidth256Bytes, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t SurfaceTiling, - dml_uint_t BytePerPixel, - enum dml_rotation_angle SourceScan, - dml_uint_t SwathWidth, - dml_uint_t ViewportHeight, - dml_uint_t ViewportXStart, - dml_uint_t ViewportYStart, - dml_bool_t GPUVMEnable, - dml_uint_t GPUVMMaxPageTableLevels, - dml_uint_t GPUVMMinPageSizeKBytes, - dml_uint_t PTEBufferSizeInRequests, - dml_uint_t Pitch, - dml_uint_t DCCMetaPitch, - dml_uint_t MacroTileWidth, - dml_uint_t MacroTileHeight, - - // Output - dml_uint_t *MetaRowByte, - dml_uint_t *PixelPTEBytesPerRow, // for bandwidth calculation - dml_uint_t *PixelPTEBytesPerRowStorage, // for PTE buffer size check - dml_uint_t *dpte_row_width_ub, - dml_uint_t *dpte_row_height, - dml_uint_t *dpte_row_height_linear, - dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, - dml_uint_t *dpte_row_width_ub_one_row_per_frame, - dml_uint_t *dpte_row_height_one_row_per_frame, - dml_uint_t *MetaRequestWidth, - dml_uint_t *MetaRequestHeight, - dml_uint_t *meta_row_width, - dml_uint_t *meta_row_height, - dml_uint_t *PixelPTEReqWidth, - dml_uint_t *PixelPTEReqHeight, - dml_uint_t *PTERequestSize, - dml_uint_t *DPDE0BytesFrame, - dml_uint_t *MetaPTEBytesFrame) -{ - dml_uint_t MPDEBytesFrame; - dml_uint_t DCCMetaSurfaceBytes; - dml_uint_t ExtraDPDEBytesFrame; - dml_uint_t PDEAndMetaPTEBytesFrame; - dml_uint_t MacroTileSizeBytes; - dml_uint_t vp_height_meta_ub; - dml_uint_t vp_height_dpte_ub; - - dml_uint_t PixelPTEReqWidth_linear = 0; // VBA_DELTA. VBA doesn't calculate this - - *MetaRequestHeight = 8 * BlockHeight256Bytes; - *MetaRequestWidth = 8 * BlockWidth256Bytes; - if (SurfaceTiling == dml_sw_linear) { - *meta_row_height = 32; - *meta_row_width = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *MetaRequestWidth - 1, *MetaRequestWidth) - dml_floor(ViewportXStart, *MetaRequestWidth)); - } else if (!dml_is_vertical_rotation(SourceScan)) { - *meta_row_height = *MetaRequestHeight; - if (ViewportStationary && NumberOfDPPs == 1) { - *meta_row_width = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *MetaRequestWidth - 1, *MetaRequestWidth) - dml_floor(ViewportXStart, *MetaRequestWidth)); - } else { - *meta_row_width = (dml_uint_t)(dml_ceil(SwathWidth - 1, *MetaRequestWidth) + *MetaRequestWidth); - } - *MetaRowByte = (dml_uint_t)(*meta_row_width * *MetaRequestHeight * BytePerPixel / 256.0); - } else { - *meta_row_height = *MetaRequestWidth; - if (ViewportStationary && NumberOfDPPs == 1) { - *meta_row_width = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + *MetaRequestHeight - 1, *MetaRequestHeight) - dml_floor(ViewportYStart, *MetaRequestHeight)); - } else { - *meta_row_width = (dml_uint_t)(dml_ceil(SwathWidth - 1, *MetaRequestHeight) + *MetaRequestHeight); - } - *MetaRowByte = (dml_uint_t)(*meta_row_width * *MetaRequestWidth * BytePerPixel / 256.0); - } - - if (ViewportStationary && (NumberOfDPPs == 1 || !dml_is_vertical_rotation(SourceScan))) { - vp_height_meta_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + 64 * BlockHeight256Bytes - 1, 64 * BlockHeight256Bytes) - dml_floor(ViewportYStart, 64 * BlockHeight256Bytes)); - } else if (!dml_is_vertical_rotation(SourceScan)) { - vp_height_meta_ub = (dml_uint_t)(dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes) + 64 * BlockHeight256Bytes); - } else { - vp_height_meta_ub = (dml_uint_t)(dml_ceil(SwathWidth - 1, 64 * BlockHeight256Bytes) + 64 * BlockHeight256Bytes); - } - - DCCMetaSurfaceBytes = (dml_uint_t)(DCCMetaPitch * vp_height_meta_ub * BytePerPixel / 256.0); - - if (GPUVMEnable == true) { - *MetaPTEBytesFrame = (dml_uint_t)((dml_ceil((dml_float_t) (DCCMetaSurfaceBytes - 4.0 * 1024.0) / (8 * 4.0 * 1024), 1) + 1) * 64); - MPDEBytesFrame = 128 * (GPUVMMaxPageTableLevels - 1); - } else { - *MetaPTEBytesFrame = 0; - MPDEBytesFrame = 0; - } - - if (DCCEnable != true) { - *MetaPTEBytesFrame = 0; - MPDEBytesFrame = 0; - *MetaRowByte = 0; - } - - MacroTileSizeBytes = MacroTileWidth * BytePerPixel * MacroTileHeight; - - if (ViewportStationary && (NumberOfDPPs == 1 || !dml_is_vertical_rotation(SourceScan))) { - vp_height_dpte_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + MacroTileHeight - 1, MacroTileHeight) - dml_floor(ViewportYStart, MacroTileHeight)); - } else if (!dml_is_vertical_rotation(SourceScan)) { - vp_height_dpte_ub = (dml_uint_t)(dml_ceil(ViewportHeight - 1, MacroTileHeight) + MacroTileHeight); - } else { - vp_height_dpte_ub = (dml_uint_t)(dml_ceil(SwathWidth - 1, MacroTileHeight) + MacroTileHeight); - } - - if (GPUVMEnable == true && GPUVMMaxPageTableLevels > 1) { - *DPDE0BytesFrame = (dml_uint_t)(64 * (dml_ceil((dml_float_t) (Pitch * vp_height_dpte_ub * BytePerPixel - MacroTileSizeBytes) / (dml_float_t) (8 * 2097152), 1) + 1)); - ExtraDPDEBytesFrame = 128 * (GPUVMMaxPageTableLevels - 2); - } else { - *DPDE0BytesFrame = 0; - ExtraDPDEBytesFrame = 0; - } - - PDEAndMetaPTEBytesFrame = *MetaPTEBytesFrame + MPDEBytesFrame + *DPDE0BytesFrame + ExtraDPDEBytesFrame; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DCCEnable = %u\n", __func__, DCCEnable); - dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); - dml_print("DML::%s: SwModeLinear = %u\n", __func__, SurfaceTiling == dml_sw_linear); - dml_print("DML::%s: BytePerPixel = %u\n", __func__, BytePerPixel); - dml_print("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, GPUVMMaxPageTableLevels); - dml_print("DML::%s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); - dml_print("DML::%s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); - dml_print("DML::%s: MacroTileHeight = %u\n", __func__, MacroTileHeight); - dml_print("DML::%s: MacroTileWidth = %u\n", __func__, MacroTileWidth); - dml_print("DML::%s: MetaPTEBytesFrame = %u\n", __func__, *MetaPTEBytesFrame); - dml_print("DML::%s: MPDEBytesFrame = %u\n", __func__, MPDEBytesFrame); - dml_print("DML::%s: DPDE0BytesFrame = %u\n", __func__, *DPDE0BytesFrame); - dml_print("DML::%s: ExtraDPDEBytesFrame= %u\n", __func__, ExtraDPDEBytesFrame); - dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, PDEAndMetaPTEBytesFrame); - dml_print("DML::%s: ViewportHeight = %u\n", __func__, ViewportHeight); - dml_print("DML::%s: SwathWidth = %u\n", __func__, SwathWidth); - dml_print("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); -#endif - - if (SurfaceTiling == dml_sw_linear) { - *PixelPTEReqHeight = 1; - *PixelPTEReqWidth = GPUVMMinPageSizeKBytes * 1024 * 8 / BytePerPixel; - PixelPTEReqWidth_linear = GPUVMMinPageSizeKBytes * 1024 * 8 / BytePerPixel; - *PTERequestSize = 64; - } else if (GPUVMMinPageSizeKBytes == 4) { - *PixelPTEReqHeight = 16 * BlockHeight256Bytes; - *PixelPTEReqWidth = 16 * BlockWidth256Bytes; - *PTERequestSize = 128; - } else { - *PixelPTEReqHeight = MacroTileHeight; - *PixelPTEReqWidth = 8 * 1024 * GPUVMMinPageSizeKBytes / (MacroTileHeight * BytePerPixel); - *PTERequestSize = 64; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); - dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u (after HostVM factor)\n", __func__, PDEAndMetaPTEBytesFrame); - dml_print("DML::%s: PixelPTEReqHeight = %u\n", __func__, *PixelPTEReqHeight); - dml_print("DML::%s: PixelPTEReqWidth = %u\n", __func__, *PixelPTEReqWidth); - dml_print("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); - dml_print("DML::%s: PTERequestSize = %u\n", __func__, *PTERequestSize); - dml_print("DML::%s: Pitch = %u\n", __func__, Pitch); -#endif - - *dpte_row_height_one_row_per_frame = vp_height_dpte_ub; - *dpte_row_width_ub_one_row_per_frame = (dml_uint_t)((dml_ceil(((dml_float_t)Pitch * (dml_float_t) *dpte_row_height_one_row_per_frame / (dml_float_t) *PixelPTEReqHeight - 1) / (dml_float_t) *PixelPTEReqWidth, 1) + 1) * (dml_float_t) *PixelPTEReqWidth); - *PixelPTEBytesPerRow_one_row_per_frame = (dml_uint_t)((dml_float_t) *dpte_row_width_ub_one_row_per_frame / (dml_float_t) *PixelPTEReqWidth * *PTERequestSize); - - if (SurfaceTiling == dml_sw_linear) { - *dpte_row_height = (dml_uint_t)(dml_min(128, 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1))); - dml_print("DML::%s: dpte_row_height term 1 = %u\n", __func__, PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch); - dml_print("DML::%s: dpte_row_height term 2 = %f\n", __func__, dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch)); - dml_print("DML::%s: dpte_row_height term 3 = %f\n", __func__, dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1)); - dml_print("DML::%s: dpte_row_height term 4 = %u\n", __func__, 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1)); - dml_print("DML::%s: dpte_row_height = %u\n", __func__, *dpte_row_height); - - *dpte_row_width_ub = (dml_uint_t)(dml_ceil(((dml_float_t) Pitch * (dml_float_t) *dpte_row_height - 1), (dml_float_t) *PixelPTEReqWidth) + *PixelPTEReqWidth); - *PixelPTEBytesPerRow = (dml_uint_t)((dml_float_t) *dpte_row_width_ub / (dml_float_t) *PixelPTEReqWidth * *PTERequestSize); - - // VBA_DELTA, VBA doesn't have programming value for pte row height linear. - *dpte_row_height_linear = 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * PixelPTEReqWidth_linear / Pitch), 1); - if (*dpte_row_height_linear > 128) - *dpte_row_height_linear = 128; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *dpte_row_width_ub); -#endif - - } else if (!dml_is_vertical_rotation(SourceScan)) { - *dpte_row_height = *PixelPTEReqHeight; - - if (GPUVMMinPageSizeKBytes > 64) { - *dpte_row_width_ub = (dml_uint_t)((dml_ceil(((dml_float_t) Pitch * (dml_float_t) *dpte_row_height / (dml_float_t) *PixelPTEReqHeight - 1) / (dml_float_t) *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth); - } else if (ViewportStationary && (NumberOfDPPs == 1)) { - *dpte_row_width_ub = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *PixelPTEReqWidth - 1, *PixelPTEReqWidth) - dml_floor(ViewportXStart, *PixelPTEReqWidth)); - } else { - *dpte_row_width_ub = (dml_uint_t)((dml_ceil((dml_float_t) (SwathWidth - 1) / (dml_float_t)*PixelPTEReqWidth, 1) + 1.0) * *PixelPTEReqWidth); - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *dpte_row_width_ub); -#endif - - ASSERT(*PixelPTEReqWidth); - if (*PixelPTEReqWidth != 0) - *PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize; - } else { - *dpte_row_height = (dml_uint_t)(dml_min(*PixelPTEReqWidth, MacroTileWidth)); - - if (ViewportStationary && (NumberOfDPPs == 1)) { - *dpte_row_width_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + *PixelPTEReqHeight - 1, *PixelPTEReqHeight) - dml_floor(ViewportYStart, *PixelPTEReqHeight)); - } else { - *dpte_row_width_ub = (dml_uint_t)((dml_ceil((dml_float_t) (SwathWidth - 1) / (dml_float_t) *PixelPTEReqHeight, 1) + 1) * *PixelPTEReqHeight); - } - - *PixelPTEBytesPerRow = (dml_uint_t)((dml_float_t) *dpte_row_width_ub / (dml_float_t) *PixelPTEReqHeight * *PTERequestSize); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *dpte_row_width_ub); -#endif - } - - if (GPUVMEnable != true) - *PixelPTEBytesPerRow = 0; - - *PixelPTEBytesPerRowStorage = *PixelPTEBytesPerRow; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); - dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); - dml_print("DML::%s: dpte_row_height = %u\n", __func__, *dpte_row_height); - dml_print("DML::%s: dpte_row_height_linear = %u\n", __func__, *dpte_row_height_linear); - dml_print("DML::%s: dpte_row_width_ub = %u\n", __func__, *dpte_row_width_ub); - dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *PixelPTEBytesPerRow); - dml_print("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *PixelPTEBytesPerRowStorage); - dml_print("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); - dml_print("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *dpte_row_height_one_row_per_frame); - dml_print("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *dpte_row_width_ub_one_row_per_frame); - dml_print("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *PixelPTEBytesPerRow_one_row_per_frame); -#endif - - dml_print("DML: vm_bytes = meta_pte_bytes_per_frame (per_pipe) = MetaPTEBytesFrame = : %i\n", *MetaPTEBytesFrame); - - return PDEAndMetaPTEBytesFrame; -} // CalculateVMAndRowBytes - -static void PixelClockAdjustmentForProgressiveToInterlaceUnit(struct dml_display_cfg_st *display_cfg, dml_bool_t ptoi_supported) -{ - dml_uint_t num_active_planes = dml_get_num_active_planes(display_cfg); - - //Progressive To Interlace Unit Effect - for (dml_uint_t k = 0; k < num_active_planes; ++k) { - display_cfg->output.PixelClockBackEnd[k] = display_cfg->timing.PixelClock[k]; - if (display_cfg->timing.Interlace[k] == 1 && ptoi_supported == true) { - display_cfg->timing.PixelClock[k] = 2 * display_cfg->timing.PixelClock[k]; - } - } -} - -static dml_float_t TruncToValidBPP( - dml_float_t LinkBitRate, - dml_uint_t Lanes, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_float_t PixelClock, - dml_float_t DesiredBPP, - dml_bool_t DSCEnable, - enum dml_output_encoder_class Output, - enum dml_output_format_class Format, - dml_uint_t DSCInputBitPerComponent, - dml_uint_t DSCSlices, - dml_uint_t AudioRate, - dml_uint_t AudioLayout, - enum dml_odm_mode ODMModeNoDSC, - enum dml_odm_mode ODMModeDSC, - - // Output - dml_uint_t *RequiredSlots) -{ - dml_float_t MaxLinkBPP; - dml_uint_t MinDSCBPP; - dml_float_t MaxDSCBPP; - dml_uint_t NonDSCBPP0; - dml_uint_t NonDSCBPP1; - dml_uint_t NonDSCBPP2; - - if (Format == dml_420) { - NonDSCBPP0 = 12; - NonDSCBPP1 = 15; - NonDSCBPP2 = 18; - MinDSCBPP = 6; - MaxDSCBPP = 1.5 * DSCInputBitPerComponent - 1.0 / 16; - } else if (Format == dml_444) { - NonDSCBPP0 = 24; - NonDSCBPP1 = 30; - NonDSCBPP2 = 36; - MinDSCBPP = 8; - MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16; - } else { - if (Output == dml_hdmi) { - NonDSCBPP0 = 24; - NonDSCBPP1 = 24; - NonDSCBPP2 = 24; - } else { - NonDSCBPP0 = 16; - NonDSCBPP1 = 20; - NonDSCBPP2 = 24; - } - if (Format == dml_n422) { - MinDSCBPP = 7; - MaxDSCBPP = 2 * DSCInputBitPerComponent - 1.0 / 16.0; - } else { - MinDSCBPP = 8; - MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16.0; - } - } - - if (Output == dml_dp2p0) { - MaxLinkBPP = LinkBitRate * Lanes / PixelClock * 128.0 / 132.0 * 383.0 / 384.0 * 65536.0 / 65540.0; - } else if (DSCEnable && Output == dml_dp) { - MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock * (1 - 2.4 / 100); - } else { - MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock; - } - - if (DSCEnable) { - if (ODMModeDSC == dml_odm_mode_combine_4to1) { - MaxLinkBPP = dml_min(MaxLinkBPP, 16); - } else if (ODMModeDSC == dml_odm_mode_combine_2to1) { - MaxLinkBPP = dml_min(MaxLinkBPP, 32); - } else if (ODMModeDSC == dml_odm_mode_split_1to2) { - MaxLinkBPP = 2 * MaxLinkBPP; - } - } else { - if (ODMModeNoDSC == dml_odm_mode_combine_4to1) { - MaxLinkBPP = dml_min(MaxLinkBPP, 16); - } else if (ODMModeNoDSC == dml_odm_mode_combine_2to1) { - MaxLinkBPP = dml_min(MaxLinkBPP, 32); - } else if (ODMModeNoDSC == dml_odm_mode_split_1to2) { - MaxLinkBPP = 2 * MaxLinkBPP; - } - } - - *RequiredSlots = (dml_uint_t)(dml_ceil(DesiredBPP / MaxLinkBPP * 64, 1)); - - if (DesiredBPP == 0) { - if (DSCEnable) { - if (MaxLinkBPP < MinDSCBPP) { - return __DML_DPP_INVALID__; - } else if (MaxLinkBPP >= MaxDSCBPP) { - return MaxDSCBPP; - } else { - return dml_floor(16.0 * MaxLinkBPP, 1.0) / 16.0; - } - } else { - if (MaxLinkBPP >= NonDSCBPP2) { - return NonDSCBPP2; - } else if (MaxLinkBPP >= NonDSCBPP1) { - return NonDSCBPP1; - } else if (MaxLinkBPP >= NonDSCBPP0) { - return NonDSCBPP0; - } else { - return __DML_DPP_INVALID__; - } - } - } else { - if (!((DSCEnable == false && (DesiredBPP == NonDSCBPP2 || DesiredBPP == NonDSCBPP1 || DesiredBPP == NonDSCBPP0)) || - (DSCEnable && DesiredBPP >= MinDSCBPP && DesiredBPP <= MaxDSCBPP))) { - return __DML_DPP_INVALID__; - } else { - return DesiredBPP; - } - } -} // TruncToValidBPP - -static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( - struct display_mode_lib_scratch_st *scratch, - struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *p) -{ - struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals_st *s = &scratch->CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals; - - s->TotalActiveWriteback = 0; - p->Watermark->UrgentWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency; - p->Watermark->USRRetrainingWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency + p->mmSOCParameters.USRRetrainingLatency + p->mmSOCParameters.SMNLatency; - p->Watermark->DRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->Watermark->UrgentWatermark; - p->Watermark->FCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->Watermark->UrgentWatermark; - p->Watermark->StutterExitWatermark = p->mmSOCParameters.SRExitTime + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; - p->Watermark->StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitTime + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; - p->Watermark->Z8StutterExitWatermark = p->mmSOCParameters.SRExitZ8Time + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; - p->Watermark->Z8StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitZ8Time + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); - dml_print("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); - dml_print("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); - dml_print("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); - dml_print("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); - dml_print("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); - dml_print("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); - dml_print("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); - dml_print("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); - dml_print("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); - dml_print("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); -#endif - - s->TotalActiveWriteback = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->WritebackEnable[k] == true) { - s->TotalActiveWriteback = s->TotalActiveWriteback + 1; - } - } - - if (s->TotalActiveWriteback <= 1) { - p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency; - } else { - p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; - } - if (p->USRRetrainingRequiredFinal) - p->Watermark->WritebackUrgentWatermark = p->Watermark->WritebackUrgentWatermark + p->mmSOCParameters.USRRetrainingLatency; - - if (s->TotalActiveWriteback <= 1) { - p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency; - p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency; - } else { - p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; - p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024 / 32 / p->SOCCLK; - } - - if (p->USRRetrainingRequiredFinal) - p->Watermark->WritebackDRAMClockChangeWatermark = p->Watermark->WritebackDRAMClockChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; - - if (p->USRRetrainingRequiredFinal) - p->Watermark->WritebackFCLKChangeWatermark = p->Watermark->WritebackFCLKChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); - dml_print("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); - dml_print("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); - dml_print("DML::%s: USRRetrainingRequiredFinal = %u\n", __func__, p->USRRetrainingRequiredFinal); - dml_print("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); -#endif - - s->TotalPixelBW = 0.0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - s->TotalPixelBW = s->TotalPixelBW + p->DPPPerSurface[k] - * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * p->VRatio[k] + p->SwathWidthC[k] * p->BytePerPixelDETC[k] * p->VRatioChroma[k]) / (p->HTotal[k] / p->PixelClock[k]); - } - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - - s->LBLatencyHidingSourceLinesY[k] = (dml_uint_t)(dml_min((dml_float_t)p->MaxLineBufferLines, dml_floor((dml_float_t)p->LineBufferSize / (dml_float_t)p->LBBitPerPixel[k] / ((dml_float_t)p->SwathWidthY[k] / dml_max(p->HRatio[k], 1.0)), 1)) - (p->VTaps[k] - 1)); - s->LBLatencyHidingSourceLinesC[k] = (dml_uint_t)(dml_min((dml_float_t)p->MaxLineBufferLines, dml_floor((dml_float_t)p->LineBufferSize / (dml_float_t)p->LBBitPerPixel[k] / ((dml_float_t)p->SwathWidthC[k] / dml_max(p->HRatioChroma[k], 1.0)), 1)) - (p->VTapsChroma[k] - 1)); - - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); - dml_print("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); - dml_print("DML::%s: k=%u, LBBitPerPixel = %u\n", __func__, k, p->LBBitPerPixel[k]); - dml_print("DML::%s: k=%u, HRatio = %f\n", __func__, k, p->HRatio[k]); - dml_print("DML::%s: k=%u, VTaps = %u\n", __func__, k, p->VTaps[k]); -#endif - - s->EffectiveLBLatencyHidingY = s->LBLatencyHidingSourceLinesY[k] / p->VRatio[k] * (p->HTotal[k] / p->PixelClock[k]); - s->EffectiveLBLatencyHidingC = s->LBLatencyHidingSourceLinesC[k] / p->VRatioChroma[k] * (p->HTotal[k] / p->PixelClock[k]); - - s->EffectiveDETBufferSizeY = p->DETBufferSizeY[k]; - if (p->UnboundedRequestEnabled) { - s->EffectiveDETBufferSizeY = s->EffectiveDETBufferSizeY + p->CompressedBufferSizeInkByte * 1024 * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * p->VRatio[k]) / (p->HTotal[k] / p->PixelClock[k]) / s->TotalPixelBW; - } - - s->LinesInDETY[k] = (dml_float_t)s->EffectiveDETBufferSizeY / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; - s->LinesInDETYRoundedDownToSwath[k] = (dml_uint_t)(dml_floor(s->LinesInDETY[k], p->SwathHeightY[k])); - s->FullDETBufferingTimeY = s->LinesInDETYRoundedDownToSwath[k] * (p->HTotal[k] / p->PixelClock[k]) / p->VRatio[k]; - - s->ActiveClockChangeLatencyHidingY = s->EffectiveLBLatencyHidingY + s->FullDETBufferingTimeY - ((dml_float_t)p->DSTXAfterScaler[k] / (dml_float_t)p->HTotal[k] + (dml_float_t)p->DSTYAfterScaler[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; - - if (p->NumberOfActiveSurfaces > 1) { - s->ActiveClockChangeLatencyHidingY = s->ActiveClockChangeLatencyHidingY - (1.0 - 1.0 / (dml_float_t)p->NumberOfActiveSurfaces) * (dml_float_t)p->SwathHeightY[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k] / p->VRatio[k]; - } - - if (p->BytePerPixelDETC[k] > 0) { - s->LinesInDETC[k] = p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k]; - s->LinesInDETCRoundedDownToSwath[k] = (dml_uint_t)(dml_floor(s->LinesInDETC[k], p->SwathHeightC[k])); - s->FullDETBufferingTimeC = s->LinesInDETCRoundedDownToSwath[k] * (p->HTotal[k] / p->PixelClock[k]) / p->VRatioChroma[k]; - s->ActiveClockChangeLatencyHidingC = s->EffectiveLBLatencyHidingC + s->FullDETBufferingTimeC - ((dml_float_t)p->DSTXAfterScaler[k] / (dml_float_t)p->HTotal[k] + (dml_float_t)p->DSTYAfterScaler[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; - if (p->NumberOfActiveSurfaces > 1) { - s->ActiveClockChangeLatencyHidingC = s->ActiveClockChangeLatencyHidingC - (1.0 - 1.0 / (dml_float_t)p->NumberOfActiveSurfaces) * (dml_float_t)p->SwathHeightC[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k] / p->VRatioChroma[k]; - } - s->ActiveClockChangeLatencyHiding = dml_min(s->ActiveClockChangeLatencyHidingY, s->ActiveClockChangeLatencyHidingC); - } else { - s->ActiveClockChangeLatencyHiding = s->ActiveClockChangeLatencyHidingY; - } - - s->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->UrgentWatermark - p->Watermark->DRAMClockChangeWatermark; - s->ActiveFCLKChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->UrgentWatermark - p->Watermark->FCLKChangeWatermark; - s->USRRetrainingLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->USRRetrainingWatermark; - - if (p->WritebackEnable[k]) { - s->WritebackLatencyHiding = (dml_float_t)p->WritebackInterfaceBufferSize * 1024.0 / ((dml_float_t)p->WritebackDestinationWidth[k] * (dml_float_t)p->WritebackDestinationHeight[k] / ((dml_float_t)p->WritebackSourceHeight[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k]) * 4.0); - if (p->WritebackPixelFormat[k] == dml_444_64) { - s->WritebackLatencyHiding = s->WritebackLatencyHiding / 2; - } - s->WritebackDRAMClockChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackDRAMClockChangeWatermark; - - s->WritebackFCLKChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackFCLKChangeWatermark; - - s->ActiveDRAMClockChangeLatencyMargin[k] = dml_min(s->ActiveDRAMClockChangeLatencyMargin[k], s->WritebackFCLKChangeLatencyMargin); - s->ActiveFCLKChangeLatencyMargin[k] = dml_min(s->ActiveFCLKChangeLatencyMargin[k], s->WritebackDRAMClockChangeLatencyMargin); - } - p->MaxActiveDRAMClockChangeLatencySupported[k] = (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) ? 0 : (s->ActiveDRAMClockChangeLatencyMargin[k] + p->mmSOCParameters.DRAMClockChangeLatency); - p->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveDRAMClockChangeLatencyMargin[k]; - } - - *p->USRRetrainingSupport = true; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && (s->USRRetrainingLatencyMargin[k] < 0)) { - *p->USRRetrainingSupport = false; - } - } - - s->FoundCriticalSurface = false; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && ((!s->FoundCriticalSurface) - || ((s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency) < *p->MaxActiveFCLKChangeLatencySupported))) { - s->FoundCriticalSurface = true; - *p->MaxActiveFCLKChangeLatencySupported = s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency; - } - } - - for (dml_uint_t i = 0; i < p->NumberOfActiveSurfaces; ++i) { - for (dml_uint_t j = 0; j < p->NumberOfActiveSurfaces; ++j) { - if (i == j || - (p->BlendingAndTiming[i] == i && p->BlendingAndTiming[j] == i) || - (p->BlendingAndTiming[j] == j && p->BlendingAndTiming[i] == j) || - (p->BlendingAndTiming[i] == p->BlendingAndTiming[j] && p->BlendingAndTiming[i] != i) || - (p->SynchronizeTimingsFinal && p->PixelClock[i] == p->PixelClock[j] && p->HTotal[i] == p->HTotal[j] && p->VTotal[i] == p->VTotal[j] && p->VActive[i] == p->VActive[j]) || - (p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && (p->DRRDisplay[i] || p->DRRDisplay[j]))) { - s->SynchronizedSurfaces[i][j] = true; - } else { - s->SynchronizedSurfaces[i][j] = false; - } - } - } - - s->FCLKChangeSupportNumber = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && (s->ActiveFCLKChangeLatencyMargin[k] < 0)) { - if (!(p->PrefetchMode[k] <= 1)) { - s->FCLKChangeSupportNumber = 3; - } else if (s->FCLKChangeSupportNumber == 0) { - s->FCLKChangeSupportNumber = ((p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && p->DRRDisplay[k]) ? 2 : 1); - s->LastSurfaceWithoutMargin = k; - } else if (((s->FCLKChangeSupportNumber == 1) && (p->DRRDisplay[k] || (!s->SynchronizedSurfaces[s->LastSurfaceWithoutMargin][k]))) || (s->FCLKChangeSupportNumber == 2)) - s->FCLKChangeSupportNumber = 3; - } - } - - if (s->FCLKChangeSupportNumber == 0) { - *p->FCLKChangeSupport = dml_fclock_change_vactive; - } else if ((s->FCLKChangeSupportNumber == 1) || (s->FCLKChangeSupportNumber == 2)) { - *p->FCLKChangeSupport = dml_fclock_change_vblank; - } else { - *p->FCLKChangeSupport = dml_fclock_change_unsupported; - } - - s->DRAMClockChangeMethod = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame) - s->DRAMClockChangeMethod = 1; - else if (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) - s->DRAMClockChangeMethod = 2; - } - - s->DRAMClockChangeSupportNumber = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (((s->DRAMClockChangeMethod == 0) && (s->ActiveDRAMClockChangeLatencyMargin[k] < 0)) || - ((s->DRAMClockChangeMethod == 1) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_full_frame)) || - ((s->DRAMClockChangeMethod == 2) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_sub_viewport) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe))) { - if (p->PrefetchMode[k] != 0) { // Don't need to support DRAM clock change, PrefetchMode 0 means needs DRAM clock change support - s->DRAMClockChangeSupportNumber = 3; - } else if (s->DRAMClockChangeSupportNumber == 0) { - s->DRAMClockChangeSupportNumber = (p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && p->DRRDisplay[k]) ? 2 : 1; - s->LastSurfaceWithoutMargin = k; - } else if (((s->DRAMClockChangeSupportNumber == 1) && (p->DRRDisplay[k] || !s->SynchronizedSurfaces[s->LastSurfaceWithoutMargin][k])) || (s->DRAMClockChangeSupportNumber == 2)) { - s->DRAMClockChangeSupportNumber = 3; - } - } - } - - if (s->DRAMClockChangeMethod == 0) { // No MALL usage - if (s->DRAMClockChangeSupportNumber == 0) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive; - } else if (s->DRAMClockChangeSupportNumber == 1) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank; - } else if (s->DRAMClockChangeSupportNumber == 2) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr; - } else { - *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; - } - } else if (s->DRAMClockChangeMethod == 1) { // Any pipe using MALL full frame - if (s->DRAMClockChangeSupportNumber == 0) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive_w_mall_full_frame; - } else if (s->DRAMClockChangeSupportNumber == 1) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_w_mall_full_frame; - } else if (s->DRAMClockChangeSupportNumber == 2) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr_w_mall_full_frame; - } else { - *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; - } - } else { // Any pipe using MALL subviewport - if (s->DRAMClockChangeSupportNumber == 0) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive_w_mall_sub_vp; - } else if (s->DRAMClockChangeSupportNumber == 1) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_w_mall_sub_vp; - } else if (s->DRAMClockChangeSupportNumber == 2) { - *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr_w_mall_sub_vp; - } else { - *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; - } - } - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - s->dst_y_pstate = (dml_uint_t)(dml_ceil((p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.UrgentLatency) / (p->HTotal[k] / p->PixelClock[k]), 1)); - s->src_y_pstate_l = (dml_uint_t)(dml_ceil(s->dst_y_pstate * p->VRatio[k], p->SwathHeightY[k])); - s->src_y_ahead_l = (dml_uint_t)(dml_floor(p->DETBufferSizeY[k] / p->BytePerPixelDETY[k] / p->SwathWidthY[k], p->SwathHeightY[k]) + s->LBLatencyHidingSourceLinesY[k]); - s->sub_vp_lines_l = s->src_y_pstate_l + s->src_y_ahead_l + p->meta_row_height[k]; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - dml_print("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - dml_print("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - dml_print("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - dml_print("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); - dml_print("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); - dml_print("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); - dml_print("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); - dml_print("DML::%s: k=%u, meta_row_height = %u\n", __func__, k, p->meta_row_height[k]); - dml_print("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); -#endif - p->SubViewportLinesNeededInMALL[k] = s->sub_vp_lines_l; - - if (p->BytePerPixelDETC[k] > 0) { - s->src_y_pstate_c = (dml_uint_t)(dml_ceil(s->dst_y_pstate * p->VRatioChroma[k], p->SwathHeightC[k])); - s->src_y_ahead_c = (dml_uint_t)(dml_floor(p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k], p->SwathHeightC[k]) + s->LBLatencyHidingSourceLinesC[k]); - s->sub_vp_lines_c = s->src_y_pstate_c + s->src_y_ahead_c + p->meta_row_height_chroma[k]; - p->SubViewportLinesNeededInMALL[k] = (dml_uint_t)(dml_max(s->sub_vp_lines_l, s->sub_vp_lines_c)); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); - dml_print("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); - dml_print("DML::%s: k=%u, meta_row_height_chroma = %u\n", __func__, k, p->meta_row_height_chroma[k]); - dml_print("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); -#endif - } - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->DRAMClockChangeSupport); - dml_print("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->FCLKChangeSupport); - dml_print("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); - dml_print("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); -#endif -} // CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport - -static void CalculateDCFCLKDeepSleep( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t BytePerPixelY[], - dml_uint_t BytePerPixelC[], - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_uint_t SwathWidthY[], - dml_uint_t SwathWidthC[], - dml_uint_t DPPPerSurface[], - dml_float_t HRatio[], - dml_float_t HRatioChroma[], - dml_float_t PixelClock[], - dml_float_t PSCL_THROUGHPUT[], - dml_float_t PSCL_THROUGHPUT_CHROMA[], - dml_float_t Dppclk[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_uint_t ReturnBusWidth, - - // Output - dml_float_t *DCFClkDeepSleep) -{ - dml_float_t DisplayPipeLineDeliveryTimeLuma; - dml_float_t DisplayPipeLineDeliveryTimeChroma; - dml_float_t DCFClkDeepSleepPerSurface[__DML_NUM_PLANES__]; - dml_float_t ReadBandwidth = 0.0; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - - if (VRatio[k] <= 1) { - DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChroma = 0; - } else { - if (VRatioChroma[k] <= 1) { - DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } - - if (BytePerPixelC[k] > 0) { - DCFClkDeepSleepPerSurface[k] = dml_max(__DML_MIN_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 32.0 / DisplayPipeLineDeliveryTimeLuma, - __DML_MIN_DCFCLK_FACTOR__ * SwathWidthC[k] * BytePerPixelC[k] / 32.0 / DisplayPipeLineDeliveryTimeChroma); - } else { - DCFClkDeepSleepPerSurface[k] = __DML_MIN_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 64.0 / DisplayPipeLineDeliveryTimeLuma; - } - DCFClkDeepSleepPerSurface[k] = dml_max(DCFClkDeepSleepPerSurface[k], PixelClock[k] / 16); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, PixelClock = %f\n", __func__, k, PixelClock[k]); - dml_print("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); -#endif - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - ReadBandwidth = ReadBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; - } - - *DCFClkDeepSleep = dml_max(8.0, __DML_MIN_DCFCLK_FACTOR__ * ReadBandwidth / (dml_float_t) ReturnBusWidth); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: __DML_MIN_DCFCLK_FACTOR__ = %f\n", __func__, __DML_MIN_DCFCLK_FACTOR__); - dml_print("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); - dml_print("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); - dml_print("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); -#endif - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - *DCFClkDeepSleep = dml_max(*DCFClkDeepSleep, DCFClkDeepSleepPerSurface[k]); - } - dml_print("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); -} // CalculateDCFCLKDeepSleep - -static void CalculateUrgentBurstFactor( - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, - dml_uint_t swath_width_luma_ub, - dml_uint_t swath_width_chroma_ub, - dml_uint_t SwathHeightY, - dml_uint_t SwathHeightC, - dml_float_t LineTime, - dml_float_t UrgentLatency, - dml_float_t CursorBufferSize, - dml_uint_t CursorWidth, - dml_uint_t CursorBPP, - dml_float_t VRatio, - dml_float_t VRatioC, - dml_float_t BytePerPixelInDETY, - dml_float_t BytePerPixelInDETC, - dml_uint_t DETBufferSizeY, - dml_uint_t DETBufferSizeC, - // Output - dml_float_t *UrgentBurstFactorCursor, - dml_float_t *UrgentBurstFactorLuma, - dml_float_t *UrgentBurstFactorChroma, - dml_bool_t *NotEnoughUrgentLatencyHiding) -{ - dml_float_t LinesInDETLuma; - dml_float_t LinesInDETChroma; - dml_uint_t LinesInCursorBuffer; - dml_float_t CursorBufferSizeInTime; - dml_float_t DETBufferSizeInTimeLuma; - dml_float_t DETBufferSizeInTimeChroma; - - *NotEnoughUrgentLatencyHiding = 0; - - if (CursorWidth > 0) { - LinesInCursorBuffer = 1 << (dml_uint_t) dml_floor(dml_log2(CursorBufferSize * 1024.0 / (CursorWidth * CursorBPP / 8.0)), 1.0); - if (VRatio > 0) { - CursorBufferSizeInTime = LinesInCursorBuffer * LineTime / VRatio; - if (CursorBufferSizeInTime - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorCursor = 0; - } else { - *UrgentBurstFactorCursor = CursorBufferSizeInTime / (CursorBufferSizeInTime - UrgentLatency); - } - } else { - *UrgentBurstFactorCursor = 1; - } - } - - LinesInDETLuma = (UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe ? 1024*1024 : DETBufferSizeY) / BytePerPixelInDETY / swath_width_luma_ub; - - if (VRatio > 0) { - DETBufferSizeInTimeLuma = dml_floor(LinesInDETLuma, SwathHeightY) * LineTime / VRatio; - if (DETBufferSizeInTimeLuma - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorLuma = 0; - } else { - *UrgentBurstFactorLuma = DETBufferSizeInTimeLuma / (DETBufferSizeInTimeLuma - UrgentLatency); - } - } else { - *UrgentBurstFactorLuma = 1; - } - - if (BytePerPixelInDETC > 0) { - LinesInDETChroma = (UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe ? 1024*1024 : DETBufferSizeC) / BytePerPixelInDETC / swath_width_chroma_ub; - - if (VRatioC > 0) { - DETBufferSizeInTimeChroma = dml_floor(LinesInDETChroma, SwathHeightC) * LineTime / VRatioC; - if (DETBufferSizeInTimeChroma - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorChroma = 0; - } else { - *UrgentBurstFactorChroma = DETBufferSizeInTimeChroma / (DETBufferSizeInTimeChroma - UrgentLatency); - } - } else { - *UrgentBurstFactorChroma = 1; - } - } -} // CalculateUrgentBurstFactor - -static void CalculatePixelDeliveryTimes( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_float_t VRatioPrefetchY[], - dml_float_t VRatioPrefetchC[], - dml_uint_t swath_width_luma_ub[], - dml_uint_t swath_width_chroma_ub[], - dml_uint_t DPPPerSurface[], - dml_float_t HRatio[], - dml_float_t HRatioChroma[], - dml_float_t PixelClock[], - dml_float_t PSCL_THROUGHPUT[], - dml_float_t PSCL_THROUGHPUT_CHROMA[], - dml_float_t Dppclk[], - dml_uint_t BytePerPixelC[], - enum dml_rotation_angle SourceScan[], - dml_uint_t NumberOfCursors[], - dml_uint_t CursorWidth[], - dml_uint_t CursorBPP[], - dml_uint_t BlockWidth256BytesY[], - dml_uint_t BlockHeight256BytesY[], - dml_uint_t BlockWidth256BytesC[], - dml_uint_t BlockHeight256BytesC[], - - // Output - dml_float_t DisplayPipeLineDeliveryTimeLuma[], - dml_float_t DisplayPipeLineDeliveryTimeChroma[], - dml_float_t DisplayPipeLineDeliveryTimeLumaPrefetch[], - dml_float_t DisplayPipeLineDeliveryTimeChromaPrefetch[], - dml_float_t DisplayPipeRequestDeliveryTimeLuma[], - dml_float_t DisplayPipeRequestDeliveryTimeChroma[], - dml_float_t DisplayPipeRequestDeliveryTimeLumaPrefetch[], - dml_float_t DisplayPipeRequestDeliveryTimeChromaPrefetch[], - dml_float_t CursorRequestDeliveryTime[], - dml_float_t CursorRequestDeliveryTimePrefetch[]) -{ - dml_float_t req_per_swath_ub; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : HRatio = %f\n", __func__, k, HRatio[k]); - dml_print("DML::%s: k=%u : VRatio = %f\n", __func__, k, VRatio[k]); - dml_print("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, HRatioChroma[k]); - dml_print("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, VRatioChroma[k]); - dml_print("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); - dml_print("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); - dml_print("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); - dml_print("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); - dml_print("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); - dml_print("DML::%s: k=%u : PixelClock = %f\n", __func__, k, PixelClock[k]); - dml_print("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); -#endif - - if (VRatio[k] <= 1) { - DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChroma[k] = 0; - } else { - if (VRatioChroma[k] <= 1) { - DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } - - if (VRatioPrefetchY[k] <= 1) { - DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0; - } else { - if (VRatioPrefetchC[k] <= 1) { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; - } else { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); - dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); - dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); - dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); -#endif - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_vertical_rotation(SourceScan[k])) { - req_per_swath_ub = swath_width_luma_ub[k] / BlockWidth256BytesY[k]; - } else { - req_per_swath_ub = swath_width_luma_ub[k] / BlockHeight256BytesY[k]; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : req_per_swath_ub = %f (Luma)\n", __func__, k, req_per_swath_ub); -#endif - - DisplayPipeRequestDeliveryTimeLuma[k] = DisplayPipeLineDeliveryTimeLuma[k] / req_per_swath_ub; - DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = DisplayPipeLineDeliveryTimeLumaPrefetch[k] / req_per_swath_ub; - if (BytePerPixelC[k] == 0) { - DisplayPipeRequestDeliveryTimeChroma[k] = 0; - DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = 0; - } else { - if (!dml_is_vertical_rotation(SourceScan[k])) { - req_per_swath_ub = swath_width_chroma_ub[k] / BlockWidth256BytesC[k]; - } else { - req_per_swath_ub = swath_width_chroma_ub[k] / BlockHeight256BytesC[k]; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : req_per_swath_ub = %f (Chroma)\n", __func__, k, req_per_swath_ub); -#endif - DisplayPipeRequestDeliveryTimeChroma[k] = DisplayPipeLineDeliveryTimeChroma[k] / req_per_swath_ub; - DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); - dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); - dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); - dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); -#endif - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - dml_uint_t cursor_req_per_width; - cursor_req_per_width = (dml_uint_t)(dml_ceil((dml_float_t) CursorWidth[k] * (dml_float_t) CursorBPP[k] / 256.0 / 8.0, 1.0)); - if (NumberOfCursors[k] > 0) { - if (VRatio[k] <= 1) { - CursorRequestDeliveryTime[k] = (dml_float_t) CursorWidth[k] / HRatio[k] / PixelClock[k] / cursor_req_per_width; - } else { - CursorRequestDeliveryTime[k] = (dml_float_t) CursorWidth[k] / PSCL_THROUGHPUT[k] / Dppclk[k] / cursor_req_per_width; - } - if (VRatioPrefetchY[k] <= 1) { - CursorRequestDeliveryTimePrefetch[k] = (dml_float_t) CursorWidth[k] / HRatio[k] / PixelClock[k] / cursor_req_per_width; - } else { - CursorRequestDeliveryTimePrefetch[k] = (dml_float_t) CursorWidth[k] / PSCL_THROUGHPUT[k] / Dppclk[k] / cursor_req_per_width; - } - } else { - CursorRequestDeliveryTime[k] = 0; - CursorRequestDeliveryTimePrefetch[k] = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u : NumberOfCursors = %u\n", __func__, k, NumberOfCursors[k]); - dml_print("DML::%s: k=%u : CursorRequestDeliveryTime = %f\n", __func__, k, CursorRequestDeliveryTime[k]); - dml_print("DML::%s: k=%u : CursorRequestDeliveryTimePrefetch = %f\n", __func__, k, CursorRequestDeliveryTimePrefetch[k]); -#endif - } -} // CalculatePixelDeliveryTimes - -static void CalculateMetaAndPTETimes( - dml_bool_t use_one_row_for_frame[], - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t GPUVMEnable, - dml_uint_t MetaChunkSize, - dml_uint_t MinMetaChunkSizeBytes, - dml_uint_t HTotal[], - dml_float_t VRatio[], - dml_float_t VRatioChroma[], - dml_float_t DestinationLinesToRequestRowInVBlank[], - dml_float_t DestinationLinesToRequestRowInImmediateFlip[], - dml_bool_t DCCEnable[], - dml_float_t PixelClock[], - dml_uint_t BytePerPixelY[], - dml_uint_t BytePerPixelC[], - enum dml_rotation_angle SourceScan[], - dml_uint_t dpte_row_height[], - dml_uint_t dpte_row_height_chroma[], - dml_uint_t meta_row_width[], - dml_uint_t meta_row_width_chroma[], - dml_uint_t meta_row_height[], - dml_uint_t meta_row_height_chroma[], - dml_uint_t meta_req_width[], - dml_uint_t meta_req_width_chroma[], - dml_uint_t meta_req_height[], - dml_uint_t meta_req_height_chroma[], - dml_uint_t dpte_group_bytes[], - dml_uint_t PTERequestSizeY[], - dml_uint_t PTERequestSizeC[], - dml_uint_t PixelPTEReqWidthY[], - dml_uint_t PixelPTEReqHeightY[], - dml_uint_t PixelPTEReqWidthC[], - dml_uint_t PixelPTEReqHeightC[], - dml_uint_t dpte_row_width_luma_ub[], - dml_uint_t dpte_row_width_chroma_ub[], - - // Output - dml_float_t DST_Y_PER_PTE_ROW_NOM_L[], - dml_float_t DST_Y_PER_PTE_ROW_NOM_C[], - dml_float_t DST_Y_PER_META_ROW_NOM_L[], - dml_float_t DST_Y_PER_META_ROW_NOM_C[], - dml_float_t TimePerMetaChunkNominal[], - dml_float_t TimePerChromaMetaChunkNominal[], - dml_float_t TimePerMetaChunkVBlank[], - dml_float_t TimePerChromaMetaChunkVBlank[], - dml_float_t TimePerMetaChunkFlip[], - dml_float_t TimePerChromaMetaChunkFlip[], - dml_float_t time_per_pte_group_nom_luma[], - dml_float_t time_per_pte_group_vblank_luma[], - dml_float_t time_per_pte_group_flip_luma[], - dml_float_t time_per_pte_group_nom_chroma[], - dml_float_t time_per_pte_group_vblank_chroma[], - dml_float_t time_per_pte_group_flip_chroma[]) -{ - dml_uint_t meta_chunk_width; - dml_uint_t min_meta_chunk_width; - dml_uint_t meta_chunk_per_row_int; - dml_uint_t meta_row_remainder; - dml_uint_t meta_chunk_threshold; - dml_uint_t meta_chunks_per_row_ub; - dml_uint_t meta_chunk_width_chroma; - dml_uint_t min_meta_chunk_width_chroma; - dml_uint_t meta_chunk_per_row_int_chroma; - dml_uint_t meta_row_remainder_chroma; - dml_uint_t meta_chunk_threshold_chroma; - dml_uint_t meta_chunks_per_row_ub_chroma; - dml_uint_t dpte_group_width_luma; - dml_uint_t dpte_groups_per_row_luma_ub; - dml_uint_t dpte_group_width_chroma; - dml_uint_t dpte_groups_per_row_chroma_ub; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - DST_Y_PER_PTE_ROW_NOM_L[k] = dpte_row_height[k] / VRatio[k]; - if (BytePerPixelC[k] == 0) { - DST_Y_PER_PTE_ROW_NOM_C[k] = 0; - } else { - DST_Y_PER_PTE_ROW_NOM_C[k] = dpte_row_height_chroma[k] / VRatioChroma[k]; - } - DST_Y_PER_META_ROW_NOM_L[k] = meta_row_height[k] / VRatio[k]; - if (BytePerPixelC[k] == 0) { - DST_Y_PER_META_ROW_NOM_C[k] = 0; - } else { - DST_Y_PER_META_ROW_NOM_C[k] = meta_row_height_chroma[k] / VRatioChroma[k]; - } - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (DCCEnable[k] == true) { - meta_chunk_width = MetaChunkSize * 1024 * 256 / BytePerPixelY[k] / meta_row_height[k]; - min_meta_chunk_width = MinMetaChunkSizeBytes * 256 / BytePerPixelY[k] / meta_row_height[k]; - meta_chunk_per_row_int = meta_row_width[k] / meta_chunk_width; - meta_row_remainder = meta_row_width[k] % meta_chunk_width; - if (!dml_is_vertical_rotation(SourceScan[k])) { - meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width[k]; - } else { - meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height[k]; - } - if (meta_row_remainder <= meta_chunk_threshold) { - meta_chunks_per_row_ub = meta_chunk_per_row_int + 1; - } else { - meta_chunks_per_row_ub = meta_chunk_per_row_int + 2; - } - TimePerMetaChunkNominal[k] = meta_row_height[k] / VRatio[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; - TimePerMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; - TimePerMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; - if (BytePerPixelC[k] == 0) { - TimePerChromaMetaChunkNominal[k] = 0; - TimePerChromaMetaChunkVBlank[k] = 0; - TimePerChromaMetaChunkFlip[k] = 0; - } else { - meta_chunk_width_chroma = MetaChunkSize * 1024 * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; - min_meta_chunk_width_chroma = MinMetaChunkSizeBytes * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; - meta_chunk_per_row_int_chroma = (dml_uint_t)((dml_float_t) meta_row_width_chroma[k] / meta_chunk_width_chroma); - meta_row_remainder_chroma = meta_row_width_chroma[k] % meta_chunk_width_chroma; - if (!dml_is_vertical_rotation(SourceScan[k])) { - meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_width_chroma[k]; - } else { - meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_height_chroma[k]; - } - if (meta_row_remainder_chroma <= meta_chunk_threshold_chroma) { - meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 1; - } else { - meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 2; - } - TimePerChromaMetaChunkNominal[k] = meta_row_height_chroma[k] / VRatioChroma[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; - TimePerChromaMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; - TimePerChromaMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; - } - } else { - TimePerMetaChunkNominal[k] = 0; - TimePerMetaChunkVBlank[k] = 0; - TimePerMetaChunkFlip[k] = 0; - TimePerChromaMetaChunkNominal[k] = 0; - TimePerChromaMetaChunkVBlank[k] = 0; - TimePerChromaMetaChunkFlip[k] = 0; - } - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (GPUVMEnable == true) { - if (!dml_is_vertical_rotation(SourceScan[k])) { - dpte_group_width_luma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeY[k] * PixelPTEReqWidthY[k]); - } else { - dpte_group_width_luma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeY[k] * PixelPTEReqHeightY[k]); - } - - if (use_one_row_for_frame[k]) { - dpte_groups_per_row_luma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_luma_ub[k] / (dml_float_t) dpte_group_width_luma / 2.0, 1.0)); - } else { - dpte_groups_per_row_luma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_luma_ub[k] / (dml_float_t) dpte_group_width_luma, 1.0)); - } - - dml_print("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, use_one_row_for_frame[k]); - dml_print("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, dpte_group_bytes[k]); - dml_print("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, PTERequestSizeY[k]); - dml_print("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, PixelPTEReqWidthY[k]); - dml_print("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, PixelPTEReqHeightY[k]); - dml_print("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, dpte_row_width_luma_ub[k]); - dml_print("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); - dml_print("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); - - time_per_pte_group_nom_luma[k] = DST_Y_PER_PTE_ROW_NOM_L[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; - time_per_pte_group_vblank_luma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; - time_per_pte_group_flip_luma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; - if (BytePerPixelC[k] == 0) { - time_per_pte_group_nom_chroma[k] = 0; - time_per_pte_group_vblank_chroma[k] = 0; - time_per_pte_group_flip_chroma[k] = 0; - } else { - if (!dml_is_vertical_rotation(SourceScan[k])) { - dpte_group_width_chroma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeC[k] * PixelPTEReqWidthC[k]); - } else { - dpte_group_width_chroma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeC[k] * PixelPTEReqHeightC[k]); - } - - if (use_one_row_for_frame[k]) { - dpte_groups_per_row_chroma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_chroma_ub[k] / (dml_float_t) dpte_group_width_chroma / 2.0, 1.0)); - } else { - dpte_groups_per_row_chroma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_chroma_ub[k] / (dml_float_t) dpte_group_width_chroma, 1.0)); - } - dml_print("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, dpte_row_width_chroma_ub[k]); - dml_print("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); - dml_print("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); - - time_per_pte_group_nom_chroma[k] = DST_Y_PER_PTE_ROW_NOM_C[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; - time_per_pte_group_vblank_chroma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; - time_per_pte_group_flip_chroma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; - } - } else { - time_per_pte_group_nom_luma[k] = 0; - time_per_pte_group_vblank_luma[k] = 0; - time_per_pte_group_flip_luma[k] = 0; - time_per_pte_group_nom_chroma[k] = 0; - time_per_pte_group_vblank_chroma[k] = 0; - time_per_pte_group_flip_chroma[k] = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, DestinationLinesToRequestRowInVBlank = %f\n", __func__, k, DestinationLinesToRequestRowInVBlank[k]); - dml_print("DML::%s: k=%u, DestinationLinesToRequestRowInImmediateFlip = %f\n", __func__, k, DestinationLinesToRequestRowInImmediateFlip[k]); - - dml_print("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, DST_Y_PER_PTE_ROW_NOM_L[k]); - dml_print("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, DST_Y_PER_PTE_ROW_NOM_C[k]); - dml_print("DML::%s: k=%u, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, DST_Y_PER_META_ROW_NOM_L[k]); - dml_print("DML::%s: k=%u, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, DST_Y_PER_META_ROW_NOM_C[k]); - dml_print("DML::%s: k=%u, TimePerMetaChunkNominal = %f\n", __func__, k, TimePerMetaChunkNominal[k]); - dml_print("DML::%s: k=%u, TimePerMetaChunkVBlank = %f\n", __func__, k, TimePerMetaChunkVBlank[k]); - dml_print("DML::%s: k=%u, TimePerMetaChunkFlip = %f\n", __func__, k, TimePerMetaChunkFlip[k]); - dml_print("DML::%s: k=%u, TimePerChromaMetaChunkNominal = %f\n", __func__, k, TimePerChromaMetaChunkNominal[k]); - dml_print("DML::%s: k=%u, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, TimePerChromaMetaChunkVBlank[k]); - dml_print("DML::%s: k=%u, TimePerChromaMetaChunkFlip = %f\n", __func__, k, TimePerChromaMetaChunkFlip[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, time_per_pte_group_nom_luma[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, time_per_pte_group_vblank_luma[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, time_per_pte_group_flip_luma[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, time_per_pte_group_nom_chroma[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, time_per_pte_group_vblank_chroma[k]); - dml_print("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, time_per_pte_group_flip_chroma[k]); -#endif - } -} // CalculateMetaAndPTETimes - -static void CalculateVMGroupAndRequestTimes( - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t GPUVMEnable, - dml_uint_t GPUVMMaxPageTableLevels, - dml_uint_t HTotal[], - dml_uint_t BytePerPixelC[], - dml_float_t DestinationLinesToRequestVMInVBlank[], - dml_float_t DestinationLinesToRequestVMInImmediateFlip[], - dml_bool_t DCCEnable[], - dml_float_t PixelClock[], - dml_uint_t dpte_row_width_luma_ub[], - dml_uint_t dpte_row_width_chroma_ub[], - dml_uint_t vm_group_bytes[], - dml_uint_t dpde0_bytes_per_frame_ub_l[], - dml_uint_t dpde0_bytes_per_frame_ub_c[], - dml_uint_t meta_pte_bytes_per_frame_ub_l[], - dml_uint_t meta_pte_bytes_per_frame_ub_c[], - - // Output - dml_float_t TimePerVMGroupVBlank[], - dml_float_t TimePerVMGroupFlip[], - dml_float_t TimePerVMRequestVBlank[], - dml_float_t TimePerVMRequestFlip[]) -{ - dml_uint_t num_group_per_lower_vm_stage; - dml_uint_t num_req_per_lower_vm_stage; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); - dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); -#endif - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, DCCEnable = %u\n", __func__, k, DCCEnable[k]); - dml_print("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); - dml_print("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); - dml_print("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); - dml_print("DML::%s: k=%u, meta_pte_bytes_per_frame_ub_l = %u\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); - dml_print("DML::%s: k=%u, meta_pte_bytes_per_frame_ub_c = %u\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); -#endif - - if (GPUVMEnable == true && (DCCEnable[k] == true || GPUVMMaxPageTableLevels > 1)) { - if (DCCEnable[k] == false) { - if (BytePerPixelC[k] > 0) { - num_group_per_lower_vm_stage = (dml_uint_t) (dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_l[k] / (dml_float_t) vm_group_bytes[k], 1.0) + - dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_c[k] / (dml_float_t) vm_group_bytes[k], 1.0)); - } else { - num_group_per_lower_vm_stage = (dml_uint_t) (dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_l[k] / (dml_float_t) vm_group_bytes[k], 1.0)); - } - } else { - if (GPUVMMaxPageTableLevels == 1) { - if (BytePerPixelC[k] > 0) { - num_group_per_lower_vm_stage = (dml_uint_t)(dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0) + - dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0)); - } else { - num_group_per_lower_vm_stage = (dml_uint_t)(dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0)); - } - } else { - if (BytePerPixelC[k] > 0) { - num_group_per_lower_vm_stage = (dml_uint_t)(2.0 + dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + - dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + - dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + - dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1)); - } else { - num_group_per_lower_vm_stage = (dml_uint_t)(1.0 + dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + - dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1)); - } - } - } - - if (DCCEnable[k] == false) { - if (BytePerPixelC[k] > 0) { - num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + dpde0_bytes_per_frame_ub_c[k] / 64; - } else { - num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64; - } - } else { - if (GPUVMMaxPageTableLevels == 1) { - if (BytePerPixelC[k] > 0) { - num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_c[k] / 64; - } else { - num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64; - } - } else { - if (BytePerPixelC[k] > 0) { - num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + dpde0_bytes_per_frame_ub_c[k] / 64 + meta_pte_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_c[k] / 64; - } else { - num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_l[k] / 64; - } - } - } - - TimePerVMGroupVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] / num_group_per_lower_vm_stage; - TimePerVMGroupFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] / num_group_per_lower_vm_stage; - TimePerVMRequestVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] / num_req_per_lower_vm_stage; - TimePerVMRequestFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] / num_req_per_lower_vm_stage; - - if (GPUVMMaxPageTableLevels > 2) { - TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; - TimePerVMGroupFlip[k] = TimePerVMGroupFlip[k] / 2; - TimePerVMRequestVBlank[k] = TimePerVMRequestVBlank[k] / 2; - TimePerVMRequestFlip[k] = TimePerVMRequestFlip[k] / 2; - } - - } else { - TimePerVMGroupVBlank[k] = 0; - TimePerVMGroupFlip[k] = 0; - TimePerVMRequestVBlank[k] = 0; - TimePerVMRequestFlip[k] = 0; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); - dml_print("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); - dml_print("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); - dml_print("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); -#endif - } -} // CalculateVMGroupAndRequestTimes - -static void CalculateStutterEfficiency(struct display_mode_lib_scratch_st *scratch, - struct CalculateStutterEfficiency_params_st *p) -{ - dml_float_t DETBufferingTimeY = 0; - dml_float_t SwathWidthYCriticalSurface = 0; - dml_float_t SwathHeightYCriticalSurface = 0; - dml_float_t VActiveTimeCriticalSurface = 0; - dml_float_t FrameTimeCriticalSurface = 0; - dml_uint_t BytePerPixelYCriticalSurface = 0; - dml_float_t LinesToFinishSwathTransferStutterCriticalSurface = 0; - dml_uint_t DETBufferSizeYCriticalSurface = 0; - dml_float_t MinTTUVBlankCriticalSurface = 0; - dml_uint_t BlockWidth256BytesYCriticalSurface = 0; - dml_bool_t SinglePlaneCriticalSurface = 0; - dml_bool_t SinglePipeCriticalSurface = 0; - dml_float_t TotalCompressedReadBandwidth = 0; - dml_float_t TotalRowReadBandwidth = 0; - dml_float_t AverageDCCCompressionRate = 0; - dml_float_t EffectiveCompressedBufferSize = 0; - dml_float_t PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = 0; - dml_float_t StutterBurstTime = 0; - dml_uint_t TotalActiveWriteback = 0; - dml_float_t LinesInDETY = 0; - dml_float_t LinesInDETYRoundedDownToSwath = 0; - dml_float_t MaximumEffectiveCompressionLuma = 0; - dml_float_t MaximumEffectiveCompressionChroma = 0; - dml_float_t TotalZeroSizeRequestReadBandwidth = 0; - dml_float_t TotalZeroSizeCompressedReadBandwidth = 0; - dml_float_t AverageDCCZeroSizeFraction = 0; - dml_float_t AverageZeroSizeCompressionRate = 0; - - dml_bool_t FoundCriticalSurface = false; - - dml_uint_t TotalNumberOfActiveOTG = 0; - dml_float_t SinglePixelClock = 0; - dml_uint_t SingleHTotal = 0; - dml_uint_t SingleVTotal = 0; - dml_bool_t SameTiming = true; - - dml_float_t LastStutterPeriod = 0.0; - dml_float_t LastZ8StutterPeriod = 0.0; - - dml_uint_t SwathSizeCriticalSurface; - dml_uint_t LastChunkOfSwathSize; - dml_uint_t MissingPartOfLastSwathOfDETSize; - - TotalZeroSizeRequestReadBandwidth = 0; - TotalZeroSizeCompressedReadBandwidth = 0; - TotalRowReadBandwidth = 0; - TotalCompressedReadBandwidth = 0; - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { - if (p->DCCEnable[k] == true) { - if ((dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockWidth256BytesY[k] > p->SwathHeightY[k]) || (!dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockHeight256BytesY[k] > p->SwathHeightY[k]) || p->DCCYMaxUncompressedBlock[k] < 256) { - MaximumEffectiveCompressionLuma = 2; - } else { - MaximumEffectiveCompressionLuma = 4; - } - TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] / dml_min(p->NetDCCRateLuma[k], MaximumEffectiveCompressionLuma); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - dml_print("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->NetDCCRateLuma[k]); - dml_print("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, MaximumEffectiveCompressionLuma); -#endif - TotalZeroSizeRequestReadBandwidth = TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->DCCFractionOfZeroSizeRequestsLuma[k]; - TotalZeroSizeCompressedReadBandwidth = TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->DCCFractionOfZeroSizeRequestsLuma[k] / MaximumEffectiveCompressionLuma; - - if (p->ReadBandwidthSurfaceChroma[k] > 0) { - if ((dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockWidth256BytesC[k] > p->SwathHeightC[k]) || (!dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockHeight256BytesC[k] > p->SwathHeightC[k]) || p->DCCCMaxUncompressedBlock[k] < 256) { - MaximumEffectiveCompressionChroma = 2; - } else { - MaximumEffectiveCompressionChroma = 4; - } - TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] / dml_min(p->NetDCCRateChroma[k], MaximumEffectiveCompressionChroma); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); - dml_print("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->NetDCCRateChroma[k]); - dml_print("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, MaximumEffectiveCompressionChroma); -#endif - TotalZeroSizeRequestReadBandwidth = TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->DCCFractionOfZeroSizeRequestsChroma[k]; - TotalZeroSizeCompressedReadBandwidth = TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->DCCFractionOfZeroSizeRequestsChroma[k] / MaximumEffectiveCompressionChroma; - } - } else { - TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] + p->ReadBandwidthSurfaceChroma[k]; - } - TotalRowReadBandwidth = TotalRowReadBandwidth + p->DPPPerSurface[k] * (p->meta_row_bw[k] + p->dpte_row_bw[k]); - } - } - - AverageDCCCompressionRate = p->TotalDataReadBandwidth / TotalCompressedReadBandwidth; - AverageDCCZeroSizeFraction = TotalZeroSizeRequestReadBandwidth / p->TotalDataReadBandwidth; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); - dml_print("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, TotalCompressedReadBandwidth); - dml_print("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, TotalZeroSizeRequestReadBandwidth); - dml_print("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, TotalZeroSizeCompressedReadBandwidth); - dml_print("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, MaximumEffectiveCompressionLuma); - dml_print("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, MaximumEffectiveCompressionChroma); - dml_print("DML::%s: AverageDCCCompressionRate = %f\n", __func__, AverageDCCCompressionRate); - dml_print("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, AverageDCCZeroSizeFraction); - dml_print("DML::%s: CompbufReservedSpace64B = %u\n", __func__, p->CompbufReservedSpace64B); - dml_print("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); - dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, p->CompressedBufferSizeInkByte); -#endif - if (AverageDCCZeroSizeFraction == 1) { - AverageZeroSizeCompressionRate = TotalZeroSizeRequestReadBandwidth / TotalZeroSizeCompressedReadBandwidth; - EffectiveCompressedBufferSize = (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageZeroSizeCompressionRate + ((dml_float_t)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 * AverageZeroSizeCompressionRate; - } else if (AverageDCCZeroSizeFraction > 0) { - AverageZeroSizeCompressionRate = TotalZeroSizeRequestReadBandwidth / TotalZeroSizeCompressedReadBandwidth; - EffectiveCompressedBufferSize = dml_min((dml_float_t)p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate, - (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate + 1 / AverageDCCCompressionRate)) + - dml_min(((dml_float_t)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate, - ((dml_float_t)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate)); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate); - dml_print("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate + 1 / AverageDCCCompressionRate)); - dml_print("DML::%s: min 3 = %f\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate); - dml_print("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate)); -#endif - } else { - EffectiveCompressedBufferSize = dml_min((dml_float_t)p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate, - (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageDCCCompressionRate) + - ((dml_float_t)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate); - dml_print("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageDCCCompressionRate); -#endif - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); - dml_print("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, AverageZeroSizeCompressionRate); - dml_print("DML::%s: EffectiveCompressedBufferSize = %f\n", __func__, EffectiveCompressedBufferSize); -#endif - - *p->StutterPeriod = 0; - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { - LinesInDETY = ((dml_float_t)p->DETBufferSizeY[k] + (p->UnboundedRequestEnabled == true ? EffectiveCompressedBufferSize : 0) * p->ReadBandwidthSurfaceLuma[k] / p->TotalDataReadBandwidth) / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; - LinesInDETYRoundedDownToSwath = dml_floor(LinesInDETY, p->SwathHeightY[k]); - DETBufferingTimeY = LinesInDETYRoundedDownToSwath * ((dml_float_t)p->HTotal[k] / p->PixelClock[k]) / p->VRatio[k]; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - dml_print("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - dml_print("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - dml_print("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); - dml_print("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, LinesInDETY); - dml_print("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, LinesInDETYRoundedDownToSwath); - dml_print("DML::%s: k=%u, HTotal = %u\n", __func__, k, p->HTotal[k]); - dml_print("DML::%s: k=%u, PixelClock = %f\n", __func__, k, p->PixelClock[k]); - dml_print("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->VRatio[k]); - dml_print("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, DETBufferingTimeY); - dml_print("DML::%s: k=%u,PixelClock = %f\n", __func__, k, p->PixelClock[k]); -#endif - - if (!FoundCriticalSurface || DETBufferingTimeY < *p->StutterPeriod) { - dml_bool_t isInterlaceTiming = p->Interlace[k] && !p->ProgressiveToInterlaceUnitInOPP; - - FoundCriticalSurface = true; - *p->StutterPeriod = DETBufferingTimeY; - FrameTimeCriticalSurface = (isInterlaceTiming ? dml_floor((dml_float_t)p->VTotal[k]/2.0, 1.0) : p->VTotal[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; - VActiveTimeCriticalSurface = (isInterlaceTiming ? dml_floor((dml_float_t)p->VActive[k]/2.0, 1.0) : p->VActive[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; - BytePerPixelYCriticalSurface = p->BytePerPixelY[k]; - SwathWidthYCriticalSurface = p->SwathWidthY[k]; - SwathHeightYCriticalSurface = p->SwathHeightY[k]; - BlockWidth256BytesYCriticalSurface = p->BlockWidth256BytesY[k]; - LinesToFinishSwathTransferStutterCriticalSurface = p->SwathHeightY[k] - (LinesInDETY - LinesInDETYRoundedDownToSwath); - DETBufferSizeYCriticalSurface = p->DETBufferSizeY[k]; - MinTTUVBlankCriticalSurface = p->MinTTUVBlank[k]; - SinglePlaneCriticalSurface = (p->ReadBandwidthSurfaceChroma[k] == 0); - SinglePipeCriticalSurface = (p->DPPPerSurface[k] == 1); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); - dml_print("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); - dml_print("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, MinTTUVBlankCriticalSurface); - dml_print("DML::%s: k=%u, FrameTimeCriticalSurface = %f\n", __func__, k, FrameTimeCriticalSurface); - dml_print("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, VActiveTimeCriticalSurface); - dml_print("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, BytePerPixelYCriticalSurface); - dml_print("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, SwathWidthYCriticalSurface); - dml_print("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, SwathHeightYCriticalSurface); - dml_print("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, BlockWidth256BytesYCriticalSurface); - dml_print("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, SinglePlaneCriticalSurface); - dml_print("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, SinglePipeCriticalSurface); - dml_print("DML::%s: k=%u, LinesToFinishSwathTransferStutterCriticalSurface = %f\n", __func__, k, LinesToFinishSwathTransferStutterCriticalSurface); -#endif - } - } - } - - PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = dml_min(*p->StutterPeriod * p->TotalDataReadBandwidth, EffectiveCompressedBufferSize); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, p->ROBBufferSizeInKByte); - dml_print("DML::%s: AverageDCCCompressionRate = %f\n", __func__, AverageDCCCompressionRate); - dml_print("DML::%s: StutterPeriod * TotalDataReadBandwidth = %f\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth); - dml_print("DML::%s: ROBBufferSizeInKByte * 1024 * AverageDCCCompressionRate + EffectiveCompressedBufferSize = %f\n", __func__, p->ROBBufferSizeInKByte * 1024 * AverageDCCCompressionRate + EffectiveCompressedBufferSize); - dml_print("DML::%s: EffectiveCompressedBufferSize = %f\n", __func__, EffectiveCompressedBufferSize); - dml_print("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f\n", __func__, PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer); - dml_print("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); - dml_print("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); - dml_print("DML::%s: TotalRowReadBandwidth = %f\n", __func__, TotalRowReadBandwidth); - dml_print("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); -#endif - - StutterBurstTime = PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / AverageDCCCompressionRate / p->ReturnBW + (*p->StutterPeriod * p->TotalDataReadBandwidth - PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64) + *p->StutterPeriod * TotalRowReadBandwidth / p->ReturnBW; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Part 1 = %f\n", __func__, PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / AverageDCCCompressionRate / p->ReturnBW); - dml_print("DML::%s: StutterPeriod * TotalDataReadBandwidth = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth)); - dml_print("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); - dml_print("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * TotalRowReadBandwidth / p->ReturnBW); - dml_print("DML::%s: StutterBurstTime = %f\n", __func__, StutterBurstTime); -#endif - StutterBurstTime = dml_max(StutterBurstTime, LinesToFinishSwathTransferStutterCriticalSurface * BytePerPixelYCriticalSurface * SwathWidthYCriticalSurface / p->ReturnBW); - - dml_print("DML::%s: Time to finish residue swath=%f\n", __func__, LinesToFinishSwathTransferStutterCriticalSurface * BytePerPixelYCriticalSurface * SwathWidthYCriticalSurface / p->ReturnBW); - - TotalActiveWriteback = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->WritebackEnable[k]) { - TotalActiveWriteback = TotalActiveWriteback + 1; - } - } - - if (TotalActiveWriteback == 0) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); - dml_print("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); - dml_print("DML::%s: StutterBurstTime = %f (final)\n", __func__, StutterBurstTime); - dml_print("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); -#endif - *p->StutterEfficiencyNotIncludingVBlank = dml_max(0., 1 - (p->SRExitTime + StutterBurstTime) / *p->StutterPeriod) * 100; - *p->Z8StutterEfficiencyNotIncludingVBlank = dml_max(0., 1 - (p->SRExitZ8Time + StutterBurstTime) / *p->StutterPeriod) * 100; - *p->NumberOfStutterBurstsPerFrame = (*p->StutterEfficiencyNotIncludingVBlank > 0 ? (dml_uint_t)(dml_ceil(VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); - *p->Z8NumberOfStutterBurstsPerFrame = (*p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? (dml_uint_t)(dml_ceil(VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); - } else { - *p->StutterEfficiencyNotIncludingVBlank = 0.; - *p->Z8StutterEfficiencyNotIncludingVBlank = 0.; - *p->NumberOfStutterBurstsPerFrame = 0; - *p->Z8NumberOfStutterBurstsPerFrame = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, VActiveTimeCriticalSurface); - dml_print("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - dml_print("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); - dml_print("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); - dml_print("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); -#endif - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { - if (p->BlendingAndTiming[k] == k) { - if (TotalNumberOfActiveOTG == 0) { - SinglePixelClock = p->PixelClock[k]; - SingleHTotal = p->HTotal[k]; - SingleVTotal = p->VTotal[k]; - } else if (SinglePixelClock != p->PixelClock[k] || SingleHTotal != p->HTotal[k] || SingleVTotal != p->VTotal[k]) { - SameTiming = false; - } - TotalNumberOfActiveOTG = TotalNumberOfActiveOTG + 1; - } - } - } - - if (*p->StutterEfficiencyNotIncludingVBlank > 0) { - LastStutterPeriod = VActiveTimeCriticalSurface - (*p->NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; - - if ((p->SynchronizeTimingsFinal || TotalNumberOfActiveOTG == 1) && SameTiming && - LastStutterPeriod + MinTTUVBlankCriticalSurface > p->StutterEnterPlusExitWatermark) { - *p->StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitTime + StutterBurstTime * VActiveTimeCriticalSurface / *p->StutterPeriod) / FrameTimeCriticalSurface) * 100; - } else { - *p->StutterEfficiency = *p->StutterEfficiencyNotIncludingVBlank; - } - } else { - *p->StutterEfficiency = 0; - } - - if (*p->Z8StutterEfficiencyNotIncludingVBlank > 0) { - LastZ8StutterPeriod = VActiveTimeCriticalSurface - (*p->NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; - if ((p->SynchronizeTimingsFinal || TotalNumberOfActiveOTG == 1) && SameTiming && LastZ8StutterPeriod + MinTTUVBlankCriticalSurface > p->Z8StutterEnterPlusExitWatermark) { - *p->Z8StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitZ8Time + StutterBurstTime * VActiveTimeCriticalSurface / *p->StutterPeriod) / FrameTimeCriticalSurface) * 100; - } else { - *p->Z8StutterEfficiency = *p->Z8StutterEfficiencyNotIncludingVBlank; - } - } else { - *p->Z8StutterEfficiency = 0.; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: LastZ8StutterPeriod = %f\n", __func__, LastZ8StutterPeriod); - dml_print("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); - dml_print("DML::%s: StutterBurstTime = %f\n", __func__, StutterBurstTime); - dml_print("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); - dml_print("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); - dml_print("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); - dml_print("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - dml_print("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); -#endif - - SwathSizeCriticalSurface = (dml_uint_t)(BytePerPixelYCriticalSurface * SwathHeightYCriticalSurface * dml_ceil(SwathWidthYCriticalSurface, BlockWidth256BytesYCriticalSurface)); - LastChunkOfSwathSize = SwathSizeCriticalSurface % (p->PixelChunkSizeInKByte * 1024); - MissingPartOfLastSwathOfDETSize = (dml_uint_t)(dml_ceil(DETBufferSizeYCriticalSurface, SwathSizeCriticalSurface) - DETBufferSizeYCriticalSurface); - - *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = !(!p->UnboundedRequestEnabled && (p->NumberOfActiveSurfaces == 1) && SinglePlaneCriticalSurface && SinglePipeCriticalSurface && (LastChunkOfSwathSize > 0) && - (LastChunkOfSwathSize <= 4096) && (MissingPartOfLastSwathOfDETSize > 0) && (MissingPartOfLastSwathOfDETSize <= LastChunkOfSwathSize)); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: SwathSizeCriticalSurface = %u\n", __func__, SwathSizeCriticalSurface); - dml_print("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, DETBufferSizeYCriticalSurface); - dml_print("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); - dml_print("DML::%s: LastChunkOfSwathSize = %u\n", __func__, LastChunkOfSwathSize); - dml_print("DML::%s: MissingPartOfLastSwathOfDETSize = %u\n", __func__, MissingPartOfLastSwathOfDETSize); - dml_print("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); -#endif -} // CalculateStutterEfficiency - -/// \CalculateSwathAndDETConfiguration -/// @brief Calculates swath width and different return buffers sizing (DET, CDB, etc.) -static void CalculateSwathAndDETConfiguration(struct display_mode_lib_scratch_st *scratch, - struct CalculateSwathAndDETConfiguration_params_st *p) -{ - dml_uint_t MaximumSwathHeightY[__DML_NUM_PLANES__]; - dml_uint_t MaximumSwathHeightC[__DML_NUM_PLANES__]; - dml_uint_t RoundedUpMaxSwathSizeBytesY[__DML_NUM_PLANES__]; - dml_uint_t RoundedUpMaxSwathSizeBytesC[__DML_NUM_PLANES__]; - dml_uint_t RoundedUpSwathSizeBytesY[__DML_NUM_PLANES__]; - dml_uint_t RoundedUpSwathSizeBytesC[__DML_NUM_PLANES__]; - dml_uint_t SwathWidthSingleDPP[__DML_NUM_PLANES__]; - dml_uint_t SwathWidthSingleDPPChroma[__DML_NUM_PLANES__]; - - dml_uint_t TotalActiveDPP = 0; - dml_bool_t NoChromaOrLinearSurfaces = true; - dml_uint_t SurfaceDoingUnboundedRequest = 0; - - dml_uint_t DETBufferSizeInKByteForSwathCalculation; - - const long TTUFIFODEPTH = 8; - const long MAXIMUMCOMPRESSION = 4; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - dml_print("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); - } -#endif - CalculateSwathWidth(p->ForceSingleDPP, - p->NumberOfActiveSurfaces, - p->SourcePixelFormat, - p->SourceScan, - p->ViewportStationary, - p->ViewportWidth, - p->ViewportHeight, - p->ViewportXStart, - p->ViewportYStart, - p->ViewportXStartC, - p->ViewportYStartC, - p->SurfaceWidthY, - p->SurfaceWidthC, - p->SurfaceHeightY, - p->SurfaceHeightC, - p->ODMMode, - p->BytePerPixY, - p->BytePerPixC, - p->Read256BytesBlockHeightY, - p->Read256BytesBlockHeightC, - p->Read256BytesBlockWidthY, - p->Read256BytesBlockWidthC, - p->BlendingAndTiming, - p->HActive, - p->HRatio, - p->DPPPerSurface, - - // Output - SwathWidthSingleDPP, - SwathWidthSingleDPPChroma, - p->SwathWidth, - p->SwathWidthChroma, - MaximumSwathHeightY, - MaximumSwathHeightC, - p->swath_width_luma_ub, - p->swath_width_chroma_ub); - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - RoundedUpMaxSwathSizeBytesY[k] = (dml_uint_t)(p->swath_width_luma_ub[k] * p->BytePerPixDETY[k] * MaximumSwathHeightY[k]); - RoundedUpMaxSwathSizeBytesC[k] = (dml_uint_t)(p->swath_width_chroma_ub[k] * p->BytePerPixDETC[k] * MaximumSwathHeightC[k]); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); - dml_print("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); - dml_print("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); - dml_print("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); - dml_print("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); - dml_print("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); - dml_print("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); -#endif - if (p->SourcePixelFormat[k] == dml_420_10) { - RoundedUpMaxSwathSizeBytesY[k] = (dml_uint_t)(dml_ceil((dml_float_t) RoundedUpMaxSwathSizeBytesY[k], 256)); - RoundedUpMaxSwathSizeBytesC[k] = (dml_uint_t)(dml_ceil((dml_float_t) RoundedUpMaxSwathSizeBytesC[k], 256)); - } - } - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - TotalActiveDPP = TotalActiveDPP + (p->ForceSingleDPP ? 1 : p->DPPPerSurface[k]); - if (p->DPPPerSurface[k] > 0) - SurfaceDoingUnboundedRequest = k; - if (p->SourcePixelFormat[k] == dml_420_8 || p->SourcePixelFormat[k] == dml_420_10 || - p->SourcePixelFormat[k] == dml_420_12 || p->SourcePixelFormat[k] == dml_rgbe_alpha - || p->SurfaceTiling[k] == dml_sw_linear) { - NoChromaOrLinearSurfaces = false; - } - } - - *p->UnboundedRequestEnabled = UnboundedRequest(p->UseUnboundedRequestingFinal, TotalActiveDPP, - NoChromaOrLinearSurfaces, p->Output[0]); - - CalculateDETBufferSize(p->DETSizeOverride, - p->UseMALLForPStateChange, - p->ForceSingleDPP, - p->NumberOfActiveSurfaces, - *p->UnboundedRequestEnabled, - p->nomDETInKByte, - p->MaxTotalDETInKByte, - p->ConfigReturnBufferSizeInKByte, - p->MinCompressedBufferSizeInKByte, - p->ConfigReturnBufferSegmentSizeInkByte, - p->CompressedBufferSegmentSizeInkByteFinal, - p->SourcePixelFormat, - p->ReadBandwidthLuma, - p->ReadBandwidthChroma, - RoundedUpMaxSwathSizeBytesY, - RoundedUpMaxSwathSizeBytesC, - p->DPPPerSurface, - - // Output - p->DETBufferSizeInKByte, // per hubp pipe - p->CompressedBufferSizeInkByte); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); - dml_print("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); - dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); - dml_print("DML::%s: UseUnboundedRequestingFinal = %u\n", __func__, p->UseUnboundedRequestingFinal); - dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); - dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); -#endif - - *p->ViewportSizeSupport = true; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - - DETBufferSizeInKByteForSwathCalculation = (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe ? 1024 : p->DETBufferSizeInKByte[k]); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); -#endif - - if (RoundedUpMaxSwathSizeBytesY[k] + RoundedUpMaxSwathSizeBytesC[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k]; - p->SwathHeightC[k] = MaximumSwathHeightC[k]; - RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k]; - RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k]; - } else if (RoundedUpMaxSwathSizeBytesY[k] >= 1.5 * RoundedUpMaxSwathSizeBytesC[k] && RoundedUpMaxSwathSizeBytesY[k] / 2 + RoundedUpMaxSwathSizeBytesC[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; - p->SwathHeightC[k] = MaximumSwathHeightC[k]; - RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k] / 2; - RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k]; - } else if (RoundedUpMaxSwathSizeBytesY[k] < 1.5 * RoundedUpMaxSwathSizeBytesC[k] && RoundedUpMaxSwathSizeBytesY[k] + RoundedUpMaxSwathSizeBytesC[k] / 2 <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k]; - p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; - RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k]; - RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k] / 2; - } else { - p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; - p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; - RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k] / 2; - RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k] / 2; - } - - if ((RoundedUpMaxSwathSizeBytesY[k] / 2 + RoundedUpMaxSwathSizeBytesC[k] / 2 > DETBufferSizeInKByteForSwathCalculation * 1024 / 2) || - p->SwathWidth[k] > p->MaximumSwathWidthLuma[k] || (p->SwathHeightC[k] > 0 && p->SwathWidthChroma[k] > p->MaximumSwathWidthChroma[k])) { - *p->ViewportSizeSupport = false; - p->ViewportSizeSupportPerSurface[k] = false; - } else { - p->ViewportSizeSupportPerSurface[k] = true; - } - - if (p->SwathHeightC[k] == 0) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u All DET for plane0\n", __func__, k); -#endif - p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024; - p->DETBufferSizeC[k] = 0; - } else if (RoundedUpSwathSizeBytesY[k] <= 1.5 * RoundedUpSwathSizeBytesC[k]) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u Half DET for plane0, half for plane1\n", __func__, k); -#endif - p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; - p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; - } else { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u 2/3 DET for plane0, 1/3 for plane1\n", __func__, k); -#endif - p->DETBufferSizeY[k] = (dml_uint_t)(dml_floor(p->DETBufferSizeInKByte[k] * 1024 * 2 / 3, 1024)); - p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 - p->DETBufferSizeY[k]; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - dml_print("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); - dml_print("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); - dml_print("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); - dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); - dml_print("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - dml_print("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); - dml_print("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); -#endif - - } - - *p->compbuf_reserved_space_64b = 2 * p->PixelChunkSizeInKByte * 1024 / 64; - if (*p->UnboundedRequestEnabled) { - *p->compbuf_reserved_space_64b = dml_max(*p->compbuf_reserved_space_64b, - (dml_float_t)(p->ROBBufferSizeInKByte * 1024/64) - - (dml_float_t)(RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest] * TTUFIFODEPTH / MAXIMUMCOMPRESSION/64)); - } - *p->compbuf_reserved_space_zs = 2 * p->PixelChunkSizeInKByte * 1024 / 256; -} // CalculateSwathAndDETConfiguration - -static void CalculateSwathWidth( - dml_bool_t ForceSingleDPP, - dml_uint_t NumberOfActiveSurfaces, - enum dml_source_format_class SourcePixelFormat[], - enum dml_rotation_angle SourceScan[], - dml_bool_t ViewportStationary[], - dml_uint_t ViewportWidth[], - dml_uint_t ViewportHeight[], - dml_uint_t ViewportXStart[], - dml_uint_t ViewportYStart[], - dml_uint_t ViewportXStartC[], - dml_uint_t ViewportYStartC[], - dml_uint_t SurfaceWidthY[], - dml_uint_t SurfaceWidthC[], - dml_uint_t SurfaceHeightY[], - dml_uint_t SurfaceHeightC[], - enum dml_odm_mode ODMMode[], - dml_uint_t BytePerPixY[], - dml_uint_t BytePerPixC[], - dml_uint_t Read256BytesBlockHeightY[], - dml_uint_t Read256BytesBlockHeightC[], - dml_uint_t Read256BytesBlockWidthY[], - dml_uint_t Read256BytesBlockWidthC[], - dml_uint_t BlendingAndTiming[], - dml_uint_t HActive[], - dml_float_t HRatio[], - dml_uint_t DPPPerSurface[], - - // Output - dml_uint_t SwathWidthSingleDPPY[], - dml_uint_t SwathWidthSingleDPPC[], - dml_uint_t SwathWidthY[], // per-pipe - dml_uint_t SwathWidthC[], // per-pipe - dml_uint_t MaximumSwathHeightY[], - dml_uint_t MaximumSwathHeightC[], - dml_uint_t swath_width_luma_ub[], // per-pipe - dml_uint_t swath_width_chroma_ub[]) // per-pipe -{ - enum dml_odm_mode MainSurfaceODMMode; - dml_uint_t surface_width_ub_l; - dml_uint_t surface_height_ub_l; - dml_uint_t surface_width_ub_c = 0; - dml_uint_t surface_height_ub_c = 0; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); -#endif - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_vertical_rotation(SourceScan[k])) { - SwathWidthSingleDPPY[k] = ViewportWidth[k]; - } else { - SwathWidthSingleDPPY[k] = ViewportHeight[k]; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u ViewportWidth=%u\n", __func__, k, ViewportWidth[k]); - dml_print("DML::%s: k=%u ViewportHeight=%u\n", __func__, k, ViewportHeight[k]); - dml_print("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); -#endif - - MainSurfaceODMMode = ODMMode[k]; - for (dml_uint_t j = 0; j < NumberOfActiveSurfaces; ++j) { - if (BlendingAndTiming[k] == j) { - MainSurfaceODMMode = ODMMode[j]; - } - } - - if (ForceSingleDPP) { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - } else { - if (MainSurfaceODMMode == dml_odm_mode_combine_4to1) { - SwathWidthY[k] = (dml_uint_t)(dml_min(SwathWidthSingleDPPY[k], dml_round(HActive[k] / 4.0 * HRatio[k], true))); - } else if (MainSurfaceODMMode == dml_odm_mode_combine_2to1) { - SwathWidthY[k] = (dml_uint_t)(dml_min(SwathWidthSingleDPPY[k], dml_round(HActive[k] / 2.0 * HRatio[k], true))); - } else if (DPPPerSurface[k] == 2) { - SwathWidthY[k] = SwathWidthSingleDPPY[k] / 2; - } else { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - } - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u HActive=%u\n", __func__, k, HActive[k]); - dml_print("DML::%s: k=%u HRatio=%f\n", __func__, k, HRatio[k]); - dml_print("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); - dml_print("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); - dml_print("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); -#endif - - if (SourcePixelFormat[k] == dml_420_8 || SourcePixelFormat[k] == dml_420_10 || SourcePixelFormat[k] == dml_420_12) { - SwathWidthC[k] = SwathWidthY[k] / 2; - SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k] / 2; - } else { - SwathWidthC[k] = SwathWidthY[k]; - SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k]; - } - - if (ForceSingleDPP == true) { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - SwathWidthC[k] = SwathWidthSingleDPPC[k]; - } - - surface_width_ub_l = (dml_uint_t)dml_ceil(SurfaceWidthY[k], Read256BytesBlockWidthY[k]); - surface_height_ub_l = (dml_uint_t)dml_ceil(SurfaceHeightY[k], Read256BytesBlockHeightY[k]); - - if (!dml_is_vertical_rotation(SourceScan[k])) { - MaximumSwathHeightY[k] = Read256BytesBlockHeightY[k]; - MaximumSwathHeightC[k] = Read256BytesBlockHeightC[k]; - if (ViewportStationary[k] && DPPPerSurface[k] == 1) { - swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_l, dml_floor(ViewportXStart[k] + SwathWidthY[k] + Read256BytesBlockWidthY[k] - 1, Read256BytesBlockWidthY[k]) - dml_floor(ViewportXStart[k], Read256BytesBlockWidthY[k]))); - } else { - swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_l, dml_ceil(SwathWidthY[k] - 1, Read256BytesBlockWidthY[k]) + Read256BytesBlockWidthY[k])); - } - if (BytePerPixC[k] > 0) { - surface_width_ub_c = (dml_uint_t)dml_ceil(SurfaceWidthC[k], Read256BytesBlockWidthC[k]); - if (ViewportStationary[k] && DPPPerSurface[k] == 1) { - swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_c, dml_floor(ViewportXStartC[k] + SwathWidthC[k] + Read256BytesBlockWidthC[k] - 1, Read256BytesBlockWidthC[k]) - dml_floor(ViewportXStartC[k], Read256BytesBlockWidthC[k]))); - } else { - swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_c, dml_ceil(SwathWidthC[k] - 1, Read256BytesBlockWidthC[k]) + Read256BytesBlockWidthC[k])); - } - } else { - swath_width_chroma_ub[k] = 0; - } - } else { - MaximumSwathHeightY[k] = Read256BytesBlockWidthY[k]; - MaximumSwathHeightC[k] = Read256BytesBlockWidthC[k]; - - if (ViewportStationary[k] && DPPPerSurface[k] == 1) { - swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_l, dml_floor(ViewportYStart[k] + SwathWidthY[k] + Read256BytesBlockHeightY[k] - 1, Read256BytesBlockHeightY[k]) - dml_floor(ViewportYStart[k], Read256BytesBlockHeightY[k]))); - } else { - swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_l, dml_ceil(SwathWidthY[k] - 1, Read256BytesBlockHeightY[k]) + Read256BytesBlockHeightY[k])); - } - if (BytePerPixC[k] > 0) { - surface_height_ub_c = (dml_uint_t)dml_ceil(SurfaceHeightC[k], Read256BytesBlockHeightC[k]); - if (ViewportStationary[k] && DPPPerSurface[k] == 1) { - swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_c, dml_floor(ViewportYStartC[k] + SwathWidthC[k] + Read256BytesBlockHeightC[k] - 1, Read256BytesBlockHeightC[k]) - dml_floor(ViewportYStartC[k], Read256BytesBlockHeightC[k]))); - } else { - swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_c, dml_ceil(SwathWidthC[k] - 1, Read256BytesBlockHeightC[k]) + Read256BytesBlockHeightC[k])); - } - } else { - swath_width_chroma_ub[k] = 0; - } - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); - dml_print("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); - dml_print("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); - dml_print("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); - dml_print("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); - dml_print("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); - dml_print("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); - dml_print("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); - dml_print("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, ViewportStationary[k]); - dml_print("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); - dml_print("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); - dml_print("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); - dml_print("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); - dml_print("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); -#endif - - } -} // CalculateSwathWidth - -static noinline_for_stack dml_float_t CalculateExtraLatency( - dml_uint_t RoundTripPingLatencyCycles, - dml_uint_t ReorderingBytes, - dml_float_t DCFCLK, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t PixelChunkSizeInKByte, - dml_uint_t TotalNumberOfDCCActiveDPP, - dml_uint_t MetaChunkSize, - dml_float_t ReturnBW, - dml_bool_t GPUVMEnable, - dml_bool_t HostVMEnable, - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t NumberOfDPP[], - dml_uint_t dpte_group_bytes[], - dml_float_t HostVMInefficiencyFactor, - dml_uint_t HostVMMinPageSize, - dml_uint_t HostVMMaxNonCachedPageTableLevels) -{ - dml_float_t ExtraLatencyBytes; - dml_float_t ExtraLatency; - - ExtraLatencyBytes = CalculateExtraLatencyBytes( - ReorderingBytes, - TotalNumberOfActiveDPP, - PixelChunkSizeInKByte, - TotalNumberOfDCCActiveDPP, - MetaChunkSize, - GPUVMEnable, - HostVMEnable, - NumberOfActiveSurfaces, - NumberOfDPP, - dpte_group_bytes, - HostVMInefficiencyFactor, - HostVMMinPageSize, - HostVMMaxNonCachedPageTableLevels); - - ExtraLatency = (RoundTripPingLatencyCycles + __DML_ARB_TO_RET_DELAY__) / DCFCLK + ExtraLatencyBytes / ReturnBW; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); - dml_print("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); - dml_print("DML::%s: ExtraLatencyBytes=%f\n", __func__, ExtraLatencyBytes); - dml_print("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); - dml_print("DML::%s: ExtraLatency=%f\n", __func__, ExtraLatency); -#endif - - return ExtraLatency; -} // CalculateExtraLatency - -static dml_uint_t CalculateHostVMDynamicLevels( - dml_bool_t GPUVMEnable, - dml_bool_t HostVMEnable, - dml_uint_t HostVMMinPageSize, - dml_uint_t HostVMMaxNonCachedPageTableLevels) -{ - dml_uint_t HostVMDynamicLevels = 0; - - if (GPUVMEnable && HostVMEnable) { - if (HostVMMinPageSize < 2048) - HostVMDynamicLevels = HostVMMaxNonCachedPageTableLevels; - else if (HostVMMinPageSize >= 2048 && HostVMMinPageSize < 1048576) - HostVMDynamicLevels = (dml_uint_t) dml_max(0, (dml_float_t) HostVMMaxNonCachedPageTableLevels - 1); - else - HostVMDynamicLevels = (dml_uint_t) dml_max(0, (dml_float_t) HostVMMaxNonCachedPageTableLevels - 2); - } else { - HostVMDynamicLevels = 0; - } - return HostVMDynamicLevels; -} - -static dml_uint_t CalculateExtraLatencyBytes(dml_uint_t ReorderingBytes, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t PixelChunkSizeInKByte, - dml_uint_t TotalNumberOfDCCActiveDPP, - dml_uint_t MetaChunkSize, - dml_bool_t GPUVMEnable, - dml_bool_t HostVMEnable, - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t NumberOfDPP[], - dml_uint_t dpte_group_bytes[], - dml_float_t HostVMInefficiencyFactor, - dml_uint_t HostVMMinPageSize, - dml_uint_t HostVMMaxNonCachedPageTableLevels) -{ - dml_uint_t HostVMDynamicLevels = CalculateHostVMDynamicLevels(GPUVMEnable, HostVMEnable, HostVMMinPageSize, HostVMMaxNonCachedPageTableLevels); - dml_float_t ret = ReorderingBytes + (TotalNumberOfActiveDPP * PixelChunkSizeInKByte + TotalNumberOfDCCActiveDPP * MetaChunkSize) * 1024.0; - - if (GPUVMEnable == true) { - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - ret = ret + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactor; - } - } - return (dml_uint_t)(ret); -} - -static dml_float_t CalculateUrgentLatency( - dml_float_t UrgentLatencyPixelDataOnly, - dml_float_t UrgentLatencyPixelMixedWithVMData, - dml_float_t UrgentLatencyVMDataOnly, - dml_bool_t DoUrgentLatencyAdjustment, - dml_float_t UrgentLatencyAdjustmentFabricClockComponent, - dml_float_t UrgentLatencyAdjustmentFabricClockReference, - dml_float_t FabricClock) -{ - dml_float_t ret; - - ret = dml_max3(UrgentLatencyPixelDataOnly, UrgentLatencyPixelMixedWithVMData, UrgentLatencyVMDataOnly); - if (DoUrgentLatencyAdjustment == true) { - ret = ret + UrgentLatencyAdjustmentFabricClockComponent * (UrgentLatencyAdjustmentFabricClockReference / FabricClock - 1); - } - return ret; -} - -static dml_float_t RequiredDTBCLK( - dml_bool_t DSCEnable, - dml_float_t PixelClock, - enum dml_output_format_class OutputFormat, - dml_float_t OutputBpp, - dml_uint_t DSCSlices, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_uint_t AudioRate, - dml_uint_t AudioLayout) -{ - if (DSCEnable != true) { - return dml_max(PixelClock / 4.0 * OutputBpp / 24.0, 25.0); - } else { - dml_float_t PixelWordRate = PixelClock / (OutputFormat == dml_444 ? 1 : 2); - dml_float_t HCActive = dml_ceil(DSCSlices * dml_ceil(OutputBpp * dml_ceil(HActive / DSCSlices, 1) / 8.0, 1) / 3.0, 1); - dml_float_t HCBlank = 64 + 32 * dml_ceil(AudioRate * (AudioLayout == 1 ? 1 : 0.25) * HTotal / (PixelClock * 1000), 1); - dml_float_t AverageTribyteRate = PixelWordRate * (HCActive + HCBlank) / HTotal; - dml_float_t HActiveTribyteRate = PixelWordRate * HCActive / HActive; - return dml_max4(PixelWordRate / 4.0, AverageTribyteRate / 4.0, HActiveTribyteRate / 4.0, 25.0) * 1.002; - } -} - -static void UseMinimumDCFCLK(struct display_mode_lib_scratch_st *scratch, struct UseMinimumDCFCLK_params_st *p) -{ - struct UseMinimumDCFCLK_locals_st *s = &scratch->UseMinimumDCFCLK_locals; - - s->NormalEfficiency = p->PercentOfIdealSDPPortBWReceivedAfterUrgLatency / 100.0; - for (dml_uint_t j = 0; j < 2; ++j) { - - - s->TotalMaxPrefetchFlipDPTERowBandwidth[j] = 0; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - s->TotalMaxPrefetchFlipDPTERowBandwidth[j] = s->TotalMaxPrefetchFlipDPTERowBandwidth[j] + p->NoOfDPP[j][k] * p->DPTEBytesPerRow[j][k] / (15.75 * p->HTotal[k] / p->PixelClock[k]); - } - - for (dml_uint_t k = 0; k <= p->NumberOfActiveSurfaces - 1; ++k) { - s->NoOfDPPState[k] = p->NoOfDPP[j][k]; - } - - s->DPTEBandwidth = s->TotalMaxPrefetchFlipDPTERowBandwidth[j]; - - s->DCFCLKRequiredForAverageBandwidth = dml_max(p->ProjectedDCFCLKDeepSleep[j], s->DPTEBandwidth / s->NormalEfficiency / p->ReturnBusWidth); - - s->ExtraLatencyBytes = CalculateExtraLatencyBytes(p->ReorderingBytes, p->TotalNumberOfActiveDPP[j], p->PixelChunkSizeInKByte, p->TotalNumberOfDCCActiveDPP[j], - p->MetaChunkSize, p->GPUVMEnable, p->HostVMEnable, p->NumberOfActiveSurfaces, s->NoOfDPPState, p->dpte_group_bytes, - 1, p->HostVMMinPageSize, p->HostVMMaxNonCachedPageTableLevels); - s->ExtraLatencyCycles = p->RoundTripPingLatencyCycles + __DML_ARB_TO_RET_DELAY__ + s->ExtraLatencyBytes / s->NormalEfficiency / p->ReturnBusWidth; - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - dml_float_t DCFCLKCyclesRequiredInPrefetch; - dml_float_t PrefetchTime; - - s->PixelDCFCLKCyclesRequiredInPrefetch[k] = (p->PrefetchLinesY[j][k] * p->swath_width_luma_ub_all_states[j][k] * p->BytePerPixelY[k] + p->PrefetchLinesC[j][k] * p->swath_width_chroma_ub_all_states[j][k] * p->BytePerPixelC[k]) / s->NormalEfficiency / p->ReturnBusWidth; - DCFCLKCyclesRequiredInPrefetch = 2 * s->ExtraLatencyCycles / s->NoOfDPPState[k] + p->PDEAndMetaPTEBytesPerFrame[j][k] / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth * (p->GPUVMMaxPageTableLevels > 2 ? 1 : 0) + 2 * p->DPTEBytesPerRow[j][k] / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth + 2 * p->MetaRowBytes[j][k] / s->NormalEfficiency / p->ReturnBusWidth + s->PixelDCFCLKCyclesRequiredInPrefetch[k]; - s->PrefetchPixelLinesTime[k] = dml_max(p->PrefetchLinesY[j][k], p->PrefetchLinesC[j][k]) * p->HTotal[k] / p->PixelClock[k]; - s->DynamicMetadataVMExtraLatency[k] = (p->GPUVMEnable == true && p->DynamicMetadataEnable[k] == true && p->DynamicMetadataVMEnabled == true) ? p->UrgLatency * p->GPUVMMaxPageTableLevels * (p->HostVMEnable == true ? p->HostVMMaxNonCachedPageTableLevels + 1 : 1) : 0; - - s->MinimumTWait = CalculateTWait(p->MaxPrefetchMode, - p->UseMALLForPStateChange[k], - p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal, - p->DRRDisplay[k], - p->DRAMClockChangeLatencyFinal, - p->FCLKChangeLatency, - p->UrgLatency, - p->SREnterPlusExitTime); - - PrefetchTime = (p->MaximumVStartup[j][k] - 1) * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - p->UrgLatency * ((p->GPUVMMaxPageTableLevels <= 2 ? p->GPUVMMaxPageTableLevels : p->GPUVMMaxPageTableLevels - 2) * (p->HostVMEnable == true ? p->HostVMMaxNonCachedPageTableLevels + 1 : 1) - 1) - s->DynamicMetadataVMExtraLatency[k]; - - if (PrefetchTime > 0) { - dml_float_t ExpectedVRatioPrefetch; - ExpectedVRatioPrefetch = s->PrefetchPixelLinesTime[k] / (PrefetchTime * s->PixelDCFCLKCyclesRequiredInPrefetch[k] / DCFCLKCyclesRequiredInPrefetch); - s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = s->NoOfDPPState[k] * s->PixelDCFCLKCyclesRequiredInPrefetch[k] / s->PrefetchPixelLinesTime[k] * dml_max(1.0, ExpectedVRatioPrefetch) * dml_max(1.0, ExpectedVRatioPrefetch / 4); - if (p->HostVMEnable == true || p->ImmediateFlipRequirement == true) { - s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = s->DCFCLKRequiredForPeakBandwidthPerSurface[k] + s->NoOfDPPState[k] * s->DPTEBandwidth / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth; - } - } else { - s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = p->DCFCLKPerState; - } - if (p->DynamicMetadataEnable[k] == true) { - dml_float_t TSetupPipe; - dml_float_t TdmbfPipe; - dml_float_t TdmsksPipe; - dml_float_t TdmecPipe; - dml_float_t AllowedTimeForUrgentExtraLatency; - - CalculateVUpdateAndDynamicMetadataParameters( - p->MaxInterDCNTileRepeaters, - p->RequiredDPPCLKPerSurface[j][k], - p->RequiredDISPCLK[j], - p->ProjectedDCFCLKDeepSleep[j], - p->PixelClock[k], - p->HTotal[k], - p->VTotal[k] - p->VActive[k], - p->DynamicMetadataTransmittedBytes[k], - p->DynamicMetadataLinesBeforeActiveRequired[k], - p->Interlace[k], - p->ProgressiveToInterlaceUnitInOPP, - - // Output - &TSetupPipe, - &TdmbfPipe, - &TdmecPipe, - &TdmsksPipe, - &s->dummy1, - &s->dummy2, - &s->dummy3); - - AllowedTimeForUrgentExtraLatency = p->MaximumVStartup[j][k] * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - TSetupPipe - TdmbfPipe - TdmecPipe - TdmsksPipe - s->DynamicMetadataVMExtraLatency[k]; - if (AllowedTimeForUrgentExtraLatency > 0) { - s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = dml_max(s->DCFCLKRequiredForPeakBandwidthPerSurface[k], s->ExtraLatencyCycles / AllowedTimeForUrgentExtraLatency); - } else { - s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = p->DCFCLKPerState; - } - } - } - s->DCFCLKRequiredForPeakBandwidth = 0; - for (dml_uint_t k = 0; k <= p->NumberOfActiveSurfaces - 1; ++k) { - s->DCFCLKRequiredForPeakBandwidth = s->DCFCLKRequiredForPeakBandwidth + s->DCFCLKRequiredForPeakBandwidthPerSurface[k]; - } - s->MinimumTvmPlus2Tr0 = p->UrgLatency * (p->GPUVMEnable == true ? (p->HostVMEnable == true ? (p->GPUVMMaxPageTableLevels + 2) * (p->HostVMMaxNonCachedPageTableLevels + 1) - 1 : p->GPUVMMaxPageTableLevels + 1) : 0); - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - dml_float_t MaximumTvmPlus2Tr0PlusTsw; - MaximumTvmPlus2Tr0PlusTsw = (p->MaximumVStartup[j][k] - 2) * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - s->DynamicMetadataVMExtraLatency[k]; - if (MaximumTvmPlus2Tr0PlusTsw <= s->MinimumTvmPlus2Tr0 + s->PrefetchPixelLinesTime[k] / 4) { - s->DCFCLKRequiredForPeakBandwidth = p->DCFCLKPerState; - } else { - s->DCFCLKRequiredForPeakBandwidth = dml_max3(s->DCFCLKRequiredForPeakBandwidth, - 2 * s->ExtraLatencyCycles / (MaximumTvmPlus2Tr0PlusTsw - s->MinimumTvmPlus2Tr0 - s->PrefetchPixelLinesTime[k] / 4), - (2 * s->ExtraLatencyCycles + s->PixelDCFCLKCyclesRequiredInPrefetch[k]) / (MaximumTvmPlus2Tr0PlusTsw - s->MinimumTvmPlus2Tr0)); - } - } - p->DCFCLKState[j] = dml_min(p->DCFCLKPerState, 1.05 * dml_max(s->DCFCLKRequiredForAverageBandwidth, s->DCFCLKRequiredForPeakBandwidth)); - } -} - - -static dml_bool_t UnboundedRequest(enum dml_unbounded_requesting_policy UseUnboundedRequestingFinal, - dml_uint_t TotalNumberOfActiveDPP, - dml_bool_t NoChromaOrLinear, - enum dml_output_encoder_class Output) -{ - dml_bool_t ret_val = false; - - ret_val = (UseUnboundedRequestingFinal != dml_unbounded_requesting_disable - && TotalNumberOfActiveDPP == 1 && NoChromaOrLinear); - if (UseUnboundedRequestingFinal == dml_unbounded_requesting_edp_only && Output != dml_edp) { - ret_val = false; - } - return (ret_val); -} - -static void CalculateSurfaceSizeInMall( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t MALLAllocatedForDCN, - enum dml_use_mall_for_static_screen_mode UseMALLForStaticScreen[], - dml_bool_t DCCEnable[], - dml_bool_t ViewportStationary[], - dml_uint_t ViewportXStartY[], - dml_uint_t ViewportYStartY[], - dml_uint_t ViewportXStartC[], - dml_uint_t ViewportYStartC[], - dml_uint_t ViewportWidthY[], - dml_uint_t ViewportHeightY[], - dml_uint_t BytesPerPixelY[], - dml_uint_t ViewportWidthC[], - dml_uint_t ViewportHeightC[], - dml_uint_t BytesPerPixelC[], - dml_uint_t SurfaceWidthY[], - dml_uint_t SurfaceWidthC[], - dml_uint_t SurfaceHeightY[], - dml_uint_t SurfaceHeightC[], - dml_uint_t Read256BytesBlockWidthY[], - dml_uint_t Read256BytesBlockWidthC[], - dml_uint_t Read256BytesBlockHeightY[], - dml_uint_t Read256BytesBlockHeightC[], - dml_uint_t ReadBlockWidthY[], - dml_uint_t ReadBlockWidthC[], - dml_uint_t ReadBlockHeightY[], - dml_uint_t ReadBlockHeightC[], - - // Output - dml_uint_t SurfaceSizeInMALL[], - dml_bool_t *ExceededMALLSize) -{ - dml_uint_t TotalSurfaceSizeInMALL = 0; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (ViewportStationary[k]) { - SurfaceSizeInMALL[k] = (dml_uint_t)(dml_min(dml_ceil(SurfaceWidthY[k], ReadBlockWidthY[k]), dml_floor(ViewportXStartY[k] + ViewportWidthY[k] + ReadBlockWidthY[k] - 1, ReadBlockWidthY[k]) - dml_floor(ViewportXStartY[k], ReadBlockWidthY[k])) * - dml_min(dml_ceil(SurfaceHeightY[k], ReadBlockHeightY[k]), dml_floor(ViewportYStartY[k] + ViewportHeightY[k] + ReadBlockHeightY[k] - 1, ReadBlockHeightY[k]) - dml_floor(ViewportYStartY[k], ReadBlockHeightY[k])) * - BytesPerPixelY[k]); - - if (ReadBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_min(dml_ceil(SurfaceWidthC[k], ReadBlockWidthC[k]), dml_floor(ViewportXStartC[k] + ViewportWidthC[k] + ReadBlockWidthC[k] - 1, ReadBlockWidthC[k]) - dml_floor(ViewportXStartC[k], ReadBlockWidthC[k])) * - dml_min(dml_ceil(SurfaceHeightC[k], ReadBlockHeightC[k]), dml_floor(ViewportYStartC[k] + ViewportHeightC[k] + ReadBlockHeightC[k] - 1, ReadBlockHeightC[k]) - dml_floor(ViewportYStartC[k], ReadBlockHeightC[k])) * BytesPerPixelC[k]); - } - if (DCCEnable[k] == true) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_min(dml_ceil(SurfaceWidthY[k], 8 * Read256BytesBlockWidthY[k]), dml_floor(ViewportXStartY[k] + ViewportWidthY[k] + 8 * Read256BytesBlockWidthY[k] - 1, 8 * Read256BytesBlockWidthY[k]) - dml_floor(ViewportXStartY[k], 8 * Read256BytesBlockWidthY[k])) * - dml_min(dml_ceil(SurfaceHeightY[k], 8 * Read256BytesBlockHeightY[k]), dml_floor(ViewportYStartY[k] + ViewportHeightY[k] + 8 * Read256BytesBlockHeightY[k] - 1, 8 * Read256BytesBlockHeightY[k]) - dml_floor(ViewportYStartY[k], 8 * Read256BytesBlockHeightY[k])) * BytesPerPixelY[k] / 256); - if (Read256BytesBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_min(dml_ceil(SurfaceWidthC[k], 8 * Read256BytesBlockWidthC[k]), dml_floor(ViewportXStartC[k] + ViewportWidthC[k] + 8 * Read256BytesBlockWidthC[k] - 1, 8 * Read256BytesBlockWidthC[k]) - dml_floor(ViewportXStartC[k], 8 * Read256BytesBlockWidthC[k])) * - dml_min(dml_ceil(SurfaceHeightC[k], 8 * Read256BytesBlockHeightC[k]), dml_floor(ViewportYStartC[k] + ViewportHeightC[k] + 8 * Read256BytesBlockHeightC[k] - 1, 8 * Read256BytesBlockHeightC[k]) - dml_floor(ViewportYStartC[k], 8 * Read256BytesBlockHeightC[k])) * BytesPerPixelC[k] / 256); - } - } - } else { - SurfaceSizeInMALL[k] = (dml_uint_t)(dml_ceil(dml_min(SurfaceWidthY[k], ViewportWidthY[k] + ReadBlockWidthY[k] - 1), ReadBlockWidthY[k]) * dml_ceil(dml_min(SurfaceHeightY[k], ViewportHeightY[k] + ReadBlockHeightY[k] - 1), ReadBlockHeightY[k]) * BytesPerPixelY[k]); - if (ReadBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_ceil(dml_min(SurfaceWidthC[k], ViewportWidthC[k] + ReadBlockWidthC[k] - 1), ReadBlockWidthC[k]) * - dml_ceil(dml_min(SurfaceHeightC[k], ViewportHeightC[k] + ReadBlockHeightC[k] - 1), ReadBlockHeightC[k]) * BytesPerPixelC[k]); - } - if (DCCEnable[k] == true) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_ceil(dml_min(SurfaceWidthY[k], ViewportWidthY[k] + 8 * Read256BytesBlockWidthY[k] - 1), 8 * Read256BytesBlockWidthY[k]) * - dml_ceil(dml_min(SurfaceHeightY[k], ViewportHeightY[k] + 8 * Read256BytesBlockHeightY[k] - 1), 8 * Read256BytesBlockHeightY[k]) * BytesPerPixelY[k] / 256); - - if (Read256BytesBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + - dml_ceil(dml_min(SurfaceWidthC[k], ViewportWidthC[k] + 8 * Read256BytesBlockWidthC[k] - 1), 8 * Read256BytesBlockWidthC[k]) * - dml_ceil(dml_min(SurfaceHeightC[k], ViewportHeightC[k] + 8 * Read256BytesBlockHeightC[k] - 1), 8 * Read256BytesBlockHeightC[k]) * BytesPerPixelC[k] / 256); - } - } - } - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable) - TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; - } - *ExceededMALLSize = (TotalSurfaceSizeInMALL > MALLAllocatedForDCN * 1024 * 1024); -} // CalculateSurfaceSizeInMall - -static void CalculateDETBufferSize( - dml_uint_t DETSizeOverride[], - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - dml_bool_t ForceSingleDPP, - dml_uint_t NumberOfActiveSurfaces, - dml_bool_t UnboundedRequestEnabled, - dml_uint_t nomDETInKByte, - dml_uint_t MaxTotalDETInKByte, - dml_uint_t ConfigReturnBufferSizeInKByte, - dml_uint_t MinCompressedBufferSizeInKByte, - dml_uint_t ConfigReturnBufferSegmentSizeInkByte, - dml_uint_t CompressedBufferSegmentSizeInkByteFinal, - enum dml_source_format_class SourcePixelFormat[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_uint_t RoundedUpMaxSwathSizeBytesY[], - dml_uint_t RoundedUpMaxSwathSizeBytesC[], - dml_uint_t DPPPerSurface[], - // Output - dml_uint_t DETBufferSizeInKByte[], - dml_uint_t *CompressedBufferSizeInkByte) -{ - dml_uint_t DETBufferSizePoolInKByte; - dml_uint_t NextDETBufferPieceInKByte; - dml_bool_t DETPieceAssignedToThisSurfaceAlready[__DML_NUM_PLANES__]; - dml_bool_t NextPotentialSurfaceToAssignDETPieceFound; - dml_uint_t NextSurfaceToAssignDETPiece; - dml_float_t TotalBandwidth; - dml_float_t BandwidthOfSurfacesNotAssignedDETPiece; - dml_uint_t max_minDET; - dml_uint_t minDET; - dml_uint_t minDET_pipe; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - dml_print("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); - dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); - dml_print("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); - dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - dml_print("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); - dml_print("DML::%s: CompressedBufferSegmentSizeInkByteFinal = %u\n", __func__, CompressedBufferSegmentSizeInkByteFinal); -#endif - - // Note: Will use default det size if that fits 2 swaths - if (UnboundedRequestEnabled) { - if (DETSizeOverride[0] > 0) { - DETBufferSizeInKByte[0] = DETSizeOverride[0]; - } else { - DETBufferSizeInKByte[0] = (dml_uint_t) dml_max(128.0, dml_ceil(2.0 * ((dml_float_t) RoundedUpMaxSwathSizeBytesY[0] + (dml_float_t) RoundedUpMaxSwathSizeBytesC[0]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte)); - } - *CompressedBufferSizeInkByte = ConfigReturnBufferSizeInKByte - DETBufferSizeInKByte[0]; - } else { - DETBufferSizePoolInKByte = MaxTotalDETInKByte; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - DETBufferSizeInKByte[k] = 0; - if (SourcePixelFormat[k] == dml_420_8 || SourcePixelFormat[k] == dml_420_10 || SourcePixelFormat[k] == dml_420_12) { - max_minDET = nomDETInKByte - ConfigReturnBufferSegmentSizeInkByte; - } else { - max_minDET = nomDETInKByte; - } - minDET = 128; - minDET_pipe = 0; - - // add DET resource until can hold 2 full swaths - while (minDET <= max_minDET && minDET_pipe == 0) { - if (2.0 * ((dml_float_t) RoundedUpMaxSwathSizeBytesY[k] + (dml_float_t) RoundedUpMaxSwathSizeBytesC[k]) / 1024.0 <= minDET) - minDET_pipe = minDET; - minDET = minDET + ConfigReturnBufferSegmentSizeInkByte; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u minDET = %u\n", __func__, k, minDET); - dml_print("DML::%s: k=%u max_minDET = %u\n", __func__, k, max_minDET); - dml_print("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, minDET_pipe); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); - dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); -#endif - - if (minDET_pipe == 0) { - minDET_pipe = (dml_uint_t)(dml_max(128, dml_ceil(((dml_float_t)RoundedUpMaxSwathSizeBytesY[k] + (dml_float_t)RoundedUpMaxSwathSizeBytesC[k]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte))); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, minDET_pipe); -#endif - } - - if (UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) { - DETBufferSizeInKByte[k] = 0; - } else if (DETSizeOverride[k] > 0) { - DETBufferSizeInKByte[k] = DETSizeOverride[k]; - DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * DETSizeOverride[k]; - } else if ((ForceSingleDPP ? 1 : DPPPerSurface[k]) * minDET_pipe <= DETBufferSizePoolInKByte) { - DETBufferSizeInKByte[k] = minDET_pipe; - DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * minDET_pipe; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); - dml_print("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, DETSizeOverride[k]); - dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); - dml_print("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, DETBufferSizePoolInKByte); -#endif - } - - TotalBandwidth = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) - TotalBandwidth = TotalBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: --- Before bandwidth adjustment ---\n", __func__); - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); - } - dml_print("DML::%s: --- DET allocation with bandwidth ---\n", __func__); -#endif - dml_print("DML::%s: TotalBandwidth = %f\n", __func__, TotalBandwidth); - BandwidthOfSurfacesNotAssignedDETPiece = TotalBandwidth; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - - if (UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) { - DETPieceAssignedToThisSurfaceAlready[k] = true; - } else if (DETSizeOverride[k] > 0 || (((dml_float_t) (ForceSingleDPP ? 1 : DPPPerSurface[k]) * (dml_float_t) DETBufferSizeInKByte[k] / (dml_float_t) MaxTotalDETInKByte) >= ((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / TotalBandwidth))) { - DETPieceAssignedToThisSurfaceAlready[k] = true; - BandwidthOfSurfacesNotAssignedDETPiece = BandwidthOfSurfacesNotAssignedDETPiece - ReadBandwidthLuma[k] - ReadBandwidthChroma[k]; - } else { - DETPieceAssignedToThisSurfaceAlready[k] = false; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); - dml_print("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, BandwidthOfSurfacesNotAssignedDETPiece); -#endif - } - - for (dml_uint_t j = 0; j < NumberOfActiveSurfaces; ++j) { - NextPotentialSurfaceToAssignDETPieceFound = false; - NextSurfaceToAssignDETPiece = 0; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); - dml_print("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); - dml_print("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[NextSurfaceToAssignDETPiece]); - dml_print("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); - dml_print("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, NextSurfaceToAssignDETPiece); -#endif - if (!DETPieceAssignedToThisSurfaceAlready[k] && (!NextPotentialSurfaceToAssignDETPieceFound || - ReadBandwidthLuma[k] + ReadBandwidthChroma[k] < ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece])) { - NextSurfaceToAssignDETPiece = k; - NextPotentialSurfaceToAssignDETPieceFound = true; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); - dml_print("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); -#endif - } - - if (NextPotentialSurfaceToAssignDETPieceFound) { - // Note: To show the banker's rounding behavior in VBA and also the fact that the DET buffer size varies due to precision issue - // - //dml_float_t tmp1 = ((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / - // ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * 64.0)); - //dml_float_t tmp2 = dml_round((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / - // ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * 64.0)); - // - //dml_print("DML::%s: j=%u, tmp1 = %f\n", __func__, j, tmp1); - //dml_print("DML::%s: j=%u, tmp2 = %f\n", __func__, j, tmp2); - - NextDETBufferPieceInKByte = (dml_uint_t)(dml_min( - dml_round((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / - ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte), true) - * (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte, - dml_floor((dml_float_t) DETBufferSizePoolInKByte, (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte))); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, DETBufferSizePoolInKByte); - dml_print("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, NextSurfaceToAssignDETPiece); - dml_print("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, NextSurfaceToAssignDETPiece, ReadBandwidthLuma[NextSurfaceToAssignDETPiece]); - dml_print("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, NextSurfaceToAssignDETPiece, ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); - dml_print("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, BandwidthOfSurfacesNotAssignedDETPiece); - dml_print("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, NextDETBufferPieceInKByte); - dml_print("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[NextSurfaceToAssignDETPiece]); -#endif - - DETBufferSizeInKByte[NextSurfaceToAssignDETPiece] = DETBufferSizeInKByte[NextSurfaceToAssignDETPiece] + NextDETBufferPieceInKByte / (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]); -#ifdef __DML_VBA_DEBUG__ - dml_print("to %u\n", DETBufferSizeInKByte[NextSurfaceToAssignDETPiece]); -#endif - - DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - NextDETBufferPieceInKByte; - DETPieceAssignedToThisSurfaceAlready[NextSurfaceToAssignDETPiece] = true; - BandwidthOfSurfacesNotAssignedDETPiece = BandwidthOfSurfacesNotAssignedDETPiece - (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); - } - } - *CompressedBufferSizeInkByte = MinCompressedBufferSizeInKByte; - } - *CompressedBufferSizeInkByte = *CompressedBufferSizeInkByte * CompressedBufferSegmentSizeInkByteFinal / ConfigReturnBufferSegmentSizeInkByte; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: --- After bandwidth adjustment ---\n", __func__); - dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); - } -#endif -} // CalculateDETBufferSize - - -/// @brief Calculate the bound for return buffer sizing -static void CalculateMaxDETAndMinCompressedBufferSize( - dml_uint_t ConfigReturnBufferSizeInKByte, - dml_uint_t ConfigReturnBufferSegmentSizeInKByte, - dml_uint_t ROBBufferSizeInKByte, - dml_uint_t MaxNumDPP, - dml_bool_t nomDETInKByteOverrideEnable, // VBA_DELTA, allow DV to override default DET size - dml_uint_t nomDETInKByteOverrideValue, // VBA_DELTA - - // Output - dml_uint_t *MaxTotalDETInKByte, - dml_uint_t *nomDETInKByte, - dml_uint_t *MinCompressedBufferSizeInKByte) -{ - *MaxTotalDETInKByte = ConfigReturnBufferSizeInKByte - ConfigReturnBufferSegmentSizeInKByte; - *nomDETInKByte = (dml_uint_t)(dml_floor((dml_float_t) *MaxTotalDETInKByte / (dml_float_t) MaxNumDPP, ConfigReturnBufferSegmentSizeInKByte)); - *MinCompressedBufferSizeInKByte = ConfigReturnBufferSizeInKByte - *MaxTotalDETInKByte; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - dml_print("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); - dml_print("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); - dml_print("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); - dml_print("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); - dml_print("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); -#endif - - if (nomDETInKByteOverrideEnable) { - *nomDETInKByte = nomDETInKByteOverrideValue; - dml_print("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); - } -} // CalculateMaxDETAndMinCompressedBufferSize - -/// @brief Calculate all the RQ request attributes, like row height and # swath -static void CalculateVMRowAndSwath(struct display_mode_lib_scratch_st *scratch, - struct CalculateVMRowAndSwath_params_st *p) -{ - struct CalculateVMRowAndSwath_locals_st *s = &scratch->CalculateVMRowAndSwath_locals; - - s->HostVMDynamicLevels = CalculateHostVMDynamicLevels(p->GPUVMEnable, p->HostVMEnable, p->HostVMMinPageSize, p->HostVMMaxNonCachedPageTableLevels); - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->HostVMEnable == true) { - p->vm_group_bytes[k] = 512; - p->dpte_group_bytes[k] = 512; - } else if (p->GPUVMEnable == true) { - p->vm_group_bytes[k] = 2048; - if (p->GPUVMMinPageSizeKBytes[k] >= 64 && dml_is_vertical_rotation(p->myPipe[k].SourceScan)) { - p->dpte_group_bytes[k] = 512; - } else { - p->dpte_group_bytes[k] = 2048; - } - } else { - p->vm_group_bytes[k] = 0; - p->dpte_group_bytes[k] = 0; - } - - if (p->myPipe[k].SourcePixelFormat == dml_420_8 || p->myPipe[k].SourcePixelFormat == dml_420_10 || - p->myPipe[k].SourcePixelFormat == dml_420_12 || p->myPipe[k].SourcePixelFormat == dml_rgbe_alpha) { - if ((p->myPipe[k].SourcePixelFormat == dml_420_10 || p->myPipe[k].SourcePixelFormat == dml_420_12) && !dml_is_vertical_rotation(p->myPipe[k].SourceScan)) { - s->PTEBufferSizeInRequestsForLuma[k] = (p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma) / 2; - s->PTEBufferSizeInRequestsForChroma[k] = s->PTEBufferSizeInRequestsForLuma[k]; - } else { - s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma; - s->PTEBufferSizeInRequestsForChroma[k] = p->PTEBufferSizeInRequestsChroma; - } - - s->PDEAndMetaPTEBytesFrameC = CalculateVMAndRowBytes( - p->myPipe[k].ViewportStationary, - p->myPipe[k].DCCEnable, - p->myPipe[k].DPPPerSurface, - p->myPipe[k].BlockHeight256BytesC, - p->myPipe[k].BlockWidth256BytesC, - p->myPipe[k].SourcePixelFormat, - p->myPipe[k].SurfaceTiling, - p->myPipe[k].BytePerPixelC, - p->myPipe[k].SourceScan, - p->SwathWidthC[k], - p->myPipe[k].ViewportHeightChroma, - p->myPipe[k].ViewportXStartC, - p->myPipe[k].ViewportYStartC, - p->GPUVMEnable, - p->GPUVMMaxPageTableLevels, - p->GPUVMMinPageSizeKBytes[k], - s->PTEBufferSizeInRequestsForChroma[k], - p->myPipe[k].PitchC, - p->myPipe[k].DCCMetaPitchC, - p->myPipe[k].BlockWidthC, - p->myPipe[k].BlockHeightC, - - // Output - &s->MetaRowByteC[k], - &s->PixelPTEBytesPerRowC[k], - &s->PixelPTEBytesPerRowStorageC[k], - &p->dpte_row_width_chroma_ub[k], - &p->dpte_row_height_chroma[k], - &p->dpte_row_height_linear_chroma[k], - &s->PixelPTEBytesPerRowC_one_row_per_frame[k], - &s->dpte_row_width_chroma_ub_one_row_per_frame[k], - &s->dpte_row_height_chroma_one_row_per_frame[k], - &p->meta_req_width_chroma[k], - &p->meta_req_height_chroma[k], - &p->meta_row_width_chroma[k], - &p->meta_row_height_chroma[k], - &p->PixelPTEReqWidthC[k], - &p->PixelPTEReqHeightC[k], - &p->PTERequestSizeC[k], - &p->dpde0_bytes_per_frame_ub_c[k], - &p->meta_pte_bytes_per_frame_ub_c[k]); - - p->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines ( - p->myPipe[k].VRatioChroma, - p->myPipe[k].VTapsChroma, - p->myPipe[k].InterlaceEnable, - p->myPipe[k].ProgressiveToInterlaceUnitInOPP, - p->myPipe[k].SwathHeightC, - p->myPipe[k].SourceScan, - p->myPipe[k].ViewportStationary, - p->SwathWidthC[k], - p->myPipe[k].ViewportHeightChroma, - p->myPipe[k].ViewportXStartC, - p->myPipe[k].ViewportYStartC, - - // Output - &p->VInitPreFillC[k], - &p->MaxNumSwathC[k]); - } else { - s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma; - s->PTEBufferSizeInRequestsForChroma[k] = 0; - s->PixelPTEBytesPerRowC[k] = 0; - s->PixelPTEBytesPerRowStorageC[k] = 0; - s->PDEAndMetaPTEBytesFrameC = 0; - s->MetaRowByteC[k] = 0; - p->MaxNumSwathC[k] = 0; - p->PrefetchSourceLinesC[k] = 0; - s->dpte_row_height_chroma_one_row_per_frame[k] = 0; - s->dpte_row_width_chroma_ub_one_row_per_frame[k] = 0; - s->PixelPTEBytesPerRowC_one_row_per_frame[k] = 0; - } - - s->PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes( - p->myPipe[k].ViewportStationary, - p->myPipe[k].DCCEnable, - p->myPipe[k].DPPPerSurface, - p->myPipe[k].BlockHeight256BytesY, - p->myPipe[k].BlockWidth256BytesY, - p->myPipe[k].SourcePixelFormat, - p->myPipe[k].SurfaceTiling, - p->myPipe[k].BytePerPixelY, - p->myPipe[k].SourceScan, - p->SwathWidthY[k], - p->myPipe[k].ViewportHeight, - p->myPipe[k].ViewportXStart, - p->myPipe[k].ViewportYStart, - p->GPUVMEnable, - p->GPUVMMaxPageTableLevels, - p->GPUVMMinPageSizeKBytes[k], - s->PTEBufferSizeInRequestsForLuma[k], - p->myPipe[k].PitchY, - p->myPipe[k].DCCMetaPitchY, - p->myPipe[k].BlockWidthY, - p->myPipe[k].BlockHeightY, - - // Output - &s->MetaRowByteY[k], - &s->PixelPTEBytesPerRowY[k], - &s->PixelPTEBytesPerRowStorageY[k], - &p->dpte_row_width_luma_ub[k], - &p->dpte_row_height_luma[k], - &p->dpte_row_height_linear_luma[k], - &s->PixelPTEBytesPerRowY_one_row_per_frame[k], - &s->dpte_row_width_luma_ub_one_row_per_frame[k], - &s->dpte_row_height_luma_one_row_per_frame[k], - &p->meta_req_width[k], - &p->meta_req_height[k], - &p->meta_row_width[k], - &p->meta_row_height[k], - &p->PixelPTEReqWidthY[k], - &p->PixelPTEReqHeightY[k], - &p->PTERequestSizeY[k], - &p->dpde0_bytes_per_frame_ub_l[k], - &p->meta_pte_bytes_per_frame_ub_l[k]); - - p->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines( - p->myPipe[k].VRatio, - p->myPipe[k].VTaps, - p->myPipe[k].InterlaceEnable, - p->myPipe[k].ProgressiveToInterlaceUnitInOPP, - p->myPipe[k].SwathHeightY, - p->myPipe[k].SourceScan, - p->myPipe[k].ViewportStationary, - p->SwathWidthY[k], - p->myPipe[k].ViewportHeight, - p->myPipe[k].ViewportXStart, - p->myPipe[k].ViewportYStart, - - // Output - &p->VInitPreFillY[k], - &p->MaxNumSwathY[k]); - - p->PDEAndMetaPTEBytesFrame[k] = (s->PDEAndMetaPTEBytesFrameY + s->PDEAndMetaPTEBytesFrameC) * (1 + 8 * s->HostVMDynamicLevels); - p->MetaRowByte[k] = s->MetaRowByteY[k] + s->MetaRowByteC[k]; - - if (s->PixelPTEBytesPerRowStorageY[k] <= 64 * s->PTEBufferSizeInRequestsForLuma[k] && s->PixelPTEBytesPerRowStorageC[k] <= 64 * s->PTEBufferSizeInRequestsForChroma[k]) { - p->PTEBufferSizeNotExceeded[k] = true; - } else { - p->PTEBufferSizeNotExceeded[k] = false; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY = %u\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC = %u\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); - dml_print("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); - dml_print("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); - dml_print("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); -#endif - } - s->one_row_per_frame_fits_in_buffer[k] = (s->PixelPTEBytesPerRowY_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForLuma[k] && - s->PixelPTEBytesPerRowC_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForChroma[k]); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrame = %u\n", __func__, k, p->PDEAndMetaPTEBytesFrame[k]); - dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrameY = %u\n", __func__, k, s->PDEAndMetaPTEBytesFrameY); - dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrameC = %u\n", __func__, k, s->PDEAndMetaPTEBytesFrameC); - dml_print("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); - dml_print("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); -#endif - } - - CalculateMALLUseForStaticScreen( - p->NumberOfActiveSurfaces, - p->MALLAllocatedForDCN, - p->UseMALLForStaticScreen, // mode - p->SurfaceSizeInMALL, - s->one_row_per_frame_fits_in_buffer, - // Output - p->UsesMALLForStaticScreen); // boolen - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->PTEBufferModeOverrideEn[k] == 1) { - p->PTE_BUFFER_MODE[k] = p->PTEBufferModeOverrideVal[k]; - } - p->PTE_BUFFER_MODE[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->UsesMALLForStaticScreen[k] || (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) || - (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) || (p->GPUVMMinPageSizeKBytes[k] > 64); - p->BIGK_FRAGMENT_SIZE[k] = (dml_uint_t)(dml_log2(p->GPUVMMinPageSizeKBytes[k] * 1024) - 12); - } - - for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); - dml_print("DML::%s: k=%u, UsesMALLForStaticScreen = %u\n", __func__, k, p->UsesMALLForStaticScreen[k]); -#endif - p->use_one_row_for_frame[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->UsesMALLForStaticScreen[k] || (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) || - (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) || (p->GPUVMMinPageSizeKBytes[k] > 64 && dml_is_vertical_rotation(p->myPipe[k].SourceScan)); - - p->use_one_row_for_frame_flip[k] = p->use_one_row_for_frame[k] && !(p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame); - - if (p->use_one_row_for_frame[k]) { - p->dpte_row_height_luma[k] = s->dpte_row_height_luma_one_row_per_frame[k]; - p->dpte_row_width_luma_ub[k] = s->dpte_row_width_luma_ub_one_row_per_frame[k]; - s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY_one_row_per_frame[k]; - p->dpte_row_height_chroma[k] = s->dpte_row_height_chroma_one_row_per_frame[k]; - p->dpte_row_width_chroma_ub[k] = s->dpte_row_width_chroma_ub_one_row_per_frame[k]; - s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC_one_row_per_frame[k]; - p->PTEBufferSizeNotExceeded[k] = s->one_row_per_frame_fits_in_buffer[k]; - } - - if (p->MetaRowByte[k] <= p->DCCMetaBufferSizeBytes) { - p->DCCMetaBufferSizeNotExceeded[k] = true; - } else { - p->DCCMetaBufferSizeNotExceeded[k] = false; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, MetaRowByte = %u\n", __func__, k, p->MetaRowByte[k]); - dml_print("DML::%s: k=%u, DCCMetaBufferSizeBytes = %u\n", __func__, k, p->DCCMetaBufferSizeBytes); - dml_print("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); -#endif - } - s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY[k] * (1 + 8 * s->HostVMDynamicLevels); - s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC[k] * (1 + 8 * s->HostVMDynamicLevels); - p->PixelPTEBytesPerRow[k] = s->PixelPTEBytesPerRowY[k] + s->PixelPTEBytesPerRowC[k]; - if (p->use_one_row_for_frame[k]) - p->PixelPTEBytesPerRow[k] = p->PixelPTEBytesPerRow[k] / 2; - - CalculateRowBandwidth( - p->GPUVMEnable, - p->myPipe[k].SourcePixelFormat, - p->myPipe[k].VRatio, - p->myPipe[k].VRatioChroma, - p->myPipe[k].DCCEnable, - p->myPipe[k].HTotal / p->myPipe[k].PixelClock, - s->MetaRowByteY[k], - s->MetaRowByteC[k], - p->meta_row_height[k], - p->meta_row_height_chroma[k], - s->PixelPTEBytesPerRowY[k], - s->PixelPTEBytesPerRowC[k], - p->dpte_row_height_luma[k], - p->dpte_row_height_chroma[k], - - // Output - &p->meta_row_bw[k], - &p->dpte_row_bw[k]); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); - dml_print("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); - dml_print("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->UseMALLForPStateChange[k]); - dml_print("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); - dml_print("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY = %u\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - dml_print("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); - dml_print("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC = %u\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - dml_print("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); - dml_print("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); - dml_print("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); - dml_print("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); -#endif - } -} - -static void CalculateOutputLink( - dml_float_t PHYCLKPerState, - dml_float_t PHYCLKD18PerState, - dml_float_t PHYCLKD32PerState, - dml_float_t Downspreading, - dml_bool_t IsMainSurfaceUsingTheIndicatedTiming, - enum dml_output_encoder_class Output, - enum dml_output_format_class OutputFormat, - dml_uint_t HTotal, - dml_uint_t HActive, - dml_float_t PixelClockBackEnd, - dml_float_t ForcedOutputLinkBPP, - dml_uint_t DSCInputBitPerComponent, - dml_uint_t NumberOfDSCSlices, - dml_float_t AudioSampleRate, - dml_uint_t AudioSampleLayout, - enum dml_odm_mode ODMModeNoDSC, - enum dml_odm_mode ODMModeDSC, - enum dml_dsc_enable DSCEnable, - dml_uint_t OutputLinkDPLanes, - enum dml_output_link_dp_rate OutputLinkDPRate, - - // Output - dml_bool_t *RequiresDSC, - dml_bool_t *RequiresFEC, - dml_float_t *OutBpp, - enum dml_output_type_and_rate__type *OutputType, - enum dml_output_type_and_rate__rate *OutputRate, - dml_uint_t *RequiredSlots) -{ - dml_bool_t LinkDSCEnable; - dml_uint_t dummy; - *RequiresDSC = false; - *RequiresFEC = false; - *OutBpp = 0; - - *OutputType = dml_output_type_unknown; - *OutputRate = dml_output_rate_unknown; - - if (IsMainSurfaceUsingTheIndicatedTiming) { - if (Output == dml_hdmi) { - *RequiresDSC = false; - *RequiresFEC = false; - *OutBpp = TruncToValidBPP(dml_min(600, PHYCLKPerState) * 10, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, false, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = "HDMI"; - *OutputType = dml_output_type_hdmi; - - } else if (Output == dml_dp || Output == dml_dp2p0 || Output == dml_edp) { - if (DSCEnable == dml_dsc_enable) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml_dp || Output == dml_dp2p0) { - *RequiresFEC = true; - } else { - *RequiresFEC = false; - } - } else { - *RequiresDSC = false; - LinkDSCEnable = false; - if (Output == dml_dp2p0) { - *RequiresFEC = true; - } else { - *RequiresFEC = false; - } - } - if (Output == dml_dp2p0) { - *OutBpp = 0; - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr10) && PHYCLKD32PerState >= 10000 / 32.0) { - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && PHYCLKD32PerState < 13500 / 32.0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR10"; - *OutputType = dml_output_type_dp2p0; - *OutputRate = dml_output_rate_dp_rate_uhbr10; - } - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr13p5) && *OutBpp == 0 && PHYCLKD32PerState >= 13500 / 32.0) { - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && PHYCLKD32PerState < 20000 / 32 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR13p5"; - *OutputType = dml_output_type_dp2p0; - *OutputRate = dml_output_rate_dp_rate_uhbr13p5; - } - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr20) && *OutBpp == 0 && PHYCLKD32PerState >= 20000 / 32) { - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR20"; - *OutputType = dml_output_type_dp2p0; - *OutputRate = dml_output_rate_dp_rate_uhbr20; - } - } else { // output is dp or edp - *OutBpp = 0; - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr) && PHYCLKPerState >= 270) { - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && PHYCLKPerState < 540 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR"; - *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; - *OutputRate = dml_output_rate_dp_rate_hbr; - } - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr2) && *OutBpp == 0 && PHYCLKPerState >= 540) { - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && PHYCLKPerState < 810 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR2"; - *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; - *OutputRate = dml_output_rate_dp_rate_hbr2; - } - if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr3) && *OutBpp == 0 && PHYCLKPerState >= 810) { // VBA_ERROR, vba code doesn't have hbr3 check - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR3"; - *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; - *OutputRate = dml_output_rate_dp_rate_hbr3; - } - } - } - } -} - -/// @brief Determine the ODM mode and number of DPP used per plane based on dispclk, dsc usage, odm usage policy -static void CalculateODMMode( - dml_uint_t MaximumPixelsPerLinePerDSCUnit, - dml_uint_t HActive, - enum dml_output_encoder_class Output, - enum dml_output_format_class OutputFormat, - enum dml_odm_use_policy ODMUse, - dml_float_t StateDispclk, - dml_float_t MaxDispclk, - dml_bool_t DSCEnable, - dml_uint_t TotalNumberOfActiveDPP, - dml_uint_t MaxNumDPP, - dml_float_t PixelClock, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKRampingMargin, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_uint_t NumberOfDSCSlices, - - // Output - dml_bool_t *TotalAvailablePipesSupport, - dml_uint_t *NumberOfDPP, - enum dml_odm_mode *ODMMode, - dml_float_t *RequiredDISPCLKPerSurface) -{ - - dml_float_t SurfaceRequiredDISPCLKWithoutODMCombine; - dml_float_t SurfaceRequiredDISPCLKWithODMCombineTwoToOne; - dml_float_t SurfaceRequiredDISPCLKWithODMCombineFourToOne; - - SurfaceRequiredDISPCLKWithoutODMCombine = CalculateRequiredDispclk(dml_odm_mode_bypass, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); - SurfaceRequiredDISPCLKWithODMCombineTwoToOne = CalculateRequiredDispclk(dml_odm_mode_combine_2to1, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); - SurfaceRequiredDISPCLKWithODMCombineFourToOne = CalculateRequiredDispclk(dml_odm_mode_combine_4to1, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); - *TotalAvailablePipesSupport = true; - - if (OutputFormat == dml_420) { - if (HActive > 4 * DML2_MAX_FMT_420_BUFFER_WIDTH) - *TotalAvailablePipesSupport = false; - else if (HActive > 2 * DML2_MAX_FMT_420_BUFFER_WIDTH) - ODMUse = dml_odm_use_policy_combine_4to1; - else if (HActive > DML2_MAX_FMT_420_BUFFER_WIDTH && ODMUse != dml_odm_use_policy_combine_4to1) - ODMUse = dml_odm_use_policy_combine_2to1; - if (Output == dml_hdmi && ODMUse == dml_odm_use_policy_combine_2to1) - *TotalAvailablePipesSupport = false; - if ((Output == dml_hdmi || Output == dml_dp || Output == dml_edp) && ODMUse == dml_odm_use_policy_combine_4to1) - *TotalAvailablePipesSupport = false; - } - - if (ODMUse == dml_odm_use_policy_bypass || ODMUse == dml_odm_use_policy_combine_as_needed) - *ODMMode = dml_odm_mode_bypass; - else if (ODMUse == dml_odm_use_policy_combine_2to1) - *ODMMode = dml_odm_mode_combine_2to1; - else if (ODMUse == dml_odm_use_policy_combine_4to1) - *ODMMode = dml_odm_mode_combine_4to1; - else if (ODMUse == dml_odm_use_policy_split_1to2) - *ODMMode = dml_odm_mode_split_1to2; - else if (ODMUse == dml_odm_use_policy_mso_1to2) - *ODMMode = dml_odm_mode_mso_1to2; - else if (ODMUse == dml_odm_use_policy_mso_1to4) - *ODMMode = dml_odm_mode_mso_1to4; - - *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithoutODMCombine; - *NumberOfDPP = 0; - - if (!(Output == dml_hdmi || Output == dml_dp || Output == dml_edp) && (ODMUse == dml_odm_use_policy_combine_4to1 || (ODMUse == dml_odm_use_policy_combine_as_needed && - (SurfaceRequiredDISPCLKWithODMCombineTwoToOne > StateDispclk || (DSCEnable && (HActive > 2 * MaximumPixelsPerLinePerDSCUnit)) || NumberOfDSCSlices > 8)))) { - if (TotalNumberOfActiveDPP + 4 <= MaxNumDPP) { - *ODMMode = dml_odm_mode_combine_4to1; - *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithODMCombineFourToOne; - *NumberOfDPP = 4; - } else { - *TotalAvailablePipesSupport = false; - } - } else if (Output != dml_hdmi && (ODMUse == dml_odm_use_policy_combine_2to1 || (ODMUse == dml_odm_use_policy_combine_as_needed && - ((SurfaceRequiredDISPCLKWithoutODMCombine > StateDispclk && SurfaceRequiredDISPCLKWithODMCombineTwoToOne <= StateDispclk) || - (DSCEnable && (HActive > MaximumPixelsPerLinePerDSCUnit)) || (NumberOfDSCSlices <= 8 && NumberOfDSCSlices > 4))))) { - if (TotalNumberOfActiveDPP + 2 <= MaxNumDPP) { - *ODMMode = dml_odm_mode_combine_2to1; - *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithODMCombineTwoToOne; - *NumberOfDPP = 2; - } else { - *TotalAvailablePipesSupport = false; - } - } else { - if (TotalNumberOfActiveDPP + 1 <= MaxNumDPP) { - *NumberOfDPP = 1; - } else { - *TotalAvailablePipesSupport = false; - } - } -} - -/// @brief Calculate the required DISPCLK given the odm mode and pixclk -static dml_float_t CalculateRequiredDispclk( - enum dml_odm_mode ODMMode, - dml_float_t PixelClock, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKRampingMargin, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_float_t MaxDispclk) -{ - dml_float_t RequiredDispclk = 0.; - dml_float_t PixelClockAfterODM; - - dml_float_t DISPCLKWithRampingRoundedToDFSGranularity; - dml_float_t DISPCLKWithoutRampingRoundedToDFSGranularity; - dml_float_t MaxDispclkRoundedDownToDFSGranularity; - - if (ODMMode == dml_odm_mode_combine_4to1) { - PixelClockAfterODM = PixelClock / 4; - } else if (ODMMode == dml_odm_mode_combine_2to1) { - PixelClockAfterODM = PixelClock / 2; - } else { - PixelClockAfterODM = PixelClock; - } - - DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularity(PixelClockAfterODM * (1.0 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * (1 + DISPCLKRampingMargin / 100.0), 1, DISPCLKDPPCLKVCOSpeed); - DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularity(PixelClockAfterODM * (1.0 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0), 1, DISPCLKDPPCLKVCOSpeed); - MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularity(MaxDispclk, 0, DISPCLKDPPCLKVCOSpeed); - - if (DISPCLKWithoutRampingRoundedToDFSGranularity > MaxDispclkRoundedDownToDFSGranularity) { - RequiredDispclk = DISPCLKWithoutRampingRoundedToDFSGranularity; - } else if (DISPCLKWithRampingRoundedToDFSGranularity > MaxDispclkRoundedDownToDFSGranularity) { - RequiredDispclk = MaxDispclkRoundedDownToDFSGranularity; - } else { - RequiredDispclk = DISPCLKWithRampingRoundedToDFSGranularity; - } - - return RequiredDispclk; -} - -/// @brief Determine DPPCLK if there only one DPP per plane, main factor is the pixel rate and DPP scaling parameter -static void CalculateSinglePipeDPPCLKAndSCLThroughput( - dml_float_t HRatio, - dml_float_t HRatioChroma, - dml_float_t VRatio, - dml_float_t VRatioChroma, - dml_float_t MaxDCHUBToPSCLThroughput, - dml_float_t MaxPSCLToLBThroughput, - dml_float_t PixelClock, - enum dml_source_format_class SourcePixelFormat, - dml_uint_t HTaps, - dml_uint_t HTapsChroma, - dml_uint_t VTaps, - dml_uint_t VTapsChroma, - - // Output - dml_float_t *PSCL_THROUGHPUT, - dml_float_t *PSCL_THROUGHPUT_CHROMA, - dml_float_t *DPPCLKUsingSingleDPP) -{ - dml_float_t DPPCLKUsingSingleDPPLuma; - dml_float_t DPPCLKUsingSingleDPPChroma; - - if (HRatio > 1) { - *PSCL_THROUGHPUT = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatio / dml_ceil((dml_float_t) HTaps / 6.0, 1.0)); - } else { - *PSCL_THROUGHPUT = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); - } - - DPPCLKUsingSingleDPPLuma = PixelClock * dml_max3(VTaps / 6 * dml_min(1, HRatio), HRatio * VRatio / *PSCL_THROUGHPUT, 1); - - if ((HTaps > 6 || VTaps > 6) && DPPCLKUsingSingleDPPLuma < 2 * PixelClock) - DPPCLKUsingSingleDPPLuma = 2 * PixelClock; - - if ((SourcePixelFormat != dml_420_8 && SourcePixelFormat != dml_420_10 && SourcePixelFormat != dml_420_12 && SourcePixelFormat != dml_rgbe_alpha)) { - *PSCL_THROUGHPUT_CHROMA = 0; - *DPPCLKUsingSingleDPP = DPPCLKUsingSingleDPPLuma; - } else { - if (HRatioChroma > 1) { - *PSCL_THROUGHPUT_CHROMA = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatioChroma / dml_ceil((dml_float_t) HTapsChroma / 6.0, 1.0)); - } else { - *PSCL_THROUGHPUT_CHROMA = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); - } - DPPCLKUsingSingleDPPChroma = PixelClock * dml_max3(VTapsChroma / 6 * dml_min(1, HRatioChroma), - HRatioChroma * VRatioChroma / *PSCL_THROUGHPUT_CHROMA, 1); - if ((HTapsChroma > 6 || VTapsChroma > 6) && DPPCLKUsingSingleDPPChroma < 2 * PixelClock) - DPPCLKUsingSingleDPPChroma = 2 * PixelClock; - *DPPCLKUsingSingleDPP = dml_max(DPPCLKUsingSingleDPPLuma, DPPCLKUsingSingleDPPChroma); - } -} - -/// @brief Calculate the actual dppclk freq -/// @param DPPCLKUsingSingleDPP DppClk freq required if there is only 1 DPP per plane -/// @param DPPPerSurface Number of DPP for each plane -static void CalculateDPPCLK( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, - dml_float_t DISPCLKDPPCLKVCOSpeed, - dml_float_t DPPCLKUsingSingleDPP[], - dml_uint_t DPPPerSurface[], - - // Output - dml_float_t *GlobalDPPCLK, - dml_float_t Dppclk[]) -{ - *GlobalDPPCLK = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - Dppclk[k] = DPPCLKUsingSingleDPP[k] / DPPPerSurface[k] * (1 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); - *GlobalDPPCLK = dml_max(*GlobalDPPCLK, Dppclk[k]); - } - *GlobalDPPCLK = RoundToDFSGranularity(*GlobalDPPCLK, 1, DISPCLKDPPCLKVCOSpeed); - - dml_print("DML::%s: GlobalDPPCLK = %f\n", __func__, *GlobalDPPCLK); - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - Dppclk[k] = *GlobalDPPCLK / 255.0 * dml_ceil(Dppclk[k] * 255.0 / *GlobalDPPCLK, 1.0); - dml_print("DML::%s: Dppclk[%0d] = %f\n", __func__, k, Dppclk[k]); - } -} - -static void CalculateMALLUseForStaticScreen( - dml_uint_t NumberOfActiveSurfaces, - dml_uint_t MALLAllocatedForDCNFinal, - enum dml_use_mall_for_static_screen_mode *UseMALLForStaticScreen, - dml_uint_t SurfaceSizeInMALL[], - dml_bool_t one_row_per_frame_fits_in_buffer[], - - // Output - dml_bool_t UsesMALLForStaticScreen[]) -{ - - dml_uint_t SurfaceToAddToMALL; - dml_bool_t CanAddAnotherSurfaceToMALL; - dml_uint_t TotalSurfaceSizeInMALL; - - TotalSurfaceSizeInMALL = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - UsesMALLForStaticScreen[k] = (UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable); - if (UsesMALLForStaticScreen[k]) - TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, UsesMALLForStaticScreen = %u\n", __func__, k, UsesMALLForStaticScreen[k]); - dml_print("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); -#endif - } - - SurfaceToAddToMALL = 0; - CanAddAnotherSurfaceToMALL = true; - while (CanAddAnotherSurfaceToMALL) { - CanAddAnotherSurfaceToMALL = false; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k] <= MALLAllocatedForDCNFinal * 1024 * 1024 && - !UsesMALLForStaticScreen[k] && UseMALLForStaticScreen[k] != dml_use_mall_static_screen_disable && one_row_per_frame_fits_in_buffer[k] && - (!CanAddAnotherSurfaceToMALL || SurfaceSizeInMALL[k] < SurfaceSizeInMALL[SurfaceToAddToMALL])) { - CanAddAnotherSurfaceToMALL = true; - SurfaceToAddToMALL = k; - dml_print("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, UseMALLForStaticScreen[k]); - } - } - if (CanAddAnotherSurfaceToMALL) { - UsesMALLForStaticScreen[SurfaceToAddToMALL] = true; - TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[SurfaceToAddToMALL]; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); - dml_print("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); -#endif - } - } -} - -// @brief Calculate return bw for VM only traffic -dml_float_t dml_get_return_bw_mbps_vm_only( - const struct soc_bounding_box_st *soc, - dml_bool_t use_ideal_dram_bw_strobe, - dml_bool_t HostVMEnable, - dml_float_t DCFCLK, - dml_float_t FabricClock, - dml_float_t DRAMSpeed) -{ - dml_float_t VMDataOnlyReturnBW = - dml_min3(soc->return_bus_width_bytes * DCFCLK * soc->pct_ideal_sdp_bw_after_urgent / 100.0, - FabricClock * soc->fabric_datapath_to_dcn_data_return_bytes * soc->pct_ideal_sdp_bw_after_urgent / 100.0, - DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes * - ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : soc->pct_ideal_dram_bw_after_urgent_vm_only) / 100.0); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); - dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); - dml_print("DML::%s: DCFCLK = %f\n", __func__, DCFCLK); - dml_print("DML::%s: FabricClock = %f\n", __func__, FabricClock); - dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); - dml_print("DML::%s: VMDataOnlyReturnBW = %f\n", __func__, VMDataOnlyReturnBW); -#endif - return VMDataOnlyReturnBW; -} - -// Function: dml_get_return_bw_mbps -// Megabyte per second -dml_float_t dml_get_return_bw_mbps( - const struct soc_bounding_box_st *soc, - dml_bool_t use_ideal_dram_bw_strobe, - dml_bool_t HostVMEnable, - dml_float_t DCFCLK, - dml_float_t FabricClock, - dml_float_t DRAMSpeed) -{ - dml_float_t ReturnBW = 0.; - dml_float_t IdealSDPPortBandwidth = soc->return_bus_width_bytes * DCFCLK; - dml_float_t IdealFabricBandwidth = FabricClock * soc->fabric_datapath_to_dcn_data_return_bytes; - dml_float_t IdealDRAMBandwidth = DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes; - dml_float_t PixelDataOnlyReturnBW = dml_min3(IdealSDPPortBandwidth * soc->pct_ideal_sdp_bw_after_urgent / 100, - IdealFabricBandwidth * soc->pct_ideal_fabric_bw_after_urgent / 100, - IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : - soc->pct_ideal_dram_bw_after_urgent_pixel_only) / 100); - dml_float_t PixelMixedWithVMDataReturnBW = dml_min3(IdealSDPPortBandwidth * soc->pct_ideal_sdp_bw_after_urgent / 100, - IdealFabricBandwidth * soc->pct_ideal_fabric_bw_after_urgent / 100, - IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : - soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm) / 100); - - if (HostVMEnable != true) { - ReturnBW = PixelDataOnlyReturnBW; - } else { - ReturnBW = PixelMixedWithVMDataReturnBW; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); - dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); - dml_print("DML::%s: DCFCLK = %f\n", __func__, DCFCLK); - dml_print("DML::%s: FabricClock = %f\n", __func__, FabricClock); - dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); - dml_print("DML::%s: IdealSDPPortBandwidth = %f\n", __func__, IdealSDPPortBandwidth); - dml_print("DML::%s: IdealFabricBandwidth = %f\n", __func__, IdealFabricBandwidth); - dml_print("DML::%s: IdealDRAMBandwidth = %f\n", __func__, IdealDRAMBandwidth); - dml_print("DML::%s: PixelDataOnlyReturnBW = %f\n", __func__, PixelDataOnlyReturnBW); - dml_print("DML::%s: PixelMixedWithVMDataReturnBW = %f\n", __func__, PixelMixedWithVMDataReturnBW); - dml_print("DML::%s: ReturnBW = %f MBps\n", __func__, ReturnBW); -#endif - return ReturnBW; -} - -// Function: dml_get_return_dram_bw_mbps -// Megabyte per second -static dml_float_t dml_get_return_dram_bw_mbps( - const struct soc_bounding_box_st *soc, - dml_bool_t use_ideal_dram_bw_strobe, - dml_bool_t HostVMEnable, - dml_float_t DRAMSpeed) -{ - dml_float_t ReturnDRAMBW = 0.; - dml_float_t IdealDRAMBandwidth = DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes; - dml_float_t PixelDataOnlyReturnBW = IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : - soc->pct_ideal_dram_bw_after_urgent_pixel_only) / 100; - dml_float_t PixelMixedWithVMDataReturnBW = IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : - soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm) / 100; - - if (HostVMEnable != true) { - ReturnDRAMBW = PixelDataOnlyReturnBW; - } else { - ReturnDRAMBW = PixelMixedWithVMDataReturnBW; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); - dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); - dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); - dml_print("DML::%s: IdealDRAMBandwidth = %f\n", __func__, IdealDRAMBandwidth); - dml_print("DML::%s: PixelDataOnlyReturnBW = %f\n", __func__, PixelDataOnlyReturnBW); - dml_print("DML::%s: PixelMixedWithVMDataReturnBW = %f\n", __func__, PixelMixedWithVMDataReturnBW); - dml_print("DML::%s: ReturnDRAMBW = %f MBps\n", __func__, ReturnDRAMBW); -#endif - return ReturnDRAMBW; -} - -/// @brief BACKEND -static dml_uint_t DSCDelayRequirement( - dml_bool_t DSCEnabled, - enum dml_odm_mode ODMMode, - dml_uint_t DSCInputBitPerComponent, - dml_float_t OutputBpp, - dml_uint_t HActive, - dml_uint_t HTotal, - dml_uint_t NumberOfDSCSlices, - enum dml_output_format_class OutputFormat, - enum dml_output_encoder_class Output, - dml_float_t PixelClock, - dml_float_t PixelClockBackEnd) -{ - dml_uint_t DSCDelayRequirement_val = 0; - - if (DSCEnabled == true && OutputBpp != 0) { - if (ODMMode == dml_odm_mode_combine_4to1) { - DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)(dml_ceil((dml_float_t) HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), - (dml_uint_t) (NumberOfDSCSlices / 4.0), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); - } else if (ODMMode == dml_odm_mode_combine_2to1) { - DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)(dml_ceil((dml_float_t) HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), - (dml_uint_t) (NumberOfDSCSlices / 2.0), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); - } else { - DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)((dml_float_t) dml_ceil(HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), - NumberOfDSCSlices, OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); - } - DSCDelayRequirement_val = (dml_uint_t)(DSCDelayRequirement_val + (HTotal - HActive) * dml_ceil((dml_float_t) DSCDelayRequirement_val / (dml_float_t) HActive, 1.0)); - DSCDelayRequirement_val = (dml_uint_t)(DSCDelayRequirement_val * PixelClock / PixelClockBackEnd); - - } else { - DSCDelayRequirement_val = 0; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DSCEnabled = %u\n", __func__, DSCEnabled); - dml_print("DML::%s: ODMMode = %u\n", __func__, ODMMode); - dml_print("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); - dml_print("DML::%s: HActive = %u\n", __func__, HActive); - dml_print("DML::%s: HTotal = %u\n", __func__, HTotal); - dml_print("DML::%s: PixelClock = %f\n", __func__, PixelClock); - dml_print("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); - dml_print("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); - dml_print("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); - dml_print("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); - dml_print("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); -#endif - - return DSCDelayRequirement_val; -} - -static noinline_for_stack dml_bool_t CalculateVActiveBandwithSupport(dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - dml_bool_t NotUrgentLatencyHiding[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[]) -{ - dml_bool_t NotEnoughUrgentLatencyHiding = false; - dml_bool_t CalculateVActiveBandwithSupport_val = false; - dml_float_t VActiveBandwith = 0; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (NotUrgentLatencyHiding[k]) { - NotEnoughUrgentLatencyHiding = true; - } - } - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - VActiveBandwith = VActiveBandwith + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * meta_row_bandwidth[k] + NumberOfDPP[k] * dpte_row_bandwidth[k]; - } - - CalculateVActiveBandwithSupport_val = (VActiveBandwith <= ReturnBW) && !NotEnoughUrgentLatencyHiding; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: NotEnoughUrgentLatencyHiding = %u\n", __func__, NotEnoughUrgentLatencyHiding); - dml_print("DML::%s: VActiveBandwith = %f\n", __func__, VActiveBandwith); - dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); - dml_print("DML::%s: CalculateVActiveBandwithSupport_val = %u\n", __func__, CalculateVActiveBandwithSupport_val); -#endif - return CalculateVActiveBandwithSupport_val; -} - -static void CalculatePrefetchBandwithSupport( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - dml_bool_t NotUrgentLatencyHiding[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_float_t cursor_bw_pre[], - dml_float_t prefetch_vmrow_bw[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[], - - // Output - dml_float_t *PrefetchBandwidth, - dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch, - dml_float_t *FractionOfUrgentBandwidth, - dml_bool_t *PrefetchBandwidthSupport) -{ - dml_bool_t NotEnoughUrgentLatencyHiding = false; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (NotUrgentLatencyHiding[k]) { - NotEnoughUrgentLatencyHiding = true; - } - } - - *PrefetchBandwidth = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - *PrefetchBandwidth = *PrefetchBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]), - NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - } - - *PrefetchBandwidthNotIncludingMALLPrefetch = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) - *PrefetchBandwidthNotIncludingMALLPrefetch = *PrefetchBandwidthNotIncludingMALLPrefetch - + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] - + cursor_bw[k] * UrgentBurstFactorCursor[k] - + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]), - NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] - + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) - + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - } - - *PrefetchBandwidthSupport = (*PrefetchBandwidth <= ReturnBW) && !NotEnoughUrgentLatencyHiding; - *FractionOfUrgentBandwidth = *PrefetchBandwidth / ReturnBW; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); - dml_print("DML::%s: PrefetchBandwidth = %f\n", __func__, *PrefetchBandwidth); - dml_print("DML::%s: FractionOfUrgentBandwidth = %f\n", __func__, *FractionOfUrgentBandwidth); - dml_print("DML::%s: PrefetchBandwidthSupport = %u\n", __func__, *PrefetchBandwidthSupport); -#endif -} - -static noinline_for_stack dml_float_t CalculateBandwidthAvailableForImmediateFlip( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t cursor_bw_pre[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[]) -{ - dml_float_t ret_val = ReturnBW; - - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - ret_val = ret_val - dml_max(ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], - NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + - cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u\n", __func__, k); - dml_print("DML::%s: NumberOfDPP = %u\n", __func__, NumberOfDPP[k]); - dml_print("DML::%s: ReadBandwidthLuma = %f\n", __func__, ReadBandwidthLuma[k]); - dml_print("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, UrgentBurstFactorLuma[k]); - dml_print("DML::%s: ReadBandwidthChroma = %f\n", __func__, ReadBandwidthChroma[k]); - dml_print("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, UrgentBurstFactorChroma[k]); - dml_print("DML::%s: cursor_bw = %f\n", __func__, cursor_bw[k]); - dml_print("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, UrgentBurstFactorCursor[k]); - - dml_print("DML::%s: PrefetchBandwidthLuma = %f\n", __func__, PrefetchBandwidthLuma[k]); - dml_print("DML::%s: UrgentBurstFactorLumaPre = %f\n", __func__, UrgentBurstFactorLumaPre[k]); - dml_print("DML::%s: PrefetchBandwidthChroma = %f\n", __func__, PrefetchBandwidthChroma[k]); - dml_print("DML::%s: UrgentBurstFactorChromaPre = %f\n", __func__, UrgentBurstFactorChromaPre[k]); - dml_print("DML::%s: cursor_bw_pre = %f\n", __func__, cursor_bw_pre[k]); - dml_print("DML::%s: UrgentBurstFactorCursorPre = %f\n", __func__, UrgentBurstFactorCursorPre[k]); - dml_print("DML::%s: ret_val = %f\n", __func__, ret_val); -#endif - } - - return ret_val; -} - -static void CalculateImmediateFlipBandwithSupport( - dml_uint_t NumberOfActiveSurfaces, - dml_float_t ReturnBW, - enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], - enum dml_immediate_flip_requirement ImmediateFlipRequirement[], - dml_float_t final_flip_bw[], - dml_float_t ReadBandwidthLuma[], - dml_float_t ReadBandwidthChroma[], - dml_float_t PrefetchBandwidthLuma[], - dml_float_t PrefetchBandwidthChroma[], - dml_float_t cursor_bw[], - dml_float_t meta_row_bandwidth[], - dml_float_t dpte_row_bandwidth[], - dml_float_t cursor_bw_pre[], - dml_float_t prefetch_vmrow_bw[], - dml_uint_t NumberOfDPP[], - dml_float_t UrgentBurstFactorLuma[], - dml_float_t UrgentBurstFactorChroma[], - dml_float_t UrgentBurstFactorCursor[], - dml_float_t UrgentBurstFactorLumaPre[], - dml_float_t UrgentBurstFactorChromaPre[], - dml_float_t UrgentBurstFactorCursorPre[], - - // Output - dml_float_t *TotalBandwidth, - dml_float_t *TotalBandwidthNotIncludingMALLPrefetch, - dml_float_t *FractionOfUrgentBandwidth, - dml_bool_t *ImmediateFlipBandwidthSupport) -{ - *TotalBandwidth = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) { - - - - *TotalBandwidth = *TotalBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - NumberOfDPP[k] * final_flip_bw[k] + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], - NumberOfDPP[k] * (final_flip_bw[k] + PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - } else { - *TotalBandwidth = *TotalBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]) + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], - NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k = %u\n", __func__, k); - dml_print("DML::%s: ImmediateFlipRequirement = %u\n", __func__, ImmediateFlipRequirement[k]); - dml_print("DML::%s: TotalBandwidth = %f\n", __func__, *TotalBandwidth); - dml_print("DML::%s: NumberOfDPP = %u\n", __func__, NumberOfDPP[k]); - dml_print("DML::%s: prefetch_vmrow_bw = %f\n", __func__, prefetch_vmrow_bw[k]); - dml_print("DML::%s: final_flip_bw = %f\n", __func__, final_flip_bw[k]); - dml_print("DML::%s: ReadBandwidthLuma = %f\n", __func__, ReadBandwidthLuma[k]); - dml_print("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, UrgentBurstFactorLuma[k]); - dml_print("DML::%s: ReadBandwidthChroma = %f\n", __func__, ReadBandwidthChroma[k]); - dml_print("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, UrgentBurstFactorChroma[k]); - dml_print("DML::%s: cursor_bw = %f\n", __func__, cursor_bw[k]); - dml_print("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, UrgentBurstFactorCursor[k]); - dml_print("DML::%s: PrefetchBandwidthLuma = %f\n", __func__, PrefetchBandwidthLuma[k]); - dml_print("DML::%s: UrgentBurstFactorLumaPre = %f\n", __func__, UrgentBurstFactorLumaPre[k]); - dml_print("DML::%s: PrefetchBandwidthChroma = %f\n", __func__, PrefetchBandwidthChroma[k]); - dml_print("DML::%s: UrgentBurstFactorChromaPre = %f\n", __func__, UrgentBurstFactorChromaPre[k]); - dml_print("DML::%s: cursor_bw_pre = %f\n", __func__, cursor_bw_pre[k]); - dml_print("DML::%s: UrgentBurstFactorCursorPre = %f\n", __func__, UrgentBurstFactorCursorPre[k]); - dml_print("DML::%s: meta_row_bandwidth = %f\n", __func__, meta_row_bandwidth[k]); - dml_print("DML::%s: dpte_row_bandwidth = %f\n", __func__, dpte_row_bandwidth[k]); -#endif - } - - *TotalBandwidthNotIncludingMALLPrefetch = 0; - for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { - if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { - if (ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) - *TotalBandwidthNotIncludingMALLPrefetch = *TotalBandwidthNotIncludingMALLPrefetch + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - NumberOfDPP[k] * final_flip_bw[k] + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], - NumberOfDPP[k] * (final_flip_bw[k] + PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) - + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - else - *TotalBandwidthNotIncludingMALLPrefetch = *TotalBandwidthNotIncludingMALLPrefetch + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], - NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]) - + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], - NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) - + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); - } - } - - *ImmediateFlipBandwidthSupport = (*TotalBandwidth <= ReturnBW); - *FractionOfUrgentBandwidth = *TotalBandwidth / ReturnBW; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); - dml_print("DML::%s: TotalBandwidth = %f\n", __func__, *TotalBandwidth); - dml_print("DML::%s: ImmediateFlipBandwidthSupport = %u\n", __func__, *ImmediateFlipBandwidthSupport); -#endif -} - -static dml_uint_t MicroSecToVertLines(dml_uint_t num_us, dml_uint_t h_total, dml_float_t pixel_clock) -{ - dml_uint_t lines_time_in_ns = 1000.0 * (h_total * 1000.0) / (pixel_clock * 1000.0); - - return dml_ceil(1000.0 * num_us / lines_time_in_ns, 1.0); -} - -/// @brief Calculate the maximum vstartup for mode support and mode programming consideration -/// Bounded by min of actual vblank and input vblank_nom, dont want vstartup/ready to start too early if actual vbllank is huge -static dml_uint_t CalculateMaxVStartup( - dml_uint_t plane_idx, - dml_bool_t ptoi_supported, - dml_uint_t vblank_nom_default_us, - struct dml_timing_cfg_st *timing, - dml_float_t write_back_delay_us) -{ - dml_uint_t vblank_size = 0; - dml_uint_t max_vstartup_lines = 0; - const dml_uint_t max_allowed_vblank_nom = 1023; - - dml_float_t line_time_us = (dml_float_t) timing->HTotal[plane_idx] / timing->PixelClock[plane_idx]; - dml_uint_t vblank_actual = timing->VTotal[plane_idx] - timing->VActive[plane_idx]; - - dml_uint_t vblank_nom_default_in_line = MicroSecToVertLines(vblank_nom_default_us, timing->HTotal[plane_idx], - timing->PixelClock[plane_idx]); - dml_uint_t vblank_nom_input = (dml_uint_t)dml_min(vblank_actual, vblank_nom_default_in_line); - - // vblank_nom should not be smaller than (VSync (VTotal - VActive - VFrontPorch) + 2) - // + 2 is because - // 1 -> VStartup_start should be 1 line before VSync - // 1 -> always reserve 1 line between start of VBlank to VStartup signal - dml_uint_t vblank_nom_vsync_capped = dml_max(vblank_nom_input, - timing->VTotal[plane_idx] - timing->VActive[plane_idx] - timing->VFrontPorch[plane_idx] + 2); - dml_uint_t vblank_nom_max_allowed_capped = dml_min(vblank_nom_vsync_capped, max_allowed_vblank_nom); - dml_uint_t vblank_avail = (vblank_nom_max_allowed_capped == 0) ? - vblank_nom_default_in_line : vblank_nom_max_allowed_capped; - - vblank_size = (dml_uint_t) dml_min(vblank_actual, vblank_avail); - - if (timing->Interlace[plane_idx] && !ptoi_supported) - max_vstartup_lines = (dml_uint_t) (dml_floor(vblank_size/2.0, 1.0)); - else - max_vstartup_lines = vblank_size - (dml_uint_t) dml_max(1.0, dml_ceil(write_back_delay_us/line_time_us, 1.0)); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: plane_idx = %u\n", __func__, plane_idx); - dml_print("DML::%s: VBlankNom = %u\n", __func__, timing->VBlankNom[plane_idx]); - dml_print("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); - dml_print("DML::%s: line_time_us = %f\n", __func__, line_time_us); - dml_print("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); - dml_print("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); - dml_print("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); -#endif - max_vstartup_lines = (dml_uint_t) dml_min(max_vstartup_lines, DML_MAX_VSTARTUP_START); - return max_vstartup_lines; -} - -static noinline_for_stack void set_calculate_prefetch_schedule_params(struct display_mode_lib_st *mode_lib, - struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params, - dml_uint_t j, - dml_uint_t k) -{ - CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelayPerState[k]; - CalculatePrefetchSchedule_params->EnhancedPrefetchScheduleAccelerationFinal = mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal; - CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ms.ip.dppclk_delay_subtotal + mode_lib->ms.ip.dppclk_delay_cnvc_formatter; - CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ms.ip.dppclk_delay_scl; - CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ms.ip.dppclk_delay_scl_lb_only; - CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ms.ip.dppclk_delay_cnvc_cursor; - CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ms.ip.dispclk_delay_subtotal; - CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (dml_uint_t)(mode_lib->ms.SwathWidthYThisState[k] / mode_lib->ms.cache_display_cfg.plane.HRatio[k]); - CalculatePrefetchSchedule_params->OutputFormat = mode_lib->ms.cache_display_cfg.output.OutputFormat[k]; - CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; - CalculatePrefetchSchedule_params->GPUVMPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - CalculatePrefetchSchedule_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - CalculatePrefetchSchedule_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - CalculatePrefetchSchedule_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - CalculatePrefetchSchedule_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]; - CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; - CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired[k]; - CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes[k]; - CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; - CalculatePrefetchSchedule_params->UrgentExtraLatency = mode_lib->ms.ExtraLatency; - CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; - CalculatePrefetchSchedule_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k]; - CalculatePrefetchSchedule_params->MetaRowByte = mode_lib->ms.MetaRowBytes[j][k]; - CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[j][k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[j][k]; - CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; - CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[j][k]; - CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; - CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC[k]; - CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub_this_state[k]; - CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub_this_state[k]; - CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState[k]; - CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState[k]; - CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait; - CalculatePrefetchSchedule_params->DestinationLinesForPrefetch = &mode_lib->ms.LineTimesForPrefetch[k]; - CalculatePrefetchSchedule_params->DestinationLinesToRequestVMInVBlank = &mode_lib->ms.LinesForMetaPTE[k]; - CalculatePrefetchSchedule_params->DestinationLinesToRequestRowInVBlank = &mode_lib->ms.LinesForMetaAndDPTERow[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[j][k]; - CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[j][k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; - CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.support.NoTimeForDynamicMetadata[j][k]; - CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; -} - -static noinline_for_stack void dml_prefetch_check(struct display_mode_lib_st *mode_lib) -{ - struct dml_core_mode_support_locals_st *s = &mode_lib->scratch.dml_core_mode_support_locals; - struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; - struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; - struct DmlPipe *myPipe; - dml_uint_t j, k; - - for (j = 0; j < 2; ++j) { - mode_lib->ms.TimeCalc = 24 / mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; - mode_lib->ms.swath_width_luma_ub_this_state[k] = mode_lib->ms.swath_width_luma_ub_all_states[j][k]; - mode_lib->ms.swath_width_chroma_ub_this_state[k] = mode_lib->ms.swath_width_chroma_ub_all_states[j][k]; - mode_lib->ms.SwathWidthYThisState[k] = mode_lib->ms.SwathWidthYAllStates[j][k]; - mode_lib->ms.SwathWidthCThisState[k] = mode_lib->ms.SwathWidthCAllStates[j][k]; - mode_lib->ms.SwathHeightYThisState[k] = mode_lib->ms.SwathHeightYAllStates[j][k]; - mode_lib->ms.SwathHeightCThisState[k] = mode_lib->ms.SwathHeightCAllStates[j][k]; - mode_lib->ms.UnboundedRequestEnabledThisState = mode_lib->ms.UnboundedRequestEnabledAllStates[j]; - mode_lib->ms.CompressedBufferSizeInkByteThisState = mode_lib->ms.CompressedBufferSizeInkByteAllStates[j]; - mode_lib->ms.DETBufferSizeInKByteThisState[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[j][k]; - mode_lib->ms.DETBufferSizeYThisState[k] = mode_lib->ms.DETBufferSizeYAllStates[j][k]; - mode_lib->ms.DETBufferSizeCThisState[k] = mode_lib->ms.DETBufferSizeCAllStates[j][k]; - } - - mode_lib->ms.support.VActiveBandwithSupport[j] = CalculateVActiveBandwithSupport( - mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBWPerState[j], - mode_lib->ms.NotUrgentLatencyHiding, - mode_lib->ms.ReadBandwidthLuma, - mode_lib->ms.ReadBandwidthChroma, - mode_lib->ms.cursor_bw, - mode_lib->ms.meta_row_bandwidth_this_state, - mode_lib->ms.dpte_row_bandwidth_this_state, - mode_lib->ms.NoOfDPPThisState, - mode_lib->ms.UrgentBurstFactorLuma[j], - mode_lib->ms.UrgentBurstFactorChroma[j], - mode_lib->ms.UrgentBurstFactorCursor[j]); - - s->VMDataOnlyReturnBWPerState = dml_get_return_bw_mbps_vm_only( - &mode_lib->ms.soc, - mode_lib->ms.state.use_ideal_dram_bw_strobe, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.DCFCLKState[j], - mode_lib->ms.state.fabricclk_mhz, - mode_lib->ms.state.dram_speed_mts); - - s->HostVMInefficiencyFactor = 1; - if (mode_lib->ms.cache_display_cfg.plane.GPUVMEnable && mode_lib->ms.cache_display_cfg.plane.HostVMEnable) - s->HostVMInefficiencyFactor = mode_lib->ms.ReturnBWPerState[j] / s->VMDataOnlyReturnBWPerState; - - mode_lib->ms.ExtraLatency = CalculateExtraLatency( - mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles, - s->ReorderingBytes, - mode_lib->ms.DCFCLKState[j], - mode_lib->ms.TotalNumberOfActiveDPP[j], - mode_lib->ms.ip.pixel_chunk_size_kbytes, - mode_lib->ms.TotalNumberOfDCCActiveDPP[j], - mode_lib->ms.ip.meta_chunk_size_kbytes, - mode_lib->ms.ReturnBWPerState[j], - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.num_active_planes, - mode_lib->ms.NoOfDPPThisState, - mode_lib->ms.dpte_group_bytes, - s->HostVMInefficiencyFactor, - mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, - mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels); - - s->NextMaxVStartup = s->MaxVStartupAllPlanes[j]; - s->MaxVStartup = 0; - s->AllPrefetchModeTested = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculatePrefetchMode(mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k], &s->MinPrefetchMode[k], &s->MaxPrefetchMode[k]); - s->NextPrefetchMode[k] = s->MinPrefetchMode[k]; - } - - do { - s->MaxVStartup = s->NextMaxVStartup; - s->AllPrefetchModeTested = true; - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.PrefetchMode[k] = s->NextPrefetchMode[k]; - mode_lib->ms.TWait = CalculateTWait( - mode_lib->ms.PrefetchMode[k], - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal, - mode_lib->ms.cache_display_cfg.timing.DRRDisplay[k], - mode_lib->ms.state.dram_clock_change_latency_us, - mode_lib->ms.state.fclk_change_latency_us, - mode_lib->ms.UrgLatency, - mode_lib->ms.state.sr_enter_plus_exit_time_us); - - myPipe = &s->myPipe; - myPipe->Dppclk = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; - myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK[j]; - myPipe->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; - myPipe->DCFClkDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; - myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[j][k]; - myPipe->ScalerEnabled = mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k]; - myPipe->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; - myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; - myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; - myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; - myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; - myPipe->InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; - myPipe->NumberOfCursors = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k]; - myPipe->VBlank = mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]; - myPipe->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; - myPipe->HActive = mode_lib->ms.cache_display_cfg.timing.HActive[k]; - myPipe->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; - myPipe->ODMMode = mode_lib->ms.ODMModePerState[k]; - myPipe->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; - myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; - myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; - myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Calling CalculatePrefetchSchedule for j=%u, k=%u\n", __func__, j, k); - dml_print("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[j][k]); - dml_print("DML::%s: MaxVStartup = %u\n", __func__, s->MaxVStartup); - dml_print("DML::%s: NextPrefetchMode = %u\n", __func__, s->NextPrefetchMode[k]); - dml_print("DML::%s: AllowForPStateChangeOrStutterInVBlank = %u\n", __func__, mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k]); - dml_print("DML::%s: PrefetchMode = %u\n", __func__, mode_lib->ms.PrefetchMode[k]); -#endif - - CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactor; - CalculatePrefetchSchedule_params->myPipe = myPipe; - CalculatePrefetchSchedule_params->VStartup = (dml_uint_t)(dml_min(s->MaxVStartup, s->MaximumVStartup[j][k])); - CalculatePrefetchSchedule_params->MaxVStartup = s->MaximumVStartup[j][k]; - CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; - CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; - CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; - CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; - CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; - CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; - CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; - CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; - CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; - - set_calculate_prefetch_schedule_params(mode_lib, CalculatePrefetchSchedule_params, j, k); - - mode_lib->ms.support.NoTimeForPrefetch[j][k] = - CalculatePrefetchSchedule(&mode_lib->scratch, - CalculatePrefetchSchedule_params); - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateUrgentBurstFactor( - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - mode_lib->ms.swath_width_luma_ub_this_state[k], - mode_lib->ms.swath_width_chroma_ub_this_state[k], - mode_lib->ms.SwathHeightYThisState[k], - mode_lib->ms.SwathHeightCThisState[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.UrgLatency, - mode_lib->ms.ip.cursor_buffer_size, - mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], - mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], - mode_lib->ms.VRatioPreY[j][k], - mode_lib->ms.VRatioPreC[j][k], - mode_lib->ms.BytePerPixelInDETY[k], - mode_lib->ms.BytePerPixelInDETC[k], - mode_lib->ms.DETBufferSizeYThisState[k], - mode_lib->ms.DETBufferSizeCThisState[k], - /* Output */ - &mode_lib->ms.UrgentBurstFactorCursorPre[k], - &mode_lib->ms.UrgentBurstFactorLumaPre[k], - &mode_lib->ms.UrgentBurstFactorChromaPre[k], - &mode_lib->ms.NotUrgentLatencyHidingPre[k]); - - mode_lib->ms.cursor_bw_pre[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * - mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / - mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.VRatioPreY[j][k]; - } - - { - CalculatePrefetchBandwithSupport( - mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBWPerState[j], - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - mode_lib->ms.NotUrgentLatencyHidingPre, - mode_lib->ms.ReadBandwidthLuma, - mode_lib->ms.ReadBandwidthChroma, - mode_lib->ms.RequiredPrefetchPixelDataBWLuma, - mode_lib->ms.RequiredPrefetchPixelDataBWChroma, - mode_lib->ms.cursor_bw, - mode_lib->ms.meta_row_bandwidth_this_state, - mode_lib->ms.dpte_row_bandwidth_this_state, - mode_lib->ms.cursor_bw_pre, - mode_lib->ms.prefetch_vmrow_bw, - mode_lib->ms.NoOfDPPThisState, - mode_lib->ms.UrgentBurstFactorLuma[j], - mode_lib->ms.UrgentBurstFactorChroma[j], - mode_lib->ms.UrgentBurstFactorCursor[j], - mode_lib->ms.UrgentBurstFactorLumaPre, - mode_lib->ms.UrgentBurstFactorChromaPre, - mode_lib->ms.UrgentBurstFactorCursorPre, - - /* output */ - &s->dummy_single[0], // dml_float_t *PrefetchBandwidth - &s->dummy_single[1], // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch - &mode_lib->mp.FractionOfUrgentBandwidth, // dml_float_t *FractionOfUrgentBandwidth - &mode_lib->ms.support.PrefetchSupported[j]); - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.LineTimesForPrefetch[k] < 2.0 - || mode_lib->ms.LinesForMetaPTE[k] >= 32.0 - || mode_lib->ms.LinesForMetaAndDPTERow[k] >= 16.0 - || mode_lib->ms.support.NoTimeForPrefetch[j][k] == true) { - mode_lib->ms.support.PrefetchSupported[j] = false; - } - } - - mode_lib->ms.support.DynamicMetadataSupported[j] = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.support.NoTimeForDynamicMetadata[j][k] == true) { - mode_lib->ms.support.DynamicMetadataSupported[j] = false; - } - } - - mode_lib->ms.support.VRatioInPrefetchSupported[j] = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.support.NoTimeForPrefetch[j][k] == true || - mode_lib->ms.VRatioPreY[j][k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || - mode_lib->ms.VRatioPreC[j][k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || - ((s->MaxVStartup < s->MaximumVStartup[j][k] || mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal == 0) && - (mode_lib->ms.VRatioPreY[j][k] > __DML_MAX_VRATIO_PRE__ || mode_lib->ms.VRatioPreC[j][k] > __DML_MAX_VRATIO_PRE__))) { - mode_lib->ms.support.VRatioInPrefetchSupported[j] = false; - } - } - - s->AnyLinesForVMOrRowTooLarge = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.LinesForMetaAndDPTERow[k] >= 16 || mode_lib->ms.LinesForMetaPTE[k] >= 32) { - s->AnyLinesForVMOrRowTooLarge = true; - } - } - - if (mode_lib->ms.support.PrefetchSupported[j] == true && mode_lib->ms.support.VRatioInPrefetchSupported[j] == true) { - mode_lib->ms.BandwidthAvailableForImmediateFlip = CalculateBandwidthAvailableForImmediateFlip( - mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBWPerState[j], - mode_lib->ms.ReadBandwidthLuma, - mode_lib->ms.ReadBandwidthChroma, - mode_lib->ms.RequiredPrefetchPixelDataBWLuma, - mode_lib->ms.RequiredPrefetchPixelDataBWChroma, - mode_lib->ms.cursor_bw, - mode_lib->ms.cursor_bw_pre, - mode_lib->ms.NoOfDPPThisState, - mode_lib->ms.UrgentBurstFactorLuma[j], - mode_lib->ms.UrgentBurstFactorChroma[j], - mode_lib->ms.UrgentBurstFactorCursor[j], - mode_lib->ms.UrgentBurstFactorLumaPre, - mode_lib->ms.UrgentBurstFactorChromaPre, - mode_lib->ms.UrgentBurstFactorCursorPre); - - mode_lib->ms.TotImmediateFlipBytes = 0; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (!(mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_not_required)) { - mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * (mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k] + mode_lib->ms.MetaRowBytes[j][k]); - if (mode_lib->ms.use_one_row_for_frame_flip[j][k]) { - mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * (2 * mode_lib->ms.DPTEBytesPerRow[j][k]); - } else { - mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * mode_lib->ms.DPTEBytesPerRow[j][k]; - } - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateFlipSchedule( - s->HostVMInefficiencyFactor, - mode_lib->ms.ExtraLatency, - mode_lib->ms.UrgLatency, - mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels, - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, - mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k], - mode_lib->ms.MetaRowBytes[j][k], - mode_lib->ms.DPTEBytesPerRow[j][k], - mode_lib->ms.BandwidthAvailableForImmediateFlip, - mode_lib->ms.TotImmediateFlipBytes, - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - mode_lib->ms.Tno_bw[k], - mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], - mode_lib->ms.dpte_row_height[k], - mode_lib->ms.meta_row_height[k], - mode_lib->ms.dpte_row_height_chroma[k], - mode_lib->ms.meta_row_height_chroma[k], - mode_lib->ms.use_one_row_for_frame_flip[j][k], // 24 - - /* Output */ - &mode_lib->ms.DestinationLinesToRequestVMInImmediateFlip[k], - &mode_lib->ms.DestinationLinesToRequestRowInImmediateFlip[k], - &mode_lib->ms.final_flip_bw[k], - &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); - } - - { - CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBWPerState[j], - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - mode_lib->ms.policy.ImmediateFlipRequirement, - mode_lib->ms.final_flip_bw, - mode_lib->ms.ReadBandwidthLuma, - mode_lib->ms.ReadBandwidthChroma, - mode_lib->ms.RequiredPrefetchPixelDataBWLuma, - mode_lib->ms.RequiredPrefetchPixelDataBWChroma, - mode_lib->ms.cursor_bw, - mode_lib->ms.meta_row_bandwidth_this_state, - mode_lib->ms.dpte_row_bandwidth_this_state, - mode_lib->ms.cursor_bw_pre, - mode_lib->ms.prefetch_vmrow_bw, - mode_lib->ms.NoOfDPP[j], // VBA_ERROR DPPPerSurface is not assigned at this point, should use NoOfDpp here - mode_lib->ms.UrgentBurstFactorLuma[j], - mode_lib->ms.UrgentBurstFactorChroma[j], - mode_lib->ms.UrgentBurstFactorCursor[j], - mode_lib->ms.UrgentBurstFactorLumaPre, - mode_lib->ms.UrgentBurstFactorChromaPre, - mode_lib->ms.UrgentBurstFactorCursorPre, - - /* output */ - &s->dummy_single[0], // dml_float_t *TotalBandwidth - &s->dummy_single[1], // dml_float_t *TotalBandwidthNotIncludingMALLPrefetch - &s->dummy_single[2], // dml_float_t *FractionOfUrgentBandwidth - &mode_lib->ms.support.ImmediateFlipSupportedForState[j]); // dml_bool_t *ImmediateFlipBandwidthSupport - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (!(mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_not_required) && (mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false)) - mode_lib->ms.support.ImmediateFlipSupportedForState[j] = false; - } - - } else { // if prefetch not support, assume iflip not supported - mode_lib->ms.support.ImmediateFlipSupportedForState[j] = false; - } - - if (s->MaxVStartup <= __DML_VBA_MIN_VSTARTUP__ || s->AnyLinesForVMOrRowTooLarge == false) { - s->NextMaxVStartup = s->MaxVStartupAllPlanes[j]; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - s->NextPrefetchMode[k] = s->NextPrefetchMode[k] + 1; - - if (s->NextPrefetchMode[k] <= s->MaxPrefetchMode[k]) - s->AllPrefetchModeTested = false; - } - } else { - s->NextMaxVStartup = s->NextMaxVStartup - 1; - } - } while (!((mode_lib->ms.support.PrefetchSupported[j] == true && mode_lib->ms.support.DynamicMetadataSupported[j] == true && - mode_lib->ms.support.VRatioInPrefetchSupported[j] == true && - // consider flip support is okay if when there is no hostvm and the user does't require a iflip OR the flip bw is ok - // If there is hostvm, DCN needs to support iflip for invalidation - ((s->ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j] == true)) || - (s->NextMaxVStartup == s->MaxVStartupAllPlanes[j] && s->AllPrefetchModeTested))); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.use_one_row_for_frame_this_state[k] = mode_lib->ms.use_one_row_for_frame[j][k]; - } - - s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; - s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; - s->mSOCParameters.WritebackLatency = mode_lib->ms.state.writeback_latency_us; - s->mSOCParameters.DRAMClockChangeLatency = mode_lib->ms.state.dram_clock_change_latency_us; - s->mSOCParameters.FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; - s->mSOCParameters.SRExitTime = mode_lib->ms.state.sr_exit_time_us; - s->mSOCParameters.SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; - s->mSOCParameters.SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; - s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->ms.state.sr_enter_plus_exit_z8_time_us; - s->mSOCParameters.USRRetrainingLatency = mode_lib->ms.state.usr_retraining_latency_us; - s->mSOCParameters.SMNLatency = mode_lib->ms.soc.smn_latency_us; - - CalculateWatermarks_params->USRRetrainingRequiredFinal = mode_lib->ms.policy.USRRetrainingRequiredFinal; - CalculateWatermarks_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateWatermarks_params->PrefetchMode = mode_lib->ms.PrefetchMode; - CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ms.ip.max_line_buffer_lines; - CalculateWatermarks_params->LineBufferSize = mode_lib->ms.ip.line_buffer_size_bits; - CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ms.ip.writeback_interface_buffer_size_kbytes; - CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLKState[j]; - CalculateWatermarks_params->ReturnBW = mode_lib->ms.ReturnBWPerState[j]; - CalculateWatermarks_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; - CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; - CalculateWatermarks_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; - CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateWatermarks_params->meta_row_height = mode_lib->ms.meta_row_height; - CalculateWatermarks_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; - CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; - CalculateWatermarks_params->WritebackChunkSize = mode_lib->ms.ip.writeback_chunk_size_kbytes; - CalculateWatermarks_params->SOCCLK = mode_lib->ms.state.socclk_mhz; - CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; - CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeYThisState; - CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeCThisState; - CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState; - CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState; - CalculateWatermarks_params->LBBitPerPixel = mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel; - CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; - CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; - CalculateWatermarks_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; - CalculateWatermarks_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; - CalculateWatermarks_params->VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps; - CalculateWatermarks_params->VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma; - CalculateWatermarks_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; - CalculateWatermarks_params->VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma; - CalculateWatermarks_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; - CalculateWatermarks_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; - CalculateWatermarks_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; - CalculateWatermarks_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; - CalculateWatermarks_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPPThisState; - CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; - CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; - CalculateWatermarks_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; - CalculateWatermarks_params->WritebackPixelFormat = mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat; - CalculateWatermarks_params->WritebackDestinationWidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth; - CalculateWatermarks_params->WritebackDestinationHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight; - CalculateWatermarks_params->WritebackSourceHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight; - CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabledThisState; - CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByteThisState; - - // Output - CalculateWatermarks_params->Watermark = &s->dummy_watermark; // Watermarks *Watermark - CalculateWatermarks_params->DRAMClockChangeSupport = &mode_lib->ms.support.DRAMClockChangeSupport[j]; - CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // dml_float_t *MaxActiveDRAMClockChangeLatencySupported[] - CalculateWatermarks_params->SubViewportLinesNeededInMALL = &mode_lib->ms.SubViewportLinesNeededInMALL[j]; // dml_uint_t SubViewportLinesNeededInMALL[] - CalculateWatermarks_params->FCLKChangeSupport = &mode_lib->ms.support.FCLKChangeSupport[j]; - CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // dml_float_t *MaxActiveFCLKChangeLatencySupported - CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport[j]; - CalculateWatermarks_params->ActiveDRAMClockChangeLatencyMargin = mode_lib->ms.support.ActiveDRAMClockChangeLatencyMargin; - - CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, - CalculateWatermarks_params); - - } // for j -} - -/// @brief The Mode Support function. -dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib) -{ - struct dml_core_mode_support_locals_st *s = &mode_lib->scratch.dml_core_mode_support_locals; - struct UseMinimumDCFCLK_params_st *UseMinimumDCFCLK_params = &mode_lib->scratch.UseMinimumDCFCLK_params; - struct CalculateSwathAndDETConfiguration_params_st *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; - struct CalculateVMRowAndSwath_params_st *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; - - dml_uint_t j, k, m; - - mode_lib->ms.num_active_planes = dml_get_num_active_planes(&mode_lib->ms.cache_display_cfg); - dml_print("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); - - CalculateMaxDETAndMinCompressedBufferSize( - mode_lib->ms.ip.config_return_buffer_size_in_kbytes, - mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes, - mode_lib->ms.ip.rob_buffer_size_kbytes, - mode_lib->ms.ip.max_num_dpp, - mode_lib->ms.policy.NomDETInKByteOverrideEnable, // VBA_DELTA - mode_lib->ms.policy.NomDETInKByteOverrideValue, // VBA_DELTA - - /* Output */ - &mode_lib->ms.MaxTotalDETInKByte, - &mode_lib->ms.NomDETInKByte, - &mode_lib->ms.MinCompressedBufferSizeInKByte); - - PixelClockAdjustmentForProgressiveToInterlaceUnit(&mode_lib->ms.cache_display_cfg, mode_lib->ms.ip.ptoi_supported); - - - /*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/ - - /*Scale Ratio, taps Support Check*/ - mode_lib->ms.support.ScaleRatioAndTapsSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k] == false - && ((mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) - || mode_lib->ms.cache_display_cfg.plane.HRatio[k] != 1.0 - || mode_lib->ms.cache_display_cfg.plane.HTaps[k] != 1.0 - || mode_lib->ms.cache_display_cfg.plane.VRatio[k] != 1.0 - || mode_lib->ms.cache_display_cfg.plane.VTaps[k] != 1.0)) { - mode_lib->ms.support.ScaleRatioAndTapsSupport = false; - } else if (mode_lib->ms.cache_display_cfg.plane.VTaps[k] < 1.0 || mode_lib->ms.cache_display_cfg.plane.VTaps[k] > 8.0 - || mode_lib->ms.cache_display_cfg.plane.HTaps[k] < 1.0 || mode_lib->ms.cache_display_cfg.plane.HTaps[k] > 8.0 - || (mode_lib->ms.cache_display_cfg.plane.HTaps[k] > 1.0 && (mode_lib->ms.cache_display_cfg.plane.HTaps[k] % 2) == 1) - || mode_lib->ms.cache_display_cfg.plane.HRatio[k] > mode_lib->ms.ip.max_hscl_ratio - || mode_lib->ms.cache_display_cfg.plane.VRatio[k] > mode_lib->ms.ip.max_vscl_ratio - || mode_lib->ms.cache_display_cfg.plane.HRatio[k] > mode_lib->ms.cache_display_cfg.plane.HTaps[k] - || mode_lib->ms.cache_display_cfg.plane.VRatio[k] > mode_lib->ms.cache_display_cfg.plane.VTaps[k] - || (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe - && (mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] < 1 || mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] > 8 || mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] < 1 || mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] > 8 || - (mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] > 1 && mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] % 2 == 1) || - mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k] > mode_lib->ms.ip.max_hscl_ratio || - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k] > mode_lib->ms.ip.max_vscl_ratio || - mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k] > mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] || - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k] > mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]))) { - mode_lib->ms.support.ScaleRatioAndTapsSupport = false; - } - } - - /*Source Format, Pixel Format and Scan Support Check*/ - mode_lib->ms.support.SourceFormatPixelAndScanSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear && (!(!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k])) || mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true)) { - mode_lib->ms.support.SourceFormatPixelAndScanSupport = false; - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateBytePerPixelAndBlockSizes( - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], - - /* Output */ - &mode_lib->ms.BytePerPixelY[k], - &mode_lib->ms.BytePerPixelC[k], - &mode_lib->ms.BytePerPixelInDETY[k], - &mode_lib->ms.BytePerPixelInDETC[k], - &mode_lib->ms.Read256BlockHeightY[k], - &mode_lib->ms.Read256BlockHeightC[k], - &mode_lib->ms.Read256BlockWidthY[k], - &mode_lib->ms.Read256BlockWidthC[k], - &mode_lib->ms.MacroTileHeightY[k], - &mode_lib->ms.MacroTileHeightC[k], - &mode_lib->ms.MacroTileWidthY[k], - &mode_lib->ms.MacroTileWidthC[k]); - } - - /*Bandwidth Support Check*/ - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k])) { - mode_lib->ms.SwathWidthYSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportWidth[k]; - mode_lib->ms.SwathWidthCSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma[k]; - } else { - mode_lib->ms.SwathWidthYSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; - mode_lib->ms.SwathWidthCSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; - } - } - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.ReadBandwidthLuma[k] = mode_lib->ms.SwathWidthYSingleDPP[k] * dml_ceil(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - mode_lib->ms.ReadBandwidthChroma[k] = mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * dml_ceil(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k] / 2.0; - } - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true - && mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k] == dml_444_64) { - mode_lib->ms.WriteBandwidth[k] = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] - * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] - / (mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] - * mode_lib->ms.cache_display_cfg.timing.HTotal[k] - / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 8.0; - } else if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - mode_lib->ms.WriteBandwidth[k] = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] - * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] - / (mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] - * mode_lib->ms.cache_display_cfg.timing.HTotal[k] - / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 4.0; - } else { - mode_lib->ms.WriteBandwidth[k] = 0.0; - } - } - - /*Writeback Latency support check*/ - mode_lib->ms.support.WritebackLatencySupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true && - (mode_lib->ms.WriteBandwidth[k] > mode_lib->ms.ip.writeback_interface_buffer_size_kbytes * 1024 / mode_lib->ms.state.writeback_latency_us)) { - mode_lib->ms.support.WritebackLatencySupport = false; - } - } - - /*Writeback Mode Support Check*/ - s->TotalNumberOfActiveWriteback = 0; - for (k = 0; k <= (dml_uint_t) mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - s->TotalNumberOfActiveWriteback = s->TotalNumberOfActiveWriteback + 1; - } - } - - mode_lib->ms.support.EnoughWritebackUnits = 1; - if (s->TotalNumberOfActiveWriteback > (dml_uint_t) mode_lib->ms.ip.max_num_wb) { - mode_lib->ms.support.EnoughWritebackUnits = false; - } - - /*Writeback Scale Ratio and Taps Support Check*/ - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] > mode_lib->ms.ip.writeback_max_hscl_ratio - || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] > mode_lib->ms.ip.writeback_max_vscl_ratio - || mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] < mode_lib->ms.ip.writeback_min_hscl_ratio - || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] < mode_lib->ms.ip.writeback_min_vscl_ratio - || mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] > (dml_uint_t) mode_lib->ms.ip.writeback_max_hscl_taps - || mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] > (dml_uint_t) mode_lib->ms.ip.writeback_max_vscl_taps - || mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] > (dml_uint_t) mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] - || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] > (dml_uint_t) mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] - || (mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] > 2.0 && ((mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] % 2) == 1))) { - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; - } - if (2.0 * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * (mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] - 1) * 57 > mode_lib->ms.ip.writeback_line_buffer_buffer_size) { - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; - } - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateSinglePipeDPPCLKAndSCLThroughput( - mode_lib->ms.cache_display_cfg.plane.HRatio[k], - mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - mode_lib->ms.ip.max_dchub_pscl_bw_pix_per_clk, - mode_lib->ms.ip.max_pscl_lb_bw_pix_per_clk, - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.plane.HTaps[k], - mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k], - mode_lib->ms.cache_display_cfg.plane.VTaps[k], - mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k], - /* Output */ - &mode_lib->ms.PSCL_FACTOR[k], - &mode_lib->ms.PSCL_FACTOR_CHROMA[k], - &mode_lib->ms.MinDPPCLKUsingSingleDPP[k]); - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear) { - s->MaximumSwathWidthSupportLuma = 8192; - } else if (!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelC[k] > 0 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) { - s->MaximumSwathWidthSupportLuma = 7680; - } else if (dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelC[k] > 0 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) { - s->MaximumSwathWidthSupportLuma = 4320; - } else if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_rgbe_alpha) { - s->MaximumSwathWidthSupportLuma = 3840; - } else if (dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelY[k] == 8 && mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { - s->MaximumSwathWidthSupportLuma = 3072; - } else { - s->MaximumSwathWidthSupportLuma = 6144; - } - - if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_8 || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_10 || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_12) { - s->MaximumSwathWidthSupportChroma = (dml_uint_t)(s->MaximumSwathWidthSupportLuma / 2.0); - } else { - s->MaximumSwathWidthSupportChroma = s->MaximumSwathWidthSupportLuma; - } - mode_lib->ms.MaximumSwathWidthInLineBufferLuma = mode_lib->ms.ip.line_buffer_size_bits * dml_max(mode_lib->ms.cache_display_cfg.plane.HRatio[k], 1.0) / mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel[k] / - (mode_lib->ms.cache_display_cfg.plane.VTaps[k] + dml_max(dml_ceil(mode_lib->ms.cache_display_cfg.plane.VRatio[k], 1.0) - 2, 0.0)); - if (mode_lib->ms.BytePerPixelC[k] == 0.0) { - mode_lib->ms.MaximumSwathWidthInLineBufferChroma = 0; - } else { - mode_lib->ms.MaximumSwathWidthInLineBufferChroma = - mode_lib->ms.ip.line_buffer_size_bits - * dml_max(mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], 1.0) - / mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel[k] - / (mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] - + dml_max(dml_ceil(mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], 1.0) - 2, 0.0)); - } - mode_lib->ms.MaximumSwathWidthLuma[k] = dml_min(s->MaximumSwathWidthSupportLuma, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); - mode_lib->ms.MaximumSwathWidthChroma[k] = dml_min(s->MaximumSwathWidthSupportChroma, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); - } - - /*Number Of DSC Slices*/ - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && - mode_lib->ms.cache_display_cfg.output.DSCEnable[k] != dml_dsc_disable) { - mode_lib->ms.support.NumberOfDSCSlices[k] = mode_lib->ms.cache_display_cfg.output.DSCSlices[k]; - - if (mode_lib->ms.support.NumberOfDSCSlices[k] == 0) { - if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 4800) { - mode_lib->ms.support.NumberOfDSCSlices[k] = (dml_uint_t)(dml_ceil(mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 600, 4)); - } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 2400) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 8; - } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 1200) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 4; - } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 340) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 2; - } else { - mode_lib->ms.support.NumberOfDSCSlices[k] = 1; - } - } - } else { - mode_lib->ms.support.NumberOfDSCSlices[k] = 1; - } - } - - CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; - CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; - CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = 1; - CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; - CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->Output = mode_lib->ms.cache_display_cfg.output.OutputEncoder; - CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; - CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; - CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; - CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; - CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; - CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; - CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; - CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; - CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; - CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; - CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; - CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; - CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; - CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; - CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; - CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; - CalculateSwathAndDETConfiguration_params->ODMMode = s->dummy_odm_mode; - CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; - CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; - CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; - CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; - CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = s->dummy_integer_array[0]; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_integer_array[1]; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_integer_array[2]; - CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_integer_array[3]; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_integer_array[4]; - CalculateSwathAndDETConfiguration_params->SwathHeightY = s->dummy_integer_array[5]; - CalculateSwathAndDETConfiguration_params->SwathHeightC = s->dummy_integer_array[6]; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = s->dummy_integer_array[7]; - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &s->dummy_boolean[0]; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[2]; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &s->dummy_integer[1]; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &s->dummy_integer[0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = mode_lib->ms.SingleDPPViewportSizeSupportPerSurface; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[1]; - - CalculateSwathAndDETConfiguration(&mode_lib->scratch, - CalculateSwathAndDETConfiguration_params); /* dml_bool_t *ViewportSizeSupport */ - - s->MPCCombineMethodAsNeededForPStateChangeAndVoltage = false; - s->MPCCombineMethodAsPossible = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_as_needed_for_pstate_and_voltage) - s->MPCCombineMethodAsNeededForPStateChangeAndVoltage = true; - if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_as_possible) - s->MPCCombineMethodAsPossible = true; - } - mode_lib->ms.support.MPCCombineMethodIncompatible = s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && s->MPCCombineMethodAsPossible; - - for (j = 0; j < 2; j++) { - mode_lib->ms.TotalNumberOfActiveDPP[j] = 0; - mode_lib->ms.support.TotalAvailablePipesSupport[j] = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateODMMode( - mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit, - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.policy.ODMUse[k], - mode_lib->ms.state.dispclk_mhz, - mode_lib->ms.max_state.dispclk_mhz, - false, // DSCEnable - mode_lib->ms.TotalNumberOfActiveDPP[j], - mode_lib->ms.ip.max_num_dpp, - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.soc.dcn_downspread_percent, - mode_lib->ms.ip.dispclk_ramp_margin_percent, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, - mode_lib->ms.support.NumberOfDSCSlices[k], - - /* Output */ - &s->TotalAvailablePipesSupportNoDSC, - &s->NumberOfDPPNoDSC, - &s->ODMModeNoDSC, - &s->RequiredDISPCLKPerSurfaceNoDSC); - - CalculateODMMode( - mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit, - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.policy.ODMUse[k], - mode_lib->ms.state.dispclk_mhz, - mode_lib->ms.max_state.dispclk_mhz, - true, // DSCEnable - mode_lib->ms.TotalNumberOfActiveDPP[j], - mode_lib->ms.ip.max_num_dpp, - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.soc.dcn_downspread_percent, - mode_lib->ms.ip.dispclk_ramp_margin_percent, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, - mode_lib->ms.support.NumberOfDSCSlices[k], - - /* Output */ - &s->TotalAvailablePipesSupportDSC, - &s->NumberOfDPPDSC, - &s->ODMModeDSC, - &s->RequiredDISPCLKPerSurfaceDSC); - - CalculateOutputLink( - mode_lib->ms.state.phyclk_mhz, - mode_lib->ms.state.phyclk_d18_mhz, - mode_lib->ms.state.phyclk_d32_mhz, - mode_lib->ms.soc.phy_downspread_percent, - (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k), - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k], - mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k], - mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], - mode_lib->ms.support.NumberOfDSCSlices[k], - mode_lib->ms.cache_display_cfg.output.AudioSampleRate[k], - mode_lib->ms.cache_display_cfg.output.AudioSampleLayout[k], - s->ODMModeNoDSC, - s->ODMModeDSC, - mode_lib->ms.cache_display_cfg.output.DSCEnable[k], - mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k], - mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k], - - /* Output */ - &mode_lib->ms.RequiresDSC[k], - &mode_lib->ms.RequiresFEC[k], - &mode_lib->ms.OutputBppPerState[k], - &mode_lib->ms.OutputTypePerState[k], // VBA_DELTA, VBA uses a string to represent type and rate, but DML uses enum, don't want to rely on strng - &mode_lib->ms.OutputRatePerState[k], - &mode_lib->ms.RequiredSlots[k]); - - if (mode_lib->ms.RequiresDSC[k] == false) { - mode_lib->ms.ODMModePerState[k] = s->ODMModeNoDSC; - mode_lib->ms.RequiredDISPCLKPerSurface[j][k] = s->RequiredDISPCLKPerSurfaceNoDSC; - if (!s->TotalAvailablePipesSupportNoDSC) - mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; - mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + s->NumberOfDPPNoDSC; - } else { - mode_lib->ms.ODMModePerState[k] = s->ODMModeDSC; - mode_lib->ms.RequiredDISPCLKPerSurface[j][k] = s->RequiredDISPCLKPerSurfaceDSC; - if (!s->TotalAvailablePipesSupportDSC) - mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; - mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + s->NumberOfDPPDSC; - } - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { - mode_lib->ms.MPCCombine[j][k] = false; - mode_lib->ms.NoOfDPP[j][k] = 4; - } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { - mode_lib->ms.MPCCombine[j][k] = false; - mode_lib->ms.NoOfDPP[j][k] = 2; - } else if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_disabled) { - mode_lib->ms.MPCCombine[j][k] = false; - mode_lib->ms.NoOfDPP[j][k] = 1; - } else if (RoundToDFSGranularity(mode_lib->ms.MinDPPCLKUsingSingleDPP[k] * (1 + mode_lib->ms.soc.dcn_downspread_percent / 100), - 1, mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz) <= mode_lib->ms.state.dppclk_mhz && - mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k] == true) { - mode_lib->ms.MPCCombine[j][k] = false; - mode_lib->ms.NoOfDPP[j][k] = 1; - } else if (mode_lib->ms.TotalNumberOfActiveDPP[j] < (dml_uint_t) mode_lib->ms.ip.max_num_dpp) { - mode_lib->ms.MPCCombine[j][k] = true; - mode_lib->ms.NoOfDPP[j][k] = 2; - mode_lib->ms.TotalNumberOfActiveDPP[j] = (dml_uint_t) mode_lib->ms.TotalNumberOfActiveDPP[j] + 1; - } else { - mode_lib->ms.MPCCombine[j][k] = false; - mode_lib->ms.NoOfDPP[j][k] = 1; - mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; - } - } - - mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = 0; - s->NoChromaOrLinear = true; - for (k = 0; k < (dml_uint_t) mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.NoOfDPP[j][k] == 1) - mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] + 1; - if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_8 - || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_10 - || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_12 - || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_rgbe_alpha - || mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear) { - s->NoChromaOrLinear = false; - } - } - - if (j == 1 && !UnboundedRequest(mode_lib->ms.policy.UseUnboundedRequesting, - mode_lib->ms.TotalNumberOfActiveDPP[j], s->NoChromaOrLinear, - mode_lib->ms.cache_display_cfg.output.OutputEncoder[0])) { - while (!(mode_lib->ms.TotalNumberOfActiveDPP[j] >= (dml_uint_t) mode_lib->ms.ip.max_num_dpp || mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] == 0)) { - s->BWOfNonCombinedSurfaceOfMaximumBandwidth = 0; - s->NumberOfNonCombinedSurfaceOfMaximumBandwidth = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.policy.MPCCombineUse[k] != dml_mpc_disabled && mode_lib->ms.policy.MPCCombineUse[k] != dml_mpc_as_needed_for_voltage && - mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k] > s->BWOfNonCombinedSurfaceOfMaximumBandwidth && - (mode_lib->ms.ODMModePerState[k] != dml_odm_mode_combine_2to1 && mode_lib->ms.ODMModePerState[k] != dml_odm_mode_combine_4to1) && - mode_lib->ms.MPCCombine[j][k] == false) { - s->BWOfNonCombinedSurfaceOfMaximumBandwidth = mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k]; - s->NumberOfNonCombinedSurfaceOfMaximumBandwidth = k; - } - } - mode_lib->ms.MPCCombine[j][s->NumberOfNonCombinedSurfaceOfMaximumBandwidth] = true; - mode_lib->ms.NoOfDPP[j][s->NumberOfNonCombinedSurfaceOfMaximumBandwidth] = 2; - mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + 1; - mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] - 1; - } - } - - //DISPCLK/DPPCLK - mode_lib->ms.WritebackRequiredDISPCLK = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k]) { - mode_lib->ms.WritebackRequiredDISPCLK = dml_max(mode_lib->ms.WritebackRequiredDISPCLK, - CalculateWriteBackDISPCLK(mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceWidth[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.ip.writeback_line_buffer_buffer_size, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz)); - } - } - - mode_lib->ms.RequiredDISPCLK[j] = mode_lib->ms.WritebackRequiredDISPCLK; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.RequiredDISPCLK[j] = dml_max(mode_lib->ms.RequiredDISPCLK[j], mode_lib->ms.RequiredDISPCLKPerSurface[j][k]); - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; - } - - CalculateDPPCLK(mode_lib->ms.num_active_planes, - mode_lib->ms.soc.dcn_downspread_percent, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, - mode_lib->ms.MinDPPCLKUsingSingleDPP, - mode_lib->ms.NoOfDPPThisState, - /* Output */ - &mode_lib->ms.GlobalDPPCLK, - mode_lib->ms.RequiredDPPCLKThisState); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.RequiredDPPCLKPerSurface[j][k] = mode_lib->ms.RequiredDPPCLKThisState[k]; - } - - mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] = !((mode_lib->ms.RequiredDISPCLK[j] > mode_lib->ms.state.dispclk_mhz) || (mode_lib->ms.GlobalDPPCLK > mode_lib->ms.state.dppclk_mhz)); - - if (mode_lib->ms.TotalNumberOfActiveDPP[j] > (dml_uint_t) mode_lib->ms.ip.max_num_dpp) { - mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; - } - } // j - - /* Total Available OTG, HDMIFRL, DP Support Check */ - s->TotalNumberOfActiveOTG = 0; - s->TotalNumberOfActiveHDMIFRL = 0; - s->TotalNumberOfActiveDP2p0 = 0; - s->TotalNumberOfActiveDP2p0Outputs = 0; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - s->TotalNumberOfActiveOTG = s->TotalNumberOfActiveOTG + 1; - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) - s->TotalNumberOfActiveHDMIFRL = s->TotalNumberOfActiveHDMIFRL + 1; - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0) { - s->TotalNumberOfActiveDP2p0 = s->TotalNumberOfActiveDP2p0 + 1; - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k || mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == false) { - s->TotalNumberOfActiveDP2p0Outputs = s->TotalNumberOfActiveDP2p0Outputs + 1; - } - } - } - } - - mode_lib->ms.support.NumberOfOTGSupport = (s->TotalNumberOfActiveOTG <= (dml_uint_t) mode_lib->ms.ip.max_num_otg); - mode_lib->ms.support.NumberOfHDMIFRLSupport = (s->TotalNumberOfActiveHDMIFRL <= (dml_uint_t) mode_lib->ms.ip.max_num_hdmi_frl_outputs); - mode_lib->ms.support.NumberOfDP2p0Support = (s->TotalNumberOfActiveDP2p0 <= (dml_uint_t) mode_lib->ms.ip.max_num_dp2p0_streams && s->TotalNumberOfActiveDP2p0Outputs <= (dml_uint_t) mode_lib->ms.ip.max_num_dp2p0_outputs); - - /* Display IO and DSC Support Check */ - mode_lib->ms.support.NonsupportedDSCInputBPC = false; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.output.OutputDisabled[k] == false && - !(mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 12.0 - || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 10.0 - || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 8.0 - || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] > (dml_uint_t) mode_lib->ms.ip.maximum_dsc_bits_per_component - )) { - mode_lib->ms.support.NonsupportedDSCInputBPC = true; - } - } - - mode_lib->ms.support.ExceededMultistreamSlots = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k) { - s->TotalSlots = mode_lib->ms.RequiredSlots[k]; - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[j] == k) - s->TotalSlots = s->TotalSlots + mode_lib->ms.RequiredSlots[j]; - } - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp && s->TotalSlots > 63) - mode_lib->ms.support.ExceededMultistreamSlots = true; - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 && s->TotalSlots > 64) - mode_lib->ms.support.ExceededMultistreamSlots = true; - } - } - mode_lib->ms.support.LinkCapacitySupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.output.OutputDisabled[k] == false && - mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) && mode_lib->ms.OutputBppPerState[k] == 0) { - mode_lib->ms.support.LinkCapacitySupport = false; - } - } - - mode_lib->ms.support.P2IWith420 = false; - mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = false; - mode_lib->ms.support.DSC422NativeNotSupported = false; - mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = false; - mode_lib->ms.support.LinkRateForMultistreamNotIndicated = false; - mode_lib->ms.support.BPPForMultistreamNotIndicated = false; - mode_lib->ms.support.MultistreamWithHDMIOreDP = false; - mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = false; - mode_lib->ms.support.NotEnoughLanesForMSO = false; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl)) { - if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420 && mode_lib->ms.cache_display_cfg.timing.Interlace[k] == 1 && mode_lib->ms.ip.ptoi_supported == true) - mode_lib->ms.support.P2IWith420 = true; - - if (mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable_if_necessary && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] != 0) - mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = true; - if ((mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable || mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable_if_necessary) && mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 && !mode_lib->ms.ip.dsc422_native_support) - mode_lib->ms.support.DSC422NativeNotSupported = true; - - if (((mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr2 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr3) && - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp && mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_edp) || - ((mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr10 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr13p5 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr20) && - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp2p0)) - mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = true; - - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k && mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_na) - mode_lib->ms.support.LinkRateForMultistreamNotIndicated = true; - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] == 0) - mode_lib->ms.support.BPPForMultistreamNotIndicated = true; - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == j && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] == 0) - mode_lib->ms.support.BPPForMultistreamNotIndicated = true; - } - } - - if ((mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl)) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1 && mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k) - mode_lib->ms.support.MultistreamWithHDMIOreDP = true; - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { - if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1 && mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == j) - mode_lib->ms.support.MultistreamWithHDMIOreDP = true; - } - } - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp && (mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_split_1to2 || - mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to2 || mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to4)) - mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = true; - - if ((mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to2 && mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k] < 2) || - (mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to4 && mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k] < 4)) - mode_lib->ms.support.NotEnoughLanesForMSO = true; - } - } - - mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl && - RequiredDTBCLK( - mode_lib->ms.RequiresDSC[k], - mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.OutputBppPerState[k], - mode_lib->ms.support.NumberOfDSCSlices[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.output.AudioSampleRate[k], - mode_lib->ms.cache_display_cfg.output.AudioSampleLayout[k]) > mode_lib->ms.state.dtbclk_mhz) { - mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = true; - } - } - - mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK = true; - mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1 && mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi) { - mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK = false; - } - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1 && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi)) { - mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK = false; - } - } - - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = false; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) { - if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420) { - s->DSCFormatFactor = 2; - } else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_444) { - s->DSCFormatFactor = 1; - } else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) { - s->DSCFormatFactor = 2; - } else { - s->DSCFormatFactor = 1; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); -#endif - if (mode_lib->ms.RequiresDSC[k] == true) { - if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { - if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 12.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); - dml_print("DML::%s: k=%u, DSCCLKPerState = %f\n", __func__, k, mode_lib->ms.state.dscclk_mhz); - dml_print("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); -#endif - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; - } - } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { - if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 6.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; - } - } else { - if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 3.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; - } - } - } - } - } - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: DSCCLKRequiredMoreThanSupported = %u\n", __func__, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); -#endif - - /* Check DSC Unit and Slices Support */ - mode_lib->ms.support.NotEnoughDSCUnits = false; - mode_lib->ms.support.NotEnoughDSCSlices = false; - s->TotalDSCUnitsRequired = 0; - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.RequiresDSC[k] == true) { - if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { - if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > 4 * (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; - s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 4; - if (mode_lib->ms.support.NumberOfDSCSlices[k] > 16) - mode_lib->ms.support.NotEnoughDSCSlices = true; - } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { - if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > 2 * (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; - s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 2; - if (mode_lib->ms.support.NumberOfDSCSlices[k] > 8) - mode_lib->ms.support.NotEnoughDSCSlices = true; - } else { - if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; - s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 1; - if (mode_lib->ms.support.NumberOfDSCSlices[k] > 4) - mode_lib->ms.support.NotEnoughDSCSlices = true; - } - } - } - if (s->TotalDSCUnitsRequired > (dml_uint_t) mode_lib->ms.ip.num_dsc) { - mode_lib->ms.support.NotEnoughDSCUnits = true; - } - - /*DSC Delay per state*/ - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.DSCDelayPerState[k] = DSCDelayRequirement(mode_lib->ms.RequiresDSC[k], - mode_lib->ms.ODMModePerState[k], - mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], - mode_lib->ms.OutputBppPerState[k], - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.support.NumberOfDSCSlices[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { - for (j = 0; j <= mode_lib->ms.num_active_planes - 1; j++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == m && mode_lib->ms.RequiresDSC[m] == true) { - mode_lib->ms.DSCDelayPerState[k] = mode_lib->ms.DSCDelayPerState[m]; - } - } - } - } - - //Calculate Swath, DET Configuration, DCFCLKDeepSleep - // - for (j = 0; j < 2; ++j) { - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.RequiredDPPCLKThisState[k] = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; - mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; - mode_lib->ms.ODMModeThisState[k] = mode_lib->ms.ODMModePerState[k]; - } - - CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; - CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; - CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; - CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; - CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->Output = mode_lib->ms.cache_display_cfg.output.OutputEncoder; - CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; - CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; - CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; - CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; - CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; - CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; - CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; - CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; - CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; - CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; - CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; - CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; - CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; - CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; - CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; - CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; - CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.ODMModeThisState; - CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; - CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; - CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; - CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; - CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.NoOfDPPThisState; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub_this_state; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub_this_state; - CalculateSwathAndDETConfiguration_params->SwathWidth = mode_lib->ms.SwathWidthYThisState; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = mode_lib->ms.SwathWidthCThisState; - CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState; - CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->ms.DETBufferSizeInKByteThisState; - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeYThisState; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeCThisState; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->ms.UnboundedRequestEnabledThisState; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[2]; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &s->dummy_integer[1]; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->ms.CompressedBufferSizeInkByteThisState; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = s->dummy_boolean_array[0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &mode_lib->ms.support.ViewportSizeSupport[j]; - - CalculateSwathAndDETConfiguration(&mode_lib->scratch, - CalculateSwathAndDETConfiguration_params); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.swath_width_luma_ub_all_states[j][k] = mode_lib->ms.swath_width_luma_ub_this_state[k]; - mode_lib->ms.swath_width_chroma_ub_all_states[j][k] = mode_lib->ms.swath_width_chroma_ub_this_state[k]; - mode_lib->ms.SwathWidthYAllStates[j][k] = mode_lib->ms.SwathWidthYThisState[k]; - mode_lib->ms.SwathWidthCAllStates[j][k] = mode_lib->ms.SwathWidthCThisState[k]; - mode_lib->ms.SwathHeightYAllStates[j][k] = mode_lib->ms.SwathHeightYThisState[k]; - mode_lib->ms.SwathHeightCAllStates[j][k] = mode_lib->ms.SwathHeightCThisState[k]; - mode_lib->ms.UnboundedRequestEnabledAllStates[j] = mode_lib->ms.UnboundedRequestEnabledThisState; - mode_lib->ms.CompressedBufferSizeInkByteAllStates[j] = mode_lib->ms.CompressedBufferSizeInkByteThisState; - mode_lib->ms.DETBufferSizeInKByteAllStates[j][k] = mode_lib->ms.DETBufferSizeInKByteThisState[k]; - mode_lib->ms.DETBufferSizeYAllStates[j][k] = mode_lib->ms.DETBufferSizeYThisState[k]; - mode_lib->ms.DETBufferSizeCAllStates[j][k] = mode_lib->ms.DETBufferSizeCThisState[k]; - } - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.cursor_bw[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - } - - CalculateSurfaceSizeInMall( - mode_lib->ms.num_active_planes, - mode_lib->ms.soc.mall_allocated_for_dcn_mbytes, - mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen, - mode_lib->ms.cache_display_cfg.surface.DCCEnable, - mode_lib->ms.cache_display_cfg.plane.ViewportStationary, - mode_lib->ms.cache_display_cfg.plane.ViewportXStart, - mode_lib->ms.cache_display_cfg.plane.ViewportYStart, - mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, - mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, - mode_lib->ms.cache_display_cfg.plane.ViewportWidth, - mode_lib->ms.cache_display_cfg.plane.ViewportHeight, - mode_lib->ms.BytePerPixelY, - mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma, - mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, - mode_lib->ms.Read256BlockWidthY, - mode_lib->ms.Read256BlockWidthC, - mode_lib->ms.Read256BlockHeightY, - mode_lib->ms.Read256BlockHeightC, - mode_lib->ms.MacroTileWidthY, - mode_lib->ms.MacroTileWidthC, - mode_lib->ms.MacroTileHeightY, - mode_lib->ms.MacroTileHeightC, - - /* Output */ - mode_lib->ms.SurfaceSizeInMALL, - &mode_lib->ms.support.ExceededMALLSize); - - for (j = 0; j < 2; j++) { - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.swath_width_luma_ub_this_state[k] = mode_lib->ms.swath_width_luma_ub_all_states[j][k]; - mode_lib->ms.swath_width_chroma_ub_this_state[k] = mode_lib->ms.swath_width_chroma_ub_all_states[j][k]; - mode_lib->ms.SwathWidthYThisState[k] = mode_lib->ms.SwathWidthYAllStates[j][k]; - mode_lib->ms.SwathWidthCThisState[k] = mode_lib->ms.SwathWidthCAllStates[j][k]; - mode_lib->ms.SwathHeightYThisState[k] = mode_lib->ms.SwathHeightYAllStates[j][k]; - mode_lib->ms.SwathHeightCThisState[k] = mode_lib->ms.SwathHeightCAllStates[j][k]; - mode_lib->ms.DETBufferSizeInKByteThisState[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[j][k]; - mode_lib->ms.DETBufferSizeYThisState[k] = mode_lib->ms.DETBufferSizeYAllStates[j][k]; - mode_lib->ms.DETBufferSizeCThisState[k] = mode_lib->ms.DETBufferSizeCAllStates[j][k]; - mode_lib->ms.RequiredDPPCLKThisState[k] = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; - mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; - } - - mode_lib->ms.TotalNumberOfDCCActiveDPP[j] = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { - mode_lib->ms.TotalNumberOfDCCActiveDPP[j] = mode_lib->ms.TotalNumberOfDCCActiveDPP[j] + mode_lib->ms.NoOfDPP[j][k]; - } - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->SurfParameters[k].PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; - s->SurfParameters[k].DPPPerSurface = mode_lib->ms.NoOfDPP[j][k]; - s->SurfParameters[k].SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; - s->SurfParameters[k].ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; - s->SurfParameters[k].ViewportHeightChroma = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; - s->SurfParameters[k].BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; - s->SurfParameters[k].BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; - s->SurfParameters[k].BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; - s->SurfParameters[k].BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; - s->SurfParameters[k].BlockWidthY = mode_lib->ms.MacroTileWidthY[k]; - s->SurfParameters[k].BlockHeightY = mode_lib->ms.MacroTileHeightY[k]; - s->SurfParameters[k].BlockWidthC = mode_lib->ms.MacroTileWidthC[k]; - s->SurfParameters[k].BlockHeightC = mode_lib->ms.MacroTileHeightC[k]; - s->SurfParameters[k].InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; - s->SurfParameters[k].HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; - s->SurfParameters[k].DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; - s->SurfParameters[k].SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; - s->SurfParameters[k].SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k]; - s->SurfParameters[k].BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; - s->SurfParameters[k].BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; - s->SurfParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - s->SurfParameters[k].VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - s->SurfParameters[k].VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; - s->SurfParameters[k].VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps[k]; - s->SurfParameters[k].VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]; - s->SurfParameters[k].PitchY = mode_lib->ms.cache_display_cfg.surface.PitchY[k]; - s->SurfParameters[k].DCCMetaPitchY = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; - s->SurfParameters[k].PitchC = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; - s->SurfParameters[k].DCCMetaPitchC = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; - s->SurfParameters[k].ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary[k]; - s->SurfParameters[k].ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart[k]; - s->SurfParameters[k].ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart[k]; - s->SurfParameters[k].ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC[k]; - s->SurfParameters[k].ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC[k]; - s->SurfParameters[k].FORCE_ONE_ROW_FOR_FRAME = mode_lib->ms.cache_display_cfg.plane.ForceOneRowForFrame[k]; - s->SurfParameters[k].SwathHeightY = mode_lib->ms.SwathHeightYThisState[k]; - s->SurfParameters[k].SwathHeightC = mode_lib->ms.SwathHeightCThisState[k]; - } - - CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; - CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; - CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; - CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; - CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; - CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; - CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; - CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; - CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceededPerState; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceededPerState; - CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[0]; - CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[1]; - CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; - CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; - CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[2]; // VBA_DELTA - CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[3]; // VBA_DELTA - CalculateVMRowAndSwath_params->meta_req_width = s->dummy_integer_array[4]; - CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[5]; - CalculateVMRowAndSwath_params->meta_req_height = s->dummy_integer_array[6]; - CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[7]; - CalculateVMRowAndSwath_params->meta_row_width = s->dummy_integer_array[8]; - CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[9]; - CalculateVMRowAndSwath_params->meta_row_height = mode_lib->ms.meta_row_height; - CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; - CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[10]; - CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[11]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[12]; - CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[13]; - CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[14]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[15]; - CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[16]; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[17]; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[18]; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[19]; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[20]; - CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesYThisState; - CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesCThisState; - CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; - CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; - CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY; - CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC; - CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bandwidth_this_state; - CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bandwidth_this_state; - CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRowThisState; - CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState; - CalculateVMRowAndSwath_params->MetaRowByte = mode_lib->ms.MetaRowBytesThisState; - CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame_this_state; - CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip_this_state; - CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = s->dummy_boolean_array[0]; - CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; - CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[21]; - - CalculateVMRowAndSwath(&mode_lib->scratch, - CalculateVMRowAndSwath_params); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.PrefetchLinesY[j][k] = mode_lib->ms.PrefetchLinesYThisState[k]; - mode_lib->ms.PrefetchLinesC[j][k] = mode_lib->ms.PrefetchLinesCThisState[k]; - mode_lib->ms.meta_row_bandwidth[j][k] = mode_lib->ms.meta_row_bandwidth_this_state[k]; - mode_lib->ms.dpte_row_bandwidth[j][k] = mode_lib->ms.dpte_row_bandwidth_this_state[k]; - mode_lib->ms.DPTEBytesPerRow[j][k] = mode_lib->ms.DPTEBytesPerRowThisState[k]; - mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k] = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState[k]; - mode_lib->ms.MetaRowBytes[j][k] = mode_lib->ms.MetaRowBytesThisState[k]; - mode_lib->ms.use_one_row_for_frame[j][k] = mode_lib->ms.use_one_row_for_frame_this_state[k]; - mode_lib->ms.use_one_row_for_frame_flip[j][k] = mode_lib->ms.use_one_row_for_frame_flip_this_state[k]; - } - - mode_lib->ms.support.PTEBufferSizeNotExceeded[j] = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.PTEBufferSizeNotExceededPerState[k] == false) - mode_lib->ms.support.PTEBufferSizeNotExceeded[j] = false; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: j=%u k=%u, PTEBufferSizeNotExceededPerState[%u] = %u\n", __func__, j, k, k, mode_lib->ms.PTEBufferSizeNotExceededPerState[k]); -#endif - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: PTEBufferSizeNotExceeded[%u] = %u\n", __func__, j, mode_lib->ms.support.PTEBufferSizeNotExceeded[j]); -#endif - - mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.DCCMetaBufferSizeNotExceededPerState[k] == false) - mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] = false; - } - - mode_lib->ms.UrgLatency = CalculateUrgentLatency(mode_lib->ms.state.urgent_latency_pixel_data_only_us, - mode_lib->ms.state.urgent_latency_pixel_mixed_with_vm_data_us, - mode_lib->ms.state.urgent_latency_vm_data_only_us, - mode_lib->ms.soc.do_urgent_latency_adjustment, - mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_component_us, - mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_reference_mhz, - mode_lib->ms.state.fabricclk_mhz); - - /* Getter functions work at mp interface so copy the urgent latency to mp*/ - mode_lib->mp.UrgentLatency = mode_lib->ms.UrgLatency; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateUrgentBurstFactor( - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - mode_lib->ms.swath_width_luma_ub_this_state[k], - mode_lib->ms.swath_width_chroma_ub_this_state[k], - mode_lib->ms.SwathHeightYThisState[k], - mode_lib->ms.SwathHeightCThisState[k], - (dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.UrgLatency, - mode_lib->ms.ip.cursor_buffer_size, - mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], - mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - mode_lib->ms.BytePerPixelInDETY[k], - mode_lib->ms.BytePerPixelInDETC[k], - mode_lib->ms.DETBufferSizeYThisState[k], - mode_lib->ms.DETBufferSizeCThisState[k], - /* Output */ - &mode_lib->ms.UrgentBurstFactorCursor[j][k], - &mode_lib->ms.UrgentBurstFactorLuma[j][k], - &mode_lib->ms.UrgentBurstFactorChroma[j][k], - &mode_lib->ms.NotUrgentLatencyHiding[k]); - } - - CalculateDCFCLKDeepSleep( - mode_lib->ms.num_active_planes, - mode_lib->ms.BytePerPixelY, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.cache_display_cfg.plane.VRatio, - mode_lib->ms.cache_display_cfg.plane.VRatioChroma, - mode_lib->ms.SwathWidthYThisState, - mode_lib->ms.SwathWidthCThisState, - mode_lib->ms.NoOfDPPThisState, - mode_lib->ms.cache_display_cfg.plane.HRatio, - mode_lib->ms.cache_display_cfg.plane.HRatioChroma, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - mode_lib->ms.PSCL_FACTOR, - mode_lib->ms.PSCL_FACTOR_CHROMA, - mode_lib->ms.RequiredDPPCLKThisState, - mode_lib->ms.ReadBandwidthLuma, - mode_lib->ms.ReadBandwidthChroma, - mode_lib->ms.soc.return_bus_width_bytes, - - /* Output */ - &mode_lib->ms.ProjectedDCFCLKDeepSleep[j]); - } - - //Calculate Return BW - for (j = 0; j < 2; ++j) { - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - mode_lib->ms.WritebackDelayTime[k] = mode_lib->ms.state.writeback_latency_us + CalculateWriteBackDelay( - mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / mode_lib->ms.RequiredDISPCLK[j]; - } else { - mode_lib->ms.WritebackDelayTime[k] = 0.0; - } - for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[m] == k && mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[m] == true) { - mode_lib->ms.WritebackDelayTime[k] = dml_max(mode_lib->ms.WritebackDelayTime[k], - mode_lib->ms.state.writeback_latency_us + CalculateWriteBackDelay( - mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[m], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[m], - mode_lib->ms.cache_display_cfg.timing.HTotal[m]) / mode_lib->ms.RequiredDISPCLK[j]); - } - } - } - } - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == m) { - mode_lib->ms.WritebackDelayTime[k] = mode_lib->ms.WritebackDelayTime[m]; - } - } - } - s->MaxVStartupAllPlanes[j] = 0; // max vstartup among all planes - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - s->MaximumVStartup[j][k] = CalculateMaxVStartup(k, - mode_lib->ms.ip.ptoi_supported, - mode_lib->ms.ip.vblank_nom_default_us, - &mode_lib->ms.cache_display_cfg.timing, - mode_lib->ms.WritebackDelayTime[k]); - - s->MaxVStartupAllPlanes[j] = (dml_uint_t)(dml_max(s->MaxVStartupAllPlanes[j], s->MaximumVStartup[j][k])); -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, MaxVStartupAllPlanes[%u] = %u\n", __func__, k, j, s->MaxVStartupAllPlanes[j]); - dml_print("DML::%s: k=%u, MaximumVStartup[%u][%u] = %u\n", __func__, k, j, k, s->MaximumVStartup[j][k]); -#endif - } - } - - s->ReorderingBytes = (dml_uint_t)(mode_lib->ms.soc.num_chans * dml_max3(mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_only_bytes, - mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, - mode_lib->ms.soc.urgent_out_of_order_return_per_channel_vm_only_bytes)); - - for (j = 0; j < 2; ++j) { - mode_lib->ms.DCFCLKState[j] = mode_lib->ms.state.dcfclk_mhz; - } - - /* Immediate Flip and MALL parameters */ - s->ImmediateFlipRequiredFinal = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->ImmediateFlipRequiredFinal = s->ImmediateFlipRequiredFinal || (mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); - } - - mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified || - ((mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_required) && - (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required)); - } - mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified && s->ImmediateFlipRequiredFinal; - - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe || ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == true || mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) && - (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame || mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe)); - } - - mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen || - ((mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable || mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_optimize) && (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe)) || - ((mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_disable || mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_optimize) && (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame)); - } - - s->FullFrameMALLPStateMethod = false; - s->SubViewportMALLPStateMethod = false; - s->PhantomPipeMALLPStateMethod = false; - s->SubViewportMALLRefreshGreaterThan120Hz = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame) - s->FullFrameMALLPStateMethod = true; - if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) { - s->SubViewportMALLPStateMethod = true; - if (mode_lib->ms.cache_display_cfg.timing.RefreshRate[k] > 120) - s->SubViewportMALLRefreshGreaterThan120Hz = true; - } - if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) - s->PhantomPipeMALLPStateMethod = true; - } - mode_lib->ms.support.InvalidCombinationOfMALLUseForPState = (s->SubViewportMALLPStateMethod != s->PhantomPipeMALLPStateMethod) - || (s->SubViewportMALLPStateMethod && s->FullFrameMALLPStateMethod) || s->SubViewportMALLRefreshGreaterThan120Hz; - - if (mode_lib->ms.policy.UseMinimumRequiredDCFCLK == true) { - UseMinimumDCFCLK_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - UseMinimumDCFCLK_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; - UseMinimumDCFCLK_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; - UseMinimumDCFCLK_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; - UseMinimumDCFCLK_params->MaxPrefetchMode = dml_prefetch_support_stutter; - UseMinimumDCFCLK_params->DRAMClockChangeLatencyFinal = mode_lib->ms.state.dram_clock_change_latency_us; - UseMinimumDCFCLK_params->FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; - UseMinimumDCFCLK_params->SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; - UseMinimumDCFCLK_params->ReturnBusWidth = mode_lib->ms.soc.return_bus_width_bytes; - UseMinimumDCFCLK_params->RoundTripPingLatencyCycles = mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles; - UseMinimumDCFCLK_params->ReorderingBytes = s->ReorderingBytes; - UseMinimumDCFCLK_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; - UseMinimumDCFCLK_params->MetaChunkSize = mode_lib->ms.ip.meta_chunk_size_kbytes; - UseMinimumDCFCLK_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - UseMinimumDCFCLK_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - UseMinimumDCFCLK_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - UseMinimumDCFCLK_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - UseMinimumDCFCLK_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - UseMinimumDCFCLK_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - UseMinimumDCFCLK_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; - UseMinimumDCFCLK_params->ImmediateFlipRequirement = s->ImmediateFlipRequiredFinal; - UseMinimumDCFCLK_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - UseMinimumDCFCLK_params->MaxAveragePercentOfIdealSDPPortBWDisplayCanUseInNormalSystemOperation = mode_lib->ms.soc.max_avg_sdp_bw_use_normal_percent; - UseMinimumDCFCLK_params->PercentOfIdealSDPPortBWReceivedAfterUrgLatency = mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent; - UseMinimumDCFCLK_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; - UseMinimumDCFCLK_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; - UseMinimumDCFCLK_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes; - UseMinimumDCFCLK_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired; - UseMinimumDCFCLK_params->Interlace = mode_lib->ms.cache_display_cfg.timing.Interlace; - UseMinimumDCFCLK_params->RequiredDPPCLKPerSurface = mode_lib->ms.RequiredDPPCLKPerSurface; - UseMinimumDCFCLK_params->RequiredDISPCLK = mode_lib->ms.RequiredDISPCLK; - UseMinimumDCFCLK_params->UrgLatency = mode_lib->ms.UrgLatency; - UseMinimumDCFCLK_params->NoOfDPP = mode_lib->ms.NoOfDPP; - UseMinimumDCFCLK_params->ProjectedDCFCLKDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep; - UseMinimumDCFCLK_params->MaximumVStartup = s->MaximumVStartup; - UseMinimumDCFCLK_params->TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP; - UseMinimumDCFCLK_params->TotalNumberOfDCCActiveDPP = mode_lib->ms.TotalNumberOfDCCActiveDPP; - UseMinimumDCFCLK_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - UseMinimumDCFCLK_params->PrefetchLinesY = mode_lib->ms.PrefetchLinesY; - UseMinimumDCFCLK_params->PrefetchLinesC = mode_lib->ms.PrefetchLinesC; - UseMinimumDCFCLK_params->swath_width_luma_ub_all_states = mode_lib->ms.swath_width_luma_ub_all_states; - UseMinimumDCFCLK_params->swath_width_chroma_ub_all_states = mode_lib->ms.swath_width_chroma_ub_all_states; - UseMinimumDCFCLK_params->BytePerPixelY = mode_lib->ms.BytePerPixelY; - UseMinimumDCFCLK_params->BytePerPixelC = mode_lib->ms.BytePerPixelC; - UseMinimumDCFCLK_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; - UseMinimumDCFCLK_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; - UseMinimumDCFCLK_params->PDEAndMetaPTEBytesPerFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrame; - UseMinimumDCFCLK_params->DPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow; - UseMinimumDCFCLK_params->MetaRowBytes = mode_lib->ms.MetaRowBytes; - UseMinimumDCFCLK_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable; - UseMinimumDCFCLK_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; - UseMinimumDCFCLK_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; - UseMinimumDCFCLK_params->DCFCLKPerState = mode_lib->ms.state.dcfclk_mhz; - UseMinimumDCFCLK_params->DCFCLKState = mode_lib->ms.DCFCLKState; - - UseMinimumDCFCLK(&mode_lib->scratch, - UseMinimumDCFCLK_params); - - } // UseMinimumRequiredDCFCLK == true - - for (j = 0; j < 2; ++j) { - mode_lib->ms.ReturnBWPerState[j] = dml_get_return_bw_mbps(&mode_lib->ms.soc, mode_lib->ms.state.use_ideal_dram_bw_strobe, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, mode_lib->ms.DCFCLKState[j], mode_lib->ms.state.fabricclk_mhz, - mode_lib->ms.state.dram_speed_mts); - mode_lib->ms.ReturnDRAMBWPerState[j] = dml_get_return_dram_bw_mbps(&mode_lib->ms.soc, mode_lib->ms.state.use_ideal_dram_bw_strobe, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.state.dram_speed_mts); - } - - //Re-ordering Buffer Support Check - for (j = 0; j < 2; ++j) { - if ((mode_lib->ms.ip.rob_buffer_size_kbytes - mode_lib->ms.ip.pixel_chunk_size_kbytes) * 1024 / mode_lib->ms.ReturnBWPerState[j] > - (mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles + 32) / mode_lib->ms.DCFCLKState[j] + s->ReorderingBytes / mode_lib->ms.ReturnBWPerState[j]) { - mode_lib->ms.support.ROBSupport[j] = true; - } else { - mode_lib->ms.support.ROBSupport[j] = false; - } - dml_print("DML::%s: DEBUG ROBSupport[%u] = %u (%u)\n", __func__, j, mode_lib->ms.support.ROBSupport[j], __LINE__); - } - - //Vertical Active BW support check - s->MaxTotalVActiveRDBandwidth = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->MaxTotalVActiveRDBandwidth = s->MaxTotalVActiveRDBandwidth + mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k]; - } - - for (j = 0; j < 2; ++j) { - mode_lib->ms.support.MaxTotalVerticalActiveAvailableBandwidth[j] = dml_min3(mode_lib->ms.soc.return_bus_width_bytes * mode_lib->ms.DCFCLKState[j] * mode_lib->ms.soc.max_avg_sdp_bw_use_normal_percent / 100.0, - mode_lib->ms.state.fabricclk_mhz * mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes * mode_lib->ms.soc.max_avg_fabric_bw_use_normal_percent / 100.0, - mode_lib->ms.state.dram_speed_mts * mode_lib->ms.soc.num_chans * mode_lib->ms.soc.dram_channel_width_bytes * - ((mode_lib->ms.state.use_ideal_dram_bw_strobe && !mode_lib->ms.cache_display_cfg.plane.HostVMEnable) ? - mode_lib->ms.soc.max_avg_dram_bw_use_normal_strobe_percent : mode_lib->ms.soc.max_avg_dram_bw_use_normal_percent) / 100.0); - - if (s->MaxTotalVActiveRDBandwidth <= mode_lib->ms.support.MaxTotalVerticalActiveAvailableBandwidth[j]) { - mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] = true; - } else { - mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] = false; - } - } - - /* Prefetch Check */ - dml_prefetch_check(mode_lib); - - // End of Prefetch Check - dml_print("DML::%s: Done prefetch calculation\n", __func__); - - /*Cursor Support Check*/ - mode_lib->ms.support.CursorSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] > 0.0) { - if (mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] == 64 && mode_lib->ms.ip.cursor_64bpp_support == false) { - mode_lib->ms.support.CursorSupport = false; - } - } - } - - /*Valid Pitch Check*/ - mode_lib->ms.support.PitchSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.support.AlignedYPitch[k] = dml_ceil( - dml_max(mode_lib->ms.cache_display_cfg.surface.PitchY[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k]), - mode_lib->ms.MacroTileWidthY[k]); - if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { - mode_lib->ms.support.AlignedDCCMetaPitchY[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k]), 64.0 * mode_lib->ms.Read256BlockWidthY[k]); - } else { - mode_lib->ms.support.AlignedDCCMetaPitchY[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; - } - if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe - && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8) { - mode_lib->ms.support.AlignedCPitch[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.PitchC[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k]), mode_lib->ms.MacroTileWidthC[k]); - if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { - mode_lib->ms.support.AlignedDCCMetaPitchC[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k]), 64.0 * mode_lib->ms.Read256BlockWidthC[k]); - } else { - mode_lib->ms.support.AlignedDCCMetaPitchC[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; - } - } else { - mode_lib->ms.support.AlignedCPitch[k] = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; - mode_lib->ms.support.AlignedDCCMetaPitchC[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; - } - if (mode_lib->ms.support.AlignedYPitch[k] > mode_lib->ms.cache_display_cfg.surface.PitchY[k] || mode_lib->ms.support.AlignedCPitch[k] > mode_lib->ms.cache_display_cfg.surface.PitchC[k] || - mode_lib->ms.support.AlignedDCCMetaPitchY[k] > mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k] || mode_lib->ms.support.AlignedDCCMetaPitchC[k] > mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]) { - mode_lib->ms.support.PitchSupport = false; - } - } - - mode_lib->ms.support.ViewportExceedsSurface = false; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.ViewportWidth[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k] || mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY[k]) { - mode_lib->ms.support.ViewportExceedsSurface = true; - if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 && - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_8 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe) { - if (mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k] || mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC[k]) { - mode_lib->ms.support.ViewportExceedsSurface = true; - } - } - } - } - - /*Mode Support, Voltage State and SOC Configuration*/ - for (j = 0; j < 2; j++) { // j iterator is for the combine mode off or on - dml_print("DML::%s: checking support for j=%u\n", __func__, j); - dml_print("DML::%s: state_idx=%0d max_state_idx=%0d\n", __func__, mode_lib->ms.state_idx, mode_lib->ms.max_state_idx); - - s->is_max_pwr_state = (mode_lib->ms.max_state_idx == mode_lib->ms.state_idx); - s->is_max_dram_pwr_state = (mode_lib->ms.max_state.dram_speed_mts == mode_lib->ms.state.dram_speed_mts); - - s->dram_clock_change_support = (!mode_lib->ms.policy.DRAMClockChangeRequirementFinal || - (s->is_max_dram_pwr_state && mode_lib->policy.AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported) || - mode_lib->ms.support.DRAMClockChangeSupport[j] != dml_dram_clock_change_unsupported); - s->f_clock_change_support = (!mode_lib->ms.policy.FCLKChangeRequirementFinal || - (s->is_max_pwr_state && mode_lib->policy.AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported) || - mode_lib->ms.support.FCLKChangeSupport[j] != dml_fclock_change_unsupported); - - if (mode_lib->ms.support.ScaleRatioAndTapsSupport == true - && mode_lib->ms.support.SourceFormatPixelAndScanSupport == true - && mode_lib->ms.support.ViewportSizeSupport[j] == true - && !mode_lib->ms.support.LinkRateDoesNotMatchDPVersion - && !mode_lib->ms.support.LinkRateForMultistreamNotIndicated - && !mode_lib->ms.support.BPPForMultistreamNotIndicated - && !mode_lib->ms.support.MultistreamWithHDMIOreDP - && !mode_lib->ms.support.ExceededMultistreamSlots - && !mode_lib->ms.support.MSOOrODMSplitWithNonDPLink - && !mode_lib->ms.support.NotEnoughLanesForMSO - && mode_lib->ms.support.LinkCapacitySupport == true - && !mode_lib->ms.support.P2IWith420 - && !mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP - && !mode_lib->ms.support.DSC422NativeNotSupported - && !mode_lib->ms.support.MPCCombineMethodIncompatible - && mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK == true - && mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK == true - && mode_lib->ms.support.NotEnoughDSCUnits == false - && !mode_lib->ms.support.NotEnoughDSCSlices - && !mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe - && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen - && mode_lib->ms.support.DSCCLKRequiredMoreThanSupported == false - && mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport - && mode_lib->ms.support.DTBCLKRequiredMoreThanSupported == false - && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPState - && !mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified - && mode_lib->ms.support.ROBSupport[j] == true - && mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] == true - && mode_lib->ms.support.TotalAvailablePipesSupport[j] == true - && mode_lib->ms.support.NumberOfOTGSupport == true - && mode_lib->ms.support.NumberOfHDMIFRLSupport == true - && mode_lib->ms.support.NumberOfDP2p0Support == true - && mode_lib->ms.support.EnoughWritebackUnits == true - && mode_lib->ms.support.WritebackLatencySupport == true - && mode_lib->ms.support.WritebackScaleRatioAndTapsSupport == true - && mode_lib->ms.support.CursorSupport == true - && mode_lib->ms.support.PitchSupport == true - && mode_lib->ms.support.ViewportExceedsSurface == false - && mode_lib->ms.support.PrefetchSupported[j] == true - && mode_lib->ms.support.VActiveBandwithSupport[j] == true - && mode_lib->ms.support.DynamicMetadataSupported[j] == true - && mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] == true - && mode_lib->ms.support.VRatioInPrefetchSupported[j] == true - && mode_lib->ms.support.PTEBufferSizeNotExceeded[j] == true - && mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] == true - && mode_lib->ms.support.NonsupportedDSCInputBPC == false - && !mode_lib->ms.support.ExceededMALLSize - && ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == false && !s->ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j]) - && s->dram_clock_change_support == true - && s->f_clock_change_support == true - && (!mode_lib->ms.policy.USRRetrainingRequiredFinal || mode_lib->ms.support.USRRetrainingSupport[j])) { - dml_print("DML::%s: mode is supported\n", __func__); - mode_lib->ms.support.ModeSupport[j] = true; - } else { - dml_print("DML::%s: mode is NOT supported\n", __func__); - mode_lib->ms.support.ModeSupport[j] = false; - dml_print_mode_support(mode_lib, j); - } - } - - mode_lib->ms.support.MaximumMPCCombine = 0; - mode_lib->ms.support.ModeIsSupported = 0; - if (mode_lib->ms.support.ModeSupport[0] == true || mode_lib->ms.support.ModeSupport[1] == true) { // if the mode is supported by either no combine or mpccombine - mode_lib->ms.support.ModeIsSupported = mode_lib->ms.support.ModeSupport[0] == true || mode_lib->ms.support.ModeSupport[1] == true; - - // Determine if MPC combine is necessary, depends on if using MPC combine will help dram clock change or fclk change, etc. - if ((mode_lib->ms.support.ModeSupport[0] == false && mode_lib->ms.support.ModeSupport[1] == true) || s->MPCCombineMethodAsPossible || - (s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && mode_lib->ms.policy.DRAMClockChangeRequirementFinal && - (((mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive_w_mall_sub_vp) && - !(mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive || mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive_w_mall_sub_vp)) || - ((mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr - || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr_w_mall_full_frame - || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_w_mall_sub_vp || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr_w_mall_sub_vp - ) && - mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_unsupported))) - || (s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && mode_lib->ms.policy.FCLKChangeRequirementFinal && - ((mode_lib->ms.support.FCLKChangeSupport[1] == dml_fclock_change_vactive && mode_lib->ms.support.FCLKChangeSupport[0] != dml_fclock_change_vactive) || - (mode_lib->ms.support.FCLKChangeSupport[1] == dml_fclock_change_vblank && mode_lib->ms.support.FCLKChangeSupport[0] == dml_fclock_change_unsupported)))) { - mode_lib->ms.support.MaximumMPCCombine = 1; - } else { - mode_lib->ms.support.MaximumMPCCombine = 0; - } - } - - // Since now the mode_support work on 1 particular power state, so there is only 1 state idx (index 0). - mode_lib->ms.support.ImmediateFlipSupport = mode_lib->ms.support.ImmediateFlipSupportedForState[mode_lib->ms.support.MaximumMPCCombine]; // Consider flip support if max combine support imm flip - mode_lib->ms.support.UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabledAllStates[mode_lib->ms.support.MaximumMPCCombine]; // Not used, informational - mode_lib->ms.support.CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByteAllStates[mode_lib->ms.support.MaximumMPCCombine]; // Not used, informational - - dml_print("DML::%s: ModeIsSupported = %u\n", __func__, mode_lib->ms.support.ModeIsSupported); - dml_print("DML::%s: MaximumMPCCombine = %u\n", __func__, mode_lib->ms.support.MaximumMPCCombine); - dml_print("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); - dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, mode_lib->ms.support.UnboundedRequestEnabled); - dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, mode_lib->ms.support.CompressedBufferSizeInkByte); - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.support.MPCCombineEnable[k] = mode_lib->ms.MPCCombine[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.support.DPPPerSurface[k] = mode_lib->ms.NoOfDPP[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.SwathHeightY[k] = mode_lib->ms.SwathHeightYAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.SwathHeightC[k] = mode_lib->ms.SwathHeightCAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.DETBufferSizeInKByte[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.DETBufferSizeY[k] = mode_lib->ms.DETBufferSizeYAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; - mode_lib->ms.DETBufferSizeC[k] = mode_lib->ms.DETBufferSizeCAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; - } - - mode_lib->ms.DRAMSpeed = mode_lib->ms.state.dram_speed_mts; - mode_lib->ms.FabricClock = mode_lib->ms.state.fabricclk_mhz; - mode_lib->ms.SOCCLK = mode_lib->ms.state.socclk_mhz; - mode_lib->ms.DCFCLK = mode_lib->ms.DCFCLKState[mode_lib->ms.support.MaximumMPCCombine]; - mode_lib->ms.ReturnBW = mode_lib->ms.ReturnBWPerState[mode_lib->ms.support.MaximumMPCCombine]; - mode_lib->ms.ReturnDRAMBW = mode_lib->ms.ReturnDRAMBWPerState[mode_lib->ms.support.MaximumMPCCombine]; - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - mode_lib->ms.support.ODMMode[k] = mode_lib->ms.ODMModePerState[k]; - } else { - mode_lib->ms.support.ODMMode[k] = dml_odm_mode_bypass; - } - - mode_lib->ms.support.DSCEnabled[k] = mode_lib->ms.RequiresDSC[k]; - mode_lib->ms.support.FECEnabled[k] = mode_lib->ms.RequiresFEC[k]; - mode_lib->ms.support.OutputBpp[k] = mode_lib->ms.OutputBppPerState[k]; - mode_lib->ms.support.OutputType[k] = mode_lib->ms.OutputTypePerState[k]; - mode_lib->ms.support.OutputRate[k] = mode_lib->ms.OutputRatePerState[k]; - mode_lib->ms.support.SubViewportLinesNeededInMALL[k] = mode_lib->ms.SubViewportLinesNeededInMALL[k]; - } - - return mode_lib->ms.support.ModeIsSupported; -} // dml_core_mode_support - -/// @brief This function calculates some parameters thats are needed ahead of the mode programming function all -void dml_core_mode_support_partial(struct display_mode_lib_st *mode_lib) -{ - CalculateMaxDETAndMinCompressedBufferSize( - mode_lib->ms.ip.config_return_buffer_size_in_kbytes, - mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes, - mode_lib->ms.ip.rob_buffer_size_kbytes, - mode_lib->ms.ip.max_num_dpp, - mode_lib->ms.policy.NomDETInKByteOverrideEnable, - mode_lib->ms.policy.NomDETInKByteOverrideValue, - - /* Output */ - &mode_lib->ms.MaxTotalDETInKByte, - &mode_lib->ms.NomDETInKByte, - &mode_lib->ms.MinCompressedBufferSizeInKByte); - - PixelClockAdjustmentForProgressiveToInterlaceUnit(&mode_lib->ms.cache_display_cfg, mode_lib->ms.ip.ptoi_supported); - - mode_lib->ms.ReturnBW = dml_get_return_bw_mbps(&mode_lib->ms.soc, - mode_lib->ms.state.use_ideal_dram_bw_strobe, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.DCFCLK, - mode_lib->ms.FabricClock, - mode_lib->ms.DRAMSpeed); - dml_print("DML::%s: ReturnBW = %f\n", __func__, mode_lib->ms.ReturnBW); - -} // dml_core_mode_support_partial - -/// @brief This is the mode programming function. It is assumed the display cfg is support at the given power state -void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struct dml_clk_cfg_st *clk_cfg) -{ - struct dml_core_mode_programming_locals_st *s = &mode_lib->scratch.dml_core_mode_programming_locals; - struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; - struct CalculateVMRowAndSwath_params_st *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; - struct CalculateSwathAndDETConfiguration_params_st *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; - struct CalculateStutterEfficiency_params_st *CalculateStutterEfficiency_params = &mode_lib->scratch.CalculateStutterEfficiency_params; - struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; - - struct mode_program_st *locals = &mode_lib->mp; - struct DmlPipe *myPipe; - dml_uint_t j = 0, k = 0; - dml_float_t TWait; - dml_bool_t isInterlaceTiming; - - mode_lib->ms.num_active_planes = dml_get_num_active_planes(&mode_lib->ms.cache_display_cfg); - mode_lib->mp.num_active_pipes = dml_get_num_active_pipes(&mode_lib->ms.cache_display_cfg); - dml_calc_pipe_plane_mapping(&mode_lib->ms.cache_display_cfg.hw, mode_lib->mp.pipe_plane); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: --- START --- \n", __func__); - dml_print("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); - dml_print("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); -#endif - - s->DSCFormatFactor = 0; - - // Unlike dppclk and dispclk which can be calculated in mode_programming - // DCFCLK is calculated in mode_support (which is the state bbox dcfclk or min dcfclk if min dcfclk option is used in mode support calculation) - if (clk_cfg->dcfclk_option != dml_use_override_freq) - locals->Dcfclk = mode_lib->ms.DCFCLK; - else - locals->Dcfclk = clk_cfg->dcfclk_mhz; - -#ifdef __DML_VBA_DEBUG__ - dml_print_dml_policy(&mode_lib->ms.policy); - dml_print_soc_state_bounding_box(&mode_lib->ms.state); - dml_print_soc_bounding_box(&mode_lib->ms.soc); - dml_print_clk_cfg(clk_cfg); - - dml_print("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); - dml_print("DML::%s: Using DCFCLK = %f\n", __func__, locals->Dcfclk); - dml_print("DML::%s: Using SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); -#endif - - locals->WritebackDISPCLK = 0.0; - locals->GlobalDPPCLK = 0.0; - - // DISPCLK and DPPCLK Calculation - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k]) { - locals->WritebackDISPCLK = - dml_max( - locals->WritebackDISPCLK, - CalculateWriteBackDISPCLK( - mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceWidth[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.ip.writeback_line_buffer_buffer_size, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz)); - } - } - - locals->Dispclk_calculated = locals->WritebackDISPCLK; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - locals->Dispclk_calculated = dml_max(locals->Dispclk_calculated, CalculateRequiredDispclk( - mode_lib->ms.cache_display_cfg.hw.ODMMode[k], - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.soc.dcn_downspread_percent, - mode_lib->ms.ip.dispclk_ramp_margin_percent, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, - mode_lib->ms.max_state.dispclk_mhz)); - } - } - if (clk_cfg->dispclk_option == dml_use_required_freq) - locals->Dispclk = locals->Dispclk_calculated; - else if (clk_cfg->dispclk_option == dml_use_override_freq) - locals->Dispclk = clk_cfg->dispclk_mhz; - else - locals->Dispclk = mode_lib->ms.state.dispclk_mhz; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Using Dispclk = %f\n", __func__, locals->Dispclk); -#endif - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateSinglePipeDPPCLKAndSCLThroughput( - mode_lib->ms.cache_display_cfg.plane.HRatio[k], - mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - mode_lib->ms.ip.max_dchub_pscl_bw_pix_per_clk, - mode_lib->ms.ip.max_pscl_lb_bw_pix_per_clk, - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.plane.HTaps[k], - mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k], - mode_lib->ms.cache_display_cfg.plane.VTaps[k], - mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k], - - /* Output */ - &locals->PSCL_THROUGHPUT[k], - &locals->PSCL_THROUGHPUT_CHROMA[k], - &locals->DPPCLKUsingSingleDPP[k]); - } - - CalculateDPPCLK(mode_lib->ms.num_active_planes, - mode_lib->ms.soc.dcn_downspread_percent, - mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, - locals->DPPCLKUsingSingleDPP, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - /* Output */ - &locals->GlobalDPPCLK, - locals->Dppclk_calculated); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (clk_cfg->dppclk_option[k] == dml_use_required_freq) - locals->Dppclk[k] = locals->Dppclk_calculated[k]; - else if (clk_cfg->dppclk_option[k] == dml_use_override_freq) - locals->Dppclk[k] = clk_cfg->dppclk_mhz[k]; - else - locals->Dppclk[k] = mode_lib->ms.state.dppclk_mhz; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Using Dppclk[%0d] = %f\n", __func__, k, locals->Dppclk[k]); -#endif - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateBytePerPixelAndBlockSizes( - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], - - /* Output */ - &locals->BytePerPixelY[k], - &locals->BytePerPixelC[k], - &locals->BytePerPixelDETY[k], - &locals->BytePerPixelDETC[k], - &locals->BlockHeight256BytesY[k], - &locals->BlockHeight256BytesC[k], - &locals->BlockWidth256BytesY[k], - &locals->BlockWidth256BytesC[k], - &locals->BlockHeightY[k], - &locals->BlockHeightC[k], - &locals->BlockWidthY[k], - &locals->BlockWidthC[k]); - } - - - dml_print("DML::%s: %u\n", __func__, __LINE__); - CalculateSwathWidth( - false, // ForceSingleDPP - mode_lib->ms.num_active_planes, - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat, - mode_lib->ms.cache_display_cfg.plane.SourceScan, - mode_lib->ms.cache_display_cfg.plane.ViewportStationary, - mode_lib->ms.cache_display_cfg.plane.ViewportWidth, - mode_lib->ms.cache_display_cfg.plane.ViewportHeight, - mode_lib->ms.cache_display_cfg.plane.ViewportXStart, - mode_lib->ms.cache_display_cfg.plane.ViewportYStart, - mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, - mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, - mode_lib->ms.cache_display_cfg.hw.ODMMode, - locals->BytePerPixelY, - locals->BytePerPixelC, - locals->BlockHeight256BytesY, - locals->BlockHeight256BytesC, - locals->BlockWidth256BytesY, - locals->BlockWidth256BytesC, - mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming, - mode_lib->ms.cache_display_cfg.timing.HActive, - mode_lib->ms.cache_display_cfg.plane.HRatio, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - - /* Output */ - locals->SwathWidthSingleDPPY, - locals->SwathWidthSingleDPPC, - locals->SwathWidthY, - locals->SwathWidthC, - s->dummy_integer_array[0], // dml_uint_t MaximumSwathHeightY[] - s->dummy_integer_array[1], // dml_uint_t MaximumSwathHeightC[] - locals->swath_width_luma_ub, - locals->swath_width_chroma_ub); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - locals->ReadBandwidthSurfaceLuma[k] = locals->SwathWidthSingleDPPY[k] * locals->BytePerPixelY[k] / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - locals->ReadBandwidthSurfaceChroma[k] = locals->SwathWidthSingleDPPC[k] * locals->BytePerPixelC[k] / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; - dml_print("DML::%s: ReadBandwidthSurfaceLuma[%i] = %fBps\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); - dml_print("DML::%s: ReadBandwidthSurfaceChroma[%i] = %fBps\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); - } - - CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; - CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; - CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; - CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; - CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->Output = s->dummy_output_encoder_array; - CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = locals->ReadBandwidthSurfaceLuma; - CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = locals->ReadBandwidthSurfaceChroma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = s->dummy_single_array[0]; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = s->dummy_single_array[1]; - CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; - CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; - CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; - CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; - CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; - CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; - CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; - CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; - CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; - CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; - CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; - CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; - CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; - CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = locals->BlockHeight256BytesY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = locals->BlockHeight256BytesC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = locals->BlockWidth256BytesY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = locals->BlockWidth256BytesC; - CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.cache_display_cfg.hw.ODMMode; - CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateSwathAndDETConfiguration_params->BytePerPixY = locals->BytePerPixelY; - CalculateSwathAndDETConfiguration_params->BytePerPixC = locals->BytePerPixelC; - CalculateSwathAndDETConfiguration_params->BytePerPixDETY = locals->BytePerPixelDETY; - CalculateSwathAndDETConfiguration_params->BytePerPixDETC = locals->BytePerPixelDETC; - CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; - CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; - CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_long_array[0]; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_long_array[1]; - CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_long_array[2]; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_long_array[3]; - CalculateSwathAndDETConfiguration_params->SwathHeightY = locals->SwathHeightY; - CalculateSwathAndDETConfiguration_params->SwathHeightC = locals->SwathHeightC; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = locals->DETBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = locals->DETBufferSizeY; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = locals->DETBufferSizeC; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &locals->UnboundedRequestEnabled; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &locals->compbuf_reserved_space_64b; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &locals->compbuf_reserved_space_zs; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &locals->CompressedBufferSizeInkByte; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = &s->dummy_boolean_array[0][0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[0]; - - // VBA_DELTA - // Calculate DET size, swath height here. In VBA, they are calculated in mode check stage - CalculateSwathAndDETConfiguration(&mode_lib->scratch, - CalculateSwathAndDETConfiguration_params); - - // DCFCLK Deep Sleep - CalculateDCFCLKDeepSleep( - mode_lib->ms.num_active_planes, - locals->BytePerPixelY, - locals->BytePerPixelC, - mode_lib->ms.cache_display_cfg.plane.VRatio, - mode_lib->ms.cache_display_cfg.plane.VRatioChroma, - locals->SwathWidthY, - locals->SwathWidthC, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - mode_lib->ms.cache_display_cfg.plane.HRatio, - mode_lib->ms.cache_display_cfg.plane.HRatioChroma, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - locals->PSCL_THROUGHPUT, - locals->PSCL_THROUGHPUT_CHROMA, - locals->Dppclk, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - mode_lib->ms.soc.return_bus_width_bytes, - - /* Output */ - &locals->DCFCLKDeepSleep); - - // DSCCLK - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if ((mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] != k) || !mode_lib->ms.cache_display_cfg.hw.DSCEnabled[k]) { - locals->DSCCLK_calculated[k] = 0.0; - } else { - if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420) - s->DSCFormatFactor = 2; - else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_444) - s->DSCFormatFactor = 1; - else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) - s->DSCFormatFactor = 2; - else - s->DSCFormatFactor = 1; - if (mode_lib->ms.cache_display_cfg.hw.ODMMode[k] == dml_odm_mode_combine_4to1) - locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 12 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); - else if (mode_lib->ms.cache_display_cfg.hw.ODMMode[k] == dml_odm_mode_combine_2to1) - locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 6 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); - else - locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 3 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); - } - } - - // DSC Delay - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - locals->DSCDelay[k] = DSCDelayRequirement(mode_lib->ms.cache_display_cfg.hw.DSCEnabled[k], - mode_lib->ms.cache_display_cfg.hw.ODMMode[k], - mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], - mode_lib->ms.cache_display_cfg.output.OutputBpp[k], - mode_lib->ms.cache_display_cfg.timing.HActive[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k], - mode_lib->ms.cache_display_cfg.hw.NumberOfDSCSlices[k], - mode_lib->ms.cache_display_cfg.output.OutputFormat[k], - mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], - mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) // NumberOfSurfaces - if (j != k && mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == j && mode_lib->ms.cache_display_cfg.hw.DSCEnabled[j]) - locals->DSCDelay[k] = locals->DSCDelay[j]; - - // Prefetch - CalculateSurfaceSizeInMall( - mode_lib->ms.num_active_planes, - mode_lib->ms.soc.mall_allocated_for_dcn_mbytes, - mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen, - mode_lib->ms.cache_display_cfg.surface.DCCEnable, - mode_lib->ms.cache_display_cfg.plane.ViewportStationary, - mode_lib->ms.cache_display_cfg.plane.ViewportXStart, - mode_lib->ms.cache_display_cfg.plane.ViewportYStart, - mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, - mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, - mode_lib->ms.cache_display_cfg.plane.ViewportWidth, - mode_lib->ms.cache_display_cfg.plane.ViewportHeight, - locals->BytePerPixelY, - mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma, - mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma, - locals->BytePerPixelC, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, - locals->BlockWidth256BytesY, - locals->BlockWidth256BytesC, - locals->BlockHeight256BytesY, - locals->BlockHeight256BytesC, - locals->BlockWidthY, - locals->BlockWidthC, - locals->BlockHeightY, - locals->BlockHeightC, - - /* Output */ - locals->SurfaceSizeInTheMALL, - &s->dummy_boolean[0]); /* dml_bool_t *ExceededMALLSize */ - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->SurfaceParameters[k].PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; - s->SurfaceParameters[k].DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; - s->SurfaceParameters[k].SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; - s->SurfaceParameters[k].ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; - s->SurfaceParameters[k].ViewportHeightChroma = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; - s->SurfaceParameters[k].BlockWidth256BytesY = locals->BlockWidth256BytesY[k]; - s->SurfaceParameters[k].BlockHeight256BytesY = locals->BlockHeight256BytesY[k]; - s->SurfaceParameters[k].BlockWidth256BytesC = locals->BlockWidth256BytesC[k]; - s->SurfaceParameters[k].BlockHeight256BytesC = locals->BlockHeight256BytesC[k]; - s->SurfaceParameters[k].BlockWidthY = locals->BlockWidthY[k]; - s->SurfaceParameters[k].BlockHeightY = locals->BlockHeightY[k]; - s->SurfaceParameters[k].BlockWidthC = locals->BlockWidthC[k]; - s->SurfaceParameters[k].BlockHeightC = locals->BlockHeightC[k]; - s->SurfaceParameters[k].InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; - s->SurfaceParameters[k].HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; - s->SurfaceParameters[k].DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; - s->SurfaceParameters[k].SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; - s->SurfaceParameters[k].SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k]; - s->SurfaceParameters[k].BytePerPixelY = locals->BytePerPixelY[k]; - s->SurfaceParameters[k].BytePerPixelC = locals->BytePerPixelC[k]; - s->SurfaceParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - s->SurfaceParameters[k].VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - s->SurfaceParameters[k].VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; - s->SurfaceParameters[k].VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps[k]; - s->SurfaceParameters[k].VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]; - s->SurfaceParameters[k].PitchY = mode_lib->ms.cache_display_cfg.surface.PitchY[k]; - s->SurfaceParameters[k].DCCMetaPitchY = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; - s->SurfaceParameters[k].PitchC = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; - s->SurfaceParameters[k].DCCMetaPitchC = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; - s->SurfaceParameters[k].ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary[k]; - s->SurfaceParameters[k].ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart[k]; - s->SurfaceParameters[k].ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart[k]; - s->SurfaceParameters[k].ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC[k]; - s->SurfaceParameters[k].ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC[k]; - s->SurfaceParameters[k].FORCE_ONE_ROW_FOR_FRAME = mode_lib->ms.cache_display_cfg.plane.ForceOneRowForFrame[k]; - s->SurfaceParameters[k].SwathHeightY = locals->SwathHeightY[k]; - s->SurfaceParameters[k].SwathHeightC = locals->SwathHeightC[k]; - } - - CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateVMRowAndSwath_params->myPipe = s->SurfaceParameters; - CalculateVMRowAndSwath_params->SurfaceSizeInMALL = locals->SurfaceSizeInTheMALL; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; - CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; - CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; - CalculateVMRowAndSwath_params->SwathWidthY = locals->SwathWidthY; - CalculateVMRowAndSwath_params->SwathWidthC = locals->SwathWidthC; - CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; - CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; - CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; - CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = s->dummy_boolean_array[0]; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = s->dummy_boolean_array[1]; - CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = locals->dpte_row_width_luma_ub; - CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = locals->dpte_row_width_chroma_ub; - CalculateVMRowAndSwath_params->dpte_row_height_luma = locals->dpte_row_height; - CalculateVMRowAndSwath_params->dpte_row_height_chroma = locals->dpte_row_height_chroma; - CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = locals->dpte_row_height_linear; - CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = locals->dpte_row_height_linear_chroma; - CalculateVMRowAndSwath_params->meta_req_width = locals->meta_req_width; - CalculateVMRowAndSwath_params->meta_req_width_chroma = locals->meta_req_width_chroma; - CalculateVMRowAndSwath_params->meta_req_height = locals->meta_req_height; - CalculateVMRowAndSwath_params->meta_req_height_chroma = locals->meta_req_height_chroma; - CalculateVMRowAndSwath_params->meta_row_width = locals->meta_row_width; - CalculateVMRowAndSwath_params->meta_row_width_chroma = locals->meta_row_width_chroma; - CalculateVMRowAndSwath_params->meta_row_height = locals->meta_row_height; - CalculateVMRowAndSwath_params->meta_row_height_chroma = locals->meta_row_height_chroma; - CalculateVMRowAndSwath_params->vm_group_bytes = locals->vm_group_bytes; - CalculateVMRowAndSwath_params->dpte_group_bytes = locals->dpte_group_bytes; - CalculateVMRowAndSwath_params->PixelPTEReqWidthY = locals->PixelPTEReqWidthY; - CalculateVMRowAndSwath_params->PixelPTEReqHeightY = locals->PixelPTEReqHeightY; - CalculateVMRowAndSwath_params->PTERequestSizeY = locals->PTERequestSizeY; - CalculateVMRowAndSwath_params->PixelPTEReqWidthC = locals->PixelPTEReqWidthC; - CalculateVMRowAndSwath_params->PixelPTEReqHeightC = locals->PixelPTEReqHeightC; - CalculateVMRowAndSwath_params->PTERequestSizeC = locals->PTERequestSizeC; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = locals->dpde0_bytes_per_frame_ub_l; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = locals->meta_pte_bytes_per_frame_ub_l; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = locals->dpde0_bytes_per_frame_ub_c; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = locals->meta_pte_bytes_per_frame_ub_c; - CalculateVMRowAndSwath_params->PrefetchSourceLinesY = locals->PrefetchSourceLinesY; - CalculateVMRowAndSwath_params->PrefetchSourceLinesC = locals->PrefetchSourceLinesC; - CalculateVMRowAndSwath_params->VInitPreFillY = locals->VInitPreFillY; - CalculateVMRowAndSwath_params->VInitPreFillC = locals->VInitPreFillC; - CalculateVMRowAndSwath_params->MaxNumSwathY = locals->MaxNumSwathY; - CalculateVMRowAndSwath_params->MaxNumSwathC = locals->MaxNumSwathC; - CalculateVMRowAndSwath_params->meta_row_bw = locals->meta_row_bw; - CalculateVMRowAndSwath_params->dpte_row_bw = locals->dpte_row_bw; - CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = locals->PixelPTEBytesPerRow; - CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = locals->PDEAndMetaPTEBytesFrame; - CalculateVMRowAndSwath_params->MetaRowByte = locals->MetaRowByte; - CalculateVMRowAndSwath_params->use_one_row_for_frame = locals->use_one_row_for_frame; - CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = locals->use_one_row_for_frame_flip; - CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = locals->UsesMALLForStaticScreen; - CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = locals->PTE_BUFFER_MODE; - CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = locals->BIGK_FRAGMENT_SIZE; - - CalculateVMRowAndSwath(&mode_lib->scratch, - CalculateVMRowAndSwath_params); - - s->ReorderBytes = (dml_uint_t)(mode_lib->ms.soc.num_chans * dml_max3( - mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_only_bytes, - mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, - mode_lib->ms.soc.urgent_out_of_order_return_per_channel_vm_only_bytes)); - - s->VMDataOnlyReturnBW = dml_get_return_bw_mbps_vm_only(&mode_lib->ms.soc, - mode_lib->ms.state.use_ideal_dram_bw_strobe, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - locals->Dcfclk, - mode_lib->ms.FabricClock, - mode_lib->ms.DRAMSpeed); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: locals->Dcfclk = %f\n", __func__, locals->Dcfclk); - dml_print("DML::%s: mode_lib->ms.soc.return_bus_width_bytes = %u\n", __func__, mode_lib->ms.soc.return_bus_width_bytes); - dml_print("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); - dml_print("DML::%s: mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes = %u\n", __func__, mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes); - dml_print("DML::%s: mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent = %f\n", __func__, mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent); - dml_print("DML::%s: mode_lib->ms.DRAMSpeed = %f\n", __func__, mode_lib->ms.DRAMSpeed); - dml_print("DML::%s: mode_lib->ms.soc.num_chans = %u\n", __func__, mode_lib->ms.soc.num_chans); - dml_print("DML::%s: mode_lib->ms.soc.dram_channel_width_bytes = %u\n", __func__, mode_lib->ms.soc.dram_channel_width_bytes); - dml_print("DML::%s: mode_lib->ms.state_idx = %u\n", __func__, mode_lib->ms.state_idx); - dml_print("DML::%s: mode_lib->ms.max_state_idx = %u\n", __func__, mode_lib->ms.max_state_idx); - dml_print("DML::%s: mode_lib->ms.state.use_ideal_dram_bw_strobe = %u\n", __func__, mode_lib->ms.state.use_ideal_dram_bw_strobe); - dml_print("DML::%s: VMDataOnlyReturnBW = %f\n", __func__, s->VMDataOnlyReturnBW); - dml_print("DML::%s: ReturnBW = %f\n", __func__, mode_lib->ms.ReturnBW); -#endif - - s->HostVMInefficiencyFactor = 1.0; - if (mode_lib->ms.cache_display_cfg.plane.GPUVMEnable && mode_lib->ms.cache_display_cfg.plane.HostVMEnable) - s->HostVMInefficiencyFactor = mode_lib->ms.ReturnBW / s->VMDataOnlyReturnBW; - - s->TotalDCCActiveDPP = 0; - s->TotalActiveDPP = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->TotalActiveDPP = s->TotalActiveDPP + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; - if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]) - s->TotalDCCActiveDPP = s->TotalDCCActiveDPP + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; - } - - locals->UrgentExtraLatency = CalculateExtraLatency( - mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles, - s->ReorderBytes, - locals->Dcfclk, - s->TotalActiveDPP, - mode_lib->ms.ip.pixel_chunk_size_kbytes, - s->TotalDCCActiveDPP, - mode_lib->ms.ip.meta_chunk_size_kbytes, - mode_lib->ms.ReturnBW, - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.num_active_planes, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - locals->dpte_group_bytes, - s->HostVMInefficiencyFactor, - mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, - mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels); - - locals->TCalc = 24.0 / locals->DCFCLKDeepSleep; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - locals->WritebackDelay[k] = - mode_lib->ms.state.writeback_latency_us - + CalculateWriteBackDelay( - mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / locals->Dispclk; - } else - locals->WritebackDelay[k] = 0; - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[j] == k - && mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[j] == true) { - locals->WritebackDelay[k] = - dml_max( - locals->WritebackDelay[k], - mode_lib->ms.state.writeback_latency_us - + CalculateWriteBackDelay( - mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[j], - mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[j], - mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / locals->Dispclk); - } - } - } - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) - for (j = 0; j < mode_lib->ms.num_active_planes; ++j) - if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == j) - locals->WritebackDelay[k] = locals->WritebackDelay[j]; - - locals->UrgentLatency = CalculateUrgentLatency(mode_lib->ms.state.urgent_latency_pixel_data_only_us, - mode_lib->ms.state.urgent_latency_pixel_mixed_with_vm_data_us, - mode_lib->ms.state.urgent_latency_vm_data_only_us, - mode_lib->ms.soc.do_urgent_latency_adjustment, - mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_component_us, - mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_reference_mhz, - mode_lib->ms.FabricClock); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateUrgentBurstFactor(mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - locals->swath_width_luma_ub[k], - locals->swath_width_chroma_ub[k], - locals->SwathHeightY[k], - locals->SwathHeightC[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - locals->UrgentLatency, - mode_lib->ms.ip.cursor_buffer_size, - mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], - mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - locals->BytePerPixelDETY[k], - locals->BytePerPixelDETC[k], - locals->DETBufferSizeY[k], - locals->DETBufferSizeC[k], - - /* output */ - &locals->UrgBurstFactorCursor[k], - &locals->UrgBurstFactorLuma[k], - &locals->UrgBurstFactorChroma[k], - &locals->NoUrgentLatencyHiding[k]); - - locals->cursor_bw[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / - ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; - } - - s->VStartupLines = __DML_VBA_MIN_VSTARTUP__; - s->MaxVStartupAllPlanes = 0; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->MaxVStartupLines[k] = CalculateMaxVStartup(k, - mode_lib->ms.ip.ptoi_supported, - mode_lib->ms.ip.vblank_nom_default_us, - &mode_lib->ms.cache_display_cfg.timing, - locals->WritebackDelay[k]); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - dml_print("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, locals->WritebackDelay[k]); -#endif - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) - s->MaxVStartupAllPlanes = (dml_uint_t)(dml_max(s->MaxVStartupAllPlanes, s->MaxVStartupLines[k])); - - s->ImmediateFlipRequirementFinal = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->ImmediateFlipRequirementFinal = s->ImmediateFlipRequirementFinal || (mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: ImmediateFlipRequirementFinal = %u\n", __func__, s->ImmediateFlipRequirementFinal); -#endif - - // The prefetch scheduling should only be calculated once as per AllowForPStateChangeOrStutterInVBlank requirement - // If the AllowForPStateChangeOrStutterInVBlank requirement is not strict (i.e. only try those power saving feature - // if possible, then will try to program for the best power saving features in order of difficulty (dram, fclk, stutter) - s->iteration = 0; - s->MaxTotalRDBandwidth = 0; - s->AllPrefetchModeTested = false; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculatePrefetchMode(mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k], &s->MinPrefetchMode[k], &s->MaxPrefetchMode[k]); - s->NextPrefetchMode[k] = s->MinPrefetchMode[k]; - } - - do { - s->MaxTotalRDBandwidthNoUrgentBurst = 0.0; - s->DestinationLineTimesForPrefetchLessThan2 = false; - s->VRatioPrefetchMoreThanMax = false; - - dml_print("DML::%s: Start one iteration: VStartupLines = %u\n", __func__, s->VStartupLines); - - s->AllPrefetchModeTested = true; - s->MaxTotalRDBandwidth = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - locals->PrefetchMode[k] = s->NextPrefetchMode[k]; - TWait = CalculateTWait( - locals->PrefetchMode[k], - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal, - mode_lib->ms.cache_display_cfg.timing.DRRDisplay[k], - mode_lib->ms.state.dram_clock_change_latency_us, - mode_lib->ms.state.fclk_change_latency_us, - locals->UrgentLatency, - mode_lib->ms.state.sr_enter_plus_exit_time_us); - - myPipe = &s->myPipe; - myPipe->Dppclk = locals->Dppclk[k]; - myPipe->Dispclk = locals->Dispclk; - myPipe->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; - myPipe->DCFClkDeepSleep = locals->DCFCLKDeepSleep; - myPipe->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; - myPipe->ScalerEnabled = mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k]; - myPipe->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; - myPipe->BlockWidth256BytesY = locals->BlockWidth256BytesY[k]; - myPipe->BlockHeight256BytesY = locals->BlockHeight256BytesY[k]; - myPipe->BlockWidth256BytesC = locals->BlockWidth256BytesC[k]; - myPipe->BlockHeight256BytesC = locals->BlockHeight256BytesC[k]; - myPipe->InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; - myPipe->NumberOfCursors = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k]; - myPipe->VBlank = mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]; - myPipe->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; - myPipe->HActive = mode_lib->ms.cache_display_cfg.timing.HActive[k]; - myPipe->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; - myPipe->ODMMode = mode_lib->ms.cache_display_cfg.hw.ODMMode[k]; - myPipe->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; - myPipe->BytePerPixelY = locals->BytePerPixelY[k]; - myPipe->BytePerPixelC = locals->BytePerPixelC[k]; - myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); - dml_print("DML::%s: AllowForPStateChangeOrStutterInVBlank = %u\n", __func__, mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k]); - dml_print("DML::%s: PrefetchMode[k] = %u (Min=%u Max=%u)\n", __func__, locals->PrefetchMode[k], s->MinPrefetchMode[k], s->MaxPrefetchMode[k]); -#endif - - CalculatePrefetchSchedule_params->EnhancedPrefetchScheduleAccelerationFinal = mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal; - CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactor; - CalculatePrefetchSchedule_params->myPipe = myPipe; - CalculatePrefetchSchedule_params->DSCDelay = locals->DSCDelay[k]; - CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ms.ip.dppclk_delay_subtotal + mode_lib->ms.ip.dppclk_delay_cnvc_formatter; - CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ms.ip.dppclk_delay_scl; - CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ms.ip.dppclk_delay_scl_lb_only; - CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ms.ip.dppclk_delay_cnvc_cursor; - CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ms.ip.dispclk_delay_subtotal; - CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (dml_uint_t)(locals->SwathWidthY[k] / mode_lib->ms.cache_display_cfg.plane.HRatio[k]); - CalculatePrefetchSchedule_params->OutputFormat = mode_lib->ms.cache_display_cfg.output.OutputFormat[k]; - CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; - CalculatePrefetchSchedule_params->VStartup = (dml_uint_t)(dml_min(s->VStartupLines, s->MaxVStartupLines[k])); - CalculatePrefetchSchedule_params->MaxVStartup = s->MaxVStartupLines[k]; - CalculatePrefetchSchedule_params->GPUVMPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; - CalculatePrefetchSchedule_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; - CalculatePrefetchSchedule_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; - CalculatePrefetchSchedule_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; - CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; - CalculatePrefetchSchedule_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]; - CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; - CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired[k]; - CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes[k]; - CalculatePrefetchSchedule_params->UrgentLatency = locals->UrgentLatency; - CalculatePrefetchSchedule_params->UrgentExtraLatency = locals->UrgentExtraLatency; - CalculatePrefetchSchedule_params->TCalc = locals->TCalc; - CalculatePrefetchSchedule_params->PDEAndMetaPTEBytesFrame = locals->PDEAndMetaPTEBytesFrame[k]; - CalculatePrefetchSchedule_params->MetaRowByte = locals->MetaRowByte[k]; - CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = locals->PixelPTEBytesPerRow[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesY = locals->PrefetchSourceLinesY[k]; - CalculatePrefetchSchedule_params->VInitPreFillY = locals->VInitPreFillY[k]; - CalculatePrefetchSchedule_params->MaxNumSwathY = locals->MaxNumSwathY[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesC = locals->PrefetchSourceLinesC[k]; - CalculatePrefetchSchedule_params->VInitPreFillC = locals->VInitPreFillC[k]; - CalculatePrefetchSchedule_params->MaxNumSwathC = locals->MaxNumSwathC[k]; - CalculatePrefetchSchedule_params->swath_width_luma_ub = locals->swath_width_luma_ub[k]; - CalculatePrefetchSchedule_params->swath_width_chroma_ub = locals->swath_width_chroma_ub[k]; - CalculatePrefetchSchedule_params->SwathHeightY = locals->SwathHeightY[k]; - CalculatePrefetchSchedule_params->SwathHeightC = locals->SwathHeightC[k]; - CalculatePrefetchSchedule_params->TWait = TWait; - CalculatePrefetchSchedule_params->DSTXAfterScaler = &locals->DSTXAfterScaler[k]; - CalculatePrefetchSchedule_params->DSTYAfterScaler = &locals->DSTYAfterScaler[k]; - CalculatePrefetchSchedule_params->DestinationLinesForPrefetch = &locals->DestinationLinesForPrefetch[k]; - CalculatePrefetchSchedule_params->DestinationLinesToRequestVMInVBlank = &locals->DestinationLinesToRequestVMInVBlank[k]; - CalculatePrefetchSchedule_params->DestinationLinesToRequestRowInVBlank = &locals->DestinationLinesToRequestRowInVBlank[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchY = &locals->VRatioPrefetchY[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchC = &locals->VRatioPrefetchC[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWLuma = &locals->RequiredPrefetchPixDataBWLuma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWChroma = &locals->RequiredPrefetchPixDataBWChroma[k]; - CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &locals->NotEnoughTimeForDynamicMetadata[k]; - CalculatePrefetchSchedule_params->Tno_bw = &locals->Tno_bw[k]; - CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &locals->prefetch_vmrow_bw[k]; - CalculatePrefetchSchedule_params->Tdmdl_vm = &locals->Tdmdl_vm[k]; - CalculatePrefetchSchedule_params->Tdmdl = &locals->Tdmdl[k]; - CalculatePrefetchSchedule_params->TSetup = &locals->TSetup[k]; - CalculatePrefetchSchedule_params->VUpdateOffsetPix = &locals->VUpdateOffsetPix[k]; - CalculatePrefetchSchedule_params->VUpdateWidthPix = &locals->VUpdateWidthPix[k]; - CalculatePrefetchSchedule_params->VReadyOffsetPix = &locals->VReadyOffsetPix[k]; - - locals->NoTimeToPrefetch[k] = - CalculatePrefetchSchedule(&mode_lib->scratch, - CalculatePrefetchSchedule_params); - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, locals->NoTimeToPrefetch[k]); -#endif - locals->VStartup[k] = (dml_uint_t)(dml_min(s->VStartupLines, s->MaxVStartupLines[k])); - locals->VStartupMin[k] = locals->VStartup[k]; - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateUrgentBurstFactor( - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], - locals->swath_width_luma_ub[k], - locals->swath_width_chroma_ub[k], - locals->SwathHeightY[k], - locals->SwathHeightC[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - locals->UrgentLatency, - mode_lib->ms.ip.cursor_buffer_size, - mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], - mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], - locals->VRatioPrefetchY[k], - locals->VRatioPrefetchC[k], - locals->BytePerPixelDETY[k], - locals->BytePerPixelDETC[k], - locals->DETBufferSizeY[k], - locals->DETBufferSizeC[k], - /* Output */ - &locals->UrgBurstFactorCursorPre[k], - &locals->UrgBurstFactorLumaPre[k], - &locals->UrgBurstFactorChromaPre[k], - &locals->NoUrgentLatencyHidingPre[k]); - - locals->cursor_bw_pre[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * locals->VRatioPrefetchY[k]; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]); - dml_print("DML::%s: k=%0u UrgBurstFactorLuma=%f\n", __func__, k, locals->UrgBurstFactorLuma[k]); - dml_print("DML::%s: k=%0u UrgBurstFactorChroma=%f\n", __func__, k, locals->UrgBurstFactorChroma[k]); - dml_print("DML::%s: k=%0u UrgBurstFactorLumaPre=%f\n", __func__, k, locals->UrgBurstFactorLumaPre[k]); - dml_print("DML::%s: k=%0u UrgBurstFactorChromaPre=%f\n", __func__, k, locals->UrgBurstFactorChromaPre[k]); - - dml_print("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, locals->VRatioPrefetchY[k]); - dml_print("DML::%s: k=%0u VRatioY=%f\n", __func__, k, mode_lib->ms.cache_display_cfg.plane.VRatio[k]); - - dml_print("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, locals->prefetch_vmrow_bw[k]); - dml_print("DML::%s: k=%0u ReadBandwidthSurfaceLuma=%f\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); - dml_print("DML::%s: k=%0u ReadBandwidthSurfaceChroma=%f\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); - dml_print("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, locals->cursor_bw[k]); - dml_print("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, locals->meta_row_bw[k]); - dml_print("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, locals->dpte_row_bw[k]); - dml_print("DML::%s: k=%0u RequiredPrefetchPixDataBWLuma=%f\n", __func__, k, locals->RequiredPrefetchPixDataBWLuma[k]); - dml_print("DML::%s: k=%0u RequiredPrefetchPixDataBWChroma=%f\n", __func__, k, locals->RequiredPrefetchPixDataBWChroma[k]); - dml_print("DML::%s: k=%0u cursor_bw_pre=%f\n", __func__, k, locals->cursor_bw_pre[k]); - dml_print("DML::%s: k=%0u MaxTotalRDBandwidthNoUrgentBurst=%f\n", __func__, k, s->MaxTotalRDBandwidthNoUrgentBurst); -#endif - if (locals->DestinationLinesForPrefetch[k] < 2) - s->DestinationLineTimesForPrefetchLessThan2 = true; - - if (locals->VRatioPrefetchY[k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || - locals->VRatioPrefetchC[k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || - ((s->VStartupLines < s->MaxVStartupLines[k] || mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal == 0) && - (locals->VRatioPrefetchY[k] > __DML_MAX_VRATIO_PRE__ || locals->VRatioPrefetchC[k] > __DML_MAX_VRATIO_PRE__))) - s->VRatioPrefetchMoreThanMax = true; - - //dml_bool_t DestinationLinesToRequestVMInVBlankEqualOrMoreThan32 = false; - //dml_bool_t DestinationLinesToRequestRowInVBlankEqualOrMoreThan16 = false; - //if (locals->DestinationLinesToRequestVMInVBlank[k] >= 32) { - // DestinationLinesToRequestVMInVBlankEqualOrMoreThan32 = true; - //} - - //if (locals->DestinationLinesToRequestRowInVBlank[k] >= 16) { - // DestinationLinesToRequestRowInVBlankEqualOrMoreThan16 = true; - //} - } - - locals->FractionOfUrgentBandwidth = s->MaxTotalRDBandwidthNoUrgentBurst / mode_lib->ms.ReturnBW; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: MaxTotalRDBandwidthNoUrgentBurst=%f \n", __func__, s->MaxTotalRDBandwidthNoUrgentBurst); - dml_print("DML::%s: ReturnBW=%f \n", __func__, mode_lib->ms.ReturnBW); - dml_print("DML::%s: FractionOfUrgentBandwidth=%f \n", __func__, locals->FractionOfUrgentBandwidth); -#endif - - CalculatePrefetchBandwithSupport( - mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBW, - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - locals->NoUrgentLatencyHidingPre, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->RequiredPrefetchPixDataBWLuma, - locals->RequiredPrefetchPixDataBWChroma, - locals->cursor_bw, - locals->meta_row_bw, - locals->dpte_row_bw, - locals->cursor_bw_pre, - locals->prefetch_vmrow_bw, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - locals->UrgBurstFactorLuma, - locals->UrgBurstFactorChroma, - locals->UrgBurstFactorCursor, - locals->UrgBurstFactorLumaPre, - locals->UrgBurstFactorChromaPre, - locals->UrgBurstFactorCursorPre, - - /* output */ - &s->MaxTotalRDBandwidth, // dml_float_t *PrefetchBandwidth - &s->MaxTotalRDBandwidthNotIncludingMALLPrefetch, // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch - &s->dummy_single[0], // dml_float_t *FractionOfUrgentBandwidth - &locals->PrefetchModeSupported); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) - s->dummy_unit_vector[k] = 1.0; - - CalculatePrefetchBandwithSupport(mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBW, - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - locals->NoUrgentLatencyHidingPre, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->RequiredPrefetchPixDataBWLuma, - locals->RequiredPrefetchPixDataBWChroma, - locals->cursor_bw, - locals->meta_row_bw, - locals->dpte_row_bw, - locals->cursor_bw_pre, - locals->prefetch_vmrow_bw, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - - /* output */ - &s->NonUrgentMaxTotalRDBandwidth, // dml_float_t *PrefetchBandwidth - &s->NonUrgentMaxTotalRDBandwidthNotIncludingMALLPrefetch, // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch - &locals->FractionOfUrgentBandwidth, - &s->dummy_boolean[0]); // dml_bool_t *PrefetchBandwidthSupport - - - - if (s->VRatioPrefetchMoreThanMax != false || s->DestinationLineTimesForPrefetchLessThan2 != false) { - dml_print("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); - dml_print("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); - locals->PrefetchModeSupported = false; - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (locals->NoTimeToPrefetch[k] == true || locals->NotEnoughTimeForDynamicMetadata[k]) { - dml_print("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, locals->NoTimeToPrefetch[k]); - dml_print("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, locals->NotEnoughTimeForDynamicMetadata[k]); - locals->PrefetchModeSupported = false; - } - } - - - if (locals->PrefetchModeSupported == true && mode_lib->ms.support.ImmediateFlipSupport == true) { - locals->BandwidthAvailableForImmediateFlip = CalculateBandwidthAvailableForImmediateFlip( - mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBW, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->RequiredPrefetchPixDataBWLuma, - locals->RequiredPrefetchPixDataBWChroma, - locals->cursor_bw, - locals->cursor_bw_pre, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - locals->UrgBurstFactorLuma, - locals->UrgBurstFactorChroma, - locals->UrgBurstFactorCursor, - locals->UrgBurstFactorLumaPre, - locals->UrgBurstFactorChromaPre, - locals->UrgBurstFactorCursorPre); - - locals->TotImmediateFlipBytes = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) { - locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * (locals->PDEAndMetaPTEBytesFrame[k] + locals->MetaRowByte[k]); - if (locals->use_one_row_for_frame_flip[k]) { - locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * (2 * locals->PixelPTEBytesPerRow[k]); - } else { - locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * locals->PixelPTEBytesPerRow[k]; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k = %u\n", __func__, k); - dml_print("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]); - dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, locals->PDEAndMetaPTEBytesFrame[k]); - dml_print("DML::%s: MetaRowByte = %u\n", __func__, locals->MetaRowByte[k]); - dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, locals->PixelPTEBytesPerRow[k]); - dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, locals->TotImmediateFlipBytes); -#endif - } - } - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - CalculateFlipSchedule( - s->HostVMInefficiencyFactor, - locals->UrgentExtraLatency, - locals->UrgentLatency, - mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, - mode_lib->ms.cache_display_cfg.plane.HostVMEnable, - mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels, - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, - locals->PDEAndMetaPTEBytesFrame[k], - locals->MetaRowByte[k], - locals->PixelPTEBytesPerRow[k], - locals->BandwidthAvailableForImmediateFlip, - locals->TotImmediateFlipBytes, - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], - mode_lib->ms.cache_display_cfg.plane.VRatio[k], - mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], - locals->Tno_bw[k], - mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], - locals->dpte_row_height[k], - locals->meta_row_height[k], - locals->dpte_row_height_chroma[k], - locals->meta_row_height_chroma[k], - locals->use_one_row_for_frame_flip[k], - - /* Output */ - &locals->DestinationLinesToRequestVMInImmediateFlip[k], - &locals->DestinationLinesToRequestRowInImmediateFlip[k], - &locals->final_flip_bw[k], - &locals->ImmediateFlipSupportedForPipe[k]); - } - - CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBW, - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - mode_lib->ms.policy.ImmediateFlipRequirement, - locals->final_flip_bw, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->RequiredPrefetchPixDataBWLuma, - locals->RequiredPrefetchPixDataBWChroma, - locals->cursor_bw, - locals->meta_row_bw, - locals->dpte_row_bw, - locals->cursor_bw_pre, - locals->prefetch_vmrow_bw, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - locals->UrgBurstFactorLuma, - locals->UrgBurstFactorChroma, - locals->UrgBurstFactorCursor, - locals->UrgBurstFactorLumaPre, - locals->UrgBurstFactorChromaPre, - locals->UrgBurstFactorCursorPre, - - /* output */ - &locals->total_dcn_read_bw_with_flip, // dml_float_t *TotalBandwidth - &locals->total_dcn_read_bw_with_flip_not_including_MALL_prefetch, // dml_float_t TotalBandwidthNotIncludingMALLPrefetch - &s->dummy_single[0], // dml_float_t *FractionOfUrgentBandwidth - &locals->ImmediateFlipSupported); // dml_bool_t *ImmediateFlipBandwidthSupport - - CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, - mode_lib->ms.ReturnBW, - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - mode_lib->ms.policy.ImmediateFlipRequirement, - locals->final_flip_bw, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->RequiredPrefetchPixDataBWLuma, - locals->RequiredPrefetchPixDataBWChroma, - locals->cursor_bw, - locals->meta_row_bw, - locals->dpte_row_bw, - locals->cursor_bw_pre, - locals->prefetch_vmrow_bw, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - s->dummy_unit_vector, - - /* output */ - &locals->non_urgent_total_dcn_read_bw_with_flip, // dml_float_t *TotalBandwidth - &locals->non_urgent_total_dcn_read_bw_with_flip_not_including_MALL_prefetch, // dml_float_t TotalBandwidthNotIncludingMALLPrefetch - &locals->FractionOfUrgentBandwidthImmediateFlip, // dml_float_t *FractionOfUrgentBandwidth - &s->dummy_boolean[0]); // dml_bool_t *ImmediateFlipBandwidthSupport - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required && locals->ImmediateFlipSupportedForPipe[k] == false) { - locals->ImmediateFlipSupported = false; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Pipe %0d not supporting iflip\n", __func__, k); -#endif - } - } - } else { - locals->ImmediateFlipSupported = false; - locals->total_dcn_read_bw_with_flip = s->MaxTotalRDBandwidth; - locals->total_dcn_read_bw_with_flip_not_including_MALL_prefetch = s->MaxTotalRDBandwidthNotIncludingMALLPrefetch; - locals->non_urgent_total_dcn_read_bw_with_flip = s->NonUrgentMaxTotalRDBandwidth; - locals->non_urgent_total_dcn_read_bw_with_flip_not_including_MALL_prefetch = s->NonUrgentMaxTotalRDBandwidthNotIncludingMALLPrefetch; - } - - /* consider flip support is okay if the flip bw is ok or (when user does't require a iflip and there is no host vm) */ - locals->PrefetchAndImmediateFlipSupported = (locals->PrefetchModeSupported == true && - ((!mode_lib->ms.support.ImmediateFlipSupport && !mode_lib->ms.cache_display_cfg.plane.HostVMEnable && !s->ImmediateFlipRequirementFinal) || - locals->ImmediateFlipSupported)) ? true : false; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: PrefetchModeSupported = %u\n", __func__, locals->PrefetchModeSupported); - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) - dml_print("DML::%s: ImmediateFlipRequirement[%u] = %u\n", __func__, k, mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); - dml_print("DML::%s: HostVMEnable = %u\n", __func__, mode_lib->ms.cache_display_cfg.plane.HostVMEnable); - dml_print("DML::%s: ImmediateFlipSupport = %u (from mode_support)\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); - dml_print("DML::%s: ImmediateFlipSupported = %u\n", __func__, locals->ImmediateFlipSupported); - dml_print("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, locals->PrefetchAndImmediateFlipSupported); -#endif - dml_print("DML::%s: Done one iteration: VStartupLines=%u, MaxVStartupAllPlanes=%u\n", __func__, s->VStartupLines, s->MaxVStartupAllPlanes); - - s->VStartupLines = s->VStartupLines + 1; - - if (s->VStartupLines > s->MaxVStartupAllPlanes) { - s->VStartupLines = __DML_VBA_MIN_VSTARTUP__; - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - s->NextPrefetchMode[k] = s->NextPrefetchMode[k] + 1; - - if (s->NextPrefetchMode[k] <= s->MaxPrefetchMode[k]) - s->AllPrefetchModeTested = false; - dml_print("DML::%s: VStartupLines=%u, reaches max vstartup, try next prefetch mode=%u\n", __func__, s->VStartupLines-1, s->AllPrefetchModeTested); - } - } else { - s->AllPrefetchModeTested = false; - } - s->iteration++; - if (s->iteration > 2500) { - dml_print("ERROR: DML::%s: Too many errors, exit now\n", __func__); - ASSERT(0); - } - } while (!(locals->PrefetchAndImmediateFlipSupported || s->AllPrefetchModeTested)); - - if (locals->PrefetchAndImmediateFlipSupported) { - dml_print("DML::%s: Good, Prefetch and flip scheduling solution found at VStartupLines=%u (MaxVStartupAllPlanes=%u)\n", __func__, s->VStartupLines-1, s->MaxVStartupAllPlanes); - } else { - dml_print("DML::%s: Bad, Prefetch and flip scheduling solution did NOT find solution! (MaxVStartupAllPlanes=%u)\n", __func__, s->MaxVStartupAllPlanes); - } - - //Watermarks and NB P-State/DRAM Clock Change Support - { - s->mmSOCParameters.UrgentLatency = locals->UrgentLatency; - s->mmSOCParameters.ExtraLatency = locals->UrgentExtraLatency; - s->mmSOCParameters.WritebackLatency = mode_lib->ms.state.writeback_latency_us; - s->mmSOCParameters.DRAMClockChangeLatency = mode_lib->ms.state.dram_clock_change_latency_us; - s->mmSOCParameters.FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; - s->mmSOCParameters.SRExitTime = mode_lib->ms.state.sr_exit_time_us; - s->mmSOCParameters.SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; - s->mmSOCParameters.SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; - s->mmSOCParameters.SREnterPlusExitZ8Time = mode_lib->ms.state.sr_enter_plus_exit_z8_time_us; - s->mmSOCParameters.USRRetrainingLatency = mode_lib->ms.state.usr_retraining_latency_us; - s->mmSOCParameters.SMNLatency = mode_lib->ms.soc.smn_latency_us; - - CalculateWatermarks_params->USRRetrainingRequiredFinal = mode_lib->ms.policy.USRRetrainingRequiredFinal; - CalculateWatermarks_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateWatermarks_params->PrefetchMode = locals->PrefetchMode; - CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ms.ip.max_line_buffer_lines; - CalculateWatermarks_params->LineBufferSize = mode_lib->ms.ip.line_buffer_size_bits; - CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ms.ip.writeback_interface_buffer_size_kbytes; - CalculateWatermarks_params->DCFCLK = locals->Dcfclk; - CalculateWatermarks_params->ReturnBW = mode_lib->ms.ReturnBW; - CalculateWatermarks_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; - CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; - CalculateWatermarks_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; - CalculateWatermarks_params->dpte_group_bytes = locals->dpte_group_bytes; - CalculateWatermarks_params->meta_row_height = locals->meta_row_height; - CalculateWatermarks_params->meta_row_height_chroma = locals->meta_row_height_chroma; - CalculateWatermarks_params->mmSOCParameters = s->mmSOCParameters; - CalculateWatermarks_params->WritebackChunkSize = mode_lib->ms.ip.writeback_chunk_size_kbytes; - CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; - CalculateWatermarks_params->DCFClkDeepSleep = locals->DCFCLKDeepSleep; - CalculateWatermarks_params->DETBufferSizeY = locals->DETBufferSizeY; - CalculateWatermarks_params->DETBufferSizeC = locals->DETBufferSizeC; - CalculateWatermarks_params->SwathHeightY = locals->SwathHeightY; - CalculateWatermarks_params->SwathHeightC = locals->SwathHeightC; - CalculateWatermarks_params->LBBitPerPixel = mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel; - CalculateWatermarks_params->SwathWidthY = locals->SwathWidthY; - CalculateWatermarks_params->SwathWidthC = locals->SwathWidthC; - CalculateWatermarks_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; - CalculateWatermarks_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; - CalculateWatermarks_params->VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps; - CalculateWatermarks_params->VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma; - CalculateWatermarks_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; - CalculateWatermarks_params->VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma; - CalculateWatermarks_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; - CalculateWatermarks_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; - CalculateWatermarks_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; - CalculateWatermarks_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; - CalculateWatermarks_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; - CalculateWatermarks_params->BytePerPixelDETY = locals->BytePerPixelDETY; - CalculateWatermarks_params->BytePerPixelDETC = locals->BytePerPixelDETC; - CalculateWatermarks_params->DSTXAfterScaler = locals->DSTXAfterScaler; - CalculateWatermarks_params->DSTYAfterScaler = locals->DSTYAfterScaler; - CalculateWatermarks_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; - CalculateWatermarks_params->WritebackPixelFormat = mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat; - CalculateWatermarks_params->WritebackDestinationWidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth; - CalculateWatermarks_params->WritebackDestinationHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight; - CalculateWatermarks_params->WritebackSourceHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight; - CalculateWatermarks_params->UnboundedRequestEnabled = locals->UnboundedRequestEnabled; - CalculateWatermarks_params->CompressedBufferSizeInkByte = locals->CompressedBufferSizeInkByte; - - // Output - CalculateWatermarks_params->Watermark = &locals->Watermark; // Watermarks *Watermark - CalculateWatermarks_params->DRAMClockChangeSupport = &locals->DRAMClockChangeSupport; - CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = locals->MaxActiveDRAMClockChangeLatencySupported; // dml_float_t *MaxActiveDRAMClockChangeLatencySupported[] - CalculateWatermarks_params->SubViewportLinesNeededInMALL = locals->SubViewportLinesNeededInMALL; // dml_uint_t SubViewportLinesNeededInMALL[] - CalculateWatermarks_params->FCLKChangeSupport = &locals->FCLKChangeSupport; - CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &locals->MaxActiveFCLKChangeLatencySupported; // dml_float_t *MaxActiveFCLKChangeLatencySupported - CalculateWatermarks_params->USRRetrainingSupport = &locals->USRRetrainingSupport; - - CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( - &mode_lib->scratch, - CalculateWatermarks_params); - - /* Copy the calculated watermarks to mp.Watermark as the getter functions are - * implemented by the DML team to copy the calculated values from the mp.Watermark interface. - * &mode_lib->mp.Watermark and &locals->Watermark are the same address, memcpy may lead to - * unexpected behavior. memmove should be used. - */ - memmove(&mode_lib->mp.Watermark, CalculateWatermarks_params->Watermark, sizeof(struct Watermarks)); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - locals->WritebackAllowDRAMClockChangeEndPosition[k] = dml_max(0, locals->VStartupMin[k] * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / - mode_lib->ms.cache_display_cfg.timing.PixelClock[k] - locals->Watermark.WritebackDRAMClockChangeWatermark); - locals->WritebackAllowFCLKChangeEndPosition[k] = dml_max(0, locals->VStartupMin[k] * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / - mode_lib->ms.cache_display_cfg.timing.PixelClock[k] - locals->Watermark.WritebackFCLKChangeWatermark); - } else { - locals->WritebackAllowDRAMClockChangeEndPosition[k] = 0; - locals->WritebackAllowFCLKChangeEndPosition[k] = 0; - } - } - } - - //Display Pipeline Delivery Time in Prefetch, Groups - CalculatePixelDeliveryTimes( - mode_lib->ms.num_active_planes, - mode_lib->ms.cache_display_cfg.plane.VRatio, - mode_lib->ms.cache_display_cfg.plane.VRatioChroma, - locals->VRatioPrefetchY, - locals->VRatioPrefetchC, - locals->swath_width_luma_ub, - locals->swath_width_chroma_ub, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - mode_lib->ms.cache_display_cfg.plane.HRatio, - mode_lib->ms.cache_display_cfg.plane.HRatioChroma, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - locals->PSCL_THROUGHPUT, - locals->PSCL_THROUGHPUT_CHROMA, - locals->Dppclk, - locals->BytePerPixelC, - mode_lib->ms.cache_display_cfg.plane.SourceScan, - mode_lib->ms.cache_display_cfg.plane.NumberOfCursors, - mode_lib->ms.cache_display_cfg.plane.CursorWidth, - mode_lib->ms.cache_display_cfg.plane.CursorBPP, - locals->BlockWidth256BytesY, - locals->BlockHeight256BytesY, - locals->BlockWidth256BytesC, - locals->BlockHeight256BytesC, - - /* Output */ - locals->DisplayPipeLineDeliveryTimeLuma, - locals->DisplayPipeLineDeliveryTimeChroma, - locals->DisplayPipeLineDeliveryTimeLumaPrefetch, - locals->DisplayPipeLineDeliveryTimeChromaPrefetch, - locals->DisplayPipeRequestDeliveryTimeLuma, - locals->DisplayPipeRequestDeliveryTimeChroma, - locals->DisplayPipeRequestDeliveryTimeLumaPrefetch, - locals->DisplayPipeRequestDeliveryTimeChromaPrefetch, - locals->CursorRequestDeliveryTime, - locals->CursorRequestDeliveryTimePrefetch); - - CalculateMetaAndPTETimes( - locals->use_one_row_for_frame, - mode_lib->ms.num_active_planes, - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.ip.meta_chunk_size_kbytes, - mode_lib->ms.ip.min_meta_chunk_size_bytes, - mode_lib->ms.cache_display_cfg.timing.HTotal, - mode_lib->ms.cache_display_cfg.plane.VRatio, - mode_lib->ms.cache_display_cfg.plane.VRatioChroma, - locals->DestinationLinesToRequestRowInVBlank, - locals->DestinationLinesToRequestRowInImmediateFlip, - mode_lib->ms.cache_display_cfg.surface.DCCEnable, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - locals->BytePerPixelY, - locals->BytePerPixelC, - mode_lib->ms.cache_display_cfg.plane.SourceScan, - locals->dpte_row_height, - locals->dpte_row_height_chroma, - locals->meta_row_width, - locals->meta_row_width_chroma, - locals->meta_row_height, - locals->meta_row_height_chroma, - locals->meta_req_width, - locals->meta_req_width_chroma, - locals->meta_req_height, - locals->meta_req_height_chroma, - locals->dpte_group_bytes, - locals->PTERequestSizeY, - locals->PTERequestSizeC, - locals->PixelPTEReqWidthY, - locals->PixelPTEReqHeightY, - locals->PixelPTEReqWidthC, - locals->PixelPTEReqHeightC, - locals->dpte_row_width_luma_ub, - locals->dpte_row_width_chroma_ub, - - /* Output */ - locals->DST_Y_PER_PTE_ROW_NOM_L, - locals->DST_Y_PER_PTE_ROW_NOM_C, - locals->DST_Y_PER_META_ROW_NOM_L, - locals->DST_Y_PER_META_ROW_NOM_C, - locals->TimePerMetaChunkNominal, - locals->TimePerChromaMetaChunkNominal, - locals->TimePerMetaChunkVBlank, - locals->TimePerChromaMetaChunkVBlank, - locals->TimePerMetaChunkFlip, - locals->TimePerChromaMetaChunkFlip, - locals->time_per_pte_group_nom_luma, - locals->time_per_pte_group_vblank_luma, - locals->time_per_pte_group_flip_luma, - locals->time_per_pte_group_nom_chroma, - locals->time_per_pte_group_vblank_chroma, - locals->time_per_pte_group_flip_chroma); - - CalculateVMGroupAndRequestTimes( - mode_lib->ms.num_active_planes, - mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, - mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, - mode_lib->ms.cache_display_cfg.timing.HTotal, - locals->BytePerPixelC, - locals->DestinationLinesToRequestVMInVBlank, - locals->DestinationLinesToRequestVMInImmediateFlip, - mode_lib->ms.cache_display_cfg.surface.DCCEnable, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - locals->dpte_row_width_luma_ub, - locals->dpte_row_width_chroma_ub, - locals->vm_group_bytes, - locals->dpde0_bytes_per_frame_ub_l, - locals->dpde0_bytes_per_frame_ub_c, - locals->meta_pte_bytes_per_frame_ub_l, - locals->meta_pte_bytes_per_frame_ub_c, - - /* Output */ - locals->TimePerVMGroupVBlank, - locals->TimePerVMGroupFlip, - locals->TimePerVMRequestVBlank, - locals->TimePerVMRequestFlip); - - // Min TTUVBlank - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (locals->PrefetchMode[k] == 0) { - locals->MinTTUVBlank[k] = dml_max4( - locals->Watermark.DRAMClockChangeWatermark, - locals->Watermark.FCLKChangeWatermark, - locals->Watermark.StutterEnterPlusExitWatermark, - locals->Watermark.UrgentWatermark); - } else if (locals->PrefetchMode[k] == 1) { - locals->MinTTUVBlank[k] = dml_max3( - locals->Watermark.FCLKChangeWatermark, - locals->Watermark.StutterEnterPlusExitWatermark, - locals->Watermark.UrgentWatermark); - } else if (locals->PrefetchMode[k] == 2) { - locals->MinTTUVBlank[k] = dml_max( - locals->Watermark.StutterEnterPlusExitWatermark, - locals->Watermark.UrgentWatermark); - } else { - locals->MinTTUVBlank[k] = locals->Watermark.UrgentWatermark; - } - if (!mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]) - locals->MinTTUVBlank[k] = locals->TCalc + locals->MinTTUVBlank[k]; - } - - // DCC Configuration - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); -#endif - CalculateDCCConfiguration( - mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], - mode_lib->ms.policy.DCCProgrammingAssumesScanDirectionUnknownFinal, - mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC[k], - mode_lib->ms.NomDETInKByte, - locals->BlockHeight256BytesY[k], - locals->BlockHeight256BytesC[k], - mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], - locals->BytePerPixelY[k], - locals->BytePerPixelC[k], - locals->BytePerPixelDETY[k], - locals->BytePerPixelDETC[k], - mode_lib->ms.cache_display_cfg.plane.SourceScan[k], - /* Output */ - &locals->DCCYMaxUncompressedBlock[k], - &locals->DCCCMaxUncompressedBlock[k], - &locals->DCCYMaxCompressedBlock[k], - &locals->DCCCMaxCompressedBlock[k], - &locals->DCCYIndependentBlock[k], - &locals->DCCCIndependentBlock[k]); - } - - // VStartup Adjustment - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->Tvstartup_margin = (s->MaxVStartupLines[k] - locals->VStartupMin[k]) * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, locals->MinTTUVBlank[k]); -#endif - - locals->MinTTUVBlank[k] = locals->MinTTUVBlank[k] + s->Tvstartup_margin; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); - dml_print("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - dml_print("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, locals->MinTTUVBlank[k]); -#endif - - locals->Tdmdl[k] = locals->Tdmdl[k] + s->Tvstartup_margin; - if (mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k] && mode_lib->ms.ip.dynamic_metadata_vm_enabled) { - locals->Tdmdl_vm[k] = locals->Tdmdl_vm[k] + s->Tvstartup_margin; - } - - isInterlaceTiming = (mode_lib->ms.cache_display_cfg.timing.Interlace[k] && !mode_lib->ms.ip.ptoi_supported); - - // The actual positioning of the vstartup - locals->VStartup[k] = (isInterlaceTiming ? (2 * s->MaxVStartupLines[k]) : s->MaxVStartupLines[k]); - - s->dlg_vblank_start = ((isInterlaceTiming ? dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) / 2.0, 1.0) : - mode_lib->ms.cache_display_cfg.timing.VTotal[k]) - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]); - s->LSetup = dml_floor(4.0 * locals->TSetup[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0) / 4.0; - s->blank_lines_remaining = (mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]) - locals->VStartup[k]; - - if (s->blank_lines_remaining < 0) { - dml_print("ERROR: Vstartup is larger than vblank!?\n"); - s->blank_lines_remaining = 0; - ASSERT(0); - } - locals->MIN_DST_Y_NEXT_START[k] = s->dlg_vblank_start + s->blank_lines_remaining + s->LSetup; - - // debug only - s->old_MIN_DST_Y_NEXT_START = ((isInterlaceTiming ? dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) / 2.0, 1.0) : - mode_lib->ms.cache_display_cfg.timing.VTotal[k]) - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) - + dml_max(1.0, dml_ceil((dml_float_t) locals->WritebackDelay[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0)) - + dml_floor(4.0 * locals->TSetup[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0) / 4.0; - - if (((locals->VUpdateOffsetPix[k] + locals->VUpdateWidthPix[k] + locals->VReadyOffsetPix[k]) / (double) mode_lib->ms.cache_display_cfg.timing.HTotal[k]) <= - (isInterlaceTiming ? - dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k] - locals->VStartup[k]) / 2.0, 1.0) : - (int) (mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k] - locals->VStartup[k]))) { - locals->VREADY_AT_OR_AFTER_VSYNC[k] = true; - } else { - locals->VREADY_AT_OR_AFTER_VSYNC[k] = false; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, locals->VStartup[k]); - dml_print("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, locals->VStartupMin[k]); - dml_print("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, locals->VUpdateOffsetPix[k]); - dml_print("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, locals->VUpdateWidthPix[k]); - dml_print("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, locals->VReadyOffsetPix[k]); - dml_print("DML::%s: k=%u, HTotal = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.HTotal[k]); - dml_print("DML::%s: k=%u, VTotal = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VTotal[k]); - dml_print("DML::%s: k=%u, VActive = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VActive[k]); - dml_print("DML::%s: k=%u, VFrontPorch = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]); - dml_print("DML::%s: k=%u, TSetup = %f\n", __func__, k, locals->TSetup[k]); - dml_print("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, locals->MIN_DST_Y_NEXT_START[k]); - dml_print("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f (old)\n", __func__, k, s->old_MIN_DST_Y_NEXT_START); - dml_print("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, locals->VREADY_AT_OR_AFTER_VSYNC[k]); -#endif - } - - //Maximum Bandwidth Used - s->TotalWRBandwidth = 0; - s->WRBandwidth = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true && mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k] == dml_444_32) { - s->WRBandwidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] / - (mode_lib->ms.cache_display_cfg.timing.HTotal[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 4; - } else if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { - s->WRBandwidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] / - (mode_lib->ms.cache_display_cfg.timing.HTotal[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 8; - } - s->TotalWRBandwidth = s->TotalWRBandwidth + s->WRBandwidth; - } - - locals->TotalDataReadBandwidth = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - locals->TotalDataReadBandwidth = locals->TotalDataReadBandwidth + locals->ReadBandwidthSurfaceLuma[k] + locals->ReadBandwidthSurfaceChroma[k]; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, locals->TotalDataReadBandwidth); - dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); - dml_print("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); -#endif - } - - locals->TotalDataReadBandwidthNotIncludingMALLPrefetch = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { - locals->TotalDataReadBandwidthNotIncludingMALLPrefetch = locals->TotalDataReadBandwidthNotIncludingMALLPrefetch - + locals->ReadBandwidthSurfaceLuma[k] + locals->ReadBandwidthSurfaceChroma[k]; - } - } - - CalculateStutterEfficiency_params->CompressedBufferSizeInkByte = locals->CompressedBufferSizeInkByte; - CalculateStutterEfficiency_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; - CalculateStutterEfficiency_params->UnboundedRequestEnabled = locals->UnboundedRequestEnabled; - CalculateStutterEfficiency_params->MetaFIFOSizeInKEntries = mode_lib->ms.ip.meta_fifo_size_in_kentries; - CalculateStutterEfficiency_params->ZeroSizeBufferEntries = mode_lib->ms.ip.zero_size_buffer_entries; - CalculateStutterEfficiency_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; - CalculateStutterEfficiency_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateStutterEfficiency_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; - CalculateStutterEfficiency_params->TotalDataReadBandwidth = locals->TotalDataReadBandwidth; - CalculateStutterEfficiency_params->DCFCLK = locals->Dcfclk; - CalculateStutterEfficiency_params->ReturnBW = mode_lib->ms.ReturnBW; - CalculateStutterEfficiency_params->CompbufReservedSpace64B = locals->compbuf_reserved_space_64b; - CalculateStutterEfficiency_params->CompbufReservedSpaceZs = locals->compbuf_reserved_space_zs; - CalculateStutterEfficiency_params->SRExitTime = mode_lib->ms.state.sr_exit_time_us; - CalculateStutterEfficiency_params->SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; - CalculateStutterEfficiency_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; - CalculateStutterEfficiency_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; - CalculateStutterEfficiency_params->StutterEnterPlusExitWatermark = locals->Watermark.StutterEnterPlusExitWatermark; - CalculateStutterEfficiency_params->Z8StutterEnterPlusExitWatermark = locals->Watermark.Z8StutterEnterPlusExitWatermark; - CalculateStutterEfficiency_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; - CalculateStutterEfficiency_params->Interlace = mode_lib->ms.cache_display_cfg.timing.Interlace; - CalculateStutterEfficiency_params->MinTTUVBlank = locals->MinTTUVBlank; - CalculateStutterEfficiency_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; - CalculateStutterEfficiency_params->DETBufferSizeY = locals->DETBufferSizeY; - CalculateStutterEfficiency_params->BytePerPixelY = locals->BytePerPixelY; - CalculateStutterEfficiency_params->BytePerPixelDETY = locals->BytePerPixelDETY; - CalculateStutterEfficiency_params->SwathWidthY = locals->SwathWidthY; - CalculateStutterEfficiency_params->SwathHeightY = locals->SwathHeightY; - CalculateStutterEfficiency_params->SwathHeightC = locals->SwathHeightC; - CalculateStutterEfficiency_params->NetDCCRateLuma = mode_lib->ms.cache_display_cfg.surface.DCCRateLuma; - CalculateStutterEfficiency_params->NetDCCRateChroma = mode_lib->ms.cache_display_cfg.surface.DCCRateChroma; - CalculateStutterEfficiency_params->DCCFractionOfZeroSizeRequestsLuma = mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsLuma; - CalculateStutterEfficiency_params->DCCFractionOfZeroSizeRequestsChroma = mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsChroma; - CalculateStutterEfficiency_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; - CalculateStutterEfficiency_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; - CalculateStutterEfficiency_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; - CalculateStutterEfficiency_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; - CalculateStutterEfficiency_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; - CalculateStutterEfficiency_params->BlockHeight256BytesY = locals->BlockHeight256BytesY; - CalculateStutterEfficiency_params->BlockWidth256BytesY = locals->BlockWidth256BytesY; - CalculateStutterEfficiency_params->BlockHeight256BytesC = locals->BlockHeight256BytesC; - CalculateStutterEfficiency_params->BlockWidth256BytesC = locals->BlockWidth256BytesC; - CalculateStutterEfficiency_params->DCCYMaxUncompressedBlock = locals->DCCYMaxUncompressedBlock; - CalculateStutterEfficiency_params->DCCCMaxUncompressedBlock = locals->DCCCMaxUncompressedBlock; - CalculateStutterEfficiency_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; - CalculateStutterEfficiency_params->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable; - CalculateStutterEfficiency_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; - CalculateStutterEfficiency_params->ReadBandwidthSurfaceLuma = locals->ReadBandwidthSurfaceLuma; - CalculateStutterEfficiency_params->ReadBandwidthSurfaceChroma = locals->ReadBandwidthSurfaceChroma; - CalculateStutterEfficiency_params->meta_row_bw = locals->meta_row_bw; - CalculateStutterEfficiency_params->dpte_row_bw = locals->dpte_row_bw; - CalculateStutterEfficiency_params->StutterEfficiencyNotIncludingVBlank = &locals->StutterEfficiencyNotIncludingVBlank; - CalculateStutterEfficiency_params->StutterEfficiency = &locals->StutterEfficiency; - CalculateStutterEfficiency_params->NumberOfStutterBurstsPerFrame = &locals->NumberOfStutterBurstsPerFrame; - CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &locals->Z8StutterEfficiencyNotIncludingVBlank; - CalculateStutterEfficiency_params->Z8StutterEfficiency = &locals->Z8StutterEfficiency; - CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &locals->Z8NumberOfStutterBurstsPerFrame; - CalculateStutterEfficiency_params->StutterPeriod = &locals->StutterPeriod; - CalculateStutterEfficiency_params->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = &locals->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE; - - // Stutter Efficiency - CalculateStutterEfficiency(&mode_lib->scratch, - CalculateStutterEfficiency_params); - -#ifdef __DML_VBA_ALLOW_DELTA__ - { - dml_float_t dummy_single[2]; - dml_uint_t dummy_integer[1]; - dml_bool_t dummy_boolean[1]; - - // Calculate z8 stutter eff assuming 0 reserved space - CalculateStutterEfficiency( - locals->CompressedBufferSizeInkByte, - mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, - locals->UnboundedRequestEnabled, - mode_lib->ms.ip.meta_fifo_size_in_kentries, - mode_lib->ms.ip.zero_size_buffer_entries, - mode_lib->ms.ip.pixel_chunk_size_kbytes, - mode_lib->ms.num_active_planes, - mode_lib->ms.ip.rob_buffer_size_kbytes, - locals->TotalDataReadBandwidth, - locals->Dcfclk, - mode_lib->ms.ReturnBW, - 0, //mode_lib->ms.ip.compbuf_reserved_space_64b, - 0, //mode_lib->ms.ip.compbuf_reserved_space_zs, - mode_lib->ms.state.sr_exit_time_us, - mode_lib->ms.state.sr_exit_z8_time_us, - mode_lib->ms.policy.SynchronizeTimingsFinal, - mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming, - locals->Watermark.StutterEnterPlusExitWatermark, - locals->Watermark.Z8StutterEnterPlusExitWatermark, - mode_lib->ms.ip.ptoi_supported, - mode_lib->ms.cache_display_cfg.timing.Interlace, - locals->MinTTUVBlank, - mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, - mode_lib->ms.DETBufferSizeY, - locals->BytePerPixelY, - locals->BytePerPixelDETY, - locals->SwathWidthY, - mode_lib->ms.SwathHeightY, - mode_lib->ms.SwathHeightC, - mode_lib->ms.cache_display_cfg.surface.DCCRateLuma, - mode_lib->ms.cache_display_cfg.surface.DCCRateChroma, - mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsLuma, - mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsChroma, - mode_lib->ms.cache_display_cfg.timing.HTotal, - mode_lib->ms.cache_display_cfg.timing.VTotal, - mode_lib->ms.cache_display_cfg.timing.PixelClock, - mode_lib->ms.cache_display_cfg.plane.VRatio, - mode_lib->ms.cache_display_cfg.plane.SourceScan, - locals->BlockHeight256BytesY, - locals->BlockWidth256BytesY, - locals->BlockHeight256BytesC, - locals->BlockWidth256BytesC, - locals->DCCYMaxUncompressedBlock, - locals->DCCCMaxUncompressedBlock, - mode_lib->ms.cache_display_cfg.timing.VActive, - mode_lib->ms.cache_display_cfg.surface.DCCEnable, - mode_lib->ms.cache_display_cfg.writeback.WritebackEnable, - locals->ReadBandwidthSurfaceLuma, - locals->ReadBandwidthSurfaceChroma, - locals->meta_row_bw, - locals->dpte_row_bw, - - /* Output */ - &dummy_single[0], - &dummy_single[1], - &dummy_integer[0], - &locals->Z8StutterEfficiencyNotIncludingVBlankBestCase, - &locals->Z8StutterEfficiencyBestCase, - &locals->Z8NumberOfStutterBurstsPerFrameBestCase, - &locals->StutterPeriodBestCase, - &dummy_boolean[0]); - } -#else - locals->Z8StutterEfficiencyNotIncludingVBlankBestCase = locals->Z8StutterEfficiencyNotIncludingVBlank; - locals->Z8StutterEfficiencyBestCase = locals->Z8StutterEfficiency; - locals->Z8NumberOfStutterBurstsPerFrameBestCase = locals->Z8NumberOfStutterBurstsPerFrame; - locals->StutterPeriodBestCase = locals->StutterPeriod; -#endif - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: --- END --- \n", __func__); -#endif -} // dml_core_mode_programming - -/// Function: dml_core_get_row_heights -/// @brief Get row height for DPTE and META with minimal input. -void dml_core_get_row_heights( - dml_uint_t *dpte_row_height, - dml_uint_t *meta_row_height, - const struct display_mode_lib_st *mode_lib, - dml_bool_t is_plane1, - enum dml_source_format_class SourcePixelFormat, - enum dml_swizzle_mode SurfaceTiling, - enum dml_rotation_angle ScanDirection, - dml_uint_t pitch, - dml_uint_t GPUVMMinPageSizeKBytes) -{ - dml_uint_t BytePerPixelY; - dml_uint_t BytePerPixelC; - dml_float_t BytePerPixelInDETY; - dml_float_t BytePerPixelInDETC; - dml_uint_t BlockHeight256BytesY; - dml_uint_t BlockHeight256BytesC; - dml_uint_t BlockWidth256BytesY; - dml_uint_t BlockWidth256BytesC; - dml_uint_t MacroTileWidthY; - dml_uint_t MacroTileWidthC; - dml_uint_t MacroTileHeightY; - dml_uint_t MacroTileHeightC; - - dml_uint_t BytePerPixel; - dml_uint_t BlockHeight256Bytes; - dml_uint_t BlockWidth256Bytes; - dml_uint_t MacroTileWidth; - dml_uint_t MacroTileHeight; - dml_uint_t PTEBufferSizeInRequests; - - dml_uint_t dummy_integer[16]; - - CalculateBytePerPixelAndBlockSizes( - SourcePixelFormat, - SurfaceTiling, - - /* Output */ - &BytePerPixelY, - &BytePerPixelC, - &BytePerPixelInDETY, - &BytePerPixelInDETC, - &BlockHeight256BytesY, - &BlockHeight256BytesC, - &BlockWidth256BytesY, - &BlockWidth256BytesC, - &MacroTileHeightY, - &MacroTileHeightC, - &MacroTileWidthY, - &MacroTileWidthC); - - BytePerPixel = is_plane1 ? BytePerPixelC : BytePerPixelY; - BlockHeight256Bytes = is_plane1 ? BlockHeight256BytesC : BlockHeight256BytesY; - BlockWidth256Bytes = is_plane1 ? BlockWidth256BytesC : BlockWidth256BytesY; - MacroTileWidth = is_plane1 ? MacroTileWidthC : MacroTileWidthY; - MacroTileHeight = is_plane1 ? MacroTileHeightC : MacroTileHeightY; - PTEBufferSizeInRequests = is_plane1 ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; -#ifdef __DML_RQ_DLG_CALC_DEBUG__ - dml_print("DML_DLG: %s: is_plane1 = %u\n", __func__, is_plane1); - dml_print("DML_DLG: %s: BytePerPixel = %u\n", __func__, BytePerPixel); - dml_print("DML_DLG: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); - dml_print("DML_DLG: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); - dml_print("DML_DLG: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); - dml_print("DML_DLG: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); - dml_print("DML_DLG: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); - dml_print("DML_DLG: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); - dml_print("DML_DLG: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); - dml_print("DML_DLG: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); -#endif - - // just supply with enough parameters to calculate meta and dte - CalculateVMAndRowBytes( - 0, // dml_bool_t ViewportStationary, - 1, // dml_bool_t DCCEnable, - 1, // dml_uint_t NumberOfDPPs, - BlockHeight256Bytes, - BlockWidth256Bytes, - SourcePixelFormat, - SurfaceTiling, - BytePerPixel, - ScanDirection, - 0, // dml_uint_t SwathWidth, - 0, // dml_uint_t ViewportHeight, (Note: DML calculates one_row_for_frame height regardless, would need test input if that height is useful) - 0, // dml_uint_t ViewportXStart, - 0, // dml_uint_t ViewportYStart, - 1, // dml_bool_t GPUVMEnable, - 4, // dml_uint_t GPUVMMaxPageTableLevels, - GPUVMMinPageSizeKBytes, - PTEBufferSizeInRequests, - pitch, - 0, // dml_uint_t DCCMetaPitch, - MacroTileWidth, - MacroTileHeight, - - // /* Output */ - &dummy_integer[0], // dml_uint_t *MetaRowByte, - &dummy_integer[1], // dml_uint_t *PixelPTEBytesPerRow, - &dummy_integer[2], // dml_uint_t *PixelPTEBytesPerRowStorage, - &dummy_integer[3], // dml_uint_t *dpte_row_width_ub, - dpte_row_height, - &dummy_integer[4], // dml_uint_t *dpte_row_height_linear - &dummy_integer[5], // dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, - &dummy_integer[6], // dml_uint_t *dpte_row_width_ub_one_row_per_frame, - &dummy_integer[7], // dml_uint_t *dpte_row_height_one_row_per_frame, - &dummy_integer[8], // dml_uint_t *MetaRequestWidth, - &dummy_integer[9], // dml_uint_t *MetaRequestHeight, - &dummy_integer[10], // dml_uint_t *meta_row_width, - meta_row_height, - &dummy_integer[11], // dml_uint_t *PixelPTEReqWidth, - &dummy_integer[12], // dml_uint_t *PixelPTEReqHeight, - &dummy_integer[13], // dml_uint_t *PTERequestSize, - &dummy_integer[14], // dml_uint_t *DPDE0BytesFrame, - &dummy_integer[15]); // dml_uint_t *MetaPTEBytesFrame) - -#ifdef __DML_RQ_DLG_CALC_DEBUG__ - dml_print("DML_DLG: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); - dml_print("DML_DLG: %s: meta_row_height = %u\n", __func__, *meta_row_height); -#endif -} - -static struct soc_state_bounding_box_st dml_get_soc_state_bounding_box( - const struct soc_states_st *states, - dml_uint_t state_idx) -{ - dml_print("DML::%s: state_idx=%u (num_states=%u)\n", __func__, state_idx, states->num_states); - - if (state_idx >= (dml_uint_t)states->num_states) { - dml_print("DML::%s: ERROR: Invalid state_idx=%u! num_states=%u\n", __func__, state_idx, states->num_states); - ASSERT(0); - } - return (states->state_array[state_idx]); -} - -/// @brief Copy the parameters to a calculation struct, it actually only need when the DML needs to have -/// the intelligence to re-calculate when any of display cfg, bbox, or policy changes since last calculated. -/// -static void cache_ip_soc_cfg(struct display_mode_lib_st *mode_lib, - dml_uint_t state_idx) -{ - mode_lib->ms.state_idx = state_idx; - mode_lib->ms.max_state_idx = mode_lib->states.num_states - 1; - mode_lib->ms.soc = mode_lib->soc; - mode_lib->ms.ip = mode_lib->ip; - mode_lib->ms.policy = mode_lib->policy; - mode_lib->ms.state = dml_get_soc_state_bounding_box(&mode_lib->states, state_idx); - mode_lib->ms.max_state = dml_get_soc_state_bounding_box(&mode_lib->states, mode_lib->states.num_states - 1); -} - -static void cache_display_cfg(struct display_mode_lib_st *mode_lib, - const struct dml_display_cfg_st *display_cfg) -{ - mode_lib->ms.cache_display_cfg = *display_cfg; -} - -static void fetch_socbb_params(struct display_mode_lib_st *mode_lib) -{ - struct soc_state_bounding_box_st *state = &mode_lib->ms.state; - - // Default values, SOCCLK, DRAMSpeed, and FabricClock will be reassigned to the same state value in mode_check step - // If UseMinimumRequiredDCFCLK is used, the DCFCLK will be the min dcflk for the mode support - mode_lib->ms.SOCCLK = (dml_float_t)state->socclk_mhz; - mode_lib->ms.DRAMSpeed = (dml_float_t)state->dram_speed_mts; - mode_lib->ms.FabricClock = (dml_float_t)state->fabricclk_mhz; - mode_lib->ms.DCFCLK = (dml_float_t)state->dcfclk_mhz; -} - -/// @brief Use display_cfg directly for mode_support calculation -/// Calculated values and informational output are stored in mode_lib.vba data struct -/// The display configuration is described with pipes struct and num_pipes -/// This function is used when physical resource mapping is not finalized (for example, -/// don't know how many pipes to represent a surface) -/// @param mode_lib Contains the bounding box and policy setting. -/// @param state_idx Power state index -/// @param display_cfg Display configurations. A display -dml_bool_t dml_mode_support( - struct display_mode_lib_st *mode_lib, - dml_uint_t state_idx, - const struct dml_display_cfg_st *display_cfg) -{ - dml_bool_t is_mode_support; - - dml_print("DML::%s: ------------- START ----------\n", __func__); - cache_ip_soc_cfg(mode_lib, state_idx); - cache_display_cfg(mode_lib, display_cfg); - - fetch_socbb_params(mode_lib); - - dml_print("DML::%s: state_idx = %u\n", __func__, state_idx); - - is_mode_support = dml_core_mode_support(mode_lib); - - dml_print("DML::%s: is_mode_support = %u\n", __func__, is_mode_support); - dml_print("DML::%s: ------------- DONE ----------\n", __func__); - return is_mode_support; -} - -/// @Brief A function to calculate the programming values for DCN DCHUB (Assume mode is supported) -/// The output will be stored in the mode_lib.mp (mode_program_st) data struct and those can be accessed via the getter functions -/// Calculated values include: watermarks, dlg, rq reg, different clock frequency -/// This function returns 1 when there is no error. -/// Note: In this function, it is assumed that DCFCLK, SOCCLK freq are the state values, and mode_program will just use the DML calculated DPPCLK and DISPCLK -/// @param mode_lib mode_lib data struct that house all the input/output/bbox and calculation values. -/// @param state_idx Power state idx chosen -/// @param display_cfg Display Configuration -/// @param call_standalone Calling mode_programming without calling mode support. Some of the "support" struct member will be pre-calculated before doing mode programming -/// TODO: Add clk_cfg input, could be useful for standalone mode -dml_bool_t dml_mode_programming( - struct display_mode_lib_st *mode_lib, - dml_uint_t state_idx, - const struct dml_display_cfg_st *display_cfg, - bool call_standalone) -{ - struct dml_clk_cfg_st clk_cfg; - memset(&clk_cfg, 0, sizeof(clk_cfg)); - - clk_cfg.dcfclk_option = dml_use_required_freq; - clk_cfg.dispclk_option = dml_use_required_freq; - for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; ++k) - clk_cfg.dppclk_option[k] = dml_use_required_freq; - - dml_print("DML::%s: ------------- START ----------\n", __func__); - dml_print("DML::%s: state_idx = %u\n", __func__, state_idx); - dml_print("DML::%s: call_standalone = %u\n", __func__, call_standalone); - - cache_ip_soc_cfg(mode_lib, state_idx); - cache_display_cfg(mode_lib, display_cfg); - - fetch_socbb_params(mode_lib); - if (call_standalone) { - mode_lib->ms.support.ImmediateFlipSupport = 1; // assume mode support say immediate flip ok at max state/combine - dml_core_mode_support_partial(mode_lib); - } - - dml_core_mode_programming(mode_lib, &clk_cfg); - - dml_print("DML::%s: ------------- DONE ----------\n", __func__); - dml_print("DML::%s: PrefetchAndImmediateFlipSupported = %0d\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); - return mode_lib->mp.PrefetchAndImmediateFlipSupported; -} - -static dml_uint_t mode_support_pwr_states( - dml_uint_t *lowest_state_idx, - struct display_mode_lib_st *mode_lib, - const struct dml_display_cfg_st *display_cfg, - dml_uint_t start_state_idx, - dml_uint_t end_state_idx) -{ - dml_uint_t state_idx = 0; - dml_bool_t mode_is_supported = 0; - *lowest_state_idx = end_state_idx; - - if (end_state_idx < start_state_idx) - ASSERT(0); - - if (end_state_idx >= mode_lib->states.num_states) // idx is 0-based - ASSERT(0); - - for (state_idx = start_state_idx; state_idx <= end_state_idx; state_idx++) { - if (dml_mode_support(mode_lib, state_idx, display_cfg)) { - dml_print("DML::%s: Mode is supported at power state_idx = %u\n", __func__, state_idx); - mode_is_supported = 1; - *lowest_state_idx = state_idx; - break; - } - } - - return mode_is_supported; -} - -dml_uint_t dml_mode_support_ex(struct dml_mode_support_ex_params_st *in_out_params) -{ - dml_uint_t result; - - result = mode_support_pwr_states(&in_out_params->out_lowest_state_idx, - in_out_params->mode_lib, - in_out_params->in_display_cfg, - in_out_params->in_start_state_idx, - in_out_params->mode_lib->states.num_states - 1); - - if (result) - *in_out_params->out_evaluation_info = in_out_params->mode_lib->ms.support; - - return result; -} - -dml_bool_t dml_get_is_phantom_pipe(struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) -{ - dml_uint_t plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; - dml_print("DML::%s: pipe_idx=%d UseMALLForPStateChange=%0d\n", __func__, pipe_idx, mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[plane_idx]); - return (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[plane_idx] == dml_use_mall_pstate_change_phantom_pipe); -} - -#define dml_get_per_surface_var_func(variable, type, interval_var) type dml_get_##variable(struct display_mode_lib_st *mode_lib, dml_uint_t surface_idx) \ -{ \ - dml_uint_t plane_idx; \ - plane_idx = mode_lib->mp.pipe_plane[surface_idx]; \ - return (type) interval_var[plane_idx]; \ -} - -#define dml_get_var_func(var, type, internal_var) type dml_get_##var(struct display_mode_lib_st *mode_lib) \ -{ \ - return (type) internal_var; \ -} - -dml_get_var_func(wm_urgent, dml_float_t, mode_lib->mp.Watermark.UrgentWatermark); -dml_get_var_func(wm_stutter_exit, dml_float_t, mode_lib->mp.Watermark.StutterExitWatermark); -dml_get_var_func(wm_stutter_enter_exit, dml_float_t, mode_lib->mp.Watermark.StutterEnterPlusExitWatermark); -dml_get_var_func(wm_memory_trip, dml_float_t, mode_lib->mp.UrgentLatency); -dml_get_var_func(wm_fclk_change, dml_float_t, mode_lib->mp.Watermark.FCLKChangeWatermark); -dml_get_var_func(wm_usr_retraining, dml_float_t, mode_lib->mp.Watermark.USRRetrainingWatermark); -dml_get_var_func(wm_dram_clock_change, dml_float_t, mode_lib->mp.Watermark.DRAMClockChangeWatermark); -dml_get_var_func(wm_z8_stutter_enter_exit, dml_float_t, mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark); -dml_get_var_func(wm_z8_stutter, dml_float_t, mode_lib->mp.Watermark.Z8StutterExitWatermark); -dml_get_var_func(fraction_of_urgent_bandwidth, dml_float_t, mode_lib->mp.FractionOfUrgentBandwidth); -dml_get_var_func(fraction_of_urgent_bandwidth_imm_flip, dml_float_t, mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip); -dml_get_var_func(urgent_latency, dml_float_t, mode_lib->mp.UrgentLatency); -dml_get_var_func(clk_dcf_deepsleep, dml_float_t, mode_lib->mp.DCFCLKDeepSleep); -dml_get_var_func(wm_writeback_dram_clock_change, dml_float_t, mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); -dml_get_var_func(wm_writeback_urgent, dml_float_t, mode_lib->mp.Watermark.WritebackUrgentWatermark); -dml_get_var_func(stutter_efficiency, dml_float_t, mode_lib->mp.StutterEfficiency); -dml_get_var_func(stutter_efficiency_no_vblank, dml_float_t, mode_lib->mp.StutterEfficiencyNotIncludingVBlank); -dml_get_var_func(stutter_efficiency_z8, dml_float_t, mode_lib->mp.Z8StutterEfficiency); -dml_get_var_func(stutter_num_bursts_z8, dml_float_t, mode_lib->mp.Z8NumberOfStutterBurstsPerFrame); -dml_get_var_func(stutter_period, dml_float_t, mode_lib->mp.StutterPeriod); -dml_get_var_func(stutter_efficiency_z8_bestcase, dml_float_t, mode_lib->mp.Z8StutterEfficiencyBestCase); -dml_get_var_func(stutter_num_bursts_z8_bestcase, dml_float_t, mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase); -dml_get_var_func(stutter_period_bestcase, dml_float_t, mode_lib->mp.StutterPeriodBestCase); -dml_get_var_func(urgent_extra_latency, dml_float_t, mode_lib->mp.UrgentExtraLatency); -dml_get_var_func(fclk_change_latency, dml_float_t, mode_lib->mp.MaxActiveFCLKChangeLatencySupported); -dml_get_var_func(dispclk_calculated, dml_float_t, mode_lib->mp.Dispclk_calculated); -dml_get_var_func(total_data_read_bw, dml_float_t, mode_lib->mp.TotalDataReadBandwidth); -dml_get_var_func(return_bw, dml_float_t, mode_lib->ms.ReturnBW); -dml_get_var_func(return_dram_bw, dml_float_t, mode_lib->ms.ReturnDRAMBW); -dml_get_var_func(tcalc, dml_float_t, mode_lib->mp.TCalc); -dml_get_var_func(comp_buffer_size_kbytes, dml_uint_t, mode_lib->mp.CompressedBufferSizeInkByte); -dml_get_var_func(pixel_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.pixel_chunk_size_kbytes); -dml_get_var_func(alpha_pixel_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.alpha_pixel_chunk_size_kbytes); -dml_get_var_func(meta_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.meta_chunk_size_kbytes); -dml_get_var_func(min_pixel_chunk_size_in_byte, dml_uint_t, mode_lib->ms.ip.min_pixel_chunk_size_bytes); -dml_get_var_func(min_meta_chunk_size_in_byte, dml_uint_t, mode_lib->ms.ip.min_meta_chunk_size_bytes); -dml_get_var_func(total_immediate_flip_bytes, dml_uint_t, mode_lib->mp.TotImmediateFlipBytes); - -dml_get_per_surface_var_func(dsc_delay, dml_uint_t, mode_lib->mp.DSCDelay); // this is the dsc latency -dml_get_per_surface_var_func(dppclk_calculated, dml_float_t, mode_lib->mp.Dppclk_calculated); -dml_get_per_surface_var_func(dscclk_calculated, dml_float_t, mode_lib->mp.DSCCLK_calculated); -dml_get_per_surface_var_func(min_ttu_vblank_in_us, dml_float_t, mode_lib->mp.MinTTUVBlank); -dml_get_per_surface_var_func(vratio_prefetch_l, dml_float_t, mode_lib->mp.VRatioPrefetchY); -dml_get_per_surface_var_func(vratio_prefetch_c, dml_float_t, mode_lib->mp.VRatioPrefetchC); -dml_get_per_surface_var_func(dst_x_after_scaler, dml_uint_t, mode_lib->mp.DSTXAfterScaler); -dml_get_per_surface_var_func(dst_y_after_scaler, dml_uint_t, mode_lib->mp.DSTYAfterScaler); -dml_get_per_surface_var_func(dst_y_per_vm_vblank, dml_float_t, mode_lib->mp.DestinationLinesToRequestVMInVBlank); -dml_get_per_surface_var_func(dst_y_per_row_vblank, dml_float_t, mode_lib->mp.DestinationLinesToRequestRowInVBlank); -dml_get_per_surface_var_func(dst_y_prefetch, dml_float_t, mode_lib->mp.DestinationLinesForPrefetch); -dml_get_per_surface_var_func(dst_y_per_vm_flip, dml_float_t, mode_lib->mp.DestinationLinesToRequestVMInImmediateFlip); -dml_get_per_surface_var_func(dst_y_per_row_flip, dml_float_t, mode_lib->mp.DestinationLinesToRequestRowInImmediateFlip); -dml_get_per_surface_var_func(dst_y_per_pte_row_nom_l, dml_float_t, mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L); -dml_get_per_surface_var_func(dst_y_per_pte_row_nom_c, dml_float_t, mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C); -dml_get_per_surface_var_func(dst_y_per_meta_row_nom_l, dml_float_t, mode_lib->mp.DST_Y_PER_META_ROW_NOM_L); -dml_get_per_surface_var_func(dst_y_per_meta_row_nom_c, dml_float_t, mode_lib->mp.DST_Y_PER_META_ROW_NOM_C); -dml_get_per_surface_var_func(refcyc_per_vm_group_vblank_in_us, dml_float_t, mode_lib->mp.TimePerVMGroupVBlank); -dml_get_per_surface_var_func(refcyc_per_vm_group_flip_in_us, dml_float_t, mode_lib->mp.TimePerVMGroupFlip); -dml_get_per_surface_var_func(refcyc_per_vm_req_vblank_in_us, dml_float_t, mode_lib->mp.TimePerVMRequestVBlank); -dml_get_per_surface_var_func(refcyc_per_vm_req_flip_in_us, dml_float_t, mode_lib->mp.TimePerVMRequestFlip); -dml_get_per_surface_var_func(refcyc_per_vm_dmdata_in_us, dml_float_t, mode_lib->mp.Tdmdl_vm); -dml_get_per_surface_var_func(dmdata_dl_delta_in_us, dml_float_t, mode_lib->mp.Tdmdl); -dml_get_per_surface_var_func(refcyc_per_line_delivery_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeLuma); -dml_get_per_surface_var_func(refcyc_per_line_delivery_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeChroma); -dml_get_per_surface_var_func(refcyc_per_line_delivery_pre_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch); -dml_get_per_surface_var_func(refcyc_per_line_delivery_pre_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch); -dml_get_per_surface_var_func(refcyc_per_req_delivery_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma); -dml_get_per_surface_var_func(refcyc_per_req_delivery_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma); -dml_get_per_surface_var_func(refcyc_per_req_delivery_pre_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch); -dml_get_per_surface_var_func(refcyc_per_req_delivery_pre_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch); -dml_get_per_surface_var_func(refcyc_per_cursor_req_delivery_in_us, dml_float_t, mode_lib->mp.CursorRequestDeliveryTime); -dml_get_per_surface_var_func(refcyc_per_cursor_req_delivery_pre_in_us, dml_float_t, mode_lib->mp.CursorRequestDeliveryTimePrefetch); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_nom_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkNominal); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_nom_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkNominal); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_vblank_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkVBlank); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_vblank_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkVBlank); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_flip_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkFlip); -dml_get_per_surface_var_func(refcyc_per_meta_chunk_flip_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkFlip); -dml_get_per_surface_var_func(refcyc_per_pte_group_nom_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_nom_luma); -dml_get_per_surface_var_func(refcyc_per_pte_group_nom_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_nom_chroma); -dml_get_per_surface_var_func(refcyc_per_pte_group_vblank_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_vblank_luma); -dml_get_per_surface_var_func(refcyc_per_pte_group_vblank_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_vblank_chroma); -dml_get_per_surface_var_func(refcyc_per_pte_group_flip_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_flip_luma); -dml_get_per_surface_var_func(refcyc_per_pte_group_flip_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_flip_chroma); -dml_get_per_surface_var_func(dpte_group_size_in_bytes, dml_uint_t, mode_lib->mp.dpte_group_bytes); -dml_get_per_surface_var_func(vm_group_size_in_bytes, dml_uint_t, mode_lib->mp.vm_group_bytes); -dml_get_per_surface_var_func(swath_height_l, dml_uint_t, mode_lib->ms.SwathHeightY); -dml_get_per_surface_var_func(swath_height_c, dml_uint_t, mode_lib->ms.SwathHeightC); -dml_get_per_surface_var_func(dpte_row_height_l, dml_uint_t, mode_lib->mp.dpte_row_height); -dml_get_per_surface_var_func(dpte_row_height_c, dml_uint_t, mode_lib->mp.dpte_row_height_chroma); -dml_get_per_surface_var_func(dpte_row_height_linear_l, dml_uint_t, mode_lib->mp.dpte_row_height_linear); -dml_get_per_surface_var_func(dpte_row_height_linear_c, dml_uint_t, mode_lib->mp.dpte_row_height_linear_chroma); -dml_get_per_surface_var_func(meta_row_height_l, dml_uint_t, mode_lib->mp.meta_row_height); -dml_get_per_surface_var_func(meta_row_height_c, dml_uint_t, mode_lib->mp.meta_row_height_chroma); - -dml_get_per_surface_var_func(vstartup_calculated, dml_uint_t, mode_lib->mp.VStartup); -dml_get_per_surface_var_func(vupdate_offset, dml_uint_t, mode_lib->mp.VUpdateOffsetPix); -dml_get_per_surface_var_func(vupdate_width, dml_uint_t, mode_lib->mp.VUpdateWidthPix); -dml_get_per_surface_var_func(vready_offset, dml_uint_t, mode_lib->mp.VReadyOffsetPix); -dml_get_per_surface_var_func(vready_at_or_after_vsync, dml_uint_t, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC); -dml_get_per_surface_var_func(min_dst_y_next_start, dml_uint_t, mode_lib->mp.MIN_DST_Y_NEXT_START); -dml_get_per_surface_var_func(det_stored_buffer_size_l_bytes, dml_uint_t, mode_lib->ms.DETBufferSizeY); -dml_get_per_surface_var_func(det_stored_buffer_size_c_bytes, dml_uint_t, mode_lib->ms.DETBufferSizeC); -dml_get_per_surface_var_func(use_mall_for_static_screen, dml_uint_t, mode_lib->mp.UsesMALLForStaticScreen); -dml_get_per_surface_var_func(surface_size_for_mall, dml_uint_t, mode_lib->mp.SurfaceSizeInTheMALL); -dml_get_per_surface_var_func(dcc_max_uncompressed_block_l, dml_uint_t, mode_lib->mp.DCCYMaxUncompressedBlock); -dml_get_per_surface_var_func(dcc_max_compressed_block_l, dml_uint_t, mode_lib->mp.DCCYMaxCompressedBlock); -dml_get_per_surface_var_func(dcc_independent_block_l, dml_uint_t, mode_lib->mp.DCCYIndependentBlock); -dml_get_per_surface_var_func(dcc_max_uncompressed_block_c, dml_uint_t, mode_lib->mp.DCCCMaxUncompressedBlock); -dml_get_per_surface_var_func(dcc_max_compressed_block_c, dml_uint_t, mode_lib->mp.DCCCMaxCompressedBlock); -dml_get_per_surface_var_func(dcc_independent_block_c, dml_uint_t, mode_lib->mp.DCCCIndependentBlock); -dml_get_per_surface_var_func(max_active_dram_clock_change_latency_supported, dml_uint_t, mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported); -dml_get_per_surface_var_func(pte_buffer_mode, dml_uint_t, mode_lib->mp.PTE_BUFFER_MODE); -dml_get_per_surface_var_func(bigk_fragment_size, dml_uint_t, mode_lib->mp.BIGK_FRAGMENT_SIZE); -dml_get_per_surface_var_func(dpte_bytes_per_row, dml_uint_t, mode_lib->mp.PixelPTEBytesPerRow); -dml_get_per_surface_var_func(meta_bytes_per_row, dml_uint_t, mode_lib->mp.MetaRowByte); -dml_get_per_surface_var_func(det_buffer_size_kbytes, dml_uint_t, mode_lib->ms.DETBufferSizeInKByte); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h deleted file mode 100644 index a38ed89c47a9..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h +++ /dev/null @@ -1,204 +0,0 @@ -/* - * Copyright 2022 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DISPLAY_MODE_CORE_H__ -#define __DISPLAY_MODE_CORE_H__ - -#include "display_mode_core_structs.h" - -struct display_mode_lib_st; - -dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib); -void dml_core_mode_support_partial(struct display_mode_lib_st *mode_lib); -void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struct dml_clk_cfg_st *clk_cfg); - -void dml_core_get_row_heights( - dml_uint_t *dpte_row_height, - dml_uint_t *meta_row_height, - const struct display_mode_lib_st *mode_lib, - dml_bool_t is_plane1, - enum dml_source_format_class SourcePixelFormat, - enum dml_swizzle_mode SurfaceTiling, - enum dml_rotation_angle ScanDirection, - dml_uint_t pitch, - dml_uint_t GPUVMMinPageSizeKBytes); - -dml_float_t dml_get_return_bw_mbps_vm_only( - const struct soc_bounding_box_st *soc, - dml_bool_t use_ideal_dram_bw_strobe, - dml_bool_t HostVMEnable, - dml_float_t DCFCLK, - dml_float_t FabricClock, - dml_float_t DRAMSpeed); - -dml_float_t dml_get_return_bw_mbps( - const struct soc_bounding_box_st *soc, - dml_bool_t use_ideal_dram_bw_strobe, - dml_bool_t HostVMEnable, - dml_float_t DCFCLK, - dml_float_t FabricClock, - dml_float_t DRAMSpeed); - -dml_bool_t dml_mode_support( - struct display_mode_lib_st *mode_lib, - dml_uint_t state_idx, - const struct dml_display_cfg_st *display_cfg); - -dml_bool_t dml_mode_programming( - struct display_mode_lib_st *mode_lib, - dml_uint_t state_idx, - const struct dml_display_cfg_st *display_cfg, - bool call_standalone); - -dml_uint_t dml_mode_support_ex( - struct dml_mode_support_ex_params_st *in_out_params); - -dml_bool_t dml_get_is_phantom_pipe(struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx); - -#define dml_get_per_surface_var_decl(variable, type) type dml_get_##variable(struct display_mode_lib_st *mode_lib, dml_uint_t surface_idx) -#define dml_get_var_decl(var, type) type dml_get_##var(struct display_mode_lib_st *mode_lib) - -dml_get_var_decl(wm_urgent, dml_float_t); -dml_get_var_decl(wm_stutter_exit, dml_float_t); -dml_get_var_decl(wm_stutter_enter_exit, dml_float_t); -dml_get_var_decl(wm_memory_trip, dml_float_t); -dml_get_var_decl(wm_dram_clock_change, dml_float_t); -dml_get_var_decl(wm_z8_stutter_enter_exit, dml_float_t); -dml_get_var_decl(wm_z8_stutter, dml_float_t); -dml_get_var_decl(urgent_latency, dml_float_t); -dml_get_var_decl(clk_dcf_deepsleep, dml_float_t); -dml_get_var_decl(wm_fclk_change, dml_float_t); -dml_get_var_decl(wm_usr_retraining, dml_float_t); -dml_get_var_decl(urgent_latency, dml_float_t); - -dml_get_var_decl(wm_writeback_dram_clock_change, dml_float_t); -dml_get_var_decl(wm_writeback_urgent, dml_float_t); -dml_get_var_decl(stutter_efficiency_no_vblank, dml_float_t); -dml_get_var_decl(stutter_efficiency, dml_float_t); -dml_get_var_decl(stutter_efficiency_z8, dml_float_t); -dml_get_var_decl(stutter_num_bursts_z8, dml_float_t); -dml_get_var_decl(stutter_period, dml_float_t); -dml_get_var_decl(stutter_efficiency_z8_bestcase, dml_float_t); -dml_get_var_decl(stutter_num_bursts_z8_bestcase, dml_float_t); -dml_get_var_decl(stutter_period_bestcase, dml_float_t); -dml_get_var_decl(urgent_latency, dml_float_t); -dml_get_var_decl(urgent_extra_latency, dml_float_t); -dml_get_var_decl(fclk_change_latency, dml_float_t); -dml_get_var_decl(nonurgent_latency, dml_float_t); -dml_get_var_decl(dispclk_calculated, dml_float_t); -dml_get_var_decl(total_data_read_bw, dml_float_t); -dml_get_var_decl(return_bw, dml_float_t); -dml_get_var_decl(return_dram_bw, dml_float_t); -dml_get_var_decl(tcalc, dml_float_t); -dml_get_var_decl(fraction_of_urgent_bandwidth, dml_float_t); -dml_get_var_decl(fraction_of_urgent_bandwidth_imm_flip, dml_float_t); -dml_get_var_decl(comp_buffer_size_kbytes, dml_uint_t); -dml_get_var_decl(pixel_chunk_size_in_kbyte, dml_uint_t); -dml_get_var_decl(alpha_pixel_chunk_size_in_kbyte, dml_uint_t); -dml_get_var_decl(meta_chunk_size_in_kbyte, dml_uint_t); -dml_get_var_decl(min_pixel_chunk_size_in_byte, dml_uint_t); -dml_get_var_decl(min_meta_chunk_size_in_byte, dml_uint_t); -dml_get_var_decl(total_immediate_flip_bytes, dml_uint_t); - -dml_get_per_surface_var_decl(dsc_delay, dml_uint_t); -dml_get_per_surface_var_decl(dppclk_calculated, dml_float_t); -dml_get_per_surface_var_decl(dscclk_calculated, dml_float_t); -dml_get_per_surface_var_decl(min_ttu_vblank_in_us, dml_float_t); -dml_get_per_surface_var_decl(vratio_prefetch_l, dml_float_t); -dml_get_per_surface_var_decl(vratio_prefetch_c, dml_float_t); -dml_get_per_surface_var_decl(dst_x_after_scaler, dml_uint_t); -dml_get_per_surface_var_decl(dst_y_after_scaler, dml_uint_t); -dml_get_per_surface_var_decl(dst_y_per_vm_vblank, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_row_vblank, dml_float_t); -dml_get_per_surface_var_decl(dst_y_prefetch, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_vm_flip, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_row_flip, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_pte_row_nom_l, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_pte_row_nom_c, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_meta_row_nom_l, dml_float_t); -dml_get_per_surface_var_decl(dst_y_per_meta_row_nom_c, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_vm_group_vblank_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_vm_group_flip_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_vm_req_vblank_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_vm_req_flip_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_vm_dmdata_in_us, dml_float_t); -dml_get_per_surface_var_decl(dmdata_dl_delta_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_line_delivery_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_line_delivery_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_line_delivery_pre_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_line_delivery_pre_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_req_delivery_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_req_delivery_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_req_delivery_pre_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_req_delivery_pre_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_cursor_req_delivery_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_cursor_req_delivery_pre_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_nom_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_nom_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_vblank_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_vblank_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_flip_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_meta_chunk_flip_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_nom_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_nom_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_vblank_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_vblank_c_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_flip_l_in_us, dml_float_t); -dml_get_per_surface_var_decl(refcyc_per_pte_group_flip_c_in_us, dml_float_t); - -dml_get_per_surface_var_decl(dpte_group_size_in_bytes, dml_uint_t); -dml_get_per_surface_var_decl(vm_group_size_in_bytes, dml_uint_t); -dml_get_per_surface_var_decl(swath_height_l, dml_uint_t); -dml_get_per_surface_var_decl(swath_height_c, dml_uint_t); -dml_get_per_surface_var_decl(dpte_row_height_l, dml_uint_t); -dml_get_per_surface_var_decl(dpte_row_height_c, dml_uint_t); -dml_get_per_surface_var_decl(dpte_row_height_linear_l, dml_uint_t); -dml_get_per_surface_var_decl(dpte_row_height_linear_c, dml_uint_t); -dml_get_per_surface_var_decl(meta_row_height_l, dml_uint_t); -dml_get_per_surface_var_decl(meta_row_height_c, dml_uint_t); -dml_get_per_surface_var_decl(vstartup_calculated, dml_uint_t); -dml_get_per_surface_var_decl(vupdate_offset, dml_uint_t); -dml_get_per_surface_var_decl(vupdate_width, dml_uint_t); -dml_get_per_surface_var_decl(vready_offset, dml_uint_t); -dml_get_per_surface_var_decl(vready_at_or_after_vsync, dml_uint_t); -dml_get_per_surface_var_decl(min_dst_y_next_start, dml_uint_t); -dml_get_per_surface_var_decl(det_stored_buffer_size_l_bytes, dml_uint_t); -dml_get_per_surface_var_decl(det_stored_buffer_size_c_bytes, dml_uint_t); -dml_get_per_surface_var_decl(use_mall_for_static_screen, dml_uint_t); -dml_get_per_surface_var_decl(surface_size_for_mall, dml_uint_t); -dml_get_per_surface_var_decl(dcc_max_uncompressed_block_l, dml_uint_t); -dml_get_per_surface_var_decl(dcc_max_uncompressed_block_c, dml_uint_t); -dml_get_per_surface_var_decl(dcc_max_compressed_block_l, dml_uint_t); -dml_get_per_surface_var_decl(dcc_max_compressed_block_c, dml_uint_t); -dml_get_per_surface_var_decl(dcc_independent_block_l, dml_uint_t); -dml_get_per_surface_var_decl(dcc_independent_block_c, dml_uint_t); -dml_get_per_surface_var_decl(max_active_dram_clock_change_latency_supported, dml_uint_t); -dml_get_per_surface_var_decl(pte_buffer_mode, dml_uint_t); -dml_get_per_surface_var_decl(bigk_fragment_size, dml_uint_t); -dml_get_per_surface_var_decl(dpte_bytes_per_row, dml_uint_t); -dml_get_per_surface_var_decl(meta_bytes_per_row, dml_uint_t); -dml_get_per_surface_var_decl(det_buffer_size_kbytes, dml_uint_t); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h deleted file mode 100644 index d57717b023eb..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h +++ /dev/null @@ -1,2033 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DISPLAY_MODE_CORE_STRUCT_H__ -#define __DISPLAY_MODE_CORE_STRUCT_H__ - -#include "display_mode_lib_defines.h" -#include "dml_top_display_cfg_types.h" - -enum dml_project_id { - dml_project_invalid = 0, - dml_project_default = 1, - dml_project_dcn32 = dml_project_default, - dml_project_dcn321 = 2, - dml_project_dcn35 = 3, - dml_project_dcn351 = 4, - dml_project_dcn401 = 5, - dml_project_dcn36 = 6, -}; -enum dml_prefetch_modes { - dml_prefetch_support_uclk_fclk_and_stutter_if_possible = 0, - dml_prefetch_support_uclk_fclk_and_stutter = 1, - dml_prefetch_support_fclk_and_stutter = 2, - dml_prefetch_support_stutter = 3, - dml_prefetch_support_none = 4 -}; -enum dml_use_mall_for_pstate_change_mode { - dml_use_mall_pstate_change_disable = 0, - dml_use_mall_pstate_change_full_frame = 1, - dml_use_mall_pstate_change_sub_viewport = 2, - dml_use_mall_pstate_change_phantom_pipe = 3, - dml_use_mall_pstate_change_phantom_pipe_no_data_return = 4, - dml_use_mall_pstate_change_imall = 5 -}; -enum dml_use_mall_for_static_screen_mode { - dml_use_mall_static_screen_disable = 0, - dml_use_mall_static_screen_enable = 1, - dml_use_mall_static_screen_optimize = 2 -}; -enum dml_output_encoder_class { - dml_dp = 0, - dml_edp = 1, - dml_dp2p0 = 2, - dml_hdmi = 3, - dml_hdmifrl = 4, - dml_none = 5 -}; -enum dml_output_link_dp_rate{ - dml_dp_rate_na = 0, - dml_dp_rate_hbr = 1, - dml_dp_rate_hbr2 = 2, - dml_dp_rate_hbr3 = 3, - dml_dp_rate_uhbr10 = 4, - dml_dp_rate_uhbr13p5 = 5, - dml_dp_rate_uhbr20 = 6 -}; -enum dml_output_type_and_rate__type{ - dml_output_type_unknown = 0, - dml_output_type_dp = 1, - dml_output_type_edp = 2, - dml_output_type_dp2p0 = 3, - dml_output_type_hdmi = 4, - dml_output_type_hdmifrl = 5 -}; -enum dml_output_type_and_rate__rate { - dml_output_rate_unknown = 0, - dml_output_rate_dp_rate_hbr = 1, - dml_output_rate_dp_rate_hbr2 = 2, - dml_output_rate_dp_rate_hbr3 = 3, - dml_output_rate_dp_rate_uhbr10 = 4, - dml_output_rate_dp_rate_uhbr13p5 = 5, - dml_output_rate_dp_rate_uhbr20 = 6, - dml_output_rate_hdmi_rate_3x3 = 7, - dml_output_rate_hdmi_rate_6x3 = 8, - dml_output_rate_hdmi_rate_6x4 = 9, - dml_output_rate_hdmi_rate_8x4 = 10, - dml_output_rate_hdmi_rate_10x4 = 11, - dml_output_rate_hdmi_rate_12x4 = 12 -}; -enum dml_output_format_class { - dml_444 = 0, - dml_s422 = 1, - dml_n422 = 2, - dml_420 = 3 -}; -enum dml_source_format_class { - dml_444_8 = 0, - dml_444_16 = 1, - dml_444_32 = 2, - dml_444_64 = 3, - dml_420_8 = 4, - dml_420_10 = 5, - dml_420_12 = 6, - dml_422_8 = 7, - dml_422_10 = 8, - dml_rgbe_alpha = 9, - dml_rgbe = 10, - dml_mono_8 = 11, - dml_mono_16 = 12 -}; -enum dml_output_bpc_class { - dml_out_6 = 0, - dml_out_8 = 1, - dml_out_10 = 2, - dml_out_12 = 3, - dml_out_16 = 4 -}; -enum dml_output_standard_class { - dml_std_cvt = 0, - dml_std_cea = 1, - dml_std_cvtr2 = 2 -}; -enum dml_rotation_angle { - dml_rotation_0 = 0, - dml_rotation_90 = 1, - dml_rotation_180 = 2, - dml_rotation_270 = 3, - dml_rotation_0m = 4, - dml_rotation_90m = 5, - dml_rotation_180m = 6, - dml_rotation_270m = 7 -}; -enum dml_swizzle_mode { - dml_sw_linear = 0, - dml_sw_256b_s = 1, - dml_sw_256b_d = 2, - dml_sw_256b_r = 3, - dml_sw_4kb_z = 4, - dml_sw_4kb_s = 5, - dml_sw_4kb_d = 6, - dml_sw_4kb_r = 7, - dml_sw_64kb_z = 8, - dml_sw_64kb_s = 9, - dml_sw_64kb_d = 10, - dml_sw_64kb_r = 11, - dml_sw_256kb_z = 12, - dml_sw_256kb_s = 13, - dml_sw_256kb_d = 14, - dml_sw_256kb_r = 15, - dml_sw_64kb_z_t = 16, - dml_sw_64kb_s_t = 17, - dml_sw_64kb_d_t = 18, - dml_sw_64kb_r_t = 19, - dml_sw_4kb_z_x = 20, - dml_sw_4kb_s_x = 21, - dml_sw_4kb_d_x = 22, - dml_sw_4kb_r_x = 23, - dml_sw_64kb_z_x = 24, - dml_sw_64kb_s_x = 25, - dml_sw_64kb_d_x = 26, - dml_sw_64kb_r_x = 27, - dml_sw_256kb_z_x = 28, - dml_sw_256kb_s_x = 29, - dml_sw_256kb_d_x = 30, - dml_sw_256kb_r_x = 31, - dml_sw_256b_2d = 32, - dml_sw_4kb_2d = 33, - dml_sw_64kb_2d = 34, - dml_sw_256kb_2d = 35 -}; -enum dml_lb_depth { - dml_lb_6 = 0, - dml_lb_8 = 1, - dml_lb_10 = 2, - dml_lb_12 = 3, - dml_lb_16 = 4 -}; -enum dml_voltage_state { - dml_vmin_lv = 0, - dml_vmin = 1, - dml_vmid = 2, - dml_vnom = 3, - dml_vmax = 4 -}; -enum dml_source_macro_tile_size { - dml_4k_tile = 0, - dml_64k_tile = 1, - dml_256k_tile = 2 -}; -enum dml_cursor_bpp { - dml_cur_2bit = 0, - dml_cur_32bit = 1, - dml_cur_64bit = 2 -}; -enum dml_dram_clock_change_support { - dml_dram_clock_change_vactive = 0, - dml_dram_clock_change_vblank = 1, - dml_dram_clock_change_vblank_drr = 2, - dml_dram_clock_change_vactive_w_mall_full_frame = 3, - dml_dram_clock_change_vactive_w_mall_sub_vp = 4, - dml_dram_clock_change_vblank_w_mall_full_frame = 5, - dml_dram_clock_change_vblank_drr_w_mall_full_frame = 6, - dml_dram_clock_change_vblank_w_mall_sub_vp = 7, - dml_dram_clock_change_vblank_drr_w_mall_sub_vp = 8, - dml_dram_clock_change_unsupported = 9 -}; -enum dml_fclock_change_support { - dml_fclock_change_vactive = 0, - dml_fclock_change_vblank = 1, - dml_fclock_change_unsupported = 2 -}; -enum dml_dsc_enable { - dml_dsc_disable = 0, - dml_dsc_enable = 1, - dml_dsc_enable_if_necessary = 2 -}; -enum dml_mpc_use_policy { - dml_mpc_disabled = 0, - dml_mpc_as_possible = 1, - dml_mpc_as_needed_for_voltage = 2, - dml_mpc_as_needed_for_pstate_and_voltage = 3, - dml_mpc_as_needed = 4, - dml_mpc_2to1 = 5 -}; -enum dml_odm_use_policy { - dml_odm_use_policy_bypass = 0, - dml_odm_use_policy_combine_as_needed = 1, - dml_odm_use_policy_combine_2to1 = 2, - dml_odm_use_policy_combine_3to1 = 3, - dml_odm_use_policy_combine_4to1 = 4, - dml_odm_use_policy_split_1to2 = 5, - dml_odm_use_policy_mso_1to2 = 6, - dml_odm_use_policy_mso_1to4 = 7 -}; -enum dml_odm_mode { - dml_odm_mode_bypass = 0, - dml_odm_mode_combine_2to1 = 1, - dml_odm_mode_combine_3to1 = 2, - dml_odm_mode_combine_4to1 = 3, - dml_odm_mode_split_1to2 = 4, - dml_odm_mode_mso_1to2 = 5, - dml_odm_mode_mso_1to4 = 6 -}; -enum dml_writeback_configuration { - dml_whole_buffer_for_single_stream_no_interleave = 0, - dml_whole_buffer_for_single_stream_interleave = 1 -}; -enum dml_immediate_flip_requirement { - dml_immediate_flip_not_required = 0, - dml_immediate_flip_required = 1, - dml_immediate_flip_if_possible = 2 -}; -enum dml_unbounded_requesting_policy { - dml_unbounded_requesting_enable = 0, - dml_unbounded_requesting_edp_only = 1, - dml_unbounded_requesting_disable = 2 -}; -enum dml_clk_cfg_policy { - dml_use_required_freq = 0, - dml_use_override_freq = 1, - dml_use_state_freq = 2 -}; - - -struct soc_state_bounding_box_st { - dml_float_t socclk_mhz; - dml_float_t dscclk_mhz; - dml_float_t phyclk_mhz; - dml_float_t phyclk_d18_mhz; - dml_float_t phyclk_d32_mhz; - dml_float_t dtbclk_mhz; - dml_float_t fabricclk_mhz; - dml_float_t dcfclk_mhz; - dml_float_t dispclk_mhz; - dml_float_t dppclk_mhz; - dml_float_t dram_speed_mts; - dml_float_t urgent_latency_pixel_data_only_us; - dml_float_t urgent_latency_pixel_mixed_with_vm_data_us; - dml_float_t urgent_latency_vm_data_only_us; - dml_float_t writeback_latency_us; - dml_float_t urgent_latency_adjustment_fabric_clock_component_us; - dml_float_t urgent_latency_adjustment_fabric_clock_reference_mhz; - dml_float_t sr_exit_time_us; - dml_float_t sr_enter_plus_exit_time_us; - dml_float_t sr_exit_z8_time_us; - dml_float_t sr_enter_plus_exit_z8_time_us; - dml_float_t dram_clock_change_latency_us; - dml_float_t fclk_change_latency_us; - dml_float_t usr_retraining_latency_us; - dml_bool_t use_ideal_dram_bw_strobe; - dml_float_t g6_temp_read_blackout_us; - - struct { - dml_uint_t urgent_ramp_uclk_cycles; - dml_uint_t trip_to_memory_uclk_cycles; - dml_uint_t meta_trip_to_memory_uclk_cycles; - dml_uint_t maximum_latency_when_urgent_uclk_cycles; - dml_uint_t average_latency_when_urgent_uclk_cycles; - dml_uint_t maximum_latency_when_non_urgent_uclk_cycles; - dml_uint_t average_latency_when_non_urgent_uclk_cycles; - } dml_dcn401_uclk_dpm_dependent_soc_qos_params; -}; - -struct soc_bounding_box_st { - dml_float_t dprefclk_mhz; - dml_float_t xtalclk_mhz; - dml_float_t pcierefclk_mhz; - dml_float_t refclk_mhz; - dml_float_t amclk_mhz; - dml_uint_t max_outstanding_reqs; - dml_float_t pct_ideal_sdp_bw_after_urgent; - dml_float_t pct_ideal_fabric_bw_after_urgent; - dml_float_t pct_ideal_dram_bw_after_urgent_pixel_only; - dml_float_t pct_ideal_dram_bw_after_urgent_pixel_and_vm; - dml_float_t pct_ideal_dram_bw_after_urgent_vm_only; - dml_float_t pct_ideal_dram_bw_after_urgent_strobe; - dml_float_t max_avg_sdp_bw_use_normal_percent; - dml_float_t max_avg_fabric_bw_use_normal_percent; - dml_float_t max_avg_dram_bw_use_normal_percent; - dml_float_t max_avg_dram_bw_use_normal_strobe_percent; - - dml_float_t svp_prefetch_pct_ideal_sdp_bw_after_urgent; - dml_float_t svp_prefetch_pct_ideal_fabric_bw_after_urgent; - dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_pixel_only; - dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_pixel_and_vm; - dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_vm_only; - dml_float_t svp_prefetch_max_avg_sdp_bw_use_normal_percent; - dml_float_t svp_prefetch_max_avg_fabric_bw_use_normal_percent; - dml_float_t svp_prefetch_max_avg_dram_bw_use_normal_percent; - - dml_uint_t round_trip_ping_latency_dcfclk_cycles; - dml_uint_t urgent_out_of_order_return_per_channel_pixel_only_bytes; - dml_uint_t urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; - dml_uint_t urgent_out_of_order_return_per_channel_vm_only_bytes; - dml_uint_t num_chans; - dml_uint_t return_bus_width_bytes; - dml_uint_t dram_channel_width_bytes; - dml_uint_t fabric_datapath_to_dcn_data_return_bytes; - dml_uint_t hostvm_min_page_size_kbytes; - dml_uint_t gpuvm_min_page_size_kbytes; - dml_float_t phy_downspread_percent; - dml_float_t dcn_downspread_percent; - dml_float_t smn_latency_us; - dml_uint_t mall_allocated_for_dcn_mbytes; - dml_float_t dispclk_dppclk_vco_speed_mhz; - dml_bool_t do_urgent_latency_adjustment; - - dml_uint_t mem_word_bytes; - dml_uint_t num_dcc_mcaches; - dml_uint_t mcache_size_bytes; - dml_uint_t mcache_line_size_bytes; - - struct { - dml_bool_t UseNewDCN401SOCParameters; - dml_uint_t df_qos_response_time_fclk_cycles; - dml_uint_t max_round_trip_to_furthest_cs_fclk_cycles; - dml_uint_t mall_overhead_fclk_cycles; - dml_uint_t meta_trip_adder_fclk_cycles; - dml_uint_t average_transport_distance_fclk_cycles; - dml_float_t umc_urgent_ramp_latency_margin; - dml_float_t umc_max_latency_margin; - dml_float_t umc_average_latency_margin; - dml_float_t fabric_max_transport_latency_margin; - dml_float_t fabric_average_transport_latency_margin; - } dml_dcn401_soc_qos_params; - -}; - -struct ip_params_st { - dml_uint_t vblank_nom_default_us; - dml_uint_t rob_buffer_size_kbytes; - dml_uint_t config_return_buffer_size_in_kbytes; - dml_uint_t config_return_buffer_segment_size_in_kbytes; - dml_uint_t compressed_buffer_segment_size_in_kbytes; - dml_uint_t meta_fifo_size_in_kentries; - dml_uint_t zero_size_buffer_entries; - dml_uint_t dpte_buffer_size_in_pte_reqs_luma; - dml_uint_t dpte_buffer_size_in_pte_reqs_chroma; - dml_uint_t dcc_meta_buffer_size_bytes; - dml_bool_t gpuvm_enable; - dml_bool_t hostvm_enable; - dml_uint_t gpuvm_max_page_table_levels; - dml_uint_t hostvm_max_page_table_levels; - dml_uint_t pixel_chunk_size_kbytes; - dml_uint_t alpha_pixel_chunk_size_kbytes; - dml_uint_t min_pixel_chunk_size_bytes; - dml_uint_t meta_chunk_size_kbytes; - dml_uint_t min_meta_chunk_size_bytes; - dml_uint_t writeback_chunk_size_kbytes; - dml_uint_t line_buffer_size_bits; - dml_uint_t max_line_buffer_lines; - dml_uint_t writeback_interface_buffer_size_kbytes; - dml_uint_t max_num_dpp; - dml_uint_t max_num_otg; - dml_uint_t max_num_wb; - dml_uint_t max_dchub_pscl_bw_pix_per_clk; - dml_uint_t max_pscl_lb_bw_pix_per_clk; - dml_uint_t max_lb_vscl_bw_pix_per_clk; - dml_uint_t max_vscl_hscl_bw_pix_per_clk; - dml_float_t max_hscl_ratio; - dml_float_t max_vscl_ratio; - dml_uint_t max_hscl_taps; - dml_uint_t max_vscl_taps; - dml_uint_t num_dsc; - dml_uint_t maximum_dsc_bits_per_component; - dml_uint_t maximum_pixels_per_line_per_dsc_unit; - dml_bool_t dsc422_native_support; - dml_bool_t cursor_64bpp_support; - dml_float_t dispclk_ramp_margin_percent; - dml_uint_t dppclk_delay_subtotal; - dml_uint_t dppclk_delay_scl; - dml_uint_t dppclk_delay_scl_lb_only; - dml_uint_t dppclk_delay_cnvc_formatter; - dml_uint_t dppclk_delay_cnvc_cursor; - dml_uint_t cursor_buffer_size; - dml_uint_t cursor_chunk_size; - dml_uint_t dispclk_delay_subtotal; - dml_bool_t dynamic_metadata_vm_enabled; - dml_uint_t max_inter_dcn_tile_repeaters; - dml_uint_t max_num_hdmi_frl_outputs; - dml_uint_t max_num_dp2p0_outputs; - dml_uint_t max_num_dp2p0_streams; - dml_bool_t dcc_supported; - dml_bool_t ptoi_supported; - dml_float_t writeback_max_hscl_ratio; - dml_float_t writeback_max_vscl_ratio; - dml_float_t writeback_min_hscl_ratio; - dml_float_t writeback_min_vscl_ratio; - dml_uint_t writeback_max_hscl_taps; - dml_uint_t writeback_max_vscl_taps; - dml_uint_t writeback_line_buffer_buffer_size; -}; - -struct DmlPipe { - dml_float_t Dppclk; - dml_float_t Dispclk; - dml_float_t PixelClock; - dml_float_t DCFClkDeepSleep; - dml_uint_t DPPPerSurface; - dml_bool_t ScalerEnabled; - enum dml_rotation_angle SourceScan; - dml_uint_t ViewportHeight; - dml_uint_t ViewportHeightChroma; - dml_uint_t BlockWidth256BytesY; - dml_uint_t BlockHeight256BytesY; - dml_uint_t BlockWidth256BytesC; - dml_uint_t BlockHeight256BytesC; - dml_uint_t BlockWidthY; - dml_uint_t BlockHeightY; - dml_uint_t BlockWidthC; - dml_uint_t BlockHeightC; - dml_uint_t InterlaceEnable; - dml_uint_t NumberOfCursors; - dml_uint_t VBlank; - dml_uint_t HTotal; - dml_uint_t HActive; - dml_bool_t DCCEnable; - enum dml_odm_mode ODMMode; - enum dml_source_format_class SourcePixelFormat; - enum dml_swizzle_mode SurfaceTiling; - dml_uint_t BytePerPixelY; - dml_uint_t BytePerPixelC; - dml_bool_t ProgressiveToInterlaceUnitInOPP; - dml_float_t VRatio; - dml_float_t VRatioChroma; - dml_uint_t VTaps; - dml_uint_t VTapsChroma; - dml_uint_t PitchY; - dml_uint_t DCCMetaPitchY; - dml_uint_t PitchC; - dml_uint_t DCCMetaPitchC; - dml_bool_t ViewportStationary; - dml_uint_t ViewportXStart; - dml_uint_t ViewportYStart; - dml_uint_t ViewportXStartC; - dml_uint_t ViewportYStartC; - dml_bool_t FORCE_ONE_ROW_FOR_FRAME; - dml_uint_t SwathHeightY; - dml_uint_t SwathHeightC; -}; - -struct Watermarks { - dml_float_t UrgentWatermark; - dml_float_t WritebackUrgentWatermark; - dml_float_t DRAMClockChangeWatermark; - dml_float_t FCLKChangeWatermark; - dml_float_t WritebackDRAMClockChangeWatermark; - dml_float_t WritebackFCLKChangeWatermark; - dml_float_t StutterExitWatermark; - dml_float_t StutterEnterPlusExitWatermark; - dml_float_t Z8StutterExitWatermark; - dml_float_t Z8StutterEnterPlusExitWatermark; - dml_float_t USRRetrainingWatermark; -}; - -struct SOCParametersList { - dml_float_t UrgentLatency; - dml_float_t ExtraLatency; - dml_float_t WritebackLatency; - dml_float_t DRAMClockChangeLatency; - dml_float_t FCLKChangeLatency; - dml_float_t SRExitTime; - dml_float_t SREnterPlusExitTime; - dml_float_t SRExitZ8Time; - dml_float_t SREnterPlusExitZ8Time; - dml_float_t USRRetrainingLatency; - dml_float_t SMNLatency; -}; - -/// @brief Struct that represent Plane configration of a display cfg -struct dml_plane_cfg_st { - // - // Pipe/Surface Parameters - // - dml_bool_t GPUVMEnable; /// > 23) & 255) - 128; - - x &= ~(255 << 23); - x += 127 << 23; - *exp_ptr = x; - - in = ((-1.0f / 3) * in + 2) * in - 2.0f / 3; - - return (in + log_2); -} - -dml_bool_t dml_util_is_420(enum dml_source_format_class source_format) -{ - dml_bool_t val = false; - - switch (source_format) { - case dml_444_16: - val = 0; - break; - case dml_444_32: - val = 0; - break; - case dml_444_64: - val = 0; - break; - case dml_420_8: - val = 1; - break; - case dml_420_10: - val = 1; - break; - case dml_422_8: - val = 0; - break; - case dml_422_10: - val = 0; - break; - default: - ASSERT(0); - break; - } - return val; -} - -static inline float dcn_bw_pow(float a, float exp) -{ - float temp; - /*ASSERT(exp == (int)exp);*/ - if ((int)exp == 0) - return 1; - temp = dcn_bw_pow(a, (int)(exp / 2)); - if (((int)exp % 2) == 0) { - return temp * temp; - } else { - if ((int)exp > 0) - return a * temp * temp; - else - return (temp * temp) / a; - } -} - -static inline float dcn_bw_ceil2(const float arg, const float significance) -{ - ASSERT(significance != 0); - - return ((int)(arg / significance + 0.99999)) * significance; -} - -static inline float dcn_bw_floor2(const float arg, const float significance) -{ - ASSERT(significance != 0); - - return ((int)(arg / significance)) * significance; -} - -dml_float_t dml_ceil(dml_float_t x, dml_float_t granularity) -{ - if (granularity == 0) - return 0; - //return (dml_float_t) (ceil(x / granularity) * granularity); - return (dml_float_t)dcn_bw_ceil2(x, granularity); -} - -dml_float_t dml_floor(dml_float_t x, dml_float_t granularity) -{ - if (granularity == 0) - return 0; - //return (dml_float_t) (floor(x / granularity) * granularity); - return (dml_float_t)dcn_bw_floor2(x, granularity); -} - -dml_float_t dml_min(dml_float_t x, dml_float_t y) -{ - if (x != x) - return y; - if (y != y) - return x; - if (x < y) - return x; - else - return y; -} - -dml_float_t dml_min3(dml_float_t x, dml_float_t y, dml_float_t z) -{ - return dml_min(dml_min(x, y), z); -} - -dml_float_t dml_min4(dml_float_t x, dml_float_t y, dml_float_t z, dml_float_t w) -{ - return dml_min(dml_min(dml_min(x, y), z), w); -} - -dml_float_t dml_max(dml_float_t x, dml_float_t y) -{ - if (x != x) - return y; - if (y != y) - return x; -if (x > y) - return x; - else - return y; -} -dml_float_t dml_max3(dml_float_t x, dml_float_t y, dml_float_t z) -{ - return dml_max(dml_max(x, y), z); -} -dml_float_t dml_max4(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d) -{ - return dml_max(dml_max(a, b), dml_max(c, d)); -} -dml_float_t dml_max5(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d, dml_float_t e) -{ - return dml_max(dml_max4(a, b, c, d), e); -} -dml_float_t dml_log(dml_float_t x, dml_float_t base) -{ - return (dml_float_t) (_log(x) / _log(base)); -} - -dml_float_t dml_log2(dml_float_t x) -{ - return (dml_float_t) (_log(x) / _log(2)); -} - -dml_float_t dml_round(dml_float_t val, dml_bool_t bankers_rounding) -{ -// if (bankers_rounding) -// return (dml_float_t) lrint(val); -// else { -// return round(val); - double round_pt = 0.5; - double ceil = dml_ceil(val, 1); - double floor = dml_floor(val, 1); - - if (val - floor >= round_pt) - return ceil; - else - return floor; -// } -} - -dml_float_t dml_pow(dml_float_t base, int exp) -{ - return (dml_float_t) dcn_bw_pow(base, exp); -} - -dml_uint_t dml_round_to_multiple(dml_uint_t num, dml_uint_t multiple, dml_bool_t up) -{ - dml_uint_t remainder; - - if (multiple == 0) - return num; - - remainder = num % multiple; - if (remainder == 0) - return num; - - if (up) - return (num + multiple - remainder); - else - return (num - remainder); -} - -void dml_print_data_rq_regs_st(const dml_display_plane_rq_regs_st *rq_regs) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DISPLAY_PLANE_RQ_REGS_ST\n"); - dml_print("DML: chunk_size = 0x%x\n", rq_regs->chunk_size); - dml_print("DML: min_chunk_size = 0x%x\n", rq_regs->min_chunk_size); - dml_print("DML: meta_chunk_size = 0x%x\n", rq_regs->meta_chunk_size); - dml_print("DML: min_meta_chunk_size = 0x%x\n", rq_regs->min_meta_chunk_size); - dml_print("DML: dpte_group_size = 0x%x\n", rq_regs->dpte_group_size); - dml_print("DML: mpte_group_size = 0x%x\n", rq_regs->mpte_group_size); - dml_print("DML: swath_height = 0x%x\n", rq_regs->swath_height); - dml_print("DML: pte_row_height_linear = 0x%x\n", rq_regs->pte_row_height_linear); - dml_print("DML: ===================================== \n"); -} - -void dml_print_rq_regs_st(const dml_display_rq_regs_st *rq_regs) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DISPLAY_RQ_REGS_ST\n"); - dml_print("DML: \n"); - dml_print_data_rq_regs_st(&rq_regs->rq_regs_l); - dml_print("DML: \n"); - dml_print_data_rq_regs_st(&rq_regs->rq_regs_c); - dml_print("DML: drq_expansion_mode = 0x%x\n", rq_regs->drq_expansion_mode); - dml_print("DML: prq_expansion_mode = 0x%x\n", rq_regs->prq_expansion_mode); - dml_print("DML: mrq_expansion_mode = 0x%x\n", rq_regs->mrq_expansion_mode); - dml_print("DML: crq_expansion_mode = 0x%x\n", rq_regs->crq_expansion_mode); - dml_print("DML: plane1_base_address = 0x%x\n", rq_regs->plane1_base_address); - dml_print("DML: ===================================== \n"); -} - -void dml_print_dlg_regs_st(const dml_display_dlg_regs_st *dlg_regs) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DISPLAY_DLG_REGS_ST \n"); - dml_print("DML: refcyc_h_blank_end = 0x%x\n", dlg_regs->refcyc_h_blank_end); - dml_print("DML: dlg_vblank_end = 0x%x\n", dlg_regs->dlg_vblank_end); - dml_print("DML: min_dst_y_next_start = 0x%x\n", dlg_regs->min_dst_y_next_start); - dml_print("DML: refcyc_per_htotal = 0x%x\n", dlg_regs->refcyc_per_htotal); - dml_print("DML: refcyc_x_after_scaler = 0x%x\n", dlg_regs->refcyc_x_after_scaler); - dml_print("DML: dst_y_after_scaler = 0x%x\n", dlg_regs->dst_y_after_scaler); - dml_print("DML: dst_y_prefetch = 0x%x\n", dlg_regs->dst_y_prefetch); - dml_print("DML: dst_y_per_vm_vblank = 0x%x\n", dlg_regs->dst_y_per_vm_vblank); - dml_print("DML: dst_y_per_row_vblank = 0x%x\n", dlg_regs->dst_y_per_row_vblank); - dml_print("DML: dst_y_per_vm_flip = 0x%x\n", dlg_regs->dst_y_per_vm_flip); - dml_print("DML: dst_y_per_row_flip = 0x%x\n", dlg_regs->dst_y_per_row_flip); - dml_print("DML: ref_freq_to_pix_freq = 0x%x\n", dlg_regs->ref_freq_to_pix_freq); - dml_print("DML: vratio_prefetch = 0x%x\n", dlg_regs->vratio_prefetch); - dml_print("DML: vratio_prefetch_c = 0x%x\n", dlg_regs->vratio_prefetch_c); - dml_print("DML: refcyc_per_pte_group_vblank_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_vblank_l); - dml_print("DML: refcyc_per_pte_group_vblank_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_vblank_c); - dml_print("DML: refcyc_per_meta_chunk_vblank_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_vblank_l); - dml_print("DML: refcyc_per_meta_chunk_vblank_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_vblank_c); - dml_print("DML: refcyc_per_pte_group_flip_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_flip_l); - dml_print("DML: refcyc_per_pte_group_flip_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_flip_c); - dml_print("DML: refcyc_per_meta_chunk_flip_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_flip_l); - dml_print("DML: refcyc_per_meta_chunk_flip_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_flip_c); - dml_print("DML: dst_y_per_pte_row_nom_l = 0x%x\n", dlg_regs->dst_y_per_pte_row_nom_l); - dml_print("DML: dst_y_per_pte_row_nom_c = 0x%x\n", dlg_regs->dst_y_per_pte_row_nom_c); - dml_print("DML: refcyc_per_pte_group_nom_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_nom_l); - dml_print("DML: refcyc_per_pte_group_nom_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_nom_c); - dml_print("DML: dst_y_per_meta_row_nom_l = 0x%x\n", dlg_regs->dst_y_per_meta_row_nom_l); - dml_print("DML: dst_y_per_meta_row_nom_c = 0x%x\n", dlg_regs->dst_y_per_meta_row_nom_c); - dml_print("DML: refcyc_per_meta_chunk_nom_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_nom_l); - dml_print("DML: refcyc_per_meta_chunk_nom_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_nom_c); - dml_print("DML: refcyc_per_line_delivery_pre_l = 0x%x\n", dlg_regs->refcyc_per_line_delivery_pre_l); - dml_print("DML: refcyc_per_line_delivery_pre_c = 0x%x\n", dlg_regs->refcyc_per_line_delivery_pre_c); - dml_print("DML: refcyc_per_line_delivery_l = 0x%x\n", dlg_regs->refcyc_per_line_delivery_l); - dml_print("DML: refcyc_per_line_delivery_c = 0x%x\n", dlg_regs->refcyc_per_line_delivery_c); - dml_print("DML: refcyc_per_vm_group_vblank = 0x%x\n", dlg_regs->refcyc_per_vm_group_vblank); - dml_print("DML: refcyc_per_vm_group_flip = 0x%x\n", dlg_regs->refcyc_per_vm_group_flip); - dml_print("DML: refcyc_per_vm_req_vblank = 0x%x\n", dlg_regs->refcyc_per_vm_req_vblank); - dml_print("DML: refcyc_per_vm_req_flip = 0x%x\n", dlg_regs->refcyc_per_vm_req_flip); - dml_print("DML: chunk_hdl_adjust_cur0 = 0x%x\n", dlg_regs->chunk_hdl_adjust_cur0); - dml_print("DML: dst_y_offset_cur1 = 0x%x\n", dlg_regs->dst_y_offset_cur1); - dml_print("DML: chunk_hdl_adjust_cur1 = 0x%x\n", dlg_regs->chunk_hdl_adjust_cur1); - dml_print("DML: vready_after_vcount0 = 0x%x\n", dlg_regs->vready_after_vcount0); - dml_print("DML: dst_y_delta_drq_limit = 0x%x\n", dlg_regs->dst_y_delta_drq_limit); - dml_print("DML: refcyc_per_vm_dmdata = 0x%x\n", dlg_regs->refcyc_per_vm_dmdata); - dml_print("DML: ===================================== \n"); -} - -void dml_print_ttu_regs_st(const dml_display_ttu_regs_st *ttu_regs) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DISPLAY_TTU_REGS_ST \n"); - dml_print("DML: qos_level_low_wm = 0x%x\n", ttu_regs->qos_level_low_wm); - dml_print("DML: qos_level_high_wm = 0x%x\n", ttu_regs->qos_level_high_wm); - dml_print("DML: min_ttu_vblank = 0x%x\n", ttu_regs->min_ttu_vblank); - dml_print("DML: qos_level_flip = 0x%x\n", ttu_regs->qos_level_flip); - dml_print("DML: refcyc_per_req_delivery_pre_l = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_l); - dml_print("DML: refcyc_per_req_delivery_l = 0x%x\n", ttu_regs->refcyc_per_req_delivery_l); - dml_print("DML: refcyc_per_req_delivery_pre_c = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_c); - dml_print("DML: refcyc_per_req_delivery_c = 0x%x\n", ttu_regs->refcyc_per_req_delivery_c); - dml_print("DML: refcyc_per_req_delivery_cur0 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_cur0); - dml_print("DML: refcyc_per_req_delivery_pre_cur0 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_cur0); - dml_print("DML: refcyc_per_req_delivery_cur1 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_cur1); - dml_print("DML: refcyc_per_req_delivery_pre_cur1 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_cur1); - dml_print("DML: qos_level_fixed_l = 0x%x\n", ttu_regs->qos_level_fixed_l); - dml_print("DML: qos_ramp_disable_l = 0x%x\n", ttu_regs->qos_ramp_disable_l); - dml_print("DML: qos_level_fixed_c = 0x%x\n", ttu_regs->qos_level_fixed_c); - dml_print("DML: qos_ramp_disable_c = 0x%x\n", ttu_regs->qos_ramp_disable_c); - dml_print("DML: qos_level_fixed_cur0 = 0x%x\n", ttu_regs->qos_level_fixed_cur0); - dml_print("DML: qos_ramp_disable_cur0 = 0x%x\n", ttu_regs->qos_ramp_disable_cur0); - dml_print("DML: qos_level_fixed_cur1 = 0x%x\n", ttu_regs->qos_level_fixed_cur1); - dml_print("DML: qos_ramp_disable_cur1 = 0x%x\n", ttu_regs->qos_ramp_disable_cur1); - dml_print("DML: ===================================== \n"); -} - -void dml_print_dml_policy(const struct dml_mode_eval_policy_st *policy) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DML_MODE_EVAL_POLICY_ST\n"); - dml_print("DML: Policy: UseUnboundedRequesting = 0x%x\n", policy->UseUnboundedRequesting); - dml_print("DML: Policy: UseMinimumRequiredDCFCLK = 0x%x\n", policy->UseMinimumRequiredDCFCLK); - dml_print("DML: Policy: DRAMClockChangeRequirementFinal = 0x%x\n", policy->DRAMClockChangeRequirementFinal); - dml_print("DML: Policy: FCLKChangeRequirementFinal = 0x%x\n", policy->FCLKChangeRequirementFinal); - dml_print("DML: Policy: USRRetrainingRequiredFinal = 0x%x\n", policy->USRRetrainingRequiredFinal); - dml_print("DML: Policy: EnhancedPrefetchScheduleAccelerationFinal = 0x%x\n", policy->EnhancedPrefetchScheduleAccelerationFinal); - dml_print("DML: Policy: NomDETInKByteOverrideEnable = 0x%x\n", policy->NomDETInKByteOverrideEnable); - dml_print("DML: Policy: NomDETInKByteOverrideValue = 0x%x\n", policy->NomDETInKByteOverrideValue); - dml_print("DML: Policy: DCCProgrammingAssumesScanDirectionUnknownFinal = 0x%x\n", policy->DCCProgrammingAssumesScanDirectionUnknownFinal); - dml_print("DML: Policy: SynchronizeTimingsFinal = 0x%x\n", policy->SynchronizeTimingsFinal); - dml_print("DML: Policy: SynchronizeDRRDisplaysForUCLKPStateChangeFinal = 0x%x\n", policy->SynchronizeDRRDisplaysForUCLKPStateChangeFinal); - dml_print("DML: Policy: AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported = 0x%x\n", policy->AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported); - dml_print("DML: Policy: AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported = 0x%x\n", policy->AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported); - - for (dml_uint_t i = 0; i < DCN_DML__NUM_PLANE; i++) { - dml_print("DML: i=%0d, Policy: MPCCombineUse = 0x%x\n", i, policy->MPCCombineUse[i]); - dml_print("DML: i=%0d, Policy: ODMUse = 0x%x\n", i, policy->ODMUse[i]); - dml_print("DML: i=%0d, Policy: ImmediateFlipRequirement = 0x%x\n", i, policy->ImmediateFlipRequirement[i]); - dml_print("DML: i=%0d, Policy: AllowForPStateChangeOrStutterInVBlank = 0x%x\n", i, policy->AllowForPStateChangeOrStutterInVBlank[i]); - } - dml_print("DML: ===================================== \n"); -} - -void dml_print_mode_support(struct display_mode_lib_st *mode_lib, dml_uint_t j) -{ - dml_print("DML: MODE SUPPORT: ===============================================\n"); - dml_print("DML: MODE SUPPORT: Voltage State %d\n", j); - dml_print("DML: MODE SUPPORT: Mode Supported : %s\n", mode_lib->ms.support.ModeSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Scale Ratio And Taps : %s\n", mode_lib->ms.support.ScaleRatioAndTapsSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Source Format Pixel And Scan : %s\n", mode_lib->ms.support.SourceFormatPixelAndScanSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Viewport Size : %s\n", mode_lib->ms.support.ViewportSizeSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Link Rate Does Not Match DP Version : %s\n", mode_lib->ms.support.LinkRateDoesNotMatchDPVersion == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Link Rate For Multistream Not Indicated : %s\n", mode_lib->ms.support.LinkRateForMultistreamNotIndicated == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: BPP For Multi stream Not Indicated : %s\n", mode_lib->ms.support.BPPForMultistreamNotIndicated == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Multistream With HDMI Or eDP : %s\n", mode_lib->ms.support.MultistreamWithHDMIOreDP == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Exceeded Multistream Slots : %s\n", mode_lib->ms.support.ExceededMultistreamSlots == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: MSO Or ODM Split With Non DP Link : %s\n", mode_lib->ms.support.MSOOrODMSplitWithNonDPLink == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Not Enough Lanes For MSO : %s\n", mode_lib->ms.support.NotEnoughLanesForMSO == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: LinkCapacitySupport : %s\n", mode_lib->ms.support.LinkCapacitySupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: P2IWith420 : %s\n", mode_lib->ms.support.P2IWith420 == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DSCOnlyIfNecessaryWithBPP : %s\n", mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DSC422NativeNotSupported : %s\n", mode_lib->ms.support.DSC422NativeNotSupported == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: MPCCombineMethodIncompatible : %s\n", mode_lib->ms.support.MPCCombineMethodIncompatible == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ODMCombineTwoToOneSupportCheckOK : %s\n", mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ODMCombineFourToOneSupportCheckOK : %s\n", mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: NotEnoughDSCUnits : %s\n", mode_lib->ms.support.NotEnoughDSCUnits == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: NotEnoughDSCSlices : %s\n", mode_lib->ms.support.NotEnoughDSCSlices == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe : %s\n", mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: InvalidCombinationOfMALLUseForPStateAndStaticScreen : %s\n", mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DSCCLKRequiredMoreThanSupported : %s\n", mode_lib->ms.support.DSCCLKRequiredMoreThanSupported == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: PixelsPerLinePerDSCUnitSupport : %s\n", mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DTBCLKRequiredMoreThanSupported : %s\n", mode_lib->ms.support.DTBCLKRequiredMoreThanSupported == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: InvalidCombinationOfMALLUseForPState : %s\n", mode_lib->ms.support.InvalidCombinationOfMALLUseForPState == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified : %s\n", mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ROB Support : %s\n", mode_lib->ms.support.ROBSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DISPCLK DPPCLK Support : %s\n", mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Total Available Pipes Support : %s\n", mode_lib->ms.support.TotalAvailablePipesSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Number Of OTG Support : %s\n", mode_lib->ms.support.NumberOfOTGSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Number Of DP2p0 Support : %s\n", mode_lib->ms.support.NumberOfDP2p0Support == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Writeback Latency Support : %s\n", mode_lib->ms.support.WritebackLatencySupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Writeback Scale Ratio And Taps Support : %s\n", mode_lib->ms.support.WritebackScaleRatioAndTapsSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Cursor Support : %s\n", mode_lib->ms.support.CursorSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Pitch Support : %s\n", mode_lib->ms.support.PitchSupport == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Viewport Exceeds Surface : %s\n", mode_lib->ms.support.ViewportExceedsSurface == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Prefetch Supported : %s\n", mode_lib->ms.support.PrefetchSupported[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: VActive Bandwith Support : %s\n", mode_lib->ms.support.VActiveBandwithSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Dynamic Metadata Supported : %s\n", mode_lib->ms.support.DynamicMetadataSupported[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Total Vertical Active Bandwidth Support : %s\n", mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: VRatio In Prefetch Supported : %s\n", mode_lib->ms.support.VRatioInPrefetchSupported[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: PTE Buffer Size Not Exceeded : %s\n", mode_lib->ms.support.PTEBufferSizeNotExceeded[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: DCC Meta Buffer Size Not Exceeded : %s\n", mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Non supported DSC Input BPC : %s\n", mode_lib->ms.support.NonsupportedDSCInputBPC == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Exceeded MALL Size : %s\n", mode_lib->ms.support.ExceededMALLSize == false ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: Host VM or Immediate Flip Supported : %s\n", ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == false && !mode_lib->scratch.dml_core_mode_support_locals.ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j]) ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: dram clock change support : %s\n", mode_lib->scratch.dml_core_mode_support_locals.dram_clock_change_support == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: f_clock change support : %s\n", mode_lib->scratch.dml_core_mode_support_locals.f_clock_change_support == true ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: USR Retraining Support : %s\n", (!mode_lib->ms.policy.USRRetrainingRequiredFinal || &mode_lib->ms.support.USRRetrainingSupport[j]) ? "Supported" : "NOT Supported"); - dml_print("DML: MODE SUPPORT: ===============================================\n"); -} - -void dml_print_dml_mode_support_info(const struct dml_mode_support_info_st *support, dml_bool_t fail_only) -{ - dml_print("DML: ===================================== \n"); - dml_print("DML: DML_MODE_SUPPORT_INFO_ST\n"); - if (!fail_only || support->ModeIsSupported == 0) - dml_print("DML: support: ModeIsSupported = 0x%x\n", support->ModeIsSupported); - if (!fail_only || support->ImmediateFlipSupport == 0) - dml_print("DML: support: ImmediateFlipSupport = 0x%x\n", support->ImmediateFlipSupport); - if (!fail_only || support->WritebackLatencySupport == 0) - dml_print("DML: support: WritebackLatencySupport = 0x%x\n", support->WritebackLatencySupport); - if (!fail_only || support->ScaleRatioAndTapsSupport == 0) - dml_print("DML: support: ScaleRatioAndTapsSupport = 0x%x\n", support->ScaleRatioAndTapsSupport); - if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) - dml_print("DML: support: SourceFormatPixelAndScanSupport = 0x%x\n", support->SourceFormatPixelAndScanSupport); - if (!fail_only || support->MPCCombineMethodIncompatible == 1) - dml_print("DML: support: MPCCombineMethodIncompatible = 0x%x\n", support->MPCCombineMethodIncompatible); - if (!fail_only || support->P2IWith420 == 1) - dml_print("DML: support: P2IWith420 = 0x%x\n", support->P2IWith420); - if (!fail_only || support->DSCOnlyIfNecessaryWithBPP == 1) - dml_print("DML: support: DSCOnlyIfNecessaryWithBPP = 0x%x\n", support->DSCOnlyIfNecessaryWithBPP); - if (!fail_only || support->DSC422NativeNotSupported == 1) - dml_print("DML: support: DSC422NativeNotSupported = 0x%x\n", support->DSC422NativeNotSupported); - if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) - dml_print("DML: support: LinkRateDoesNotMatchDPVersion = 0x%x\n", support->LinkRateDoesNotMatchDPVersion); - if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) - dml_print("DML: support: LinkRateForMultistreamNotIndicated = 0x%x\n", support->LinkRateForMultistreamNotIndicated); - if (!fail_only || support->BPPForMultistreamNotIndicated == 1) - dml_print("DML: support: BPPForMultistreamNotIndicated = 0x%x\n", support->BPPForMultistreamNotIndicated); - if (!fail_only || support->MultistreamWithHDMIOreDP == 1) - dml_print("DML: support: MultistreamWithHDMIOreDP = 0x%x\n", support->MultistreamWithHDMIOreDP); - if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) - dml_print("DML: support: MSOOrODMSplitWithNonDPLink = 0x%x\n", support->MSOOrODMSplitWithNonDPLink); - if (!fail_only || support->NotEnoughLanesForMSO == 1) - dml_print("DML: support: NotEnoughLanesForMSO = 0x%x\n", support->NotEnoughLanesForMSO); - if (!fail_only || support->NumberOfOTGSupport == 0) - dml_print("DML: support: NumberOfOTGSupport = 0x%x\n", support->NumberOfOTGSupport); - if (!fail_only || support->NumberOfDP2p0Support == 0) - dml_print("DML: support: NumberOfDP2p0Support = 0x%x\n", support->NumberOfDP2p0Support); - if (!fail_only || support->NonsupportedDSCInputBPC == 1) - dml_print("DML: support: NonsupportedDSCInputBPC = 0x%x\n", support->NonsupportedDSCInputBPC); - if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) - dml_print("DML: support: WritebackScaleRatioAndTapsSupport = 0x%x\n", support->WritebackScaleRatioAndTapsSupport); - if (!fail_only || support->CursorSupport == 0) - dml_print("DML: support: CursorSupport = 0x%x\n", support->CursorSupport); - if (!fail_only || support->PitchSupport == 0) - dml_print("DML: support: PitchSupport = 0x%x\n", support->PitchSupport); - if (!fail_only || support->ViewportExceedsSurface == 1) - dml_print("DML: support: ViewportExceedsSurface = 0x%x\n", support->ViewportExceedsSurface); - if (!fail_only || support->ExceededMALLSize == 1) - dml_print("DML: support: ExceededMALLSize = 0x%x\n", support->ExceededMALLSize); - if (!fail_only || support->EnoughWritebackUnits == 0) - dml_print("DML: support: EnoughWritebackUnits = 0x%x\n", support->EnoughWritebackUnits); - if (!fail_only || support->ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified == 1) - dml_print("DML: support: ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = 0x%x\n", support->ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified); - if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) - dml_print("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = 0x%x\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); - if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) - dml_print("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = 0x%x\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); - if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) - dml_print("DML: support: InvalidCombinationOfMALLUseForPState = 0x%x\n", support->InvalidCombinationOfMALLUseForPState); - - if (!fail_only || support->ExceededMultistreamSlots == 1) - dml_print("DML: support: ExceededMultistreamSlots = 0x%x\n", support->ExceededMultistreamSlots); - if (!fail_only || support->ODMCombineTwoToOneSupportCheckOK == 0) - dml_print("DML: support: ODMCombineTwoToOneSupportCheckOK = 0x%x\n", support->ODMCombineTwoToOneSupportCheckOK); - if (!fail_only || support->ODMCombineFourToOneSupportCheckOK == 0) - dml_print("DML: support: ODMCombineFourToOneSupportCheckOK = 0x%x\n", support->ODMCombineFourToOneSupportCheckOK); - if (!fail_only || support->NotEnoughDSCUnits == 1) - dml_print("DML: support: NotEnoughDSCUnits = 0x%x\n", support->NotEnoughDSCUnits); - if (!fail_only || support->NotEnoughDSCSlices == 1) - dml_print("DML: support: NotEnoughDSCSlices = 0x%x\n", support->NotEnoughDSCSlices); - if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) - dml_print("DML: support: PixelsPerLinePerDSCUnitSupport = 0x%x\n", support->PixelsPerLinePerDSCUnitSupport); - if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) - dml_print("DML: support: DSCCLKRequiredMoreThanSupported = 0x%x\n", support->DSCCLKRequiredMoreThanSupported); - if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) - dml_print("DML: support: DTBCLKRequiredMoreThanSupported = 0x%x\n", support->DTBCLKRequiredMoreThanSupported); - if (!fail_only || support->LinkCapacitySupport == 0) - dml_print("DML: support: LinkCapacitySupport = 0x%x\n", support->LinkCapacitySupport); - - for (dml_uint_t j = 0; j < 2; j++) { - if (!fail_only || support->DRAMClockChangeSupport[j] == dml_dram_clock_change_unsupported) - dml_print("DML: support: combine=%d, DRAMClockChangeSupport = %d\n", j, support->DRAMClockChangeSupport[j]); - if (!fail_only || support->FCLKChangeSupport[j] == dml_fclock_change_unsupported) - dml_print("DML: support: combine=%d, FCLKChangeSupport = %d\n", j, support->FCLKChangeSupport[j]); - if (!fail_only || support->ROBSupport[j] == 0) - dml_print("DML: support: combine=%d, ROBSupport = %d\n", j, support->ROBSupport[j]); - if (!fail_only || support->PTEBufferSizeNotExceeded[j] == 0) - dml_print("DML: support: combine=%d, PTEBufferSizeNotExceeded = %d\n", j, support->PTEBufferSizeNotExceeded[j]); - if (!fail_only || support->DCCMetaBufferSizeNotExceeded[j] == 0) - dml_print("DML: support: combine=%d, DCCMetaBufferSizeNotExceeded = %d\n", j, support->DCCMetaBufferSizeNotExceeded[j]); - if (!fail_only || support->TotalVerticalActiveBandwidthSupport[j] == 0) - dml_print("DML: support: combine=%d, TotalVerticalActiveBandwidthSupport = %d\n", j, support->TotalVerticalActiveBandwidthSupport[j]); - if (!fail_only || support->USRRetrainingSupport[j] == 0) - dml_print("DML: support: combine=%d, USRRetrainingSupport = %d\n", j, support->USRRetrainingSupport[j]); - if (!fail_only || support->VActiveBandwithSupport[j] == 0) - dml_print("DML: support: combine=%d, VActiveBandwithSupport = %d\n", j, support->VActiveBandwithSupport[j]); - if (!fail_only || support->PrefetchSupported[j] == 0) - dml_print("DML: support: combine=%d, PrefetchSupported = %d\n", j, support->PrefetchSupported[j]); - if (!fail_only || support->DynamicMetadataSupported[j] == 0) - dml_print("DML: support: combine=%d, DynamicMetadataSupported = %d\n", j, support->DynamicMetadataSupported[j]); - if (!fail_only || support->VRatioInPrefetchSupported[j] == 0) - dml_print("DML: support: combine=%d, VRatioInPrefetchSupported = %d\n", j, support->VRatioInPrefetchSupported[j]); - if (!fail_only || support->DISPCLK_DPPCLK_Support[j] == 0) - dml_print("DML: support: combine=%d, DISPCLK_DPPCLK_Support = %d\n", j, support->DISPCLK_DPPCLK_Support[j]); - if (!fail_only || support->TotalAvailablePipesSupport[j] == 0) - dml_print("DML: support: combine=%d, TotalAvailablePipesSupport = %d\n", j, support->TotalAvailablePipesSupport[j]); - if (!fail_only || support->ModeSupport[j] == 0) - dml_print("DML: support: combine=%d, ModeSupport = %d\n", j, support->ModeSupport[j]); - if (!fail_only || support->ViewportSizeSupport[j] == 0) - dml_print("DML: support: combine=%d, ViewportSizeSupport = %d\n", j, support->ViewportSizeSupport[j]); - if (!fail_only || support->ImmediateFlipSupportedForState[j] == 0) - dml_print("DML: support: combine=%d, ImmediateFlipSupportedForState = %d\n", j, support->ImmediateFlipSupportedForState[j]); - } -} - -void dml_print_dml_display_cfg_timing(const struct dml_timing_cfg_st *timing, dml_uint_t num_plane) -{ - for (dml_uint_t i = 0; i < num_plane; i++) { - dml_print("DML: timing_cfg: plane=%d, HTotal = %d\n", i, timing->HTotal[i]); - dml_print("DML: timing_cfg: plane=%d, VTotal = %d\n", i, timing->VTotal[i]); - dml_print("DML: timing_cfg: plane=%d, HActive = %d\n", i, timing->HActive[i]); - dml_print("DML: timing_cfg: plane=%d, VActive = %d\n", i, timing->VActive[i]); - dml_print("DML: timing_cfg: plane=%d, VFrontPorch = %d\n", i, timing->VFrontPorch[i]); - dml_print("DML: timing_cfg: plane=%d, VBlankNom = %d\n", i, timing->VBlankNom[i]); - dml_print("DML: timing_cfg: plane=%d, RefreshRate = %d\n", i, timing->RefreshRate[i]); - dml_print("DML: timing_cfg: plane=%d, PixelClock = %f\n", i, timing->PixelClock[i]); - dml_print("DML: timing_cfg: plane=%d, Interlace = %d\n", i, timing->Interlace[i]); - dml_print("DML: timing_cfg: plane=%d, DRRDisplay = %d\n", i, timing->DRRDisplay[i]); - } -} - -void dml_print_dml_display_cfg_plane(const struct dml_plane_cfg_st *plane, dml_uint_t num_plane) -{ - dml_print("DML: plane_cfg: num_plane = %d\n", num_plane); - dml_print("DML: plane_cfg: GPUVMEnable = %d\n", plane->GPUVMEnable); - dml_print("DML: plane_cfg: HostVMEnable = %d\n", plane->HostVMEnable); - dml_print("DML: plane_cfg: GPUVMMaxPageTableLevels = %d\n", plane->GPUVMMaxPageTableLevels); - dml_print("DML: plane_cfg: HostVMMaxPageTableLevels = %d\n", plane->HostVMMaxPageTableLevels); - - for (dml_uint_t i = 0; i < num_plane; i++) { - dml_print("DML: plane_cfg: plane=%d, GPUVMMinPageSizeKBytes = %d\n", i, plane->GPUVMMinPageSizeKBytes[i]); - dml_print("DML: plane_cfg: plane=%d, ForceOneRowForFrame = %d\n", i, plane->ForceOneRowForFrame[i]); - dml_print("DML: plane_cfg: plane=%d, PTEBufferModeOverrideEn = %d\n", i, plane->PTEBufferModeOverrideEn[i]); - dml_print("DML: plane_cfg: plane=%d, PTEBufferMode = %d\n", i, plane->PTEBufferMode[i]); - dml_print("DML: plane_cfg: plane=%d, DETSizeOverride = %d\n", i, plane->DETSizeOverride[i]); - dml_print("DML: plane_cfg: plane=%d, UseMALLForStaticScreen = %d\n", i, plane->UseMALLForStaticScreen[i]); - dml_print("DML: plane_cfg: plane=%d, UseMALLForPStateChange = %d\n", i, plane->UseMALLForPStateChange[i]); - dml_print("DML: plane_cfg: plane=%d, BlendingAndTiming = %d\n", i, plane->BlendingAndTiming[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportWidth = %d\n", i, plane->ViewportWidth[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportHeight = %d\n", i, plane->ViewportHeight[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportWidthChroma = %d\n", i, plane->ViewportWidthChroma[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportHeightChroma = %d\n", i, plane->ViewportHeightChroma[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportXStart = %d\n", i, plane->ViewportXStart[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportXStartC = %d\n", i, plane->ViewportXStartC[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportYStart = %d\n", i, plane->ViewportYStart[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportYStartC = %d\n", i, plane->ViewportYStartC[i]); - dml_print("DML: plane_cfg: plane=%d, ViewportStationary = %d\n", i, plane->ViewportStationary[i]); - dml_print("DML: plane_cfg: plane=%d, ScalerEnabled = %d\n", i, plane->ScalerEnabled[i]); - dml_print("DML: plane_cfg: plane=%d, HRatio = %3.2f\n", i, plane->HRatio[i]); - dml_print("DML: plane_cfg: plane=%d, VRatio = %3.2f\n", i, plane->VRatio[i]); - dml_print("DML: plane_cfg: plane=%d, HRatioChroma = %3.2f\n", i, plane->HRatioChroma[i]); - dml_print("DML: plane_cfg: plane=%d, VRatioChroma = %3.2f\n", i, plane->VRatioChroma[i]); - dml_print("DML: plane_cfg: plane=%d, HTaps = %d\n", i, plane->HTaps[i]); - dml_print("DML: plane_cfg: plane=%d, VTaps = %d\n", i, plane->VTaps[i]); - dml_print("DML: plane_cfg: plane=%d, HTapsChroma = %d\n", i, plane->HTapsChroma[i]); - dml_print("DML: plane_cfg: plane=%d, VTapsChroma = %d\n", i, plane->VTapsChroma[i]); - dml_print("DML: plane_cfg: plane=%d, LBBitPerPixel = %d\n", i, plane->LBBitPerPixel[i]); - dml_print("DML: plane_cfg: plane=%d, SourceScan = %d\n", i, plane->SourceScan[i]); - dml_print("DML: plane_cfg: plane=%d, ScalerRecoutWidth = %d\n", i, plane->ScalerRecoutWidth[i]); - dml_print("DML: plane_cfg: plane=%d, NumberOfCursors = %d\n", i, plane->NumberOfCursors[i]); - dml_print("DML: plane_cfg: plane=%d, CursorWidth = %d\n", i, plane->CursorWidth[i]); - dml_print("DML: plane_cfg: plane=%d, CursorBPP = %d\n", i, plane->CursorBPP[i]); - - dml_print("DML: plane_cfg: plane=%d, DynamicMetadataEnable = %d\n", i, plane->DynamicMetadataEnable[i]); - dml_print("DML: plane_cfg: plane=%d, DynamicMetadataLinesBeforeActiveRequired = %d\n", i, plane->DynamicMetadataLinesBeforeActiveRequired[i]); - dml_print("DML: plane_cfg: plane=%d, DynamicMetadataTransmittedBytes = %d\n", i, plane->DynamicMetadataTransmittedBytes[i]); - } -} - -void dml_print_dml_display_cfg_surface(const struct dml_surface_cfg_st *surface, dml_uint_t num_plane) -{ - for (dml_uint_t i = 0; i < num_plane; i++) { - dml_print("DML: surface_cfg: plane=%d, PitchY = %d\n", i, surface->PitchY[i]); - dml_print("DML: surface_cfg: plane=%d, SurfaceWidthY = %d\n", i, surface->SurfaceWidthY[i]); - dml_print("DML: surface_cfg: plane=%d, SurfaceHeightY = %d\n", i, surface->SurfaceHeightY[i]); - dml_print("DML: surface_cfg: plane=%d, PitchC = %d\n", i, surface->PitchC[i]); - dml_print("DML: surface_cfg: plane=%d, SurfaceWidthC = %d\n", i, surface->SurfaceWidthC[i]); - dml_print("DML: surface_cfg: plane=%d, SurfaceHeightC = %d\n", i, surface->SurfaceHeightC[i]); - dml_print("DML: surface_cfg: plane=%d, DCCEnable = %d\n", i, surface->DCCEnable[i]); - dml_print("DML: surface_cfg: plane=%d, DCCMetaPitchY = %d\n", i, surface->DCCMetaPitchY[i]); - dml_print("DML: surface_cfg: plane=%d, DCCMetaPitchC = %d\n", i, surface->DCCMetaPitchC[i]); - dml_print("DML: surface_cfg: plane=%d, DCCRateLuma = %f\n", i, surface->DCCRateLuma[i]); - dml_print("DML: surface_cfg: plane=%d, DCCRateChroma = %f\n", i, surface->DCCRateChroma[i]); - dml_print("DML: surface_cfg: plane=%d, DCCFractionOfZeroSizeRequestsLuma = %f\n", i, surface->DCCFractionOfZeroSizeRequestsLuma[i]); - dml_print("DML: surface_cfg: plane=%d, DCCFractionOfZeroSizeRequestsChroma= %f\n", i, surface->DCCFractionOfZeroSizeRequestsChroma[i]); - } -} - -void dml_print_dml_display_cfg_hw_resource(const struct dml_hw_resource_st *hw, dml_uint_t num_plane) -{ - for (dml_uint_t i = 0; i < num_plane; i++) { - dml_print("DML: hw_resource: plane=%d, ODMMode = %d\n", i, hw->ODMMode[i]); - dml_print("DML: hw_resource: plane=%d, DPPPerSurface = %d\n", i, hw->DPPPerSurface[i]); - dml_print("DML: hw_resource: plane=%d, DSCEnabled = %d\n", i, hw->DSCEnabled[i]); - dml_print("DML: hw_resource: plane=%d, NumberOfDSCSlices = %d\n", i, hw->NumberOfDSCSlices[i]); - } - dml_print("DML: hw_resource: DLGRefClkFreqMHz = %f\n", hw->DLGRefClkFreqMHz); -} - -__DML_DLL_EXPORT__ void dml_print_soc_state_bounding_box(const struct soc_state_bounding_box_st *state) -{ - dml_print("DML: state_bbox: socclk_mhz = %f\n", state->socclk_mhz); - dml_print("DML: state_bbox: dscclk_mhz = %f\n", state->dscclk_mhz); - dml_print("DML: state_bbox: phyclk_mhz = %f\n", state->phyclk_mhz); - dml_print("DML: state_bbox: phyclk_d18_mhz = %f\n", state->phyclk_d18_mhz); - dml_print("DML: state_bbox: phyclk_d32_mhz = %f\n", state->phyclk_d32_mhz); - dml_print("DML: state_bbox: dtbclk_mhz = %f\n", state->dtbclk_mhz); - dml_print("DML: state_bbox: dispclk_mhz = %f\n", state->dispclk_mhz); - dml_print("DML: state_bbox: dppclk_mhz = %f\n", state->dppclk_mhz); - dml_print("DML: state_bbox: fabricclk_mhz = %f\n", state->fabricclk_mhz); - dml_print("DML: state_bbox: dcfclk_mhz = %f\n", state->dcfclk_mhz); - dml_print("DML: state_bbox: dram_speed_mts = %f\n", state->dram_speed_mts); - dml_print("DML: state_bbox: urgent_latency_pixel_data_only_us = %f\n", state->urgent_latency_pixel_data_only_us); - dml_print("DML: state_bbox: urgent_latency_pixel_mixed_with_vm_data_us = %f\n", state->urgent_latency_pixel_mixed_with_vm_data_us); - dml_print("DML: state_bbox: urgent_latency_vm_data_only_us = %f\n", state->urgent_latency_vm_data_only_us); - dml_print("DML: state_bbox: writeback_latency_us = %f\n", state->writeback_latency_us); - dml_print("DML: state_bbox: urgent_latency_adjustment_fabric_clock_component_us = %f\n", state->urgent_latency_adjustment_fabric_clock_component_us); - dml_print("DML: state_bbox: urgent_latency_adjustment_fabric_clock_reference_mhz= %f\n", state->urgent_latency_adjustment_fabric_clock_reference_mhz); - dml_print("DML: state_bbox: sr_exit_time_us = %f\n", state->sr_exit_time_us); - dml_print("DML: state_bbox: sr_enter_plus_exit_time_us = %f\n", state->sr_enter_plus_exit_time_us); - dml_print("DML: state_bbox: sr_exit_z8_time_us = %f\n", state->sr_exit_z8_time_us); - dml_print("DML: state_bbox: sr_enter_plus_exit_z8_time_us = %f\n", state->sr_enter_plus_exit_z8_time_us); - dml_print("DML: state_bbox: dram_clock_change_latency_us = %f\n", state->dram_clock_change_latency_us); - dml_print("DML: state_bbox: fclk_change_latency_us = %f\n", state->fclk_change_latency_us); - dml_print("DML: state_bbox: usr_retraining_latency_us = %f\n", state->usr_retraining_latency_us); - dml_print("DML: state_bbox: use_ideal_dram_bw_strobe = %d\n", state->use_ideal_dram_bw_strobe); -} - -__DML_DLL_EXPORT__ void dml_print_soc_bounding_box(const struct soc_bounding_box_st *soc) -{ - dml_print("DML: soc_bbox: dprefclk_mhz = %f\n", soc->dprefclk_mhz); - dml_print("DML: soc_bbox: xtalclk_mhz = %f\n", soc->xtalclk_mhz); - dml_print("DML: soc_bbox: pcierefclk_mhz = %f\n", soc->pcierefclk_mhz); - dml_print("DML: soc_bbox: refclk_mhz = %f\n", soc->refclk_mhz); - dml_print("DML: soc_bbox: amclk_mhz = %f\n", soc->amclk_mhz); - - dml_print("DML: soc_bbox: max_outstanding_reqs = %f\n", soc->max_outstanding_reqs); - dml_print("DML: soc_bbox: pct_ideal_sdp_bw_after_urgent = %f\n", soc->pct_ideal_sdp_bw_after_urgent); - dml_print("DML: soc_bbox: pct_ideal_fabric_bw_after_urgent = %f\n", soc->pct_ideal_fabric_bw_after_urgent); - dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_pixel_only = %f\n", soc->pct_ideal_dram_bw_after_urgent_pixel_only); - dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_pixel_and_vm = %f\n", soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm); - dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_vm_only = %f\n", soc->pct_ideal_dram_bw_after_urgent_vm_only); - dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_strobe = %f\n", soc->pct_ideal_dram_bw_after_urgent_strobe); - dml_print("DML: soc_bbox: max_avg_sdp_bw_use_normal_percent = %f\n", soc->max_avg_sdp_bw_use_normal_percent); - dml_print("DML: soc_bbox: max_avg_fabric_bw_use_normal_percent = %f\n", soc->max_avg_fabric_bw_use_normal_percent); - dml_print("DML: soc_bbox: max_avg_dram_bw_use_normal_percent = %f\n", soc->max_avg_dram_bw_use_normal_percent); - dml_print("DML: soc_bbox: max_avg_dram_bw_use_normal_strobe_percent = %f\n", soc->max_avg_dram_bw_use_normal_strobe_percent); - dml_print("DML: soc_bbox: round_trip_ping_latency_dcfclk_cycles = %d\n", soc->round_trip_ping_latency_dcfclk_cycles); - dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_pixel_only_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_pixel_only_bytes); - dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes); - dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_vm_only_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_vm_only_bytes); - dml_print("DML: soc_bbox: num_chans = %d\n", soc->num_chans); - dml_print("DML: soc_bbox: return_bus_width_bytes = %d\n", soc->return_bus_width_bytes); - dml_print("DML: soc_bbox: dram_channel_width_bytes = %d\n", soc->dram_channel_width_bytes); - dml_print("DML: soc_bbox: fabric_datapath_to_dcn_data_return_bytes = %d\n", soc->fabric_datapath_to_dcn_data_return_bytes); - dml_print("DML: soc_bbox: hostvm_min_page_size_kbytes = %d\n", soc->hostvm_min_page_size_kbytes); - dml_print("DML: soc_bbox: gpuvm_min_page_size_kbytes = %d\n", soc->gpuvm_min_page_size_kbytes); - dml_print("DML: soc_bbox: phy_downspread_percent = %f\n", soc->phy_downspread_percent); - dml_print("DML: soc_bbox: dcn_downspread_percent = %f\n", soc->dcn_downspread_percent); - dml_print("DML: soc_bbox: smn_latency_us = %f\n", soc->smn_latency_us); - dml_print("DML: soc_bbox: mall_allocated_for_dcn_mbytes = %d\n", soc->mall_allocated_for_dcn_mbytes); - dml_print("DML: soc_bbox: dispclk_dppclk_vco_speed_mhz = %f\n", soc->dispclk_dppclk_vco_speed_mhz); - dml_print("DML: soc_bbox: do_urgent_latency_adjustment = %d\n", soc->do_urgent_latency_adjustment); -} - -__DML_DLL_EXPORT__ void dml_print_clk_cfg(const struct dml_clk_cfg_st *clk_cfg) -{ - dml_print("DML: clk_cfg: 0-use_required, 1-use pipe.clks_cfg, 2-use state bbox\n"); - dml_print("DML: clk_cfg: dcfclk_option = %d\n", clk_cfg->dcfclk_option); - dml_print("DML: clk_cfg: dispclk_option = %d\n", clk_cfg->dispclk_option); - - dml_print("DML: clk_cfg: dcfclk_mhz = %f\n", clk_cfg->dcfclk_mhz); - dml_print("DML: clk_cfg: dispclk_mhz = %f\n", clk_cfg->dispclk_mhz); - - for (dml_uint_t i = 0; i < DCN_DML__NUM_PLANE; i++) { - dml_print("DML: clk_cfg: i=%d, dppclk_option = %d\n", i, clk_cfg->dppclk_option[i]); - dml_print("DML: clk_cfg: i=%d, dppclk_mhz = %f\n", i, clk_cfg->dppclk_mhz[i]); - } -} - -dml_bool_t dml_is_vertical_rotation(enum dml_rotation_angle Scan) -{ - dml_bool_t is_vert = false; - if (Scan == dml_rotation_90 || Scan == dml_rotation_90m || Scan == dml_rotation_270 || Scan == dml_rotation_270m) { - is_vert = true; - } else { - is_vert = false; - } - return is_vert; -} // dml_is_vertical_rotation - -dml_uint_t dml_get_cursor_bit_per_pixel(enum dml_cursor_bpp ebpp) -{ - switch (ebpp) { - case dml_cur_2bit: - return 2; - case dml_cur_32bit: - return 32; - case dml_cur_64bit: - return 64; - default: - return 0; - } -} - -/// @brief Determine the physical pipe to logical plane mapping using the display_cfg -dml_uint_t dml_get_num_active_planes(const struct dml_display_cfg_st *display_cfg) -{ - dml_uint_t num_active_planes = 0; - - for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; k++) { - if (display_cfg->plane.ViewportWidth[k] > 0) - num_active_planes = num_active_planes + 1; - } -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: num_active_planes = %d\n", __func__, num_active_planes); -#endif - return num_active_planes; -} - -/// @brief Determine the physical pipe to logical plane mapping using the display_cfg -dml_uint_t dml_get_num_active_pipes(const struct dml_display_cfg_st *display_cfg) -{ - dml_uint_t num_active_pipes = 0; - - for (dml_uint_t j = 0; j < dml_get_num_active_planes(display_cfg); j++) { - num_active_pipes = num_active_pipes + display_cfg->hw.DPPPerSurface[j]; - } - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); -#endif - return num_active_pipes; -} - -dml_uint_t dml_get_plane_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) -{ - dml_uint_t plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; - return plane_idx; -} - -dml_uint_t dml_get_pipe_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t plane_idx) -{ - dml_uint_t pipe_idx = 0; - dml_bool_t pipe_found = 0; - - ASSERT(plane_idx < __DML_NUM_PLANES__); - - for (dml_uint_t i = 0; i < __DML_NUM_PLANES__; i++) { - if (plane_idx == mode_lib->mp.pipe_plane[i]) { - pipe_idx = i; - pipe_found = 1; - break; - } - } - ASSERT(pipe_found != 0); - - return pipe_idx; -} - -void dml_calc_pipe_plane_mapping(const struct dml_hw_resource_st *hw, dml_uint_t *pipe_plane) -{ - dml_uint_t pipe_idx = 0; - - for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; ++k) { - pipe_plane[k] = __DML_PIPE_NO_PLANE__; - } - - for (dml_uint_t plane_idx = 0; plane_idx < __DML_NUM_PLANES__; plane_idx++) { - for (dml_uint_t i = 0; i < hw->DPPPerSurface[plane_idx]; i++) { - pipe_plane[pipe_idx] = plane_idx; - pipe_idx++; - } - } -} - - diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h deleted file mode 100644 index 113b0265e1d1..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h +++ /dev/null @@ -1,76 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DISPLAY_MODE_UTIL_H__ -#define __DISPLAY_MODE_UTIL_H__ - -#include "display_mode_core_structs.h" -#include "cmntypes.h" - - -#include "dml_assert.h" -#include "dml_logging.h" - -__DML_DLL_EXPORT__ dml_bool_t dml_util_is_420(enum dml_source_format_class source_format); -__DML_DLL_EXPORT__ dml_float_t dml_ceil(dml_float_t x, dml_float_t granularity); -__DML_DLL_EXPORT__ dml_float_t dml_floor(dml_float_t x, dml_float_t granularity); -__DML_DLL_EXPORT__ dml_float_t dml_min(dml_float_t x, dml_float_t y); -__DML_DLL_EXPORT__ dml_float_t dml_min3(dml_float_t x, dml_float_t y, dml_float_t z); -__DML_DLL_EXPORT__ dml_float_t dml_min4(dml_float_t x, dml_float_t y, dml_float_t z, dml_float_t w); -__DML_DLL_EXPORT__ dml_float_t dml_max(dml_float_t x, dml_float_t y); -__DML_DLL_EXPORT__ dml_float_t dml_max3(dml_float_t x, dml_float_t y, dml_float_t z); -__DML_DLL_EXPORT__ dml_float_t dml_max4(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d); -__DML_DLL_EXPORT__ dml_float_t dml_max5(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d, dml_float_t e); -__DML_DLL_EXPORT__ dml_float_t dml_log(dml_float_t x, dml_float_t base); -__DML_DLL_EXPORT__ dml_float_t dml_log2(dml_float_t x); -__DML_DLL_EXPORT__ dml_float_t dml_round(dml_float_t val, dml_bool_t bankers_rounding); -__DML_DLL_EXPORT__ dml_float_t dml_pow(dml_float_t base, int exp); -__DML_DLL_EXPORT__ dml_uint_t dml_round_to_multiple(dml_uint_t num, dml_uint_t multiple, dml_bool_t up); -__DML_DLL_EXPORT__ dml_bool_t dml_is_vertical_rotation(enum dml_rotation_angle scan); -__DML_DLL_EXPORT__ dml_uint_t dml_get_cursor_bit_per_pixel(enum dml_cursor_bpp ebpp); -__DML_DLL_EXPORT__ void dml_print_data_rq_regs_st(const dml_display_plane_rq_regs_st *data_rq_regs); -__DML_DLL_EXPORT__ void dml_print_rq_regs_st(const dml_display_rq_regs_st *rq_regs); -__DML_DLL_EXPORT__ void dml_print_dlg_regs_st(const dml_display_dlg_regs_st *dlg_regs); -__DML_DLL_EXPORT__ void dml_print_ttu_regs_st(const dml_display_ttu_regs_st *ttu_regs); -__DML_DLL_EXPORT__ void dml_print_dml_policy(const struct dml_mode_eval_policy_st *policy); -__DML_DLL_EXPORT__ void dml_print_mode_support(struct display_mode_lib_st *mode_lib, dml_uint_t j); -__DML_DLL_EXPORT__ void dml_print_dml_mode_support_info(const struct dml_mode_support_info_st *support, dml_bool_t fail_only); -__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_timing(const struct dml_timing_cfg_st *timing, dml_uint_t num_plane); -__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_plane(const struct dml_plane_cfg_st *plane, dml_uint_t num_plane); -__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_surface(const struct dml_surface_cfg_st *surface, dml_uint_t num_plane); -__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_hw_resource(const struct dml_hw_resource_st *hw, dml_uint_t num_plane); -__DML_DLL_EXPORT__ void dml_print_soc_state_bounding_box(const struct soc_state_bounding_box_st *state); -__DML_DLL_EXPORT__ void dml_print_soc_bounding_box(const struct soc_bounding_box_st *soc); -__DML_DLL_EXPORT__ void dml_print_clk_cfg(const struct dml_clk_cfg_st *clk_cfg); - -__DML_DLL_EXPORT__ dml_uint_t dml_get_num_active_planes(const struct dml_display_cfg_st *display_cfg); -__DML_DLL_EXPORT__ dml_uint_t dml_get_num_active_pipes(const struct dml_display_cfg_st *display_cfg); -__DML_DLL_EXPORT__ dml_uint_t dml_get_plane_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx); -__DML_DLL_EXPORT__ dml_uint_t dml_get_pipe_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t plane_idx); -__DML_DLL_EXPORT__ void dml_calc_pipe_plane_mapping(const struct dml_hw_resource_st *hw, dml_uint_t *pipe_plane); - - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c deleted file mode 100644 index bf5e7f4e0416..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c +++ /dev/null @@ -1,929 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml21_wrapper.h" -#include "dml2_core_dcn4_calcs.h" -#include "dml2_internal_shared_types.h" -#include "dml2_internal_types.h" -#include "dml21_utils.h" -#include "dml21_translation_helper.h" -#include "soc_and_ip_translator.h" - -static void dml21_populate_pmo_options(struct dml2_pmo_options *pmo_options, - const struct dc *in_dc, - const struct dml2_configuration_options *config) -{ - bool disable_fams2 = !in_dc->debug.fams2_config.bits.enable; - - /* ODM options */ - pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; - pmo_options->disable_dyn_odm_for_multi_stream = true; - pmo_options->disable_dyn_odm_for_stream_with_svp = true; - - pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); - - /* NOTE: DRR and SubVP Require FAMS2 */ - pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || - in_dc->debug.force_disable_subvp || - disable_fams2; - pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || - disable_fams2; - pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || - disable_fams2; - pmo_options->disable_fams2 = disable_fams2; - - pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || - in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; - pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; -} - -static enum dml2_project_id dml21_dcn_revision_to_dml2_project_id(enum dce_version dcn_version) -{ - enum dml2_project_id project_id; - switch (dcn_version) { - case DCN_VERSION_4_01: - project_id = dml2_project_dcn4x_stage2_auto_drr_svp; - break; - default: - project_id = dml2_project_invalid; - DC_ERR("unsupported dcn version for DML21!"); - break; - } - - return project_id; -} - -void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, - const struct dml2_configuration_options *config, - const struct dc *in_dc) -{ - dml_init->options.project_id = dml21_dcn_revision_to_dml2_project_id(in_dc->ctx->dce_version); - - if (config->use_native_soc_bb_construction) { - in_dc->soc_and_ip_translator->translator_funcs->get_soc_bb(&dml_init->soc_bb, in_dc, config); - in_dc->soc_and_ip_translator->translator_funcs->get_ip_caps(&dml_init->ip_caps); - } else { - dml_init->soc_bb = config->external_socbb_ip_params->soc_bb; - dml_init->ip_caps = config->external_socbb_ip_params->ip_params; - } - - dml21_populate_pmo_options(&dml_init->options.pmo_options, in_dc, config); -} - -static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream) -{ - unsigned int max_hw_v_total = stream->ctx->dc->caps.max_v_total; - - if (stream->ctx->dc->caps.vtotal_limited_by_fp2) { - max_hw_v_total -= stream->timing.v_front_porch + 1; - } - - return max_hw_v_total; -} - -static void populate_dml21_timing_config_from_stream_state(struct dml2_timing_cfg *timing, - struct dc_stream_state *stream, - struct pipe_ctx *pipe_ctx, - struct dml2_context *dml_ctx) -{ - unsigned int hblank_start, vblank_start, min_hardware_refresh_in_uhz; - uint32_t pix_clk_100hz; - - timing->h_active = stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right + pipe_ctx->dsc_padding_params.dsc_hactive_padding; - timing->v_active = stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top; - timing->h_front_porch = stream->timing.h_front_porch; - timing->v_front_porch = stream->timing.v_front_porch; - timing->pixel_clock_khz = stream->timing.pix_clk_100hz / 10; - if (pipe_ctx->dsc_padding_params.dsc_hactive_padding != 0) - timing->pixel_clock_khz = pipe_ctx->dsc_padding_params.dsc_pix_clk_100hz / 10; - if (stream->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING) - timing->pixel_clock_khz *= 2; - timing->h_total = stream->timing.h_total + pipe_ctx->dsc_padding_params.dsc_htotal_padding; - timing->v_total = stream->timing.v_total; - timing->h_sync_width = stream->timing.h_sync_width; - timing->interlaced = stream->timing.flags.INTERLACE; - - hblank_start = stream->timing.h_total - stream->timing.h_front_porch; - - timing->h_blank_end = hblank_start - stream->timing.h_addressable - pipe_ctx->dsc_padding_params.dsc_hactive_padding - - stream->timing.h_border_left - stream->timing.h_border_right; - - if (hblank_start < stream->timing.h_addressable) - timing->h_blank_end = 0; - - vblank_start = stream->timing.v_total - stream->timing.v_front_porch; - - timing->v_blank_end = vblank_start - stream->timing.v_addressable - - stream->timing.v_border_top - stream->timing.v_border_bottom; - - timing->drr_config.enabled = stream->ignore_msa_timing_param; - timing->drr_config.drr_active_variable = stream->vrr_active_variable; - timing->drr_config.drr_active_fixed = stream->vrr_active_fixed; - timing->drr_config.disallowed = !stream->allow_freesync; - - /* limit min refresh rate to DC cap */ - min_hardware_refresh_in_uhz = stream->timing.min_refresh_in_uhz; - if (stream->ctx->dc->caps.max_v_total != 0) { - if (pipe_ctx->dsc_padding_params.dsc_hactive_padding != 0) { - pix_clk_100hz = pipe_ctx->dsc_padding_params.dsc_pix_clk_100hz; - } else { - pix_clk_100hz = stream->timing.pix_clk_100hz; - } - min_hardware_refresh_in_uhz = div64_u64((pix_clk_100hz * 100000000ULL), - (timing->h_total * (long long)calc_max_hardware_v_total(stream))); - } - - timing->drr_config.min_refresh_uhz = max(stream->timing.min_refresh_in_uhz, min_hardware_refresh_in_uhz); - - if (dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase && - stream->ctx->dc->config.enable_fpo_flicker_detection == 1) - timing->drr_config.max_instant_vtotal_delta = dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase(stream, false); - else - timing->drr_config.max_instant_vtotal_delta = 0; - - if (stream->timing.flags.DSC) { - timing->dsc.enable = dml2_dsc_enable; - timing->dsc.overrides.num_slices = stream->timing.dsc_cfg.num_slices_h; - timing->dsc.dsc_compressed_bpp_x16 = stream->timing.dsc_cfg.bits_per_pixel; - } else - timing->dsc.enable = dml2_dsc_disable; - - switch (stream->timing.display_color_depth) { - case COLOR_DEPTH_666: - timing->bpc = 6; - break; - case COLOR_DEPTH_888: - timing->bpc = 8; - break; - case COLOR_DEPTH_101010: - timing->bpc = 10; - break; - case COLOR_DEPTH_121212: - timing->bpc = 12; - break; - case COLOR_DEPTH_141414: - timing->bpc = 14; - break; - case COLOR_DEPTH_161616: - timing->bpc = 16; - break; - case COLOR_DEPTH_999: - timing->bpc = 9; - break; - case COLOR_DEPTH_111111: - timing->bpc = 11; - break; - default: - timing->bpc = 8; - break; - } - - timing->vblank_nom = timing->v_total - timing->v_active; -} - -static void populate_dml21_output_config_from_stream_state(struct dml2_link_output_cfg *output, - struct dc_stream_state *stream, const struct pipe_ctx *pipe) -{ - output->output_dp_lane_count = 4; - - switch (stream->signal) { - case SIGNAL_TYPE_DISPLAY_PORT_MST: - case SIGNAL_TYPE_DISPLAY_PORT: - output->output_encoder = dml2_dp; - if (check_dp2p0_output_encoder(pipe)) - output->output_encoder = dml2_dp2p0; - break; - case SIGNAL_TYPE_EDP: - output->output_encoder = dml2_edp; - break; - case SIGNAL_TYPE_HDMI_TYPE_A: - case SIGNAL_TYPE_DVI_SINGLE_LINK: - case SIGNAL_TYPE_DVI_DUAL_LINK: - output->output_encoder = dml2_hdmi; - break; - default: - output->output_encoder = dml2_dp; - } - - switch (stream->timing.pixel_encoding) { - case PIXEL_ENCODING_RGB: - case PIXEL_ENCODING_YCBCR444: - output->output_format = dml2_444; - break; - case PIXEL_ENCODING_YCBCR420: - output->output_format = dml2_420; - break; - case PIXEL_ENCODING_YCBCR422: - if (stream->timing.flags.DSC && !stream->timing.dsc_cfg.ycbcr422_simple) - output->output_format = dml2_n422; - else - output->output_format = dml2_s422; - break; - default: - output->output_format = dml2_444; - break; - } - - switch (stream->signal) { - case SIGNAL_TYPE_NONE: - case SIGNAL_TYPE_DVI_SINGLE_LINK: - case SIGNAL_TYPE_DVI_DUAL_LINK: - case SIGNAL_TYPE_HDMI_TYPE_A: - case SIGNAL_TYPE_LVDS: - case SIGNAL_TYPE_RGB: - case SIGNAL_TYPE_DISPLAY_PORT: - case SIGNAL_TYPE_DISPLAY_PORT_MST: - case SIGNAL_TYPE_EDP: - case SIGNAL_TYPE_VIRTUAL: - default: - output->output_dp_link_rate = dml2_dp_rate_na; - break; - } - - output->audio_sample_layout = stream->audio_info.modes->sample_size; - output->audio_sample_rate = stream->audio_info.modes->max_bit_rate; - output->output_disabled = true; - - //TODO : New to DML2.1. How do we populate this ? - // output->validate_output -} - -static void populate_dml21_stream_overrides_from_stream_state( - struct dml2_stream_parameters *stream_desc, - struct dc_stream_state *stream, - struct dc_stream_status *stream_status) -{ - switch (stream->debug.force_odm_combine_segments) { - case 0: - stream_desc->overrides.odm_mode = dml2_odm_mode_auto; - break; - case 1: - stream_desc->overrides.odm_mode = dml2_odm_mode_bypass; - break; - case 2: - stream_desc->overrides.odm_mode = dml2_odm_mode_combine_2to1; - break; - case 3: - stream_desc->overrides.odm_mode = dml2_odm_mode_combine_3to1; - break; - case 4: - stream_desc->overrides.odm_mode = dml2_odm_mode_combine_4to1; - break; - default: - stream_desc->overrides.odm_mode = dml2_odm_mode_auto; - break; - } - if (!stream->ctx->dc->debug.enable_single_display_2to1_odm_policy || - stream->debug.force_odm_combine_segments > 0) - stream_desc->overrides.disable_dynamic_odm = true; - stream_desc->overrides.disable_subvp = stream->ctx->dc->debug.force_disable_subvp || - stream->hw_cursor_req || - stream_status->mall_stream_config.cursor_size_limit_subvp; -} - -static enum dml2_swizzle_mode gfx_addr3_to_dml2_swizzle_mode(enum swizzle_mode_addr3_values addr3_mode) -{ - enum dml2_swizzle_mode dml2_mode = dml2_sw_linear; - - switch (addr3_mode) { - case DC_ADDR3_SW_LINEAR: - dml2_mode = dml2_sw_linear; - break; - case DC_ADDR3_SW_256B_2D: - dml2_mode = dml2_sw_256b_2d; - break; - case DC_ADDR3_SW_4KB_2D: - dml2_mode = dml2_sw_4kb_2d; - break; - case DC_ADDR3_SW_64KB_2D: - dml2_mode = dml2_sw_64kb_2d; - break; - case DC_ADDR3_SW_256KB_2D: - dml2_mode = dml2_sw_256kb_2d; - break; - default: - /* invalid swizzle mode for DML2.1 */ - ASSERT(false); - dml2_mode = dml2_sw_linear; - } - - return dml2_mode; -} - -static enum dml2_swizzle_mode gfx9_to_dml2_swizzle_mode(enum swizzle_mode_values gfx9_mode) -{ - enum dml2_swizzle_mode dml2_mode = dml2_sw_64kb_2d; - - switch (gfx9_mode) { - case DC_SW_LINEAR: - dml2_mode = dml2_sw_linear; - break; - case DC_SW_256_D: - case DC_SW_256_R: - dml2_mode = dml2_sw_256b_2d; - break; - case DC_SW_4KB_D: - case DC_SW_4KB_R: - case DC_SW_4KB_R_X: - dml2_mode = dml2_sw_4kb_2d; - break; - case DC_SW_64KB_D: - case DC_SW_64KB_D_X: - case DC_SW_64KB_R: - case DC_SW_64KB_R_X: - dml2_mode = dml2_sw_64kb_2d; - break; - case DC_SW_256B_S: - case DC_SW_4KB_S: - case DC_SW_64KB_S: - case DC_SW_VAR_S: - case DC_SW_VAR_D: - case DC_SW_VAR_R: - case DC_SW_64KB_S_T: - case DC_SW_64KB_D_T: - case DC_SW_4KB_S_X: - case DC_SW_4KB_D_X: - case DC_SW_64KB_S_X: - case DC_SW_VAR_S_X: - case DC_SW_VAR_D_X: - case DC_SW_VAR_R_X: - default: - /* - * invalid swizzle mode for DML2.1. This could happen because - * DML21 is not intended to be used by N-1 in production. To - * properly filter out unsupported swizzle modes, we will need - * to fix capability reporting when DML2.1 is used for N-1 in - * dc. So DML will only receive DML21 supported swizzle modes. - * This implementation is not added and has a low value because - * the supported swizzle modes should already cover most of our - * N-1 test cases. - */ - return dml2_sw_64kb_2d; - } - - return dml2_mode; -} - -static void populate_dml21_dummy_surface_cfg(struct dml2_surface_cfg *surface, const struct dc_stream_state *stream) -{ - surface->plane0.width = stream->timing.h_addressable; - surface->plane0.height = stream->timing.v_addressable; - surface->plane1.width = stream->timing.h_addressable; - surface->plane1.height = stream->timing.v_addressable; - surface->plane0.pitch = ((surface->plane0.width + 127) / 128) * 128; - surface->plane1.pitch = 0; - surface->dcc.enable = false; - surface->dcc.informative.dcc_rate_plane0 = 1.0; - surface->dcc.informative.dcc_rate_plane1 = 1.0; - surface->dcc.informative.fraction_of_zero_size_request_plane0 = 0; - surface->dcc.informative.fraction_of_zero_size_request_plane1 = 0; - surface->tiling = dml2_sw_64kb_2d; -} - -static void populate_dml21_dummy_plane_cfg(struct dml2_plane_parameters *plane, const struct dc_stream_state *stream) -{ - unsigned int width, height; - - if (stream->timing.h_addressable > 3840) - width = 3840; - else - width = stream->timing.h_addressable; // 4K max - - if (stream->timing.v_addressable > 2160) - height = 2160; - else - height = stream->timing.v_addressable; // 4K max - - plane->cursor.cursor_bpp = 32; - - plane->cursor.cursor_width = 256; - plane->cursor.num_cursors = 1; - - plane->composition.viewport.plane0.width = width; - plane->composition.viewport.plane0.height = height; - plane->composition.viewport.plane1.width = 0; - plane->composition.viewport.plane1.height = 0; - - plane->composition.viewport.stationary = false; - plane->composition.viewport.plane0.x_start = 0; - plane->composition.viewport.plane0.y_start = 0; - plane->composition.viewport.plane1.x_start = 0; - plane->composition.viewport.plane1.y_start = 0; - - plane->composition.scaler_info.enabled = false; - plane->composition.rotation_angle = dml2_rotation_0; - plane->composition.scaler_info.plane0.h_ratio = 1.0; - plane->composition.scaler_info.plane0.v_ratio = 1.0; - plane->composition.scaler_info.plane1.h_ratio = 0; - plane->composition.scaler_info.plane1.v_ratio = 0; - plane->composition.scaler_info.plane0.h_taps = 1; - plane->composition.scaler_info.plane0.v_taps = 1; - plane->composition.scaler_info.plane1.h_taps = 0; - plane->composition.scaler_info.plane1.v_taps = 0; - plane->composition.scaler_info.rect_out_width = width; - plane->pixel_format = dml2_444_32; - - plane->dynamic_meta_data.enable = false; - plane->overrides.gpuvm_min_page_size_kbytes = 256; -} - -static void populate_dml21_surface_config_from_plane_state( - const struct dc *in_dc, - struct dml2_surface_cfg *surface, - const struct dc_plane_state *plane_state) -{ - surface->plane0.pitch = plane_state->plane_size.surface_pitch; - surface->plane1.pitch = plane_state->plane_size.chroma_pitch; - surface->plane0.height = plane_state->plane_size.surface_size.height; - surface->plane0.width = plane_state->plane_size.surface_size.width; - surface->plane1.height = plane_state->plane_size.chroma_size.height; - surface->plane1.width = plane_state->plane_size.chroma_size.width; - surface->dcc.enable = plane_state->dcc.enable; - surface->dcc.informative.dcc_rate_plane0 = 1.0; - surface->dcc.informative.dcc_rate_plane1 = 1.0; - surface->dcc.informative.fraction_of_zero_size_request_plane0 = plane_state->dcc.independent_64b_blks; - surface->dcc.informative.fraction_of_zero_size_request_plane1 = plane_state->dcc.independent_64b_blks_c; - surface->dcc.plane0.pitch = plane_state->dcc.meta_pitch; - surface->dcc.plane1.pitch = plane_state->dcc.meta_pitch_c; - - // Update swizzle / array mode based on the gfx_format - switch (plane_state->tiling_info.gfxversion) { - case DcGfxVersion7: - case DcGfxVersion8: - break; - case DcGfxVersion9: - case DcGfxVersion10: - case DcGfxVersion11: - surface->tiling = gfx9_to_dml2_swizzle_mode(plane_state->tiling_info.gfx9.swizzle); - break; - case DcGfxAddr3: - surface->tiling = gfx_addr3_to_dml2_swizzle_mode(plane_state->tiling_info.gfx_addr3.swizzle); - break; - } -} - -static const struct scaler_data *get_scaler_data_for_plane( - struct dml2_context *dml_ctx, - const struct dc_plane_state *in, - const struct dc_state *context) -{ - int i; - struct pipe_ctx *temp_pipe = &dml_ctx->v21.scratch.temp_pipe; - - memset(temp_pipe, 0, sizeof(struct pipe_ctx)); - - for (i = 0; i < MAX_PIPES; i++) { - const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state == in && !pipe->prev_odm_pipe) { - temp_pipe->stream = pipe->stream; - temp_pipe->plane_state = pipe->plane_state; - temp_pipe->plane_res.scl_data.taps = pipe->plane_res.scl_data.taps; - temp_pipe->stream_res = pipe->stream_res; - temp_pipe->dsc_padding_params.dsc_hactive_padding = pipe->dsc_padding_params.dsc_hactive_padding; - temp_pipe->dsc_padding_params.dsc_htotal_padding = pipe->dsc_padding_params.dsc_htotal_padding; - temp_pipe->dsc_padding_params.dsc_pix_clk_100hz = pipe->dsc_padding_params.dsc_pix_clk_100hz; - dml_ctx->config.callbacks.build_scaling_params(temp_pipe); - break; - } - } - - ASSERT(i < MAX_PIPES); - return &temp_pipe->plane_res.scl_data; -} - -static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dml_ctx, - struct dml2_plane_parameters *plane, const struct dc_plane_state *plane_state, - const struct dc_state *context, unsigned int stream_index) -{ - const struct scaler_data *scaler_data = get_scaler_data_for_plane(dml_ctx, plane_state, context); - struct dc_stream_state *stream = context->streams[stream_index]; - - plane->cursor.cursor_bpp = 32; - plane->cursor.cursor_width = 256; - plane->cursor.num_cursors = 1; - - switch (plane_state->format) { - case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr: - case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: - plane->pixel_format = dml2_420_8; - break; - case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: - case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: - plane->pixel_format = dml2_420_10; - break; - case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: - case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: - case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: - case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: - plane->pixel_format = dml2_444_64; - break; - case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555: - case SURFACE_PIXEL_FORMAT_GRPH_RGB565: - plane->pixel_format = dml2_444_16; - break; - case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS: - plane->pixel_format = dml2_444_8; - break; - case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA: - plane->pixel_format = dml2_rgbe_alpha; - break; - default: - plane->pixel_format = dml2_444_32; - break; - } - - plane->composition.viewport.plane0.height = scaler_data->viewport.height; - plane->composition.viewport.plane0.width = scaler_data->viewport.width; - plane->composition.viewport.plane1.height = scaler_data->viewport_c.height; - plane->composition.viewport.plane1.width = scaler_data->viewport_c.width; - plane->composition.viewport.plane0.x_start = scaler_data->viewport.x; - plane->composition.viewport.plane0.y_start = scaler_data->viewport.y; - plane->composition.viewport.plane1.x_start = scaler_data->viewport_c.x; - plane->composition.viewport.plane1.y_start = scaler_data->viewport_c.y; - plane->composition.viewport.stationary = false; - plane->composition.scaler_info.enabled = scaler_data->ratios.horz.value != dc_fixpt_one.value || - scaler_data->ratios.horz_c.value != dc_fixpt_one.value || - scaler_data->ratios.vert.value != dc_fixpt_one.value || - scaler_data->ratios.vert_c.value != dc_fixpt_one.value; - - if (!scaler_data->taps.h_taps) { - /* Above logic determines scaling should be enabled even when there are no taps for - * certain cases. Hence do corrective active and disable scaling. - */ - plane->composition.scaler_info.enabled = false; - } else if ((plane_state->ctx->dc->config.use_spl == true) && - (plane->composition.scaler_info.enabled == false)) { - /* To enable sharpener for 1:1, scaler must be enabled. If use_spl is set, then - * allow case where ratio is 1 but taps > 1 - */ - if ((scaler_data->taps.h_taps > 1) || (scaler_data->taps.v_taps > 1) || - (scaler_data->taps.h_taps_c > 1) || (scaler_data->taps.v_taps_c > 1)) - plane->composition.scaler_info.enabled = true; - } - - /* always_scale is only used for debug purposes not used in production but has to be - * maintained for certain complainces. */ - if (plane_state->ctx->dc->debug.always_scale == true) { - plane->composition.scaler_info.enabled = true; - } - - if (plane->composition.scaler_info.enabled == false) { - plane->composition.scaler_info.plane0.h_ratio = 1.0; - plane->composition.scaler_info.plane0.v_ratio = 1.0; - plane->composition.scaler_info.plane1.h_ratio = 1.0; - plane->composition.scaler_info.plane1.v_ratio = 1.0; - } else { - plane->composition.scaler_info.plane0.h_ratio = (double)scaler_data->ratios.horz.value / (1ULL << 32); - plane->composition.scaler_info.plane0.v_ratio = (double)scaler_data->ratios.vert.value / (1ULL << 32); - plane->composition.scaler_info.plane1.h_ratio = (double)scaler_data->ratios.horz_c.value / (1ULL << 32); - plane->composition.scaler_info.plane1.v_ratio = (double)scaler_data->ratios.vert_c.value / (1ULL << 32); - } - - if (!scaler_data->taps.h_taps) { - plane->composition.scaler_info.plane0.h_taps = 1; - plane->composition.scaler_info.plane1.h_taps = 1; - } else { - plane->composition.scaler_info.plane0.h_taps = scaler_data->taps.h_taps; - plane->composition.scaler_info.plane1.h_taps = scaler_data->taps.h_taps_c; - } - if (!scaler_data->taps.v_taps) { - plane->composition.scaler_info.plane0.v_taps = 1; - plane->composition.scaler_info.plane1.v_taps = 1; - } else { - plane->composition.scaler_info.plane0.v_taps = scaler_data->taps.v_taps; - plane->composition.scaler_info.plane1.v_taps = scaler_data->taps.v_taps_c; - } - - plane->composition.viewport.stationary = false; - - if (plane_state->mcm_luts.lut3d_data.lut3d_src == DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { - plane->tdlut.setup_for_tdlut = true; - - switch (plane_state->mcm_luts.lut3d_data.gpu_mem_params.layout) { - case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_RGB: - case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_BGR: - plane->tdlut.tdlut_addressing_mode = dml2_tdlut_sw_linear; - break; - case DC_CM2_GPU_MEM_LAYOUT_1D_PACKED_LINEAR: - plane->tdlut.tdlut_addressing_mode = dml2_tdlut_simple_linear; - break; - } - - switch (plane_state->mcm_luts.lut3d_data.gpu_mem_params.size) { - case DC_CM2_GPU_MEM_SIZE_171717: - plane->tdlut.tdlut_width_mode = dml2_tdlut_width_17_cube; - break; - case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: - default: - //plane->tdlut.tdlut_width_mode = dml2_tdlut_width_flatten; // dml2_tdlut_width_flatten undefined - break; - } - } - plane->tdlut.setup_for_tdlut |= dml_ctx->config.force_tdlut_enable; - - plane->dynamic_meta_data.enable = false; - plane->dynamic_meta_data.lines_before_active_required = 0; - plane->dynamic_meta_data.transmitted_bytes = 0; - - plane->composition.scaler_info.rect_out_width = plane_state->dst_rect.width; - plane->composition.rotation_angle = (enum dml2_rotation_angle) plane_state->rotation; - plane->stream_index = stream_index; - - plane->overrides.gpuvm_min_page_size_kbytes = 256; - - plane->immediate_flip = plane_state->flip_immediate; - - plane->composition.rect_out_height_spans_vactive = - plane_state->dst_rect.height >= stream->src.height && - stream->dst.height >= stream->timing.v_addressable; -} - -//TODO : Could be possibly moved to a common helper layer. -static bool dml21_wrapper_get_plane_id(const struct dc_state *context, unsigned int stream_id, const struct dc_plane_state *plane, unsigned int *plane_id) -{ - int i, j; - - if (!plane_id) - return false; - - for (i = 0; i < context->stream_count; i++) { - if (context->streams[i]->stream_id == stream_id) { - for (j = 0; j < context->stream_status[i].plane_count; j++) { - if (context->stream_status[i].plane_states[j] == plane) { - *plane_id = (i << 16) | j; - return true; - } - } - } - } - - return false; -} - -static unsigned int map_stream_to_dml21_display_cfg(const struct dml2_context *dml_ctx, const struct dc_stream_state *stream) -{ - int i = 0; - int location = -1; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] && dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == stream->stream_id) { - location = i; - break; - } - } - - return location; -} - -unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, - const struct dc_plane_state *plane, const struct dc_state *context) -{ - unsigned int plane_id; - int i = 0; - int location = -1; - - if (!dml21_wrapper_get_plane_id(context, stream_id, plane, &plane_id)) { - ASSERT(false); - return -1; - } - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] && dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i] == plane_id) { - location = i; - break; - } - } - - return location; -} - -static enum dml2_uclk_pstate_change_strategy dml21_force_pstate_method_to_uclk_state_change_strategy(enum dml2_force_pstate_methods force_pstate_method) -{ - enum dml2_uclk_pstate_change_strategy val = dml2_uclk_pstate_change_strategy_auto; - - switch (force_pstate_method) { - case dml2_force_pstate_method_vactive: - val = dml2_uclk_pstate_change_strategy_force_vactive; - break; - case dml2_force_pstate_method_vblank: - val = dml2_uclk_pstate_change_strategy_force_vblank; - break; - case dml2_force_pstate_method_drr: - val = dml2_uclk_pstate_change_strategy_force_drr; - break; - case dml2_force_pstate_method_subvp: - val = dml2_uclk_pstate_change_strategy_force_mall_svp; - break; - case dml2_force_pstate_method_auto: - default: - val = dml2_uclk_pstate_change_strategy_auto; - } - - return val; -} - -bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) -{ - int stream_index, plane_index; - int disp_cfg_stream_location, disp_cfg_plane_location; - struct dml2_display_cfg *dml_dispcfg = &dml_ctx->v21.display_config; - unsigned int plane_count = 0; - - memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); - - dml_dispcfg->gpuvm_enable = dml_ctx->config.gpuvm_enable; - dml_dispcfg->gpuvm_max_page_table_levels = 4; - dml_dispcfg->hostvm_enable = false; - dml_dispcfg->minimize_det_reallocation = true; - dml_dispcfg->overrides.enable_subvp_implicit_pmo = true; - - if (in_dc->debug.disable_unbounded_requesting) { - dml_dispcfg->overrides.hw.force_unbounded_requesting.enable = true; - dml_dispcfg->overrides.hw.force_unbounded_requesting.value = false; - } - - for (stream_index = 0; stream_index < context->stream_count; stream_index++) { - disp_cfg_stream_location = map_stream_to_dml21_display_cfg(dml_ctx, context->streams[stream_index]); - - if (disp_cfg_stream_location < 0) - disp_cfg_stream_location = dml_dispcfg->num_streams++; - - ASSERT(disp_cfg_stream_location >= 0 && disp_cfg_stream_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); - populate_dml21_timing_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index], dml_ctx); - populate_dml21_output_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].output, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index]); - populate_dml21_stream_overrides_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location], context->streams[stream_index], &context->stream_status[stream_index]); - - dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.fclk_pstate = dml2_twait_budgeting_setting_if_needed; - dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.uclk_pstate = dml2_twait_budgeting_setting_if_needed; - dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.stutter_enter_exit = dml2_twait_budgeting_setting_if_needed; - - dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_stream_location] = context->streams[stream_index]->stream_id; - dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_stream_location] = true; - - if (context->stream_status[stream_index].plane_count == 0) { - disp_cfg_plane_location = dml_dispcfg->num_planes++; - populate_dml21_dummy_surface_cfg(&dml_dispcfg->plane_descriptors[disp_cfg_plane_location].surface, context->streams[stream_index]); - populate_dml21_dummy_plane_cfg(&dml_dispcfg->plane_descriptors[disp_cfg_plane_location], context->streams[stream_index]); - dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; - } else { - for (plane_index = 0; plane_index < context->stream_status[stream_index].plane_count; plane_index++) { - disp_cfg_plane_location = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], context); - - if (disp_cfg_plane_location < 0) - disp_cfg_plane_location = dml_dispcfg->num_planes++; - - ASSERT(disp_cfg_plane_location >= 0 && disp_cfg_plane_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); - - populate_dml21_surface_config_from_plane_state(in_dc, &dml_dispcfg->plane_descriptors[disp_cfg_plane_location].surface, context->stream_status[stream_index].plane_states[plane_index]); - populate_dml21_plane_config_from_plane_state(dml_ctx, &dml_dispcfg->plane_descriptors[disp_cfg_plane_location], context->stream_status[stream_index].plane_states[plane_index], context, stream_index); - dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; - - if (dml21_wrapper_get_plane_id(context, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], &dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) - dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; - - /* apply forced pstate policy */ - if (dml_ctx->config.pmo.force_pstate_method_enable) { - dml_dispcfg->plane_descriptors[disp_cfg_plane_location].overrides.uclk_pstate_change_strategy = - dml21_force_pstate_method_to_uclk_state_change_strategy(dml_ctx->config.pmo.force_pstate_method_values[stream_index]); - } - - plane_count++; - } - } - } - - if (plane_count == 0) { - dml_dispcfg->overrides.all_streams_blanked = true; - } - - return true; -} - -void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context) -{ - /* TODO these should be the max of active, svp prefetch and idle should be tracked seperately */ - context->bw_ctx.bw.dcn.clk.dispclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dispclk_khz; - context->bw_ctx.bw.dcn.clk.dcfclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.dcfclk_khz; - context->bw_ctx.bw.dcn.clk.dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.uclk_khz; - context->bw_ctx.bw.dcn.clk.fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.fclk_khz; - context->bw_ctx.bw.dcn.clk.idle_dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.idle.uclk_khz; - context->bw_ctx.bw.dcn.clk.idle_fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.idle.fclk_khz; - context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.deepsleep_dcfclk_khz; - context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = in_ctx->v21.mode_programming.programming->fclk_pstate_supported; - context->bw_ctx.bw.dcn.clk.p_state_change_support = in_ctx->v21.mode_programming.programming->uclk_pstate_supported; - context->bw_ctx.bw.dcn.clk.dtbclk_en = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dtbrefclk_khz > 0; - context->bw_ctx.bw.dcn.clk.ref_dtbclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dtbrefclk_khz; - context->bw_ctx.bw.dcn.clk.socclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.socclk_khz; - context->bw_ctx.bw.dcn.clk.subvp_prefetch_dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz; - context->bw_ctx.bw.dcn.clk.subvp_prefetch_fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz; - context->bw_ctx.bw.dcn.clk.stutter_efficiency.base_efficiency = in_ctx->v21.mode_programming.programming->stutter.base_percent_efficiency; - context->bw_ctx.bw.dcn.clk.stutter_efficiency.low_power_efficiency = in_ctx->v21.mode_programming.programming->stutter.low_power_percent_efficiency; -} - -static struct dml2_dchub_watermark_regs *wm_set_index_to_dc_wm_set(union dcn_watermark_set *watermarks, const enum dml2_dchub_watermark_reg_set_index wm_index) -{ - struct dml2_dchub_watermark_regs *wm_regs = NULL; - - switch (wm_index) { - case DML2_DCHUB_WATERMARK_SET_A: - wm_regs = &watermarks->dcn4x.a; - break; - case DML2_DCHUB_WATERMARK_SET_B: - wm_regs = &watermarks->dcn4x.b; - break; - case DML2_DCHUB_WATERMARK_SET_C: - wm_regs = &watermarks->dcn4x.c; - break; - case DML2_DCHUB_WATERMARK_SET_D: - wm_regs = &watermarks->dcn4x.d; - break; - case DML2_DCHUB_WATERMARK_SET_NUM: - default: - /* invalid wm set index */ - wm_regs = NULL; - } - - return wm_regs; -} - -void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx) -{ - const struct dml2_display_cfg_programming *programming = in_ctx->v21.mode_programming.programming; - - unsigned int wm_index; - - /* copy watermark sets from DML */ - for (wm_index = 0; wm_index < programming->global_regs.num_watermark_sets; wm_index++) { - struct dml2_dchub_watermark_regs *wm_regs = wm_set_index_to_dc_wm_set(watermarks, wm_index); - - if (wm_regs) - memcpy(wm_regs, - &programming->global_regs.wm_regs[wm_index], - sizeof(struct dml2_dchub_watermark_regs)); - } -} - -void dml21_map_hw_resources(struct dml2_context *dml_ctx) -{ - unsigned int i = 0; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] = dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i]; - dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] = true; - dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] = dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i]; - dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] = true; - } - -} - -void dml21_get_pipe_mcache_config( - struct dc_state *context, - struct pipe_ctx *pipe_ctx, - struct dml2_per_plane_programming *pln_prog, - struct dml2_pipe_configuration_descriptor *mcache_pipe_config) -{ - mcache_pipe_config->plane0.viewport_x_start = pipe_ctx->plane_res.scl_data.viewport.x; - mcache_pipe_config->plane0.viewport_width = pipe_ctx->plane_res.scl_data.viewport.width; - - mcache_pipe_config->plane1.viewport_x_start = pipe_ctx->plane_res.scl_data.viewport_c.x; - mcache_pipe_config->plane1.viewport_width = pipe_ctx->plane_res.scl_data.viewport_c.width; - - mcache_pipe_config->plane1_enabled = - dml21_is_plane1_enabled(pln_prog->plane_descriptor->pixel_format); -} - -void dml21_set_dc_p_state_type( - struct pipe_ctx *pipe_ctx, - struct dml2_per_stream_programming *stream_programming, - bool sub_vp_enabled) -{ - switch (stream_programming->uclk_pstate_method) { - case dml2_pstate_method_vactive: - case dml2_pstate_method_fw_vactive_drr: - pipe_ctx->p_state_type = P_STATE_V_ACTIVE; - break; - case dml2_pstate_method_vblank: - case dml2_pstate_method_fw_vblank_drr: - if (sub_vp_enabled) - pipe_ctx->p_state_type = P_STATE_V_BLANK_SUB_VP; - else - pipe_ctx->p_state_type = P_STATE_V_BLANK; - break; - case dml2_pstate_method_fw_svp: - case dml2_pstate_method_fw_svp_drr: - pipe_ctx->p_state_type = P_STATE_SUB_VP; - break; - case dml2_pstate_method_fw_drr: - if (sub_vp_enabled) - pipe_ctx->p_state_type = P_STATE_DRR_SUB_VP; - else - pipe_ctx->p_state_type = P_STATE_FPO; - break; - default: - pipe_ctx->p_state_type = P_STATE_UNKNOWN; - break; - } -} - diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h deleted file mode 100644 index 9880d3e0398e..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h +++ /dev/null @@ -1,28 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#ifndef _DML21_TRANSLATION_HELPER_H_ -#define _DML21_TRANSLATION_HELPER_H_ - -struct dc; -struct dc_state; -struct dcn_watermarks; -union dcn_watermark_set; -struct pipe_ctx; -struct dc_plane_state; - -struct dml2_context; -struct dml2_configuration_options; -struct dml2_initialize_instance_in_out; - -void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); -bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); -void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context); -void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx); -void dml21_map_hw_resources(struct dml2_context *dml_ctx); -void dml21_get_pipe_mcache_config(struct dc_state *context, struct pipe_ctx *pipe_ctx, struct dml2_per_plane_programming *pln_prog, struct dml2_pipe_configuration_descriptor *mcache_pipe_config); -void dml21_set_dc_p_state_type(struct pipe_ctx *pipe_ctx, struct dml2_per_stream_programming *stream_programming, bool sub_vp_enabled); -unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, const struct dc_plane_state *plane, const struct dc_state *context); -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c deleted file mode 100644 index ee721606b883..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c +++ /dev/null @@ -1,516 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#include "dml2_internal_shared_types.h" -#include "dml21_translation_helper.h" -#include "dml2_internal_types.h" -#include "dml21_utils.h" -#include "dml2_dc_resource_mgmt.h" - -#include "dml2_core_dcn4_calcs.h" - -int dml21_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id) -{ - int i; - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] && ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] == stream_id) - return i; - } - - return -1; -} - -int dml21_find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id) -{ - int i; - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] && ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] == plane_id) - return i; - } - - return -1; -} - -bool dml21_get_plane_id(const struct dc_state *state, const struct dc_plane_state *plane, unsigned int *plane_id) -{ - int i, j; - - if (!plane_id) - return false; - - for (i = 0; i < state->stream_count; i++) { - for (j = 0; j < state->stream_status[i].plane_count; j++) { - if (state->stream_status[i].plane_states[j] == plane) { - *plane_id = (i << 16) | j; - return true; - } - } - } - - return false; -} - -unsigned int dml21_get_dc_plane_idx_from_plane_id(unsigned int plane_id) -{ - return 0xffff & plane_id; -} - -void find_valid_pipe_idx_for_stream_index(const struct dml2_context *dml_ctx, unsigned int *dml_pipe_idx, unsigned int stream_index) -{ - unsigned int i = 0; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (dml_ctx->v21.mode_programming.programming->plane_programming[i].plane_descriptor->stream_index == stream_index) { - *dml_pipe_idx = i; - return; - } - } -} - -void find_pipe_regs_idx(const struct dml2_context *dml_ctx, - struct pipe_ctx *pipe, unsigned int *pipe_regs_idx) -{ - struct pipe_ctx *opp_head = dml_ctx->config.callbacks.get_opp_head(pipe); - - *pipe_regs_idx = dml_ctx->config.callbacks.get_odm_slice_index(opp_head); - - if (pipe->plane_state) - *pipe_regs_idx += dml_ctx->config.callbacks.get_mpc_slice_index(pipe); -} - -/* places pipe references into pipes arrays and returns number of pipes */ -int dml21_find_dc_pipes_for_plane(const struct dc *in_dc, - struct dc_state *context, - struct dml2_context *dml_ctx, - struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], - struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], - int dml_plane_idx) -{ - unsigned int dml_stream_index; - unsigned int main_stream_id; - unsigned int dc_plane_index; - struct dc_stream_state *dc_main_stream; - struct dc_stream_status *dc_main_stream_status; - struct dc_plane_state *dc_main_plane; - struct dc_stream_state *dc_phantom_stream; - struct dc_stream_status *dc_phantom_stream_status; - struct dc_plane_state *dc_phantom_plane; - int num_pipes = 0; - - memset(dc_main_pipes, 0, sizeof(struct pipe_ctx *) * __DML2_WRAPPER_MAX_STREAMS_PLANES__); - memset(dc_phantom_pipes, 0, sizeof(struct pipe_ctx *) * __DML2_WRAPPER_MAX_STREAMS_PLANES__); - - dml_stream_index = dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_idx].plane_descriptor->stream_index; - main_stream_id = dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_stream_index]; - - dc_main_stream = dml_ctx->config.callbacks.get_stream_from_id(context, main_stream_id); - dc_main_stream_status = dml_ctx->config.callbacks.get_stream_status(context, dc_main_stream); - if (!dc_main_stream_status) - return num_pipes; - - /* find main plane based on id */ - dc_plane_index = dml21_get_dc_plane_idx_from_plane_id(dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[dml_plane_idx]); - dc_main_plane = dc_main_stream_status->plane_states[dc_plane_index]; - - if (dc_main_plane) { - num_pipes = dml_ctx->config.callbacks.get_dpp_pipes_for_plane(dc_main_plane, &context->res_ctx, dc_main_pipes); - } else { - /* stream was configured with dummy plane, so get pipes from opp head */ - struct pipe_ctx *otg_master_pipe = dml_ctx->config.callbacks.get_otg_master_for_stream(&context->res_ctx, dc_main_stream); - if (otg_master_pipe != NULL) - num_pipes = dml_ctx->config.callbacks.get_opp_heads_for_otg_master(otg_master_pipe, &context->res_ctx, dc_main_pipes); - } - - /* if phantom exists, find associated pipes */ - dc_phantom_stream = dml_ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, dc_main_stream); - if (dc_phantom_stream && num_pipes > 0) { - dc_phantom_stream_status = dml_ctx->config.callbacks.get_stream_status(context, dc_phantom_stream); - - if (dc_phantom_stream_status) { - /* phantom plane will have same index as main */ - dc_phantom_plane = dc_phantom_stream_status->plane_states[dc_plane_index]; - - if (dc_phantom_plane) { - /* only care about phantom pipes if they contain the phantom plane */ - dml_ctx->config.callbacks.get_dpp_pipes_for_plane(dc_phantom_plane, &context->res_ctx, dc_phantom_pipes); - } - } - } - - return num_pipes; -} - -void dml21_pipe_populate_global_sync(struct dml2_context *dml_ctx, - struct dc_state *context, - struct pipe_ctx *pipe_ctx, - struct dml2_per_stream_programming *stream_programming) -{ - union dml2_global_sync_programming *global_sync = &stream_programming->global_sync; - - if (dml_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe_ctx) == SUBVP_PHANTOM) { - /* phantom has its own global sync */ - global_sync = &stream_programming->phantom_stream.global_sync; - } - - memcpy(&pipe_ctx->global_sync, - global_sync, - sizeof(union dml2_global_sync_programming)); -} - -void dml21_populate_mall_allocation_size(struct dc_state *context, - struct dml2_context *in_ctx, - struct dml2_per_plane_programming *pln_prog, - struct pipe_ctx *dc_pipe) -{ - - /* Reuse MALL Allocation Sizes logic from dcn32_fpu.c */ - /* Count from active, top pipes per plane only. Only add mall_ss_size_bytes for each unique plane. */ - if (dc_pipe->stream && dc_pipe->plane_state && - (dc_pipe->top_pipe == NULL || - dc_pipe->plane_state != dc_pipe->top_pipe->plane_state) && - dc_pipe->prev_odm_pipe == NULL) { - /* SS: all active surfaces stored in MALL */ - if (in_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, dc_pipe) != SUBVP_PHANTOM) { - dc_pipe->surface_size_in_mall_bytes = pln_prog->surface_size_mall_bytes; - context->bw_ctx.bw.dcn.mall_ss_size_bytes += dc_pipe->surface_size_in_mall_bytes; - } else { - /* SUBVP: phantom surfaces only stored in MALL */ - dc_pipe->surface_size_in_mall_bytes = pln_prog->svp_size_mall_bytes; - context->bw_ctx.bw.dcn.mall_subvp_size_bytes += dc_pipe->surface_size_in_mall_bytes; - } - } -} - -bool check_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx) -{ - /* If this assert is hit then we have a link encoder dynamic management issue */ - ASSERT(pipe_ctx->stream_res.hpo_dp_stream_enc ? pipe_ctx->link_res.hpo_dp_link_enc != NULL : true); - return (pipe_ctx->stream_res.hpo_dp_stream_enc && - pipe_ctx->link_res.hpo_dp_link_enc && - dc_is_dp_signal(pipe_ctx->stream->signal)); -} - - -static bool is_sub_vp_enabled(struct dc *dc, struct dc_state *context) -{ - int i; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->stream && dc_state_get_paired_subvp_stream(context, pipe_ctx->stream) && - dc_state_get_pipe_subvp_type(context, pipe_ctx) == SUBVP_MAIN) { - return true; - } - } - return false; -} - - -void dml21_program_dc_pipe(struct dml2_context *dml_ctx, struct dc_state *context, struct pipe_ctx *pipe_ctx, struct dml2_per_plane_programming *pln_prog, - struct dml2_per_stream_programming *stream_prog) -{ - unsigned int pipe_reg_index = 0; - - dml21_pipe_populate_global_sync(dml_ctx, context, pipe_ctx, stream_prog); - find_pipe_regs_idx(dml_ctx, pipe_ctx, &pipe_reg_index); - - if (dml_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe_ctx) == SUBVP_PHANTOM) { - memcpy(&pipe_ctx->hubp_regs, pln_prog->phantom_plane.pipe_regs[pipe_reg_index], sizeof(struct dml2_dchub_per_pipe_register_set)); - pipe_ctx->unbounded_req = false; - pipe_ctx->det_buffer_size_kb = 0; - } else { - memcpy(&pipe_ctx->hubp_regs, pln_prog->pipe_regs[pipe_reg_index], sizeof(struct dml2_dchub_per_pipe_register_set)); - pipe_ctx->unbounded_req = pln_prog->pipe_regs[pipe_reg_index]->rq_regs.unbounded_request_enabled; - pipe_ctx->det_buffer_size_kb = pln_prog->pipe_regs[pipe_reg_index]->det_size * 64; - } - - pipe_ctx->plane_res.bw.dppclk_khz = pln_prog->min_clocks.dcn4x.dppclk_khz; - if (context->bw_ctx.bw.dcn.clk.dppclk_khz < pipe_ctx->plane_res.bw.dppclk_khz) - context->bw_ctx.bw.dcn.clk.dppclk_khz = pipe_ctx->plane_res.bw.dppclk_khz; - - dml21_populate_mall_allocation_size(context, dml_ctx, pln_prog, pipe_ctx); - - bool sub_vp_enabled = is_sub_vp_enabled(pipe_ctx->stream->ctx->dc, context); - - dml21_set_dc_p_state_type(pipe_ctx, stream_prog, sub_vp_enabled); -} - -static struct dc_stream_state *dml21_add_phantom_stream(struct dml2_context *dml_ctx, - const struct dc *dc, - struct dc_state *context, - struct dc_stream_state *main_stream, - struct dml2_per_stream_programming *stream_programming) -{ - struct dc_stream_state *phantom_stream; - struct dml2_stream_parameters *phantom_stream_descriptor = &stream_programming->phantom_stream.descriptor; - - phantom_stream = dml_ctx->config.svp_pstate.callbacks.create_phantom_stream(dc, context, main_stream); - if (!phantom_stream) - return NULL; - - /* copy details of phantom stream from main */ - memcpy(&phantom_stream->timing, &main_stream->timing, sizeof(phantom_stream->timing)); - memcpy(&phantom_stream->src, &main_stream->src, sizeof(phantom_stream->src)); - memcpy(&phantom_stream->dst, &main_stream->dst, sizeof(phantom_stream->dst)); - - /* modify timing for phantom */ - phantom_stream->timing.v_front_porch = phantom_stream_descriptor->timing.v_front_porch; - phantom_stream->timing.v_addressable = phantom_stream_descriptor->timing.v_active; - phantom_stream->timing.v_total = phantom_stream_descriptor->timing.v_total; - phantom_stream->timing.flags.DSC = 0; // phantom always has DSC disabled - - phantom_stream->dst.y = 0; - phantom_stream->dst.height = stream_programming->phantom_stream.descriptor.timing.v_active; - - phantom_stream->src.y = 0; - phantom_stream->src.height = (double)phantom_stream_descriptor->timing.v_active * (double)main_stream->src.height / (double)main_stream->dst.height; - - phantom_stream->use_dynamic_meta = false; - - dml_ctx->config.svp_pstate.callbacks.add_phantom_stream(dc, context, phantom_stream, main_stream); - - return phantom_stream; -} - -static struct dc_plane_state *dml21_add_phantom_plane(struct dml2_context *dml_ctx, - const struct dc *dc, - struct dc_state *context, - struct dc_stream_state *phantom_stream, - struct dc_plane_state *main_plane, - struct dml2_per_plane_programming *plane_programming) -{ - struct dc_plane_state *phantom_plane; - - phantom_plane = dml_ctx->config.svp_pstate.callbacks.create_phantom_plane(dc, context, main_plane); - if (!phantom_plane) - return NULL; - - phantom_plane->format = main_plane->format; - phantom_plane->rotation = main_plane->rotation; - phantom_plane->visible = main_plane->visible; - - memcpy(&phantom_plane->address, &main_plane->address, sizeof(phantom_plane->address)); - memcpy(&phantom_plane->scaling_quality, &main_plane->scaling_quality, - sizeof(phantom_plane->scaling_quality)); - memcpy(&phantom_plane->src_rect, &main_plane->src_rect, sizeof(phantom_plane->src_rect)); - memcpy(&phantom_plane->dst_rect, &main_plane->dst_rect, sizeof(phantom_plane->dst_rect)); - memcpy(&phantom_plane->clip_rect, &main_plane->clip_rect, sizeof(phantom_plane->clip_rect)); - memcpy(&phantom_plane->plane_size, &main_plane->plane_size, - sizeof(phantom_plane->plane_size)); - memcpy(&phantom_plane->tiling_info, &main_plane->tiling_info, - sizeof(phantom_plane->tiling_info)); - memcpy(&phantom_plane->dcc, &main_plane->dcc, sizeof(phantom_plane->dcc)); - - phantom_plane->format = main_plane->format; - phantom_plane->rotation = main_plane->rotation; - phantom_plane->visible = main_plane->visible; - - /* Shadow pipe has small viewport. */ - phantom_plane->clip_rect.y = 0; - phantom_plane->clip_rect.height = phantom_stream->src.height; - - dml_ctx->config.svp_pstate.callbacks.add_phantom_plane(dc, phantom_stream, phantom_plane, context); - - return phantom_plane; -} - -void dml21_handle_phantom_streams_planes(const struct dc *dc, struct dc_state *context, struct dml2_context *dml_ctx) -{ - unsigned int dml_stream_index, dml_plane_index, dc_plane_index; - struct dc_stream_state *main_stream; - struct dc_stream_status *main_stream_status; - struct dc_stream_state *phantom_stream; - struct dc_plane_state *main_plane; - bool phantoms_added = false; - - /* create phantom streams and planes and add to context */ - for (dml_stream_index = 0; dml_stream_index < dml_ctx->v21.mode_programming.programming->display_config.num_streams; dml_stream_index++) { - /* iterate through DML streams looking for phantoms */ - if (dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_index].phantom_stream.enabled) { - /* find associated dc stream */ - main_stream = dml_ctx->config.callbacks.get_stream_from_id(context, - dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_stream_index]); - - main_stream_status = dml_ctx->config.callbacks.get_stream_status(context, main_stream); - - if (!main_stream_status || main_stream_status->plane_count == 0) - continue; - - /* create phantom stream for subvp enabled stream */ - phantom_stream = dml21_add_phantom_stream(dml_ctx, - dc, - context, - main_stream, - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_index]); - - if (!phantom_stream) - continue; - - /* iterate through DML planes associated with this stream */ - for (dml_plane_index = 0; dml_plane_index < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_plane_index++) { - if (dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_index].plane_descriptor->stream_index == dml_stream_index) { - /* find associated dc plane */ - dc_plane_index = dml21_get_dc_plane_idx_from_plane_id(dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[dml_plane_index]); - main_plane = main_stream_status->plane_states[dc_plane_index]; - - /* create phantom planes for subvp enabled plane */ - dml21_add_phantom_plane(dml_ctx, - dc, - context, - phantom_stream, - main_plane, - &dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_index]); - - phantoms_added = true; - } - } - } - } - - if (phantoms_added) - dml2_map_dc_pipes(dml_ctx, context, NULL, &dml_ctx->v21.dml_to_dc_pipe_mapping, dc->current_state); -} - -void dml21_build_fams2_programming(const struct dc *dc, - struct dc_state *context, - struct dml2_context *dml_ctx) -{ - int i, j, k; - unsigned int num_fams2_streams = 0; - - /* reset fams2 data */ - memset(&context->bw_ctx.bw.dcn.fams2_stream_base_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); - memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); - memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2, 0, sizeof(union dmub_fams2_stream_static_sub_state_v2) * DML2_MAX_PLANES); - memset(&context->bw_ctx.bw.dcn.fams2_global_config, 0, sizeof(struct dmub_cmd_fams2_global_config)); - - if (dml_ctx->v21.mode_programming.programming->fams2_required) { - for (i = 0; i < context->stream_count; i++) { - int dml_stream_idx; - struct dc_stream_state *phantom_stream; - struct dc_stream_status *phantom_status; - enum fams2_stream_type type = 0; - - union dmub_cmd_fams2_config *static_base_state = &context->bw_ctx.bw.dcn.fams2_stream_base_params[num_fams2_streams]; - union dmub_cmd_fams2_config *static_sub_state = &context->bw_ctx.bw.dcn.fams2_stream_sub_params[num_fams2_streams]; - - struct dc_stream_state *stream = context->streams[i]; - - if (context->stream_status[i].plane_count == 0 || - dml_ctx->config.svp_pstate.callbacks.get_stream_subvp_type(context, stream) == SUBVP_PHANTOM) { - /* can ignore blanked or phantom streams */ - continue; - } - - dml_stream_idx = dml21_helper_find_dml_pipe_idx_by_stream_id(dml_ctx, stream->stream_id); - if (dml_stream_idx < 0) { - ASSERT(dml_stream_idx >= 0); - continue; - } - - /* copy static state from PMO */ - memcpy(static_base_state, - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_base_params, - sizeof(union dmub_cmd_fams2_config)); - - if (dc->debug.fams_version.major == 3) { - memcpy(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[num_fams2_streams], - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params_v2, - sizeof(union dmub_fams2_stream_static_sub_state_v2)); - } else { - memcpy(static_sub_state, - &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, - sizeof(union dmub_cmd_fams2_config)); - } - - switch (dc->debug.fams_version.minor) { - case 1: - default: - type = static_base_state->stream_v1.base.type; - - /* get information from context */ - static_base_state->stream_v1.base.num_planes = context->stream_status[i].plane_count; - static_base_state->stream_v1.base.otg_inst = context->stream_status[i].primary_otg_inst; - - /* populate pipe masks for planes */ - for (j = 0; j < context->stream_status[i].plane_count; j++) { - for (k = 0; k < dc->res_pool->pipe_count; k++) { - if (context->res_ctx.pipe_ctx[k].stream && - context->res_ctx.pipe_ctx[k].stream->stream_id == stream->stream_id && - context->res_ctx.pipe_ctx[k].plane_state == context->stream_status[i].plane_states[j]) { - static_base_state->stream_v1.base.pipe_mask |= (1 << k); - static_base_state->stream_v1.base.plane_pipe_masks[j] |= (1 << k); - } - } - } - } - - - /* get per method programming */ - switch (type) { - case FAMS2_STREAM_TYPE_VBLANK: - case FAMS2_STREAM_TYPE_VACTIVE: - case FAMS2_STREAM_TYPE_DRR: - break; - case FAMS2_STREAM_TYPE_SUBVP: - phantom_stream = dml_ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, stream); - if (!phantom_stream) - break; - - phantom_status = dml_ctx->config.callbacks.get_stream_status(context, phantom_stream); - - /* phantom status should always be present */ - ASSERT(phantom_status); - if (!phantom_status) - break; - - switch (dc->debug.fams_version.minor) { - case 1: - default: - static_sub_state->stream_v1.sub_state.subvp.phantom_otg_inst = phantom_status->primary_otg_inst; - - /* populate pipe masks for phantom planes */ - for (j = 0; j < phantom_status->plane_count; j++) { - for (k = 0; k < dc->res_pool->pipe_count; k++) { - if (context->res_ctx.pipe_ctx[k].stream && - context->res_ctx.pipe_ctx[k].stream->stream_id == phantom_stream->stream_id && - context->res_ctx.pipe_ctx[k].plane_state == phantom_status->plane_states[j]) { - switch (dc->debug.fams_version.minor) { - case 1: - default: - static_sub_state->stream_v1.sub_state.subvp.phantom_pipe_mask |= (1 << k); - static_sub_state->stream_v1.sub_state.subvp.phantom_plane_pipe_masks[j] |= (1 << k); - } - } - } - } - } - break; - default: - ASSERT(false); - break; - } - - num_fams2_streams++; - } - } - - if (num_fams2_streams > 0) { - /* copy FAMS2 configuration */ - memcpy(&context->bw_ctx.bw.dcn.fams2_global_config, - &dml_ctx->v21.mode_programming.programming->fams2_global_config, - sizeof(struct dmub_cmd_fams2_global_config)); - - context->bw_ctx.bw.dcn.fams2_global_config.num_streams = num_fams2_streams; - } - - context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = context->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable; -} - -bool dml21_is_plane1_enabled(enum dml2_source_format_class source_format) -{ - return source_format >= dml2_420_8 && source_format <= dml2_rgbe_alpha; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h deleted file mode 100644 index 4bff52eaaef8..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h +++ /dev/null @@ -1,50 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#ifndef _DML21_UTILS_H_ -#define _DML21_UTILS_H_ - -struct dc_state; -struct dc_plane_state; -struct pipe_ctx; - -struct dml2_context; -struct dml2_display_rq_regs; -struct dml2_display_dlg_regs; -struct dml2_display_ttu_regs; - -int dml21_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id); -int dml21_find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id); -bool dml21_get_plane_id(const struct dc_state *state, const struct dc_plane_state *plane, unsigned int *plane_id); -void dml21_pipe_populate_global_sync(struct dml2_context *dml_ctx, - struct dc_state *context, - struct pipe_ctx *pipe_ctx, - struct dml2_per_stream_programming *stream_programming); -void dml21_populate_mall_allocation_size(struct dc_state *context, - struct dml2_context *in_ctx, - struct dml2_per_plane_programming *pln_prog, - struct pipe_ctx *dc_pipe); -bool check_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx); -void find_valid_pipe_idx_for_stream_index(const struct dml2_context *dml_ctx, unsigned int *dml_pipe_idx, unsigned int stream_index); -void find_pipe_regs_idx(const struct dml2_context *dml_ctx, - struct pipe_ctx *pipe, unsigned int *pipe_regs_idx); -int dml21_find_dc_pipes_for_plane(const struct dc *in_dc, - struct dc_state *context, - struct dml2_context *dml_ctx, - struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], - struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], - int dml_plane_idx); -void dml21_program_dc_pipe(struct dml2_context *dml_ctx, - struct dc_state *context, - struct pipe_ctx *pipe_ctx, - struct dml2_per_plane_programming *pln_prog, - struct dml2_per_stream_programming *stream_prog); -void dml21_handle_phantom_streams_planes(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); -unsigned int dml21_get_dc_plane_idx_from_plane_id(unsigned int plane_id); -void dml21_build_fams2_programming(const struct dc *dc, - struct dc_state *context, - struct dml2_context *dml_ctx); -bool dml21_is_plane1_enabled(enum dml2_source_format_class source_format); -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c deleted file mode 100644 index 43d333d6608e..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c +++ /dev/null @@ -1,468 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_internal_types.h" -#include "dml_top.h" -#include "dml2_core_dcn4_calcs.h" -#include "dml2_internal_shared_types.h" -#include "dml21_utils.h" -#include "dml21_translation_helper.h" -#include "dml2_dc_resource_mgmt.h" - -#define INVALID -1 - -static bool dml21_allocate_memory(struct dml2_context **dml_ctx) -{ - *dml_ctx = vzalloc(sizeof(struct dml2_context)); - if (!(*dml_ctx)) - return false; - - (*dml_ctx)->v21.dml_init.dml2_instance = vzalloc(sizeof(struct dml2_instance)); - if (!((*dml_ctx)->v21.dml_init.dml2_instance)) - return false; - - (*dml_ctx)->v21.mode_support.dml2_instance = (*dml_ctx)->v21.dml_init.dml2_instance; - (*dml_ctx)->v21.mode_programming.dml2_instance = (*dml_ctx)->v21.dml_init.dml2_instance; - - (*dml_ctx)->v21.mode_support.display_config = &(*dml_ctx)->v21.display_config; - (*dml_ctx)->v21.mode_programming.display_config = (*dml_ctx)->v21.mode_support.display_config; - - (*dml_ctx)->v21.mode_programming.programming = vzalloc(sizeof(struct dml2_display_cfg_programming)); - if (!((*dml_ctx)->v21.mode_programming.programming)) - return false; - - return true; -} - -static void dml21_populate_configuration_options(const struct dc *in_dc, - struct dml2_context *dml_ctx, - const struct dml2_configuration_options *config) -{ - dml_ctx->config = *config; - - /* UCLK P-State options */ - if (in_dc->debug.dml21_force_pstate_method) { - dml_ctx->config.pmo.force_pstate_method_enable = true; - for (int i = 0; i < MAX_PIPES; i++) - dml_ctx->config.pmo.force_pstate_method_values[i] = in_dc->debug.dml21_force_pstate_method_values[i]; - } else { - dml_ctx->config.pmo.force_pstate_method_enable = false; - } -} - -static void dml21_init(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) -{ - - dml_ctx->architecture = dml2_architecture_21; - - dml21_populate_configuration_options(in_dc, dml_ctx, config); - - DC_FP_START(); - - dml21_populate_dml_init_params(&dml_ctx->v21.dml_init, &dml_ctx->config, in_dc); - - dml2_initialize_instance(&dml_ctx->v21.dml_init); - - DC_FP_END(); -} - -bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) -{ - /* Allocate memory for initializing DML21 instance */ - if (!dml21_allocate_memory(dml_ctx)) - return false; - - dml21_init(in_dc, *dml_ctx, config); - - return true; -} - -void dml21_destroy(struct dml2_context *dml2) -{ - vfree(dml2->v21.dml_init.dml2_instance); - vfree(dml2->v21.mode_programming.programming); -} - -static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, - struct dml2_context *in_ctx, unsigned int pipe_cnt) -{ - unsigned int dml_prog_idx = 0, dc_pipe_index = 0, num_dpps_required = 0; - struct dml2_per_plane_programming *pln_prog = NULL; - struct dml2_per_stream_programming *stream_prog = NULL; - struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; - int num_pipes; - unsigned int dml_phantom_prog_idx; - - context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; - - /* copy global DCHUBBUB arbiter registers */ - memcpy(&context->bw_ctx.bw.dcn.arb_regs, &in_ctx->v21.mode_programming.programming->global_regs.arb_regs, sizeof(struct dml2_display_arb_regs)); - - /* legacy only */ - context->bw_ctx.bw.dcn.compbuf_size_kb = (int)in_ctx->v21.mode_programming.programming->global_regs.arb_regs.compbuf_size * 64; - - context->bw_ctx.bw.dcn.mall_ss_size_bytes = 0; - context->bw_ctx.bw.dcn.mall_ss_psr_active_size_bytes = 0; - context->bw_ctx.bw.dcn.mall_subvp_size_bytes = 0; - - /* phantom's start after main planes */ - dml_phantom_prog_idx = in_ctx->v21.mode_programming.programming->display_config.num_planes; - - for (dml_prog_idx = 0; dml_prog_idx < DML2_MAX_PLANES; dml_prog_idx++) { - pln_prog = &in_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; - - if (!pln_prog->plane_descriptor) - continue; - - stream_prog = &in_ctx->v21.mode_programming.programming->stream_programming[pln_prog->plane_descriptor->stream_index]; - num_dpps_required = pln_prog->num_dpps_required; - - if (num_dpps_required == 0) { - continue; - } - num_pipes = dml21_find_dc_pipes_for_plane(dc, context, in_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); - - if (num_pipes <= 0) - continue; - - /* program each pipe */ - for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { - dml21_program_dc_pipe(in_ctx, context, dc_main_pipes[dc_pipe_index], pln_prog, stream_prog); - - if (pln_prog->phantom_plane.valid && dc_phantom_pipes[dc_pipe_index]) { - dml21_program_dc_pipe(in_ctx, context, dc_phantom_pipes[dc_pipe_index], pln_prog, stream_prog); - } - } - - /* copy per plane mcache allocation */ - memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_prog_idx], &pln_prog->mcache_allocation, sizeof(struct dml2_mcache_surface_allocation)); - if (pln_prog->phantom_plane.valid) { - memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_phantom_prog_idx], - &pln_prog->phantom_plane.mcache_allocation, - sizeof(struct dml2_mcache_surface_allocation)); - - dml_phantom_prog_idx++; - } - } - - /* assign global clocks */ - context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; - context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; - if (in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.num_clk_values > 1) { - context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = - in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.clk_values_khz[in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.num_clk_values] * 1000; - } else { - context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.clk_values_khz[0] * 1000; - } - - if (in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.num_clk_values > 1) { - context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = - in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.clk_values_khz[in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.num_clk_values] * 1000; - } else { - context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.clk_values_khz[0] * 1000; - } - - /* get global mall allocation */ - if (dc->res_pool->funcs->calculate_mall_ways_from_bytes) { - context->bw_ctx.bw.dcn.clk.num_ways = dc->res_pool->funcs->calculate_mall_ways_from_bytes(dc, context->bw_ctx.bw.dcn.mall_subvp_size_bytes); - } else { - context->bw_ctx.bw.dcn.clk.num_ways = 0; - } -} - -static void dml21_prepare_mcache_params(struct dml2_context *dml_ctx, struct dc_state *context, struct dc_mcache_params *mcache_params) -{ - int dc_plane_idx = 0; - int dml_prog_idx, stream_idx, plane_idx; - struct dml2_per_plane_programming *pln_prog = NULL; - - for (stream_idx = 0; stream_idx < context->stream_count; stream_idx++) { - for (plane_idx = 0; plane_idx < context->stream_status[stream_idx].plane_count; plane_idx++) { - dml_prog_idx = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_idx]->stream_id, context->stream_status[stream_idx].plane_states[plane_idx], context); - if (dml_prog_idx == INVALID) { - continue; - } - pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; - mcache_params[dc_plane_idx].valid = pln_prog->mcache_allocation.valid; - mcache_params[dc_plane_idx].num_mcaches_plane0 = pln_prog->mcache_allocation.num_mcaches_plane0; - mcache_params[dc_plane_idx].num_mcaches_plane1 = pln_prog->mcache_allocation.num_mcaches_plane1; - mcache_params[dc_plane_idx].requires_dedicated_mall_mcache = pln_prog->mcache_allocation.requires_dedicated_mall_mcache; - mcache_params[dc_plane_idx].last_slice_sharing.plane0_plane1 = pln_prog->mcache_allocation.last_slice_sharing.plane0_plane1; - memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane0, - pln_prog->mcache_allocation.mcache_x_offsets_plane0, - sizeof(int) * (DML2_MAX_MCACHES + 1)); - memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane1, - pln_prog->mcache_allocation.mcache_x_offsets_plane1, - sizeof(int) * (DML2_MAX_MCACHES + 1)); - dc_plane_idx++; - } - } -} - -static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) -{ - bool result = false; - struct dml2_build_mode_programming_in_out *mode_programming = &dml_ctx->v21.mode_programming; - struct dc_mcache_params mcache_params[MAX_PLANES] = {0}; - - memset(&dml_ctx->v21.display_config, 0, sizeof(struct dml2_display_cfg)); - memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); - memset(&dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params, 0, sizeof(struct dml2_core_mode_programming_in_out)); - - if (!context) - return true; - - if (context->stream_count == 0) { - dml21_build_fams2_programming(in_dc, context, dml_ctx); - return true; - } - - /* scrub phantom's from current dc_state */ - dml_ctx->config.svp_pstate.callbacks.remove_phantom_streams_and_planes(in_dc, context); - dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); - - /* Populate stream, plane mappings and other fields in display config. */ - DC_FP_START(); - result = dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); - DC_FP_END(); - if (!result) - return false; - - DC_FP_START(); - result = dml2_build_mode_programming(mode_programming); - DC_FP_END(); - if (!result) - return false; - - /* Check and map HW resources */ - if (result && !dml_ctx->config.skip_hw_state_mapping) { - dml21_map_hw_resources(dml_ctx); - dml2_map_dc_pipes(dml_ctx, context, NULL, &dml_ctx->v21.dml_to_dc_pipe_mapping, in_dc->current_state); - /* if subvp phantoms are present, expand them into dc context */ - dml21_handle_phantom_streams_planes(in_dc, context, dml_ctx); - - if (in_dc->res_pool->funcs->program_mcache_pipe_config) { - //Prepare mcache params for each plane based on mcache output from DML - dml21_prepare_mcache_params(dml_ctx, context, mcache_params); - - //populate mcache regs to each pipe - dml_ctx->config.callbacks.allocate_mcache(context, mcache_params); - } - } - - /* Copy DML CLK, WM and REG outputs to bandwidth context */ - if (result && !dml_ctx->config.skip_hw_state_mapping) { - dml21_calculate_rq_and_dlg_params(in_dc, context, &context->res_ctx, dml_ctx, in_dc->res_pool->pipe_count); - dml21_copy_clocks_to_dc_state(dml_ctx, context); - dml21_extract_watermark_sets(in_dc, &context->bw_ctx.bw.dcn.watermarks, dml_ctx); - dml21_build_fams2_programming(in_dc, context, dml_ctx); - } - - return true; -} - -static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) -{ - bool is_supported = false; - struct dml2_initialize_instance_in_out *dml_init = &dml_ctx->v21.dml_init; - struct dml2_check_mode_supported_in_out *mode_support = &dml_ctx->v21.mode_support; - - memset(&dml_ctx->v21.display_config, 0, sizeof(struct dml2_display_cfg)); - memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); - memset(&dml_ctx->v21.mode_programming.dml2_instance->scratch.check_mode_supported_locals.mode_support_params, 0, sizeof(struct dml2_core_mode_support_in_out)); - - if (!context || context->stream_count == 0) - return true; - - /* Scrub phantom's from current dc_state */ - dml_ctx->config.svp_pstate.callbacks.remove_phantom_streams_and_planes(in_dc, context); - dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); - - mode_support->dml2_instance = dml_init->dml2_instance; - DC_FP_START(); - dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); - dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params.programming = dml_ctx->v21.mode_programming.programming; - is_supported = dml2_check_mode_supported(mode_support); - DC_FP_END(); - if (!is_supported) - return false; - - return true; -} - -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, - enum dc_validate_mode validate_mode) -{ - bool out = false; - - /* Use dml21_check_mode_support for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ - if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) - out = dml21_check_mode_support(in_dc, context, dml_ctx); - else - out = dml21_mode_check_and_programming(in_dc, context, dml_ctx); - - return out; -} - -void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) -{ - unsigned int dml_prog_idx, dml_phantom_prog_idx, dc_pipe_index; - int num_pipes; - struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; - - struct dml2_per_plane_programming *pln_prog = NULL; - struct dml2_plane_mcache_configuration_descriptor *mcache_config = NULL; - struct prepare_mcache_programming_locals *l = &dml_ctx->v21.scratch.prepare_mcache_locals; - - if (context->stream_count == 0) { - return; - } - - memset(&l->build_mcache_programming_params, 0, sizeof(struct dml2_build_mcache_programming_in_out)); - l->build_mcache_programming_params.dml2_instance = dml_ctx->v21.dml_init.dml2_instance; - - /* phantom's start after main planes */ - dml_phantom_prog_idx = dml_ctx->v21.mode_programming.programming->display_config.num_planes; - - /* Build mcache programming parameters per plane per pipe */ - for (dml_prog_idx = 0; dml_prog_idx < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_prog_idx++) { - pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; - - mcache_config = &l->build_mcache_programming_params.mcache_configurations[dml_prog_idx]; - memset(mcache_config, 0, sizeof(struct dml2_plane_mcache_configuration_descriptor)); - mcache_config->plane_descriptor = pln_prog->plane_descriptor; - mcache_config->mcache_allocation = &context->bw_ctx.bw.dcn.mcache_allocations[dml_prog_idx]; - mcache_config->num_pipes = pln_prog->num_dpps_required; - l->build_mcache_programming_params.num_configurations++; - - if (pln_prog->num_dpps_required == 0) { - continue; - } - - num_pipes = dml21_find_dc_pipes_for_plane(in_dc, context, dml_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); - if (num_pipes <= 0 || dc_main_pipes[0]->stream == NULL || - dc_main_pipes[0]->plane_state == NULL) - continue; - - /* get config for each pipe */ - for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { - ASSERT(dc_main_pipes[dc_pipe_index]); - dml21_get_pipe_mcache_config(context, dc_main_pipes[dc_pipe_index], pln_prog, &mcache_config->pipe_configurations[dc_pipe_index]); - } - - /* get config for each phantom pipe */ - if (pln_prog->phantom_plane.valid && - dc_phantom_pipes[0] && - dc_main_pipes[0]->stream && - dc_phantom_pipes[0]->plane_state) { - mcache_config = &l->build_mcache_programming_params.mcache_configurations[dml_phantom_prog_idx]; - memset(mcache_config, 0, sizeof(struct dml2_plane_mcache_configuration_descriptor)); - mcache_config->plane_descriptor = pln_prog->plane_descriptor; - mcache_config->mcache_allocation = &context->bw_ctx.bw.dcn.mcache_allocations[dml_phantom_prog_idx]; - mcache_config->num_pipes = pln_prog->num_dpps_required; - l->build_mcache_programming_params.num_configurations++; - - for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { - ASSERT(dc_phantom_pipes[dc_pipe_index]); - dml21_get_pipe_mcache_config(context, dc_phantom_pipes[dc_pipe_index], pln_prog, &mcache_config->pipe_configurations[dc_pipe_index]); - } - - /* increment phantom index */ - dml_phantom_prog_idx++; - } - } - - /* Call to generate mcache programming per plane per pipe for the given display configuration */ - dml2_build_mcache_programming(&l->build_mcache_programming_params); - - /* get per plane per pipe mcache programming */ - for (dml_prog_idx = 0; dml_prog_idx < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_prog_idx++) { - pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; - - num_pipes = dml21_find_dc_pipes_for_plane(in_dc, context, dml_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); - if (num_pipes <= 0 || dc_main_pipes[0]->stream == NULL || - dc_main_pipes[0]->plane_state == NULL) - continue; - - /* get config for each pipe */ - for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { - ASSERT(dc_main_pipes[dc_pipe_index]); - if (l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_prog_idx][dc_pipe_index]) { - memcpy(&dc_main_pipes[dc_pipe_index]->mcache_regs, - l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_prog_idx][dc_pipe_index], - sizeof(struct dml2_hubp_pipe_mcache_regs)); - } - } - - /* get config for each phantom pipe */ - if (pln_prog->phantom_plane.valid && - dc_phantom_pipes[0] && - dc_main_pipes[0]->stream && - dc_phantom_pipes[0]->plane_state) { - for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { - ASSERT(dc_phantom_pipes[dc_pipe_index]); - if (l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_phantom_prog_idx][dc_pipe_index]) { - memcpy(&dc_phantom_pipes[dc_pipe_index]->mcache_regs, - l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_phantom_prog_idx][dc_pipe_index], - sizeof(struct dml2_hubp_pipe_mcache_regs)); - } - } - /* increment phantom index */ - dml_phantom_prog_idx++; - } - } -} - -void dml21_copy(struct dml2_context *dst_dml_ctx, - struct dml2_context *src_dml_ctx) -{ - /* Preserve references to internals */ - struct dml2_instance *dst_dml2_instance = dst_dml_ctx->v21.dml_init.dml2_instance; - struct dml2_display_cfg_programming *dst_dml2_programming = dst_dml_ctx->v21.mode_programming.programming; - - /* Copy context */ - memcpy(dst_dml_ctx, src_dml_ctx, sizeof(struct dml2_context)); - - /* Copy Internals */ - memcpy(dst_dml2_instance, src_dml_ctx->v21.dml_init.dml2_instance, sizeof(struct dml2_instance)); - memcpy(dst_dml2_programming, src_dml_ctx->v21.mode_programming.programming, sizeof(struct dml2_display_cfg_programming)); - - /* Restore references to internals */ - dst_dml_ctx->v21.dml_init.dml2_instance = dst_dml2_instance; - - dst_dml_ctx->v21.mode_support.dml2_instance = dst_dml2_instance; - dst_dml_ctx->v21.mode_programming.dml2_instance = dst_dml2_instance; - - dst_dml_ctx->v21.mode_support.display_config = &dst_dml_ctx->v21.display_config; - dst_dml_ctx->v21.mode_programming.display_config = dst_dml_ctx->v21.mode_support.display_config; - - dst_dml_ctx->v21.mode_programming.programming = dst_dml2_programming; - - DC_FP_START(); - - /* need to initialize copied instance for internal references to be correct */ - dml2_initialize_instance(&dst_dml_ctx->v21.dml_init); - - DC_FP_END(); -} - -bool dml21_create_copy(struct dml2_context **dst_dml_ctx, - struct dml2_context *src_dml_ctx) -{ - /* Allocate memory for initializing DML21 instance */ - if (!dml21_allocate_memory(dst_dml_ctx)) - return false; - - dml21_copy(*dst_dml_ctx, src_dml_ctx); - - return true; -} - -void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) -{ - dml21_init(in_dc, dml_ctx, config); -} - diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h deleted file mode 100644 index 15f92029d2e5..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h +++ /dev/null @@ -1,135 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#ifndef _DML21_WRAPPER_H_ -#define _DML21_WRAPPER_H_ - -#include "os_types.h" -#include "dml_top_soc_parameter_types.h" -#include "dml_top_display_cfg_types.h" - -struct dc; -struct dc_state; -struct dml2_configuration_options; -struct dml2_context; -enum dc_validate_mode; - -/** - * dml2_create - Creates dml21_context. - * @in_dc: dc. - * @dml2: Created dml21 context. - * @config: dml21 configuration options. - * - * Create of DML21 is done as part of dc_state creation. - * DML21 IP, SOC and STATES are initialized at - * creation time. - * - * Return: True if dml2 is successfully created, false otherwise. - */ -bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config); -void dml21_destroy(struct dml2_context *dml2); -void dml21_copy(struct dml2_context *dst_dml_ctx, - struct dml2_context *src_dml_ctx); -bool dml21_create_copy(struct dml2_context **dst_dml_ctx, - struct dml2_context *src_dml_ctx); -void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config); - -/** - * dml21_validate - Determines if a display configuration is supported or not. - * @in_dc: dc. - * @context: dc_state to be validated. - * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX - * will not populate context.res_ctx. - * - * Based on fast_validate option internally would call: - * - * -dml21_mode_check_and_programming - for DC_VALIDATE_MODE_AND_PROGRAMMING option - * Calculates if dc_state can be supported on the input display - * configuration. If supported, generates the necessary HW - * programming for the new dc_state. - * - * -dml21_check_mode_support - for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX option - * Calculates if dc_state can be supported for the input display - * config. - - * Context: Two threads may not invoke this function concurrently unless they reference - * separate dc_states for validation. - * Return: True if mode is supported, false otherwise. - */ -bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, - enum dc_validate_mode validate_mode); - -/* Prepare hubp mcache_regs for hubp mcache ID and split coordinate programming */ -void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); - -/* Structure for inputting external SOCBB and DCNIP values for tool based debugging. */ -struct socbb_ip_params_external { - struct dml2_ip_capabilities ip_params; - struct dml2_soc_bb soc_bb; -}; - -/*mcache parameters decided by dml*/ -struct dc_mcache_params { - bool valid; - /* - * For iMALL, dedicated mall mcaches are required (sharing of last - * slice possible), for legacy phantom or phantom without return - * the only mall mcaches need to be valid. - */ - bool requires_dedicated_mall_mcache; - unsigned int num_mcaches_plane0; - unsigned int num_mcaches_plane1; - /* - * Generally, plane0/1 slices must use a disjoint set of caches - * but in some cases the final segement of the two planes can - * use the same cache. If plane0_plane1 is set, then this is - * allowed. - * - * Similarly, the caches allocated to MALL prefetcher are generally - * disjoint, but if mall_prefetch is set, then the final segment - * between the main and the mall pixel requestor can use the same - * cache. - * - * Note that both bits may be set at the same time. - */ - struct { - bool mall_comb_mcache_p0; - bool mall_comb_mcache_p1; - bool plane0_plane1; - } last_slice_sharing; - /* - * A plane is divided into vertical slices of mcaches, - * which wrap on the surface width. - * - * For example, if the surface width is 7680, and split into - * three slices of equal width, the boundary array would contain - * [2560, 5120, 7680] - * - * The assignments are - * 0 = [0 .. 2559] - * 1 = [2560 .. 5119] - * 2 = [5120 .. 7679] - * 0 = [7680 .. INF] - * The final element implicitly is the same as the first, and - * at first seems invalid since it is never referenced (since) - * it is outside the surface. However, its useful when shifting - * (see below). - * - * For any given valid mcache assignment, a shifted version, wrapped - * on the surface width boundary is also assumed to be valid. - * - * For example, shifting [2560, 5120, 7680] by -50 results in - * [2510, 5170, 7630]. - * - * The assignments are now: - * 0 = [0 .. 2509] - * 1 = [2510 .. 5169] - * 2 = [5170 .. 7629] - * 0 = [7630 .. INF] - */ - int mcache_x_offsets_plane0[DML2_MAX_MCACHES + 1]; - int mcache_x_offsets_plane1[DML2_MAX_MCACHES + 1]; -}; -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h deleted file mode 100644 index 793e1c038efd..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h +++ /dev/null @@ -1,373 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#ifndef __DML_DML_DCN4_SOC_BB__ -#define __DML_DML_DCN4_SOC_BB__ - -#include "dml_top_soc_parameter_types.h" - -static const struct dml2_soc_qos_parameters dml_dcn4_variant_a_soc_qos_params = { - .derate_table = { - .system_active_urgent = { - .dram_derate_percent_pixel = 22, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 76, - .dcfclk_derate_percent = 100, - }, - .system_active_average = { - .dram_derate_percent_pixel = 17, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 57, - .dcfclk_derate_percent = 75, - }, - .dcn_mall_prefetch_urgent = { - .dram_derate_percent_pixel = 40, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 83, - .dcfclk_derate_percent = 100, - }, - .dcn_mall_prefetch_average = { - .dram_derate_percent_pixel = 33, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 62, - .dcfclk_derate_percent = 83, - }, - .system_idle_average = { - .dram_derate_percent_pixel = 70, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 83, - .dcfclk_derate_percent = 100, - }, - }, - .writeback = { - .base_latency_us = 12, - .scaling_factor_us = 0, - .scaling_factor_mhz = 0, - }, - .qos_params = { - .dcn4x = { - .df_qos_response_time_fclk_cycles = 300, - .max_round_trip_to_furthest_cs_fclk_cycles = 350, - .mall_overhead_fclk_cycles = 50, - .meta_trip_adder_fclk_cycles = 36, - .average_transport_distance_fclk_cycles = 257, - .umc_urgent_ramp_latency_margin = 50, - .umc_max_latency_margin = 30, - .umc_average_latency_margin = 20, - .fabric_max_transport_latency_margin = 20, - .fabric_average_transport_latency_margin = 10, - - .per_uclk_dpm_params = { - { - .minimum_uclk_khz = 97 * 1000, - .urgent_ramp_uclk_cycles = 472, - .trip_to_memory_uclk_cycles = 827, - .meta_trip_to_memory_uclk_cycles = 827, - .maximum_latency_when_urgent_uclk_cycles = 72, - .average_latency_when_urgent_uclk_cycles = 61, - .maximum_latency_when_non_urgent_uclk_cycles = 827, - .average_latency_when_non_urgent_uclk_cycles = 118, - }, - }, - }, - }, - .qos_type = dml2_qos_param_type_dcn4x, -}; - -static const struct dml2_soc_bb dml2_socbb_dcn401 = { - .clk_table = { - .uclk = { - .clk_values_khz = {97000}, - .num_clk_values = 1, - }, - .fclk = { - .clk_values_khz = {300000, 2500000}, - .num_clk_values = 2, - }, - .dcfclk = { - .clk_values_khz = {200000, 1564000}, - .num_clk_values = 2, - }, - .dispclk = { - .clk_values_khz = {100000, 2000000}, - .num_clk_values = 2, - }, - .dppclk = { - .clk_values_khz = {100000, 2000000}, - .num_clk_values = 2, - }, - .dtbclk = { - .clk_values_khz = {100000, 1564000}, - .num_clk_values = 2, - }, - .phyclk = { - .clk_values_khz = {810000, 810000}, - .num_clk_values = 2, - }, - .socclk = { - .clk_values_khz = {300000, 1200000}, - .num_clk_values = 2, - }, - .dscclk = { - .clk_values_khz = {666667, 666667}, - .num_clk_values = 2, - }, - .phyclk_d18 = { - .clk_values_khz = {625000, 625000}, - .num_clk_values = 2, - }, - .phyclk_d32 = { - .clk_values_khz = {625000, 625000}, - .num_clk_values = 2, - }, - .dram_config = { - .channel_width_bytes = 2, - .channel_count = 16, - .transactions_per_clock = 16, - }, - }, - - .qos_parameters = { - .derate_table = { - .system_active_urgent = { - .dram_derate_percent_pixel = 22, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 76, - .dcfclk_derate_percent = 100, - }, - .system_active_average = { - .dram_derate_percent_pixel = 15, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 57, - .dcfclk_derate_percent = 75, - }, - .dcn_mall_prefetch_urgent = { - .dram_derate_percent_pixel = 40, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 83, - .dcfclk_derate_percent = 100, - }, - .dcn_mall_prefetch_average = { - .dram_derate_percent_pixel = 30, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 62, - .dcfclk_derate_percent = 83, - }, - .system_idle_average = { - .dram_derate_percent_pixel = 70, - .dram_derate_percent_vm = 0, - .dram_derate_percent_pixel_and_vm = 0, - .fclk_derate_percent = 83, - .dcfclk_derate_percent = 100, - }, - }, - .writeback = { - .base_latency_us = 0, - .scaling_factor_us = 0, - .scaling_factor_mhz = 0, - }, - .qos_params = { - .dcn4x = { - .df_qos_response_time_fclk_cycles = 300, - .max_round_trip_to_furthest_cs_fclk_cycles = 350, - .mall_overhead_fclk_cycles = 50, - .meta_trip_adder_fclk_cycles = 36, - .average_transport_distance_fclk_cycles = 260, - .umc_urgent_ramp_latency_margin = 50, - .umc_max_latency_margin = 30, - .umc_average_latency_margin = 20, - .fabric_max_transport_latency_margin = 20, - .fabric_average_transport_latency_margin = 10, - - .per_uclk_dpm_params = { - { - // State 1 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 472, - .trip_to_memory_uclk_cycles = 827, - .meta_trip_to_memory_uclk_cycles = 827, - .maximum_latency_when_urgent_uclk_cycles = 72, - .average_latency_when_urgent_uclk_cycles = 72, - .maximum_latency_when_non_urgent_uclk_cycles = 827, - .average_latency_when_non_urgent_uclk_cycles = 117, - }, - { - // State 2 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 546, - .trip_to_memory_uclk_cycles = 848, - .meta_trip_to_memory_uclk_cycles = 848, - .maximum_latency_when_urgent_uclk_cycles = 146, - .average_latency_when_urgent_uclk_cycles = 146, - .maximum_latency_when_non_urgent_uclk_cycles = 848, - .average_latency_when_non_urgent_uclk_cycles = 133, - }, - { - // State 3 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 564, - .trip_to_memory_uclk_cycles = 853, - .meta_trip_to_memory_uclk_cycles = 853, - .maximum_latency_when_urgent_uclk_cycles = 164, - .average_latency_when_urgent_uclk_cycles = 164, - .maximum_latency_when_non_urgent_uclk_cycles = 853, - .average_latency_when_non_urgent_uclk_cycles = 136, - }, - { - // State 4 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 613, - .trip_to_memory_uclk_cycles = 869, - .meta_trip_to_memory_uclk_cycles = 869, - .maximum_latency_when_urgent_uclk_cycles = 213, - .average_latency_when_urgent_uclk_cycles = 213, - .maximum_latency_when_non_urgent_uclk_cycles = 869, - .average_latency_when_non_urgent_uclk_cycles = 149, - }, - { - // State 5 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 632, - .trip_to_memory_uclk_cycles = 874, - .meta_trip_to_memory_uclk_cycles = 874, - .maximum_latency_when_urgent_uclk_cycles = 232, - .average_latency_when_urgent_uclk_cycles = 232, - .maximum_latency_when_non_urgent_uclk_cycles = 874, - .average_latency_when_non_urgent_uclk_cycles = 153, - }, - { - // State 6 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 665, - .trip_to_memory_uclk_cycles = 885, - .meta_trip_to_memory_uclk_cycles = 885, - .maximum_latency_when_urgent_uclk_cycles = 265, - .average_latency_when_urgent_uclk_cycles = 265, - .maximum_latency_when_non_urgent_uclk_cycles = 885, - .average_latency_when_non_urgent_uclk_cycles = 161, - }, - { - // State 7 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 689, - .trip_to_memory_uclk_cycles = 895, - .meta_trip_to_memory_uclk_cycles = 895, - .maximum_latency_when_urgent_uclk_cycles = 289, - .average_latency_when_urgent_uclk_cycles = 289, - .maximum_latency_when_non_urgent_uclk_cycles = 895, - .average_latency_when_non_urgent_uclk_cycles = 167, - }, - { - // State 8 - .minimum_uclk_khz = 0, - .urgent_ramp_uclk_cycles = 716, - .trip_to_memory_uclk_cycles = 902, - .meta_trip_to_memory_uclk_cycles = 902, - .maximum_latency_when_urgent_uclk_cycles = 316, - .average_latency_when_urgent_uclk_cycles = 316, - .maximum_latency_when_non_urgent_uclk_cycles = 902, - .average_latency_when_non_urgent_uclk_cycles = 174, - }, - }, - }, - }, - .qos_type = dml2_qos_param_type_dcn4x, - }, - - .power_management_parameters = { - .dram_clk_change_blackout_us = 400, - .fclk_change_blackout_us = 0, - .g7_ppt_blackout_us = 0, - .stutter_enter_plus_exit_latency_us = 54, - .stutter_exit_latency_us = 41, - .z8_stutter_enter_plus_exit_latency_us = 0, - .z8_stutter_exit_latency_us = 0, - /* - .g6_temp_read_blackout_us = { - 23.00, - 10.00, - 10.00, - 8.00, - 8.00, - 5.00, - 5.00, - 5.00, - }, - */ - }, - - .vmin_limit = { - .dispclk_khz = 600 * 1000, - }, - - .dprefclk_mhz = 720, - .xtalclk_mhz = 100, - .pcie_refclk_mhz = 100, - .dchub_refclk_mhz = 50, - .mall_allocated_for_dcn_mbytes = 64, - .max_outstanding_reqs = 512, - .fabric_datapath_to_dcn_data_return_bytes = 64, - .return_bus_width_bytes = 64, - .hostvm_min_page_size_kbytes = 0, - .gpuvm_min_page_size_kbytes = 256, - .phy_downspread_percent = 0.38, - .dcn_downspread_percent = 0.38, - .dispclk_dppclk_vco_speed_mhz = 4500, - .do_urgent_latency_adjustment = 0, - .mem_word_bytes = 32, - .num_dcc_mcaches = 8, - .mcache_size_bytes = 2048, - .mcache_line_size_bytes = 32, - .max_fclk_for_uclk_dpm_khz = 1250 * 1000, -}; - -static const struct dml2_ip_capabilities dml2_dcn401_max_ip_caps = { - .pipe_count = 4, - .otg_count = 4, - .num_dsc = 4, - .max_num_dp2p0_streams = 4, - .max_num_hdmi_frl_outputs = 1, - .max_num_dp2p0_outputs = 4, - .rob_buffer_size_kbytes = 192, - .config_return_buffer_size_in_kbytes = 1344, - .config_return_buffer_segment_size_in_kbytes = 64, - .meta_fifo_size_in_kentries = 22, - .compressed_buffer_segment_size_in_kbytes = 64, - .cursor_buffer_size = 24, - .max_flip_time_us = 80, - .max_flip_time_lines = 32, - .hostvm_mode = 0, - .subvp_drr_scheduling_margin_us = 100, - .subvp_prefetch_end_to_mall_start_us = 15, - .subvp_fw_processing_delay = 15, - .max_vactive_det_fill_delay_us = 400, - - .fams2 = { - .max_allow_delay_us = 100 * 1000, - .scheduling_delay_us = 550, - .vertical_interrupt_ack_delay_us = 40, - .allow_programming_delay_us = 18, - .min_allow_width_us = 20, - .subvp_df_throttle_delay_us = 100, - .subvp_programming_delay_us = 200, - .subvp_prefetch_to_mall_delay_us = 18, - .drr_programming_delay_us = 35, - - .lock_timeout_us = 5000, - .recovery_timeout_us = 5000, - .flip_programming_delay_us = 300, - }, -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h deleted file mode 100644 index 281d7ad230d8..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h +++ /dev/null @@ -1,10 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_EXTERNAL_LIB_DEPS__ -#define __DML2_EXTERNAL_LIB_DEPS__ - -#include "os_types.h" - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h deleted file mode 100644 index a64ec4dcf11a..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h +++ /dev/null @@ -1,46 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML_TOP_H__ -#define __DML_TOP_H__ - -#include "dml_top_types.h" - -/* - * Top Level Interface for DML2 - */ - -/* - * Returns the size of the DML instance for the caller to allocate - */ -unsigned int dml2_get_instance_size_bytes(void); - -/* - * Initializes the DML instance (i.e. with configuration, soc BB, IP params, etc...) - */ -bool dml2_initialize_instance(struct dml2_initialize_instance_in_out *in_out); - -/* - * Determines if the input mode is supported (boolean) on the SoC at all. Does not return - * information on how mode should be programmed. - */ -bool dml2_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out); - -/* - * Determines the full (optimized) programming for the input mode. Returns minimum - * clocks as well as dchub register programming values for all pipes, additional meta - * such as ODM or MPCC combine factors. - */ -bool dml2_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out); - -/* - * Determines the correct per pipe mcache register programming for a valid mode. - * The mcache allocation must have been calculated (successfully) in a previous - * call to dml2_build_mode_programming. - * The actual hubp viewport dimensions be what the actual registers will be - * programmed to (i.e. based on scaler setup). - */ -bool dml2_build_mcache_programming(struct dml2_build_mcache_programming_in_out *in_out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h deleted file mode 100644 index 8e5a30287220..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h +++ /dev/null @@ -1,189 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __dml2_TOP_DCHUB_REGISTERS_H__ -#define __dml2_TOP_DCHUB_REGISTERS_H__ - -#include "dml2_external_lib_deps.h" -// These types are uint32_t as they represent actual calculated register values for HW - -struct dml2_display_dlg_regs { - uint32_t refcyc_h_blank_end; - uint32_t dlg_vblank_end; - uint32_t min_dst_y_next_start; - uint32_t refcyc_per_htotal; - uint32_t refcyc_x_after_scaler; - uint32_t dst_y_after_scaler; - uint32_t dst_y_prefetch; - uint32_t dst_y_per_vm_vblank; - uint32_t dst_y_per_row_vblank; - uint32_t dst_y_per_vm_flip; - uint32_t dst_y_per_row_flip; - uint32_t ref_freq_to_pix_freq; - uint32_t vratio_prefetch; - uint32_t vratio_prefetch_c; - uint32_t refcyc_per_tdlut_group; - uint32_t refcyc_per_pte_group_vblank_l; - uint32_t refcyc_per_pte_group_vblank_c; - uint32_t refcyc_per_pte_group_flip_l; - uint32_t refcyc_per_pte_group_flip_c; - uint32_t dst_y_per_pte_row_nom_l; - uint32_t dst_y_per_pte_row_nom_c; - uint32_t refcyc_per_pte_group_nom_l; - uint32_t refcyc_per_pte_group_nom_c; - uint32_t refcyc_per_line_delivery_pre_l; - uint32_t refcyc_per_line_delivery_pre_c; - uint32_t refcyc_per_line_delivery_l; - uint32_t refcyc_per_line_delivery_c; - uint32_t refcyc_per_vm_group_vblank; - uint32_t refcyc_per_vm_group_flip; - uint32_t refcyc_per_vm_req_vblank; - uint32_t refcyc_per_vm_req_flip; - uint32_t dst_y_offset_cur0; - uint32_t chunk_hdl_adjust_cur0; - uint32_t vready_after_vcount0; - uint32_t dst_y_delta_drq_limit; - uint32_t refcyc_per_vm_dmdata; - uint32_t dmdata_dl_delta; - - // MRQ - uint32_t refcyc_per_meta_chunk_vblank_l; - uint32_t refcyc_per_meta_chunk_vblank_c; - uint32_t refcyc_per_meta_chunk_flip_l; - uint32_t refcyc_per_meta_chunk_flip_c; - uint32_t dst_y_per_meta_row_nom_l; - uint32_t dst_y_per_meta_row_nom_c; - uint32_t refcyc_per_meta_chunk_nom_l; - uint32_t refcyc_per_meta_chunk_nom_c; -}; - -struct dml2_display_ttu_regs { - uint32_t qos_level_low_wm; - uint32_t qos_level_high_wm; - uint32_t min_ttu_vblank; - uint32_t qos_level_flip; - uint32_t refcyc_per_req_delivery_l; - uint32_t refcyc_per_req_delivery_c; - uint32_t refcyc_per_req_delivery_cur0; - uint32_t refcyc_per_req_delivery_pre_l; - uint32_t refcyc_per_req_delivery_pre_c; - uint32_t refcyc_per_req_delivery_pre_cur0; - uint32_t qos_level_fixed_l; - uint32_t qos_level_fixed_c; - uint32_t qos_level_fixed_cur0; - uint32_t qos_ramp_disable_l; - uint32_t qos_ramp_disable_c; - uint32_t qos_ramp_disable_cur0; -}; - -struct dml2_display_arb_regs { - uint32_t max_req_outstanding; - uint32_t min_req_outstanding; - uint32_t sat_level_us; - uint32_t hvm_max_qos_commit_threshold; - uint32_t hvm_min_req_outstand_commit_threshold; - uint32_t compbuf_reserved_space_kbytes; - uint32_t compbuf_size; - uint32_t sdpif_request_rate_limit; - uint32_t allow_sdpif_rate_limit_when_cstate_req; - uint32_t dcfclk_deep_sleep_hysteresis; - uint32_t pstate_stall_threshold; -}; - -struct dml2_cursor_dlg_regs{ - uint32_t dst_x_offset; // CURSOR0_DST_X_OFFSET - uint32_t dst_y_offset; // CURSOR0_DST_Y_OFFSET - uint32_t chunk_hdl_adjust; // CURSOR0_CHUNK_HDL_ADJUST - - uint32_t qos_level_fixed; - uint32_t qos_ramp_disable; -}; - -struct dml2_display_plane_rq_regs { - uint32_t chunk_size; - uint32_t min_chunk_size; - uint32_t dpte_group_size; - uint32_t mpte_group_size; - uint32_t swath_height; - uint32_t pte_row_height_linear; - - // MRQ - uint32_t meta_chunk_size; - uint32_t min_meta_chunk_size; -}; - -struct dml2_display_rq_regs { - struct dml2_display_plane_rq_regs rq_regs_l; - struct dml2_display_plane_rq_regs rq_regs_c; - uint32_t drq_expansion_mode; - uint32_t prq_expansion_mode; - uint32_t crq_expansion_mode; - uint32_t plane1_base_address; - uint32_t unbounded_request_enabled; - - // MRQ - uint32_t mrq_expansion_mode; -}; - -struct dml2_display_mcache_regs { - uint32_t mcache_id_first; - uint32_t mcache_id_second; - uint32_t split_location; -}; - -struct dml2_hubp_pipe_mcache_regs { - struct { - struct dml2_display_mcache_regs p0; - struct dml2_display_mcache_regs p1; - } main; - struct { - struct dml2_display_mcache_regs p0; - struct dml2_display_mcache_regs p1; - } mall; -}; - -struct dml2_dchub_per_pipe_register_set { - struct dml2_display_rq_regs rq_regs; - struct dml2_display_ttu_regs ttu_regs; - struct dml2_display_dlg_regs dlg_regs; - - uint32_t det_size; -}; - -struct dml2_dchub_watermark_regs { - /* watermarks */ - uint32_t urgent; - uint32_t sr_enter; - uint32_t sr_exit; - uint32_t sr_enter_z8; - uint32_t sr_exit_z8; - uint32_t sr_enter_low_power; - uint32_t sr_exit_low_power; - uint32_t uclk_pstate; - uint32_t fclk_pstate; - uint32_t temp_read_or_ppt; - uint32_t usr; - /* qos */ - uint32_t refcyc_per_trip_to_mem; - uint32_t refcyc_per_meta_trip_to_mem; - uint32_t frac_urg_bw_flip; - uint32_t frac_urg_bw_nom; - uint32_t frac_urg_bw_mall; -}; - -enum dml2_dchub_watermark_reg_set_index { - DML2_DCHUB_WATERMARK_SET_A = 0, - DML2_DCHUB_WATERMARK_SET_B = 1, - DML2_DCHUB_WATERMARK_SET_C = 2, - DML2_DCHUB_WATERMARK_SET_D = 3, - DML2_DCHUB_WATERMARK_SET_NUM = 4, -}; - -struct dml2_dchub_global_register_set { - struct dml2_display_arb_regs arb_regs; - struct dml2_dchub_watermark_regs wm_regs[DML2_DCHUB_WATERMARK_SET_NUM]; - unsigned int num_watermark_sets; -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h deleted file mode 100644 index 13749c9fcf18..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h +++ /dev/null @@ -1,520 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML_TOP_DISPLAY_CFG_TYPES_H__ -#define __DML_TOP_DISPLAY_CFG_TYPES_H__ - -#include "dml2_external_lib_deps.h" - -#define DML2_MAX_PLANES 8 -#define DML2_MAX_DCN_PIPES 8 -#define DML2_MAX_MCACHES 8 // assume plane is going to be supported by a max of 8 mcaches -#define DML2_MAX_WRITEBACK 3 - -enum dml2_swizzle_mode { - dml2_sw_linear, // SW_LINEAR accepts 256 byte aligned pitch and also 128 byte aligned pitch if DCC is not enabled - dml2_sw_256b_2d, - dml2_sw_4kb_2d, - dml2_sw_64kb_2d, - dml2_sw_256kb_2d, - - dml2_gfx11_sw_linear, - dml2_gfx11_sw_64kb_d, - dml2_gfx11_sw_64kb_d_t, - dml2_gfx11_sw_64kb_d_x, - dml2_gfx11_sw_64kb_r_x, - dml2_gfx11_sw_256kb_d_x, - dml2_gfx11_sw_256kb_r_x, - -}; - -enum dml2_source_format_class { - dml2_444_8 = 0, - dml2_444_16 = 1, - dml2_444_32 = 2, - dml2_444_64 = 3, - dml2_420_8 = 4, - dml2_420_10 = 5, - dml2_420_12 = 6, - dml2_rgbe_alpha = 9, - dml2_rgbe = 10, - dml2_mono_8 = 11, - dml2_mono_16 = 12, - dml2_422_planar_8 = 13, - dml2_422_planar_10 = 14, - dml2_422_planar_12 = 15, - dml2_422_packed_8 = 16, - dml2_422_packed_10 = 17, - dml2_422_packed_12 = 18 -}; - -enum dml2_sample_positioning { - dml2_interstitial = 0, - dml2_cosited = 1 -}; - -enum dml2_rotation_angle { - dml2_rotation_0 = 0, - dml2_rotation_90 = 1, - dml2_rotation_180 = 2, - dml2_rotation_270 = 3 -}; - -enum dml2_output_format_class { - dml2_444 = 0, - dml2_s422 = 1, - dml2_n422 = 2, - dml2_420 = 3 -}; - -enum dml2_output_encoder_class { - dml2_dp = 0, - dml2_edp = 1, - dml2_dp2p0 = 2, - dml2_hdmi = 3, - dml2_hdmifrl = 4, - dml2_none = 5 -}; - -enum dml2_output_link_dp_rate { - dml2_dp_rate_na = 0, - dml2_dp_rate_hbr = 1, - dml2_dp_rate_hbr2 = 2, - dml2_dp_rate_hbr3 = 3, - dml2_dp_rate_uhbr10 = 4, - dml2_dp_rate_uhbr13p5 = 5, - dml2_dp_rate_uhbr20 = 6 -}; - -enum dml2_uclk_pstate_change_strategy { - dml2_uclk_pstate_change_strategy_auto = 0, - dml2_uclk_pstate_change_strategy_force_vactive = 1, - dml2_uclk_pstate_change_strategy_force_vblank = 2, - dml2_uclk_pstate_change_strategy_force_drr = 3, - dml2_uclk_pstate_change_strategy_force_mall_svp = 4, - dml2_uclk_pstate_change_strategy_force_mall_full_frame = 5, -}; - -enum dml2_svp_mode_override { - dml2_svp_mode_override_auto = 0, - dml2_svp_mode_override_main_pipe = 1, - dml2_svp_mode_override_phantom_pipe = 2, //does not need to be defined explicitly, main overrides result in implicit phantom additions - dml2_svp_mode_override_phantom_pipe_no_data_return = 3, - dml2_svp_mode_override_imall = 4 -}; - -enum dml2_refresh_from_mall_mode_override { - dml2_refresh_from_mall_mode_override_auto = 0, - dml2_refresh_from_mall_mode_override_force_disable = 1, - dml2_refresh_from_mall_mode_override_force_enable = 2 -}; - -enum dml2_odm_mode { - dml2_odm_mode_auto = 0, - dml2_odm_mode_bypass, - dml2_odm_mode_combine_2to1, - dml2_odm_mode_combine_3to1, - dml2_odm_mode_combine_4to1, - dml2_odm_mode_split_1to2, - dml2_odm_mode_mso_1to2, - dml2_odm_mode_mso_1to4 -}; - -enum dml2_scaling_transform { - dml2_scaling_transform_explicit = 0, - dml2_scaling_transform_fullscreen, - dml2_scaling_transform_aspect_ratio, - dml2_scaling_transform_centered -}; - -enum dml2_dsc_enable_option { - dml2_dsc_disable = 0, - dml2_dsc_enable = 1, - dml2_dsc_enable_if_necessary = 2 -}; - -enum dml2_tdlut_addressing_mode { - dml2_tdlut_sw_linear = 0, - dml2_tdlut_simple_linear = 1 -}; - -enum dml2_tdlut_width_mode { - dml2_tdlut_width_17_cube = 0, - dml2_tdlut_width_33_cube = 1 -}; - -enum dml2_twait_budgeting_setting { - dml2_twait_budgeting_setting_ignore = 0,// Ignore this budget in twait - - dml2_twait_budgeting_setting_if_needed, // Budget for it only if needed - //(i.e. UCLK/FCLK DPM cannot be supported in active) - - dml2_twait_budgeting_setting_try, // Budget for it as long as there is an SoC state that - // can support it -}; - -struct dml2_get_cursor_dlg_reg{ - unsigned int cursor_x_position; - unsigned int cursor_hotspot_x; - unsigned int cursor_primary_offset; - unsigned int cursor_secondary_offset; - bool cursor_stereo_en; - bool cursor_2x_magnify; - double hratio; - double pixel_rate_mhz; - double dlg_refclk_mhz; -}; - -/// @brief Surface Parameters -struct dml2_surface_cfg { - enum dml2_swizzle_mode tiling; - - struct { - unsigned long pitch; // In elements, two pixels per element in 422 packed format - unsigned long width; - unsigned long height; - } plane0; - - - struct { - unsigned long pitch; - unsigned long width; - unsigned long height; - } plane1; - - struct { - bool enable; - struct { - unsigned long pitch; - } plane0; - struct { - unsigned long pitch; - } plane1; - - struct { - double dcc_rate_plane0; - double dcc_rate_plane1; - double fraction_of_zero_size_request_plane0; - double fraction_of_zero_size_request_plane1; - } informative; - } dcc; -}; - - -struct dml2_composition_cfg { - enum dml2_rotation_angle rotation_angle; - bool mirrored; - enum dml2_scaling_transform scaling_transform; - bool rect_out_height_spans_vactive; - - struct { - bool stationary; - struct { - unsigned long width; - unsigned long height; - unsigned long x_start; - unsigned long y_start; - } plane0; - - struct { - unsigned long width; - unsigned long height; - unsigned long x_start; - unsigned long y_start; - } plane1; - } viewport; - - struct { - bool enabled; - bool easf_enabled; - bool isharp_enabled; - bool upsp_enabled; - enum dml2_sample_positioning upsp_sample_positioning; - unsigned int upsp_vtaps; - struct { - double h_ratio; - double v_ratio; - unsigned int h_taps; - unsigned int v_taps; - } plane0; - - struct { - double h_ratio; - double v_ratio; - unsigned int h_taps; - unsigned int v_taps; - } plane1; - - unsigned long rect_out_width; - } scaler_info; -}; - -struct dml2_timing_cfg { - unsigned long h_total; - unsigned long v_total; - unsigned long h_blank_end; - unsigned long v_blank_end; - unsigned long h_front_porch; - unsigned long v_front_porch; - unsigned long h_sync_width; - unsigned long pixel_clock_khz; - unsigned long h_active; - unsigned long v_active; - unsigned int bpc; //FIXME: review with Jun - struct { - enum dml2_dsc_enable_option enable; - unsigned int dsc_compressed_bpp_x16; - struct { - // for dv to specify num dsc slices to use - unsigned int num_slices; - } overrides; - } dsc; - bool interlaced; - struct { - /* static */ - bool enabled; - unsigned long min_refresh_uhz; - unsigned int max_instant_vtotal_delta; - /* dynamic */ - bool disallowed; - bool drr_active_variable; - bool drr_active_fixed; - } drr_config; - unsigned long vblank_nom; -}; - -struct dml2_link_output_cfg { - enum dml2_output_format_class output_format; - enum dml2_output_encoder_class output_encoder; - unsigned int output_dp_lane_count; - enum dml2_output_link_dp_rate output_dp_link_rate; - unsigned long audio_sample_rate; - unsigned long audio_sample_layout; - bool output_disabled; // The stream does not go to a backend for output to a physical - //connector (e.g. writeback only, phantom pipe) goes to writeback - bool validate_output; // Do not validate the link configuration for this display stream. -}; - -struct dml2_writeback_info { - enum dml2_source_format_class pixel_format; - unsigned long input_width; - unsigned long input_height; - unsigned long output_width; - unsigned long output_height; - unsigned long v_taps; - unsigned long h_taps; - unsigned long v_taps_chroma; - unsigned long h_taps_chroma; - double h_ratio; - double v_ratio; -}; - -struct dml2_writeback_cfg { - unsigned int active_writebacks_per_stream; - struct dml2_writeback_info writeback_stream[DML2_MAX_WRITEBACK]; -}; - -struct dml2_plane_parameters { - unsigned int stream_index; // Identifies which plane will be composed - - enum dml2_source_format_class pixel_format; - /* - * The surface and composition structures use - * the terms plane0 and plane1. These planes - * are expected to hold the following data based - * on the pixel format. - * - * RGB or YUV Non-Planar Types: - * dml2_444_8 - * dml2_444_16 - * dml2_444_32 - * dml2_444_64 - * dml2_rgbe - * - * plane0 = argb or rgbe - * plane1 = not used - * - * YUV Planar-Types: - * dml2_420_8 - * dml2_420_10 - * dml2_420_12 - * - * plane0 = luma - * plane1 = chroma - * - * RGB Planar Types: - * dml2_rgbe_alpha - * - * plane0 = rgbe - * plane1 = alpha - * - * Mono Non-Planar Types: - * dml2_mono_8 - * dml2_mono_16 - * - * plane0 = luma - * plane1 = not used - */ - - struct dml2_surface_cfg surface; - struct dml2_composition_cfg composition; - - struct { - bool enable; - unsigned long lines_before_active_required; - unsigned long transmitted_bytes; - } dynamic_meta_data; - - struct { - unsigned int num_cursors; - unsigned long cursor_width; - unsigned long cursor_bpp; - } cursor; - - // For TDLUT, SW would assume TDLUT is setup and enable all the time and - // budget for worst case addressing/width mode - struct { - bool setup_for_tdlut; - enum dml2_tdlut_addressing_mode tdlut_addressing_mode; - enum dml2_tdlut_width_mode tdlut_width_mode; - bool tdlut_mpc_width_flag; - } tdlut; - - bool immediate_flip; - - struct { - // Logical overrides to power management policies (usually) - enum dml2_uclk_pstate_change_strategy uclk_pstate_change_strategy; - enum dml2_refresh_from_mall_mode_override refresh_from_mall; - unsigned int det_size_override_kb; - unsigned int mpcc_combine_factor; - - // reserved_vblank_time_ns is the minimum time to reserve in vblank for Twait - // The actual reserved vblank time used for the corresponding stream in mode_programming would be at least as much as this per-plane override. - long reserved_vblank_time_ns; - unsigned int max_vactive_det_fill_delay_us; // 0 = no reserved time, +ve = explicit max delay - unsigned int vactive_latency_to_hide_for_pstate_admissibility_us; - unsigned int gpuvm_min_page_size_kbytes; - unsigned int hostvm_min_page_size_kbytes; - - enum dml2_svp_mode_override legacy_svp_config; //TODO remove in favor of svp_config - - struct { - // HW specific overrides, there's almost no reason to mess with these - // generally used for debugging or simulation - bool force_one_row_for_frame; - struct { - bool enable; - bool value; - } force_pte_buffer_mode; - double dppclk_mhz; - } hw; - } overrides; -}; - -struct dml2_stream_parameters { - struct dml2_timing_cfg timing; - struct dml2_link_output_cfg output; - struct dml2_writeback_cfg writeback; - - struct { - enum dml2_odm_mode odm_mode; - bool disable_dynamic_odm; - bool disable_subvp; - int minimum_vblank_idle_requirement_us; - bool minimize_active_latency_hiding; - - struct { - struct { - enum dml2_twait_budgeting_setting uclk_pstate; - enum dml2_twait_budgeting_setting fclk_pstate; - enum dml2_twait_budgeting_setting stutter_enter_exit; - } twait_budgeting; - } hw; - } overrides; -}; - -struct dml2_display_cfg { - bool gpuvm_enable; - bool ffbm_enable; - bool hostvm_enable; - - // Allocate DET proportionally between streams based on pixel rate - // and then allocate proportionally between planes. - bool minimize_det_reallocation; - - unsigned int gpuvm_max_page_table_levels; - unsigned int hostvm_max_non_cached_page_table_levels; - - struct dml2_plane_parameters plane_descriptors[DML2_MAX_PLANES]; - struct dml2_stream_parameters stream_descriptors[DML2_MAX_PLANES]; - - unsigned int num_planes; - unsigned int num_streams; - - struct { - struct { - // HW specific overrides, there's almost no reason to mess with these - // generally used for debugging or simulation - struct { - bool enable; - bool value; - } force_unbounded_requesting; - - struct { - bool enable; - bool value; - } force_nom_det_size_kbytes; - - bool mode_support_check_disable; - bool mcache_admissibility_check_disable; - bool surface_viewport_size_check_disable; - double dlg_ref_clk_mhz; - double dispclk_mhz; - double dcfclk_mhz; - bool optimize_tdlut_scheduling; // TBD: for DV, will set this to 1, to ensure tdlut schedule is calculated based on address/width mode - } hw; - - struct { - bool uclk_pstate_change_disable; - bool fclk_pstate_change_disable; - bool g6_temp_read_pstate_disable; - bool g7_ppt_pstate_disable; - } power_management; - - bool enhanced_prefetch_schedule_acceleration; - bool dcc_programming_assumes_scan_direction_unknown; - bool synchronize_timings; - bool synchronize_ddr_displays_for_uclk_pstate_change; - bool max_outstanding_when_urgent_expected_disable; - bool enable_subvp_implicit_pmo; //enables PMO to switch pipe uclk strategy to subvp, and generate phantom programming - unsigned int best_effort_min_active_latency_hiding_us; - bool all_streams_blanked; - } overrides; -}; - -struct dml2_pipe_configuration_descriptor { - struct { - unsigned int viewport_x_start; - unsigned int viewport_width; - } plane0; - - struct { - unsigned int viewport_x_start; - unsigned int viewport_width; - } plane1; - - bool plane1_enabled; - bool imall_enabled; -}; - -struct dml2_plane_mcache_configuration_descriptor { - const struct dml2_plane_parameters *plane_descriptor; - const struct dml2_mcache_surface_allocation *mcache_allocation; - - struct dml2_pipe_configuration_descriptor pipe_configurations[DML2_MAX_DCN_PIPES]; - char num_pipes; -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h deleted file mode 100644 index 8f624a912e78..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML_TOP_POLICY_TYPES_H__ -#define __DML_TOP_POLICY_TYPES_H__ - -struct dml2_policy_parameters { - unsigned long odm_combine_dispclk_threshold_khz; - unsigned int max_immediate_flip_latency; -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h deleted file mode 100644 index 4a9a0d5a09b7..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h +++ /dev/null @@ -1,215 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML_TOP_SOC_PARAMETER_TYPES_H__ -#define __DML_TOP_SOC_PARAMETER_TYPES_H__ - -#include "dml2_external_lib_deps.h" - -#define DML_MAX_CLK_TABLE_SIZE 20 - -struct dml2_soc_derate_values { - unsigned int dram_derate_percent_pixel; - unsigned int dram_derate_percent_vm; - unsigned int dram_derate_percent_pixel_and_vm; - - unsigned int fclk_derate_percent; - unsigned int dcfclk_derate_percent; -}; - -struct dml2_soc_derates { - struct dml2_soc_derate_values system_active_urgent; - struct dml2_soc_derate_values system_active_average; - struct dml2_soc_derate_values dcn_mall_prefetch_urgent; - struct dml2_soc_derate_values dcn_mall_prefetch_average; - struct dml2_soc_derate_values system_idle_average; -}; - -struct dml2_dcn32x_soc_qos_params { - struct { - unsigned int base_latency_us; - unsigned int base_latency_pixel_vm_us; - unsigned int base_latency_vm_us; - unsigned int scaling_factor_fclk_us; - unsigned int scaling_factor_mhz; - } urgent_latency_us; - - unsigned int loaded_round_trip_latency_fclk_cycles; - unsigned int urgent_out_of_order_return_per_channel_pixel_only_bytes; - unsigned int urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; - unsigned int urgent_out_of_order_return_per_channel_vm_only_bytes; -}; - -struct dml2_dcn4_uclk_dpm_dependent_qos_params { - unsigned long minimum_uclk_khz; - unsigned int urgent_ramp_uclk_cycles; - unsigned int trip_to_memory_uclk_cycles; - unsigned int meta_trip_to_memory_uclk_cycles; - unsigned int maximum_latency_when_urgent_uclk_cycles; - unsigned int average_latency_when_urgent_uclk_cycles; - unsigned int maximum_latency_when_non_urgent_uclk_cycles; - unsigned int average_latency_when_non_urgent_uclk_cycles; -}; - -struct dml2_dcn4x_soc_qos_params { - unsigned int df_qos_response_time_fclk_cycles; - unsigned int max_round_trip_to_furthest_cs_fclk_cycles; - unsigned int mall_overhead_fclk_cycles; - unsigned int meta_trip_adder_fclk_cycles; - unsigned int average_transport_distance_fclk_cycles; - double umc_urgent_ramp_latency_margin; - double umc_max_latency_margin; - double umc_average_latency_margin; - double fabric_max_transport_latency_margin; - double fabric_average_transport_latency_margin; - struct dml2_dcn4_uclk_dpm_dependent_qos_params per_uclk_dpm_params[DML_MAX_CLK_TABLE_SIZE]; -}; - -enum dml2_qos_param_type { - dml2_qos_param_type_dcn3, - dml2_qos_param_type_dcn4x -}; - -struct dml2_soc_qos_parameters { - struct dml2_soc_derates derate_table; - struct { - unsigned int base_latency_us; - unsigned int scaling_factor_us; - unsigned int scaling_factor_mhz; - } writeback; - - union { - struct dml2_dcn32x_soc_qos_params dcn32x; - struct dml2_dcn4x_soc_qos_params dcn4x; - } qos_params; - - enum dml2_qos_param_type qos_type; -}; - -struct dml2_soc_power_management_parameters { - double dram_clk_change_blackout_us; - double dram_clk_change_read_only_us; - double dram_clk_change_write_only_us; - double fclk_change_blackout_us; - double g7_ppt_blackout_us; - double g7_temperature_read_blackout_us; - double stutter_enter_plus_exit_latency_us; - double stutter_exit_latency_us; - double low_power_stutter_enter_plus_exit_latency_us; - double low_power_stutter_exit_latency_us; - double z8_stutter_enter_plus_exit_latency_us; - double z8_stutter_exit_latency_us; - double z8_min_idle_time; - double g6_temp_read_blackout_us[DML_MAX_CLK_TABLE_SIZE]; - double type_b_dram_clk_change_blackout_us; - double type_b_ppt_blackout_us; -}; - -struct dml2_clk_table { - unsigned long clk_values_khz[DML_MAX_CLK_TABLE_SIZE]; - unsigned char num_clk_values; -}; - -struct dml2_dram_params { - unsigned int channel_width_bytes; - unsigned int channel_count; - unsigned int transactions_per_clock; -}; - -struct dml2_soc_state_table { - struct dml2_clk_table uclk; - struct dml2_clk_table fclk; - struct dml2_clk_table dcfclk; - struct dml2_clk_table dispclk; - struct dml2_clk_table dppclk; - struct dml2_clk_table dtbclk; - struct dml2_clk_table phyclk; - struct dml2_clk_table socclk; - struct dml2_clk_table dscclk; - struct dml2_clk_table phyclk_d18; - struct dml2_clk_table phyclk_d32; - - struct dml2_dram_params dram_config; -}; - -struct dml2_soc_vmin_clock_limits { - unsigned long dispclk_khz; - unsigned long dcfclk_khz; -}; - -struct dml2_soc_bb { - struct dml2_soc_state_table clk_table; - struct dml2_soc_qos_parameters qos_parameters; - struct dml2_soc_power_management_parameters power_management_parameters; - struct dml2_soc_vmin_clock_limits vmin_limit; - - double lower_bound_bandwidth_dchub; - double fraction_of_urgent_bandwidth_nominal_target; - double fraction_of_urgent_bandwidth_flip_target; - unsigned int dprefclk_mhz; - unsigned int xtalclk_mhz; - unsigned int pcie_refclk_mhz; - unsigned int dchub_refclk_mhz; - unsigned int mall_allocated_for_dcn_mbytes; - unsigned int max_outstanding_reqs; - unsigned long fabric_datapath_to_dcn_data_return_bytes; - unsigned long return_bus_width_bytes; - unsigned long hostvm_min_page_size_kbytes; - unsigned long gpuvm_min_page_size_kbytes; - double phy_downspread_percent; - double dcn_downspread_percent; - double dispclk_dppclk_vco_speed_mhz; - bool no_dfs; - bool do_urgent_latency_adjustment; - unsigned int mem_word_bytes; - unsigned int num_dcc_mcaches; - unsigned int mcache_size_bytes; - unsigned int mcache_line_size_bytes; - unsigned long max_fclk_for_uclk_dpm_khz; -}; - -struct dml2_ip_capabilities { - unsigned int pipe_count; - unsigned int otg_count; - unsigned int TDLUT_33cube_count; - unsigned int num_dsc; - unsigned int max_num_dp2p0_streams; - unsigned int max_num_hdmi_frl_outputs; - unsigned int max_num_dp2p0_outputs; - unsigned int max_num_wb; - unsigned int rob_buffer_size_kbytes; - unsigned int config_return_buffer_size_in_kbytes; - unsigned int config_return_buffer_segment_size_in_kbytes; - unsigned int meta_fifo_size_in_kentries; - unsigned int compressed_buffer_segment_size_in_kbytes; - unsigned int cursor_buffer_size; - unsigned int max_flip_time_us; - unsigned int max_flip_time_lines; - unsigned int hostvm_mode; - unsigned int subvp_drr_scheduling_margin_us; - unsigned int subvp_prefetch_end_to_mall_start_us; - unsigned int subvp_fw_processing_delay; - unsigned int max_vactive_det_fill_delay_us; - unsigned int ppt_max_allow_delay_ns; - unsigned int temp_read_max_allow_delay_us; - unsigned int dummy_pstate_max_allow_delay_us; - /* FAMS2 delays */ - struct { - unsigned int max_allow_delay_us; - unsigned int scheduling_delay_us; - unsigned int vertical_interrupt_ack_delay_us; // delay to acknowledge vline int - unsigned int allow_programming_delay_us; // time requires to program allow - unsigned int min_allow_width_us; - unsigned int subvp_df_throttle_delay_us; - unsigned int subvp_programming_delay_us; - unsigned int subvp_prefetch_to_mall_delay_us; - unsigned int drr_programming_delay_us; - - unsigned int lock_timeout_us; - unsigned int recovery_timeout_us; - unsigned int flip_programming_delay_us; - } fams2; -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h deleted file mode 100644 index 8646ce5f1c01..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h +++ /dev/null @@ -1,750 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML_TOP_TYPES_H__ -#define __DML_TOP_TYPES_H__ - -#include "dml_top_display_cfg_types.h" -#include "dml_top_soc_parameter_types.h" -#include "dml_top_policy_types.h" -#include "dml_top_dchub_registers.h" - -#include "dmub_cmd.h" - -struct dml2_instance; - -enum dml2_project_id { - dml2_project_invalid = 0, - dml2_project_dcn4x_stage1, - dml2_project_dcn4x_stage2, - dml2_project_dcn4x_stage2_auto_drr_svp, -}; - -enum dml2_pstate_change_support { - dml2_pstate_change_vactive = 0, - dml2_pstate_change_vblank = 1, - dml2_pstate_change_vblank_and_vactive = 2, - dml2_pstate_change_drr = 3, - dml2_pstate_change_mall_svp = 4, - dml2_pstate_change_mall_full_frame = 6, - dml2_pstate_change_unsupported = 7 -}; - -enum dml2_output_type_and_rate__type { - dml2_output_type_unknown = 0, - dml2_output_type_dp = 1, - dml2_output_type_edp = 2, - dml2_output_type_dp2p0 = 3, - dml2_output_type_hdmi = 4, - dml2_output_type_hdmifrl = 5 -}; - -enum dml2_output_type_and_rate__rate { - dml2_output_rate_unknown = 0, - dml2_output_rate_dp_rate_hbr = 1, - dml2_output_rate_dp_rate_hbr2 = 2, - dml2_output_rate_dp_rate_hbr3 = 3, - dml2_output_rate_dp_rate_uhbr10 = 4, - dml2_output_rate_dp_rate_uhbr13p5 = 5, - dml2_output_rate_dp_rate_uhbr20 = 6, - dml2_output_rate_hdmi_rate_3x3 = 7, - dml2_output_rate_hdmi_rate_6x3 = 8, - dml2_output_rate_hdmi_rate_6x4 = 9, - dml2_output_rate_hdmi_rate_8x4 = 10, - dml2_output_rate_hdmi_rate_10x4 = 11, - dml2_output_rate_hdmi_rate_12x4 = 12, - dml2_output_rate_hdmi_rate_16x4 = 13, - dml2_output_rate_hdmi_rate_20x4 = 14 -}; - -struct dml2_pmo_options { - bool disable_vblank; - bool disable_svp; - bool disable_drr_var; - bool disable_drr_clamped; - bool disable_drr_var_when_var_active; - bool disable_drr_clamped_when_var_active; - bool disable_fams2; - bool disable_vactive_det_fill_bw_pad; /* dml2_project_dcn4x_stage2_auto_drr_svp and above only */ - bool disable_dyn_odm; - bool disable_dyn_odm_for_multi_stream; - bool disable_dyn_odm_for_stream_with_svp; - struct dml2_pmo_pstate_strategy *override_strategy_lists[DML2_MAX_PLANES]; - unsigned int num_override_strategies_per_list[DML2_MAX_PLANES]; -}; - -struct dml2_options { - enum dml2_project_id project_id; - struct dml2_pmo_options pmo_options; -}; - -struct dml2_initialize_instance_in_out { - struct dml2_instance *dml2_instance; - struct dml2_options options; - struct dml2_soc_bb soc_bb; - struct dml2_ip_capabilities ip_caps; - - struct { - void *explicit_ip_bb; - unsigned int explicit_ip_bb_size; - } overrides; -}; - -struct dml2_reset_instance_in_out { - struct dml2_instance *dml2_instance; -}; - -struct dml2_check_mode_supported_in_out { - /* - * Inputs - */ - struct dml2_instance *dml2_instance; - const struct dml2_display_cfg *display_config; - - /* - * Outputs - */ - bool is_supported; -}; - -struct dml2_mcache_surface_allocation { - bool valid; - /* - * For iMALL, dedicated mall mcaches are required (sharing of last - * slice possible), for legacy phantom or phantom without return - * the only mall mcaches need to be valid. - */ - bool requires_dedicated_mall_mcache; - - unsigned int num_mcaches_plane0; - unsigned int num_mcaches_plane1; - /* - * A plane is divided into vertical slices of mcaches, - * which wrap on the surface width. - * - * For example, if the surface width is 7680, and split into - * three slices of equal width, the boundary array would contain - * [2560, 5120, 7680] - * - * The assignments are - * 0 = [0 .. 2559] - * 1 = [2560 .. 5119] - * 2 = [5120 .. 7679] - * 0 = [7680 .. INF] - * The final element implicitly is the same as the first, and - * at first seems invalid since it is never referenced (since) - * it is outside the surface. However, its useful when shifting - * (see below). - * - * For any given valid mcache assignment, a shifted version, wrapped - * on the surface width boundary is also assumed to be valid. - * - * For example, shifting [2560, 5120, 7680] by -50 results in - * [2510, 5170, 7630]. - * - * The assignments are now: - * 0 = [0 .. 2509] - * 1 = [2510 .. 5169] - * 2 = [5170 .. 7629] - * 0 = [7630 .. INF] - */ - int mcache_x_offsets_plane0[DML2_MAX_MCACHES + 1]; - int mcache_x_offsets_plane1[DML2_MAX_MCACHES + 1]; - - /* - * Shift grainularity is not necessarily 1 - */ - struct { - int p0; - int p1; - } shift_granularity; - - /* - * MCacheIDs have global scope in the SoC, and they are stored here. - * These IDs are generally not valid until all planes in a display - * configuration have had their mcache requirements calculated. - */ - int global_mcache_ids_plane0[DML2_MAX_MCACHES + 1]; - int global_mcache_ids_plane1[DML2_MAX_MCACHES + 1]; - int global_mcache_ids_mall_plane0[DML2_MAX_MCACHES + 1]; - int global_mcache_ids_mall_plane1[DML2_MAX_MCACHES + 1]; - - /* - * Generally, plane0/1 slices must use a disjoint set of caches - * but in some cases the final segement of the two planes can - * use the same cache. If plane0_plane1 is set, then this is - * allowed. - * - * Similarly, the caches allocated to MALL prefetcher are generally - * disjoint, but if mall_prefetch is set, then the final segment - * between the main and the mall pixel requestor can use the same - * cache. - * - * Note that both bits may be set at the same time. - */ - struct { - bool mall_comb_mcache_p0; - bool mall_comb_mcache_p1; - bool plane0_plane1; - } last_slice_sharing; - - struct { - int meta_row_bytes_plane0; - int meta_row_bytes_plane1; - } informative; -}; - -enum dml2_pstate_type { - dml2_pstate_type_uclk, - dml2_pstate_type_ppt, - dml2_pstate_type_temp_read, - dml2_pstate_type_dummy_pstate, - dml2_pstate_type_count -}; - -enum dml2_pstate_method { - dml2_pstate_method_na = 0, - /* hw exclusive modes */ - dml2_pstate_method_vactive = 1, - dml2_pstate_method_vblank = 2, - dml2_pstate_method_reserved_hw = 5, - /* fw assisted exclusive modes */ - dml2_pstate_method_fw_svp = 6, - dml2_pstate_method_reserved_fw = 10, - /* fw assisted modes requiring drr modulation */ - dml2_pstate_method_fw_vactive_drr = 11, - dml2_pstate_method_fw_vblank_drr = 12, - dml2_pstate_method_fw_svp_drr = 13, - dml2_pstate_method_reserved_fw_drr_clamped = 20, - dml2_pstate_method_fw_drr = 21, - dml2_pstate_method_reserved_fw_drr_var = 22, - dml2_pstate_method_count -}; - -struct dml2_per_plane_programming { - const struct dml2_plane_parameters *plane_descriptor; - - union { - struct { - unsigned long dppclk_khz; - } dcn4x; - } min_clocks; - - struct dml2_mcache_surface_allocation mcache_allocation; - - // If a stream is using automatic or forced odm combine - // and the stream for this plane has num_odms_required > 1 - // num_dpps_required is always equal to num_odms_required for - // ALL planes of the stream - - // If a stream is using odm split, then this value is always 1 - unsigned int num_dpps_required; - - enum dml2_pstate_method uclk_pstate_support_method; - - // MALL size requirements for MALL SS and SubVP - unsigned int surface_size_mall_bytes; - unsigned int svp_size_mall_bytes; - - struct dml2_dchub_per_pipe_register_set *pipe_regs[DML2_MAX_PLANES]; - - struct { - bool valid; - struct dml2_plane_parameters descriptor; - struct dml2_mcache_surface_allocation mcache_allocation; - struct dml2_dchub_per_pipe_register_set *pipe_regs[DML2_MAX_PLANES]; - } phantom_plane; -}; - -union dml2_global_sync_programming { - struct { - unsigned int vstartup_lines; - unsigned int vupdate_offset_pixels; - unsigned int vupdate_vupdate_width_pixels; - unsigned int vready_offset_pixels; - unsigned int pstate_keepout_start_lines; - } dcn4x; -}; - -struct dml2_per_stream_programming { - const struct dml2_stream_parameters *stream_descriptor; - - union { - struct { - unsigned long dscclk_khz; - unsigned long dtbclk_khz; - unsigned long phyclk_khz; - } dcn4x; - } min_clocks; - - union dml2_global_sync_programming global_sync; - - unsigned int num_odms_required; - - enum dml2_pstate_method uclk_pstate_method; - - struct { - bool enabled; - struct dml2_stream_parameters descriptor; - union dml2_global_sync_programming global_sync; - } phantom_stream; - - union dmub_cmd_fams2_config fams2_base_params; - union { - union dmub_cmd_fams2_config fams2_sub_params; - union dmub_fams2_stream_static_sub_state_v2 fams2_sub_params_v2; - }; -}; - -//----------------- -// Mode Support Information -//----------------- - -struct dml2_mode_support_info { - bool ModeIsSupported; //pipe_count = ip_params->max_num_dpp; - ip_caps->otg_count = ip_params->max_num_otg; - ip_caps->num_dsc = ip_params->num_dsc; - ip_caps->max_num_dp2p0_streams = ip_params->max_num_dp2p0_streams; - ip_caps->max_num_dp2p0_outputs = ip_params->max_num_dp2p0_outputs; - ip_caps->max_num_hdmi_frl_outputs = ip_params->max_num_hdmi_frl_outputs; - ip_caps->rob_buffer_size_kbytes = ip_params->rob_buffer_size_kbytes; - ip_caps->config_return_buffer_size_in_kbytes = ip_params->config_return_buffer_size_in_kbytes; - ip_caps->config_return_buffer_segment_size_in_kbytes = ip_params->config_return_buffer_segment_size_in_kbytes; - ip_caps->meta_fifo_size_in_kentries = ip_params->meta_fifo_size_in_kentries; - ip_caps->compressed_buffer_segment_size_in_kbytes = ip_params->compressed_buffer_segment_size_in_kbytes; - ip_caps->cursor_buffer_size = ip_params->cursor_buffer_size; - ip_caps->max_flip_time_us = ip_params->max_flip_time_us; - ip_caps->max_flip_time_lines = ip_params->max_flip_time_lines; - ip_caps->hostvm_mode = ip_params->hostvm_mode; - - // FIXME_STAGE2: cleanup after adding all dv override to ip_caps - ip_caps->subvp_drr_scheduling_margin_us = 100; - ip_caps->subvp_prefetch_end_to_mall_start_us = 15; - ip_caps->subvp_fw_processing_delay = 16; - -} - -static void patch_ip_params_with_ip_caps(struct dml2_core_ip_params *ip_params, const struct dml2_ip_capabilities *ip_caps) -{ - ip_params->max_num_dpp = ip_caps->pipe_count; - ip_params->max_num_otg = ip_caps->otg_count; - ip_params->num_dsc = ip_caps->num_dsc; - ip_params->max_num_dp2p0_streams = ip_caps->max_num_dp2p0_streams; - ip_params->max_num_dp2p0_outputs = ip_caps->max_num_dp2p0_outputs; - ip_params->max_num_hdmi_frl_outputs = ip_caps->max_num_hdmi_frl_outputs; - ip_params->rob_buffer_size_kbytes = ip_caps->rob_buffer_size_kbytes; - ip_params->config_return_buffer_size_in_kbytes = ip_caps->config_return_buffer_size_in_kbytes; - ip_params->config_return_buffer_segment_size_in_kbytes = ip_caps->config_return_buffer_segment_size_in_kbytes; - ip_params->meta_fifo_size_in_kentries = ip_caps->meta_fifo_size_in_kentries; - ip_params->compressed_buffer_segment_size_in_kbytes = ip_caps->compressed_buffer_segment_size_in_kbytes; - ip_params->cursor_buffer_size = ip_caps->cursor_buffer_size; - ip_params->max_flip_time_us = ip_caps->max_flip_time_us; - ip_params->max_flip_time_lines = ip_caps->max_flip_time_lines; - ip_params->hostvm_mode = ip_caps->hostvm_mode; -} - -bool core_dcn4_initialize(struct dml2_core_initialize_in_out *in_out) -{ - struct dml2_core_instance *core = in_out->instance; - - if (!in_out->minimum_clock_table) - return false; - else - core->minimum_clock_table = in_out->minimum_clock_table; - - if (in_out->explicit_ip_bb && in_out->explicit_ip_bb_size > 0) { - memcpy(&core->clean_me_up.mode_lib.ip, in_out->explicit_ip_bb, in_out->explicit_ip_bb_size); - - // FIXME_STAGE2: - // DV still uses stage1 ip_param_st for each variant, need to patch the ip_caps with ip_param info - // Should move DV to use ip_caps but need move more overrides to ip_caps - patch_ip_caps_with_explicit_ip_params(in_out->ip_caps, in_out->explicit_ip_bb); - core->clean_me_up.mode_lib.ip.subvp_pstate_allow_width_us = core_dcn4_ip_caps_base.subvp_pstate_allow_width_us; - core->clean_me_up.mode_lib.ip.subvp_fw_processing_delay_us = core_dcn4_ip_caps_base.subvp_pstate_allow_width_us; - core->clean_me_up.mode_lib.ip.subvp_swath_height_margin_lines = core_dcn4_ip_caps_base.subvp_swath_height_margin_lines; - } else { - memcpy(&core->clean_me_up.mode_lib.ip, &core_dcn4_ip_caps_base, sizeof(struct dml2_core_ip_params)); - patch_ip_params_with_ip_caps(&core->clean_me_up.mode_lib.ip, in_out->ip_caps); - core->clean_me_up.mode_lib.ip.imall_supported = false; - } - - memcpy(&core->clean_me_up.mode_lib.soc, in_out->soc_bb, sizeof(struct dml2_soc_bb)); - memcpy(&core->clean_me_up.mode_lib.ip_caps, in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); - - return true; -} - -static void create_phantom_stream_from_main_stream(struct dml2_stream_parameters *phantom, const struct dml2_stream_parameters *main, - const struct dml2_implicit_svp_meta *meta) -{ - memcpy(phantom, main, sizeof(struct dml2_stream_parameters)); - - phantom->timing.v_total = meta->v_total; - phantom->timing.v_active = meta->v_active; - phantom->timing.v_front_porch = meta->v_front_porch; - phantom->timing.v_blank_end = phantom->timing.v_total - phantom->timing.v_front_porch - phantom->timing.v_active; - phantom->timing.vblank_nom = phantom->timing.v_total - phantom->timing.v_active; - phantom->timing.drr_config.enabled = false; -} - -static void create_phantom_plane_from_main_plane(struct dml2_plane_parameters *phantom, const struct dml2_plane_parameters *main, - const struct dml2_stream_parameters *phantom_stream, int phantom_stream_index, const struct dml2_stream_parameters *main_stream) -{ - memcpy(phantom, main, sizeof(struct dml2_plane_parameters)); - - phantom->stream_index = phantom_stream_index; - phantom->overrides.refresh_from_mall = dml2_refresh_from_mall_mode_override_force_disable; - phantom->overrides.legacy_svp_config = dml2_svp_mode_override_phantom_pipe_no_data_return; - phantom->composition.viewport.plane0.height = (long int unsigned) math_min2(math_ceil2( - (double)main->composition.scaler_info.plane0.v_ratio * (double)phantom_stream->timing.v_active, 16.0), - (double)main->composition.viewport.plane0.height); - phantom->composition.viewport.plane1.height = (long int unsigned) math_min2(math_ceil2( - (double)main->composition.scaler_info.plane1.v_ratio * (double)phantom_stream->timing.v_active, 16.0), - (double)main->composition.viewport.plane1.height); - phantom->immediate_flip = false; - phantom->dynamic_meta_data.enable = false; - phantom->cursor.num_cursors = 0; - phantom->cursor.cursor_width = 0; - phantom->tdlut.setup_for_tdlut = false; -} - -static void expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, - struct dml2_core_scratch *scratch) -{ - unsigned int stream_index, plane_index; - const struct dml2_plane_parameters *main_plane; - const struct dml2_stream_parameters *main_stream; - const struct dml2_stream_parameters *phantom_stream; - - memcpy(svp_expanded_display_cfg, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); - memset(scratch->main_stream_index_from_svp_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); - memset(scratch->svp_stream_index_from_main_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); - memset(scratch->main_plane_index_to_phantom_plane_index, 0, sizeof(int) * DML2_MAX_PLANES); - - if (!display_cfg->display_config.overrides.enable_subvp_implicit_pmo) - return; - - /* disable unbounded requesting for all planes until stage 3 has been performed */ - if (!display_cfg->stage3.performed) { - svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.enable = true; - svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.value = false; - } - // Create the phantom streams - for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { - main_stream = &display_cfg->display_config.stream_descriptors[stream_index]; - scratch->main_stream_index_from_svp_stream_index[stream_index] = stream_index; - scratch->svp_stream_index_from_main_stream_index[stream_index] = stream_index; - - if (display_cfg->stage3.stream_svp_meta[stream_index].valid) { - // Create the phantom stream - create_phantom_stream_from_main_stream(&svp_expanded_display_cfg->stream_descriptors[svp_expanded_display_cfg->num_streams], - main_stream, &display_cfg->stage3.stream_svp_meta[stream_index]); - - // Associate this phantom stream to the main stream - scratch->main_stream_index_from_svp_stream_index[svp_expanded_display_cfg->num_streams] = stream_index; - scratch->svp_stream_index_from_main_stream_index[stream_index] = svp_expanded_display_cfg->num_streams; - - // Increment num streams - svp_expanded_display_cfg->num_streams++; - } - } - - // Create the phantom planes - for (plane_index = 0; plane_index < display_cfg->display_config.num_planes; plane_index++) { - main_plane = &display_cfg->display_config.plane_descriptors[plane_index]; - - if (display_cfg->stage3.stream_svp_meta[main_plane->stream_index].valid) { - main_stream = &display_cfg->display_config.stream_descriptors[main_plane->stream_index]; - phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index]]; - create_phantom_plane_from_main_plane(&svp_expanded_display_cfg->plane_descriptors[svp_expanded_display_cfg->num_planes], - main_plane, phantom_stream, scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index], main_stream); - - // Associate this phantom plane to the main plane - scratch->phantom_plane_index_to_main_plane_index[svp_expanded_display_cfg->num_planes] = plane_index; - scratch->main_plane_index_to_phantom_plane_index[plane_index] = svp_expanded_display_cfg->num_planes; - - // Increment num planes - svp_expanded_display_cfg->num_planes++; - - // Adjust the main plane settings - svp_expanded_display_cfg->plane_descriptors[plane_index].overrides.legacy_svp_config = dml2_svp_mode_override_main_pipe; - } - } -} - -static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_instance *core, const struct display_configuation_with_meta *display_cfg, - const struct dml2_display_cfg *svp_expanded_display_cfg, struct dml2_display_cfg_programming *programming, struct dml2_core_scratch *scratch) -{ - unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index, mcache_index; - unsigned int total_main_mcaches_required = 0; - int total_pipe_regs_copied = 0; - int dml_internal_pipe_index = 0; - const struct dml2_plane_parameters *main_plane; - const struct dml2_plane_parameters *phantom_plane; - const struct dml2_stream_parameters *main_stream; - const struct dml2_stream_parameters *phantom_stream; - - // Copy the unexpanded display config to output - memcpy(&programming->display_config, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); - - // Set the global register values - dml2_core_calcs_get_arb_params(&display_cfg->display_config, &core->clean_me_up.mode_lib, &programming->global_regs.arb_regs); - // Get watermarks uses display config for ref clock override, so it doesn't matter whether we pass the pre or post expansion - // display config - dml2_core_calcs_get_watermarks(&display_cfg->display_config, &core->clean_me_up.mode_lib, &programming->global_regs.wm_regs[0]); - - // Check if FAMS2 is required - if (display_cfg->stage3.performed && display_cfg->stage3.success) { - programming->fams2_required = display_cfg->stage3.fams2_required; - - dml2_core_calcs_get_global_fams2_programming(&core->clean_me_up.mode_lib, display_cfg, &programming->fams2_global_config); - } - - // Only loop over all the main streams (the implicit svp streams will be packed as part of the main stream) - for (stream_index = 0; stream_index < programming->display_config.num_streams; stream_index++) { - main_stream = &svp_expanded_display_cfg->stream_descriptors[stream_index]; - phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[stream_index]]; - - // Set the descriptor - programming->stream_programming[stream_index].stream_descriptor = &programming->display_config.stream_descriptors[stream_index]; - - // Set the odm combine factor - programming->stream_programming[stream_index].num_odms_required = display_cfg->mode_support_result.cfg_support_info.stream_support_info[stream_index].odms_used; - - // Check if the stream has implicit SVP enabled - if (main_stream != phantom_stream) { - // If so, copy the phantom stream descriptor - programming->stream_programming[stream_index].phantom_stream.enabled = true; - memcpy(&programming->stream_programming[stream_index].phantom_stream.descriptor, phantom_stream, sizeof(struct dml2_stream_parameters)); - } else { - programming->stream_programming[stream_index].phantom_stream.enabled = false; - } - - // Due to the way DML indexes data internally, it's easier to populate the rest of the display - // stream programming in the next stage - } - - dml_internal_pipe_index = 0; - total_pipe_regs_copied = 0; - stream_already_populated_mask = 0x0; - - // Loop over all main planes - for (plane_index = 0; plane_index < programming->display_config.num_planes; plane_index++) { - main_plane = &svp_expanded_display_cfg->plane_descriptors[plane_index]; - - // Set the descriptor - programming->plane_programming[plane_index].plane_descriptor = &programming->display_config.plane_descriptors[plane_index]; - - // Set the mpc combine factor - programming->plane_programming[plane_index].num_dpps_required = core->clean_me_up.mode_lib.mp.NoOfDPP[plane_index]; - - // Setup the appropriate p-state strategy - if (display_cfg->stage3.performed && display_cfg->stage3.success) { - programming->plane_programming[plane_index].uclk_pstate_support_method = display_cfg->stage3.pstate_switch_modes[plane_index]; - } else { - programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_na; - } - - dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); - - memcpy(&programming->plane_programming[plane_index].mcache_allocation, - &display_cfg->stage2.mcache_allocations[plane_index], - sizeof(struct dml2_mcache_surface_allocation)); - total_main_mcaches_required += programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane0 + - programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane1 - - (programming->plane_programming[plane_index].mcache_allocation.last_slice_sharing.plane0_plane1 ? 1 : 0); - - for (pipe_offset = 0; pipe_offset < programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { - // Assign storage for this pipe's register values - programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; - memset(programming->plane_programming[plane_index].pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); - total_pipe_regs_copied++; - - // Populate the main plane regs - dml2_core_calcs_get_pipe_regs(svp_expanded_display_cfg, &core->clean_me_up.mode_lib, programming->plane_programming[plane_index].pipe_regs[pipe_offset], dml_internal_pipe_index); - - // Multiple planes can refer to the same stream index, so it's only necessary to populate it once - if (!(stream_already_populated_mask & (0x1 << main_plane->stream_index))) { - dml2_core_calcs_get_stream_programming(&core->clean_me_up.mode_lib, &programming->stream_programming[main_plane->stream_index], dml_internal_pipe_index); - - programming->stream_programming[main_plane->stream_index].uclk_pstate_method = programming->plane_programming[plane_index].uclk_pstate_support_method; - - /* unconditionally populate fams2 params */ - dml2_core_calcs_get_stream_fams2_programming(&core->clean_me_up.mode_lib, - display_cfg, - &programming->stream_programming[main_plane->stream_index].fams2_base_params, - &programming->stream_programming[main_plane->stream_index].fams2_sub_params, - programming->stream_programming[main_plane->stream_index].uclk_pstate_method, - plane_index); - - stream_already_populated_mask |= (0x1 << main_plane->stream_index); - } - dml_internal_pipe_index++; - } - } - - for (plane_index = programming->display_config.num_planes; plane_index < svp_expanded_display_cfg->num_planes; plane_index++) { - phantom_plane = &svp_expanded_display_cfg->plane_descriptors[plane_index]; - main_plane_index = scratch->phantom_plane_index_to_main_plane_index[plane_index]; - main_plane = &svp_expanded_display_cfg->plane_descriptors[main_plane_index]; - - programming->plane_programming[main_plane_index].phantom_plane.valid = true; - memcpy(&programming->plane_programming[main_plane_index].phantom_plane.descriptor, phantom_plane, sizeof(struct dml2_plane_parameters)); - - dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[main_plane_index].svp_size_mall_bytes, dml_internal_pipe_index); - - /* generate mcache allocation, phantoms use identical mcache configuration, but in the MALL set and unique mcache ID's beginning after all main ID's */ - memcpy(&programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation, - &programming->plane_programming[main_plane_index].mcache_allocation, - sizeof(struct dml2_mcache_surface_allocation)); - for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane0; mcache_index++) { - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index] += total_main_mcaches_required; - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane0[mcache_index] = - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index]; - } - for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane1; mcache_index++) { - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index] += total_main_mcaches_required; - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane1[mcache_index] = - programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index]; - } - - for (pipe_offset = 0; pipe_offset < programming->plane_programming[main_plane_index].num_dpps_required; pipe_offset++) { - // Assign storage for this pipe's register values - programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; - memset(programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); - total_pipe_regs_copied++; - - // Populate the phantom plane regs - dml2_core_calcs_get_pipe_regs(svp_expanded_display_cfg, &core->clean_me_up.mode_lib, programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset], dml_internal_pipe_index); - // Populate the phantom stream specific programming - if (!(stream_already_populated_mask & (0x1 << phantom_plane->stream_index))) { - dml2_core_calcs_get_global_sync_programming(&core->clean_me_up.mode_lib, &programming->stream_programming[main_plane->stream_index].phantom_stream.global_sync, dml_internal_pipe_index); - - stream_already_populated_mask |= (0x1 << phantom_plane->stream_index); - } - - dml_internal_pipe_index++; - } - } -} - -bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out) -{ - struct dml2_core_instance *core = (struct dml2_core_instance *)in_out->instance; - struct dml2_core_mode_support_locals *l = &core->scratch.mode_support_locals; - - bool result; - unsigned int i, stream_index, stream_bitmask; - int unsigned odm_count, num_odm_output_segments, dpp_count; - - expand_implict_subvp(in_out->display_cfg, &l->svp_expanded_display_cfg, &core->scratch); - - l->mode_support_ex_params.mode_lib = &core->clean_me_up.mode_lib; - l->mode_support_ex_params.in_display_cfg = &l->svp_expanded_display_cfg; - l->mode_support_ex_params.min_clk_table = in_out->min_clk_table; - l->mode_support_ex_params.min_clk_index = in_out->min_clk_index; - l->mode_support_ex_params.out_evaluation_info = &in_out->mode_support_result.cfg_support_info.clean_me_up.support_info; - - result = dml2_core_calcs_mode_support_ex(&l->mode_support_ex_params); - - in_out->mode_support_result.cfg_support_info.is_supported = result; - - if (result) { - in_out->mode_support_result.global.dispclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDISPCLK * 1000); - in_out->mode_support_result.global.dcfclk_deepsleep_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.dcfclk_deepsleep * 1000); - in_out->mode_support_result.global.socclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.SOCCLK * 1000); - - in_out->mode_support_result.global.fclk_pstate_supported = l->mode_support_ex_params.out_evaluation_info->global_fclk_change_supported; - in_out->mode_support_result.global.uclk_pstate_supported = l->mode_support_ex_params.out_evaluation_info->global_dram_clock_change_supported; - - in_out->mode_support_result.global.active.fclk_khz = (unsigned long)(core->clean_me_up.mode_lib.ms.FabricClock * 1000); - in_out->mode_support_result.global.active.dcfclk_khz = (unsigned long)(core->clean_me_up.mode_lib.ms.DCFCLK * 1000); - - - in_out->mode_support_result.global.svp_prefetch.fclk_khz = (unsigned long)core->clean_me_up.mode_lib.ms.FabricClock * 1000; - in_out->mode_support_result.global.svp_prefetch.dcfclk_khz = (unsigned long)core->clean_me_up.mode_lib.ms.DCFCLK * 1000; - - in_out->mode_support_result.global.active.average_bw_sdp_kbps = 0; - in_out->mode_support_result.global.active.urgent_bw_dram_kbps = 0; - in_out->mode_support_result.global.svp_prefetch.average_bw_sdp_kbps = 0; - in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = 0; - - in_out->mode_support_result.global.active.average_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] * 1000), 1.0); - in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] * 1000), 1.0); - in_out->mode_support_result.global.svp_prefetch.average_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] * 1000), 1.0); - in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] * 1000), 1.0); - - in_out->mode_support_result.global.active.average_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] * 1000), 1.0); - in_out->mode_support_result.global.active.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] * 1000), 1.0); - in_out->mode_support_result.global.svp_prefetch.average_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); - in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); - DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_sdp_kbps); - DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps); - DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_dram_kbps); - DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps); - - for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) { - in_out->mode_support_result.per_plane[i].dppclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDPPCLK[i] * 1000); - } - - stream_bitmask = 0; - for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) { - odm_count = 1; - dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; - num_odm_output_segments = 1; - - switch (l->mode_support_ex_params.out_evaluation_info->ODMMode[i]) { - case dml2_odm_mode_bypass: - odm_count = 1; - dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; - break; - case dml2_odm_mode_combine_2to1: - odm_count = 2; - dpp_count = 2; - break; - case dml2_odm_mode_combine_3to1: - odm_count = 3; - dpp_count = 3; - break; - case dml2_odm_mode_combine_4to1: - odm_count = 4; - dpp_count = 4; - break; - case dml2_odm_mode_split_1to2: - case dml2_odm_mode_mso_1to2: - num_odm_output_segments = 2; - break; - case dml2_odm_mode_mso_1to4: - num_odm_output_segments = 4; - break; - case dml2_odm_mode_auto: - default: - odm_count = 1; - dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; - break; - } - - in_out->mode_support_result.cfg_support_info.plane_support_info[i].dpps_used = dpp_count; - - dml2_core_calcs_get_plane_support_info(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->mode_support_result.cfg_support_info.plane_support_info[i], i); - - stream_index = l->svp_expanded_display_cfg.plane_descriptors[i].stream_index; - - in_out->mode_support_result.per_stream[stream_index].dscclk_khz = (unsigned int)core->clean_me_up.mode_lib.ms.required_dscclk_freq_mhz[i] * 1000; - DML_LOG_VERBOSE("CORE_DCN4::%s: i=%d stream_index=%d, in_out->mode_support_result.per_stream[stream_index].dscclk_khz = %u\n", __func__, i, stream_index, in_out->mode_support_result.per_stream[stream_index].dscclk_khz); - - if (!((stream_bitmask >> stream_index) & 0x1)) { - in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].odms_used = odm_count; - in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].num_odm_output_segments = num_odm_output_segments; - in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].dsc_enable = l->mode_support_ex_params.out_evaluation_info->DSCEnabled[i]; - in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].num_dsc_slices = l->mode_support_ex_params.out_evaluation_info->NumberOfDSCSlices[i]; - dml2_core_calcs_get_stream_support_info(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index], i); - in_out->mode_support_result.per_stream[stream_index].dtbclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDTBCLK[i] * 1000); - stream_bitmask |= 0x1 << stream_index; - } - } - } - - return result; -} - -static int lookup_uclk_dpm_index_by_freq(unsigned long uclk_freq_khz, struct dml2_soc_bb *soc_bb) -{ - int i; - - for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { - if (uclk_freq_khz == soc_bb->clk_table.uclk.clk_values_khz[i]) - return i; - } - return 0; -} - -bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out) -{ - struct dml2_core_instance *core = (struct dml2_core_instance *)in_out->instance; - struct dml2_core_mode_programming_locals *l = &core->scratch.mode_programming_locals; - - bool result = false; - unsigned int pipe_offset; - int dml_internal_pipe_index; - int total_pipe_regs_copied = 0; - int stream_already_populated_mask = 0; - - int main_stream_index; - unsigned int plane_index; - - expand_implict_subvp(in_out->display_cfg, &l->svp_expanded_display_cfg, &core->scratch); - - l->mode_programming_ex_params.mode_lib = &core->clean_me_up.mode_lib; - l->mode_programming_ex_params.in_display_cfg = &l->svp_expanded_display_cfg; - l->mode_programming_ex_params.min_clk_table = in_out->instance->minimum_clock_table; - l->mode_programming_ex_params.cfg_support_info = in_out->cfg_support_info; - l->mode_programming_ex_params.programming = in_out->programming; - l->mode_programming_ex_params.min_clk_index = lookup_uclk_dpm_index_by_freq(in_out->programming->min_clocks.dcn4x.active.uclk_khz, - &core->clean_me_up.mode_lib.soc); - - result = dml2_core_calcs_mode_programming_ex(&l->mode_programming_ex_params); - - if (result) { - // If the input display configuration contains implict SVP, we need to use a special packer - if (in_out->display_cfg->display_config.overrides.enable_subvp_implicit_pmo) { - pack_mode_programming_params_with_implicit_subvp(core, in_out->display_cfg, &l->svp_expanded_display_cfg, in_out->programming, &core->scratch); - } else { - memcpy(&in_out->programming->display_config, in_out->display_cfg, sizeof(struct dml2_display_cfg)); - - dml2_core_calcs_get_arb_params(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->programming->global_regs.arb_regs); - dml2_core_calcs_get_watermarks(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->programming->global_regs.wm_regs[0]); - - dml_internal_pipe_index = 0; - - for (plane_index = 0; plane_index < in_out->programming->display_config.num_planes; plane_index++) { - in_out->programming->plane_programming[plane_index].num_dpps_required = core->clean_me_up.mode_lib.mp.NoOfDPP[plane_index]; - - if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; - else if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe) - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; - else if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; - else { - if (core->clean_me_up.mode_lib.mp.MaxActiveDRAMClockChangeLatencySupported[plane_index] >= core->clean_me_up.mode_lib.soc.power_management_parameters.dram_clk_change_blackout_us) - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_vactive; - else if (core->clean_me_up.mode_lib.mp.TWait[plane_index] >= core->clean_me_up.mode_lib.soc.power_management_parameters.dram_clk_change_blackout_us) - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_vblank; - else - in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_na; - } - - dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &in_out->programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); - - memcpy(&in_out->programming->plane_programming[plane_index].mcache_allocation, - &in_out->display_cfg->stage2.mcache_allocations[plane_index], - sizeof(struct dml2_mcache_surface_allocation)); - - for (pipe_offset = 0; pipe_offset < in_out->programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { - in_out->programming->plane_programming[plane_index].plane_descriptor = &in_out->programming->display_config.plane_descriptors[plane_index]; - - // Assign storage for this pipe's register values - in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &in_out->programming->pipe_regs[total_pipe_regs_copied]; - memset(in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); - total_pipe_regs_copied++; - - // Populate - dml2_core_calcs_get_pipe_regs(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset], dml_internal_pipe_index); - - main_stream_index = in_out->programming->display_config.plane_descriptors[plane_index].stream_index; - - // Multiple planes can refer to the same stream index, so it's only necessary to populate it once - if (!(stream_already_populated_mask & (0x1 << main_stream_index))) { - in_out->programming->stream_programming[main_stream_index].stream_descriptor = &in_out->programming->display_config.stream_descriptors[main_stream_index]; - in_out->programming->stream_programming[main_stream_index].num_odms_required = in_out->cfg_support_info->stream_support_info[main_stream_index].odms_used; - dml2_core_calcs_get_stream_programming(&core->clean_me_up.mode_lib, &in_out->programming->stream_programming[main_stream_index], dml_internal_pipe_index); - - stream_already_populated_mask |= (0x1 << main_stream_index); - } - dml_internal_pipe_index++; - } - } - } - } - - return result; -} - -bool core_dcn4_populate_informative(struct dml2_core_populate_informative_in_out *in_out) -{ - struct dml2_core_internal_display_mode_lib *mode_lib = &in_out->instance->clean_me_up.mode_lib; - - if (in_out->mode_is_supported) - in_out->programming->informative.voltage_level = in_out->instance->scratch.mode_programming_locals.mode_programming_ex_params.min_clk_index; - else - in_out->programming->informative.voltage_level = in_out->instance->scratch.mode_support_locals.mode_support_ex_params.min_clk_index; - - dml2_core_calcs_get_informative(mode_lib, in_out->programming); - return true; -} - -bool core_dcn4_calculate_mcache_allocation(struct dml2_calculate_mcache_allocation_in_out *in_out) -{ - memset(in_out->mcache_allocation, 0, sizeof(struct dml2_mcache_surface_allocation)); - - dml2_core_calcs_get_mcache_allocation(&in_out->instance->clean_me_up.mode_lib, in_out->mcache_allocation, in_out->plane_index); - - if (in_out->mcache_allocation->num_mcaches_plane0 > 0) - in_out->mcache_allocation->mcache_x_offsets_plane0[in_out->mcache_allocation->num_mcaches_plane0 - 1] = in_out->plane_descriptor->surface.plane0.width; - - if (in_out->mcache_allocation->num_mcaches_plane1 > 0) - in_out->mcache_allocation->mcache_x_offsets_plane1[in_out->mcache_allocation->num_mcaches_plane1 - 1] = in_out->plane_descriptor->surface.plane1.width; - - in_out->mcache_allocation->requires_dedicated_mall_mcache = false; - - return true; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h deleted file mode 100644 index a68bb001a346..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h +++ /dev/null @@ -1,12 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_CORE_DCN4_H__ -#define __DML2_CORE_DCN4_H__ -bool core_dcn4_initialize(struct dml2_core_initialize_in_out *in_out); -bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out); -bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out); -bool core_dcn4_populate_informative(struct dml2_core_populate_informative_in_out *in_out); -bool core_dcn4_calculate_mcache_allocation(struct dml2_calculate_mcache_allocation_in_out *in_out); -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c deleted file mode 100644 index f16792f79bef..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ /dev/null @@ -1,13339 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - - -#include "dml2_internal_shared_types.h" -#include "dml2_core_dcn4_calcs.h" -#include "dml2_debug.h" -#include "lib_float_math.h" -#include "dml_top_types.h" - -#define DML2_MAX_FMT_420_BUFFER_WIDTH 4096 -#define DML_MAX_NUM_OF_SLICES_PER_DSC 4 -#define DML_MAX_COMPRESSION_RATIO 4 -//#define DML_MODE_SUPPORT_USE_DPM_DRAM_BW -//#define DML_GLOBAL_PREFETCH_CHECK -#define ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE -#define DML_MAX_VSTARTUP_START 1023 - -const char *dml2_core_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type) -{ - switch (bw_type) { - case (dml2_core_internal_bw_sdp): - return("dml2_core_internal_bw_sdp"); - case (dml2_core_internal_bw_dram): - return("dml2_core_internal_bw_dram"); - case (dml2_core_internal_bw_max): - return("dml2_core_internal_bw_max"); - default: - return("dml2_core_internal_bw_unknown"); - } -} - -const char *dml2_core_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type) -{ - switch (dml2_core_internal_soc_state_type) { - case (dml2_core_internal_soc_state_sys_idle): - return("dml2_core_internal_soc_state_sys_idle"); - case (dml2_core_internal_soc_state_sys_active): - return("dml2_core_internal_soc_state_sys_active"); - case (dml2_core_internal_soc_state_svp_prefetch): - return("dml2_core_internal_soc_state_svp_prefetch"); - case dml2_core_internal_soc_state_max: - default: - return("dml2_core_internal_soc_state_unknown"); - } -} - -static double dml2_core_div_rem(double dividend, unsigned int divisor, unsigned int *remainder) -{ - *remainder = ((dividend / divisor) - (int)(dividend / divisor) > 0); - return dividend / divisor; -} - -static void dml2_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) -{ - DML_LOG_VERBOSE("DML: ===================================== \n"); - DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); - if (!fail_only || support->ScaleRatioAndTapsSupport == 0) - DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); - if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) - DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); - if (!fail_only || support->ViewportSizeSupport == 0) - DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); - if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) - DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); - if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) - DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); - if (!fail_only || support->BPPForMultistreamNotIndicated == 1) - DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); - if (!fail_only || support->MultistreamWithHDMIOreDP == 1) - DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); - if (!fail_only || support->ExceededMultistreamSlots == 1) - DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); - if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) - DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); - if (!fail_only || support->NotEnoughLanesForMSO == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); - if (!fail_only || support->P2IWith420 == 1) - DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); - if (!fail_only || support->DSC422NativeNotSupported == 1) - DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); - if (!fail_only || support->DSCSlicesODMModeSupported == 0) - DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); - if (!fail_only || support->NotEnoughDSCUnits == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); - if (!fail_only || support->NotEnoughDSCSlices == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); - if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) - DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); - if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) - DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); - if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) - DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); - if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) - DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); - if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) - DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); - if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) - DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); - if (!fail_only || support->ROBSupport == 0) - DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); - if (!fail_only || support->OutstandingRequestsSupport == 0) - DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); - if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) - DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); - if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) - DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); - if (!fail_only || support->TotalAvailablePipesSupport == 0) - DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); - if (!fail_only || support->NumberOfOTGSupport == 0) - DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); - if (!fail_only || support->NumberOfHDMIFRLSupport == 0) - DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); - if (!fail_only || support->NumberOfDP2p0Support == 0) - DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); - if (!fail_only || support->EnoughWritebackUnits == 0) - DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); - if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) - DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); - if (!fail_only || support->WritebackLatencySupport == 0) - DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); - if (!fail_only || support->CursorSupport == 0) - DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); - if (!fail_only || support->PitchSupport == 0) - DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); - if (!fail_only || support->ViewportExceedsSurface == 1) - DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); - if (!fail_only || support->PrefetchSupported == 0) - DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); - if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) - DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); - if (!fail_only || support->AvgBandwidthSupport == 0) - DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); - if (!fail_only || support->DynamicMetadataSupported == 0) - DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); - if (!fail_only || support->VRatioInPrefetchSupported == 0) - DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); - if (!fail_only || support->PTEBufferSizeNotExceeded == 0) - DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); - if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) - DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); - if (!fail_only || support->ExceededMALLSize == 1) - DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); - if (!fail_only || support->g6_temp_read_support == 0) - DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); - if (!fail_only || support->ImmediateFlipSupport == 0) - DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); - if (!fail_only || support->LinkCapacitySupport == 0) - DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); - - if (!fail_only || support->ModeSupport == 0) - DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); - DML_LOG_VERBOSE("DML: ===================================== \n"); -} - -static void get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg) -{ - for (unsigned int k = 0; k < display_cfg->num_planes; k++) { - double bpc = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.bpc; - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_disable) { - switch (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format) { - case dml2_444: - out_bpp[k] = bpc * 3; - break; - case dml2_s422: - out_bpp[k] = bpc * 2; - break; - case dml2_n422: - out_bpp[k] = bpc * 2; - break; - case dml2_420: - default: - out_bpp[k] = bpc * 1.5; - break; - } - } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable) { - out_bpp[k] = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.dsc_compressed_bpp_x16 / 16; - } else { - out_bpp[k] = 0; - } - DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); - DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); - DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); - } -} - -static unsigned int dml_round_to_multiple(unsigned int num, unsigned int multiple, bool up) -{ - unsigned int remainder; - - if (multiple == 0) - return num; - - remainder = num % multiple; - if (remainder == 0) - return num; - - if (up) - return (num + multiple - remainder); - else - return (num - remainder); -} - -static unsigned int dml_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info) -{ - unsigned int num_active_pipes = 0; - - for (unsigned int k = 0; k < num_planes; k++) { - num_active_pipes = num_active_pipes + (unsigned int)cfg_support_info->plane_support_info[k].dpps_used; - } - - DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); - return num_active_pipes; -} - -static void dml_calc_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane) -{ - unsigned int pipe_idx = 0; - - for (unsigned int k = 0; k < DML2_MAX_PLANES; ++k) { - pipe_plane[k] = __DML2_CALCS_PIPE_NO_PLANE__; - } - - for (unsigned int plane_idx = 0; plane_idx < DML2_MAX_PLANES; plane_idx++) { - for (int i = 0; i < cfg_support_info->plane_support_info[plane_idx].dpps_used; i++) { - pipe_plane[pipe_idx] = plane_idx; - pipe_idx++; - } - } -} - -static bool dml_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg) -{ - bool is_phantom = false; - - if (plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe || - plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) { - is_phantom = true; - } - - return is_phantom; -} - -static bool dml_get_is_phantom_pipe(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) -{ - unsigned int plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; - - bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_idx]); - DML_LOG_VERBOSE("DML::%s: pipe_idx=%d legacy_svp_config=%0d is_phantom=%d\n", __func__, pipe_idx, display_cfg->plane_descriptors[plane_idx].overrides.legacy_svp_config, is_phantom); - return is_phantom; -} - -#define dml_get_per_pipe_var_func(variable, type, interval_var) static type dml_get_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) \ -{ \ -unsigned int plane_idx; \ -plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; \ -return (type) interval_var[plane_idx]; \ -} - -dml_get_per_pipe_var_func(dpte_group_size_in_bytes, unsigned int, mode_lib->mp.dpte_group_bytes); -dml_get_per_pipe_var_func(vm_group_size_in_bytes, unsigned int, mode_lib->mp.vm_group_bytes); -dml_get_per_pipe_var_func(swath_height_l, unsigned int, mode_lib->mp.SwathHeightY); -dml_get_per_pipe_var_func(swath_height_c, unsigned int, mode_lib->mp.SwathHeightC); -dml_get_per_pipe_var_func(dpte_row_height_linear_l, unsigned int, mode_lib->mp.dpte_row_height_linear); -dml_get_per_pipe_var_func(dpte_row_height_linear_c, unsigned int, mode_lib->mp.dpte_row_height_linear_chroma); - -dml_get_per_pipe_var_func(vstartup_calculated, unsigned int, mode_lib->mp.VStartup); -dml_get_per_pipe_var_func(vupdate_offset, unsigned int, mode_lib->mp.VUpdateOffsetPix); -dml_get_per_pipe_var_func(vupdate_width, unsigned int, mode_lib->mp.VUpdateWidthPix); -dml_get_per_pipe_var_func(vready_offset, unsigned int, mode_lib->mp.VReadyOffsetPix); -dml_get_per_pipe_var_func(pstate_keepout_dst_lines, unsigned int, mode_lib->mp.pstate_keepout_dst_lines); -dml_get_per_pipe_var_func(det_stored_buffer_size_l_bytes, unsigned int, mode_lib->mp.DETBufferSizeY); -dml_get_per_pipe_var_func(det_stored_buffer_size_c_bytes, unsigned int, mode_lib->mp.DETBufferSizeC); -dml_get_per_pipe_var_func(det_buffer_size_kbytes, unsigned int, mode_lib->mp.DETBufferSizeInKByte); -dml_get_per_pipe_var_func(surface_size_in_mall_bytes, unsigned int, mode_lib->mp.SurfaceSizeInTheMALL); - -#define dml_get_per_plane_var_func(variable, type, interval_var) static type dml_get_plane_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int plane_idx) \ -{ \ -return (type) interval_var[plane_idx]; \ -} - -dml_get_per_plane_var_func(num_mcaches_plane0, unsigned int, mode_lib->ms.num_mcaches_l); -dml_get_per_plane_var_func(mcache_row_bytes_plane0, unsigned int, mode_lib->ms.mcache_row_bytes_l); -dml_get_per_plane_var_func(mcache_shift_granularity_plane0, unsigned int, mode_lib->ms.mcache_shift_granularity_l); -dml_get_per_plane_var_func(num_mcaches_plane1, unsigned int, mode_lib->ms.num_mcaches_c); -dml_get_per_plane_var_func(mcache_row_bytes_plane1, unsigned int, mode_lib->ms.mcache_row_bytes_c); -dml_get_per_plane_var_func(mcache_shift_granularity_plane1, unsigned int, mode_lib->ms.mcache_shift_granularity_c); -dml_get_per_plane_var_func(mall_comb_mcache_l, unsigned int, mode_lib->ms.mall_comb_mcache_l); -dml_get_per_plane_var_func(mall_comb_mcache_c, unsigned int, mode_lib->ms.mall_comb_mcache_c); -dml_get_per_plane_var_func(lc_comb_mcache, unsigned int, mode_lib->ms.lc_comb_mcache); -dml_get_per_plane_var_func(subviewport_lines_needed_in_mall, unsigned int, mode_lib->ms.SubViewportLinesNeededInMALL); -dml_get_per_plane_var_func(max_vstartup_lines, unsigned int, mode_lib->ms.MaxVStartupLines); - -#define dml_get_per_plane_array_var_func(variable, type, interval_var) static type dml_get_plane_array_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int plane_idx, unsigned int array_idx) \ -{ \ -return (type) interval_var[plane_idx][array_idx]; \ -} - -dml_get_per_plane_array_var_func(mcache_offsets_plane0, unsigned int, mode_lib->ms.mcache_offsets_l); -dml_get_per_plane_array_var_func(mcache_offsets_plane1, unsigned int, mode_lib->ms.mcache_offsets_c); - -#define dml_get_var_func(var, type, internal_var) static type dml_get_##var(const struct dml2_core_internal_display_mode_lib *mode_lib) \ -{ \ -return (type) internal_var; \ -} - -dml_get_var_func(wm_urgent, double, mode_lib->mp.Watermark.UrgentWatermark); -dml_get_var_func(wm_stutter_exit, double, mode_lib->mp.Watermark.StutterExitWatermark); -dml_get_var_func(wm_stutter_enter_exit, double, mode_lib->mp.Watermark.StutterEnterPlusExitWatermark); -dml_get_var_func(wm_z8_stutter_exit, double, mode_lib->mp.Watermark.Z8StutterExitWatermark); -dml_get_var_func(wm_z8_stutter_enter_exit, double, mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark); -dml_get_var_func(wm_memory_trip, double, mode_lib->mp.UrgentLatency); -dml_get_var_func(meta_trip_memory_us, double, mode_lib->mp.MetaTripToMemory); - -dml_get_var_func(wm_fclk_change, double, mode_lib->mp.Watermark.FCLKChangeWatermark); -dml_get_var_func(wm_usr_retraining, double, mode_lib->mp.Watermark.USRRetrainingWatermark); -dml_get_var_func(wm_temp_read_or_ppt, double, mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us); -dml_get_var_func(wm_dram_clock_change, double, mode_lib->mp.Watermark.DRAMClockChangeWatermark); -dml_get_var_func(fraction_of_urgent_bandwidth, double, mode_lib->mp.FractionOfUrgentBandwidth); -dml_get_var_func(fraction_of_urgent_bandwidth_imm_flip, double, mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip); -dml_get_var_func(fraction_of_urgent_bandwidth_mall, double, mode_lib->mp.FractionOfUrgentBandwidthMALL); -dml_get_var_func(wm_writeback_dram_clock_change, double, mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); -dml_get_var_func(wm_writeback_fclk_change, double, mode_lib->mp.Watermark.WritebackFCLKChangeWatermark); -dml_get_var_func(stutter_efficiency, double, mode_lib->mp.StutterEfficiency); -dml_get_var_func(stutter_efficiency_no_vblank, double, mode_lib->mp.StutterEfficiencyNotIncludingVBlank); -dml_get_var_func(stutter_num_bursts, double, mode_lib->mp.NumberOfStutterBurstsPerFrame); -dml_get_var_func(stutter_efficiency_z8, double, mode_lib->mp.Z8StutterEfficiency); -dml_get_var_func(stutter_num_bursts_z8, double, mode_lib->mp.Z8NumberOfStutterBurstsPerFrame); -dml_get_var_func(stutter_period, double, mode_lib->mp.StutterPeriod); -dml_get_var_func(stutter_efficiency_z8_bestcase, double, mode_lib->mp.Z8StutterEfficiencyBestCase); -dml_get_var_func(stutter_num_bursts_z8_bestcase, double, mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase); -dml_get_var_func(stutter_period_bestcase, double, mode_lib->mp.StutterPeriodBestCase); -dml_get_var_func(fclk_change_latency, double, mode_lib->mp.MaxActiveFCLKChangeLatencySupported); -dml_get_var_func(global_dppclk_khz, double, mode_lib->mp.GlobalDPPCLK * 1000.0); - -dml_get_var_func(sys_active_avg_bw_required_sdp, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_avg_bw_required_dram, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); - -dml_get_var_func(svp_prefetch_avg_bw_required_sdp, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_avg_bw_required_dram, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(sys_active_avg_bw_available_sdp, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_avg_bw_available_dram, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); - -dml_get_var_func(svp_prefetch_avg_bw_available_sdp, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_avg_bw_available_dram, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(sys_active_urg_bw_available_sdp, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_urg_bw_available_dram, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); -dml_get_var_func(sys_active_urg_bw_available_dram_vm_only, double, mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); - -dml_get_var_func(svp_prefetch_urg_bw_available_sdp, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_urg_bw_available_dram, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); -dml_get_var_func(svp_prefetch_urg_bw_available_dram_vm_only, double, mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_svp_prefetch]); - -dml_get_var_func(urgent_latency, double, mode_lib->mp.UrgentLatency); -dml_get_var_func(max_urgent_latency_us, double, mode_lib->ms.support.max_urgent_latency_us); -dml_get_var_func(max_non_urgent_latency_us, double, mode_lib->ms.support.max_non_urgent_latency_us); -dml_get_var_func(avg_non_urgent_latency_us, double, mode_lib->ms.support.avg_non_urgent_latency_us); -dml_get_var_func(avg_urgent_latency_us, double, mode_lib->ms.support.avg_urgent_latency_us); - -dml_get_var_func(sys_active_urg_bw_required_sdp, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_urg_bw_required_dram, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); -dml_get_var_func(svp_prefetch_urg_bw_required_sdp, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_urg_bw_required_dram, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(sys_active_non_urg_required_sdp, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_non_urg_required_dram, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); -dml_get_var_func(svp_prefetch_non_urg_bw_required_sdp, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_non_urg_bw_required_dram, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(sys_active_urg_bw_required_sdp_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_urg_bw_required_dram_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); -dml_get_var_func(svp_prefetch_urg_bw_required_sdp_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_urg_bw_required_dram_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(sys_active_non_urg_required_sdp_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); -dml_get_var_func(sys_active_non_urg_required_dram_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); -dml_get_var_func(svp_prefetch_non_urg_bw_required_sdp_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); -dml_get_var_func(svp_prefetch_non_urg_bw_required_dram_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); - -dml_get_var_func(comp_buffer_size_kbytes, unsigned int, mode_lib->mp.CompressedBufferSizeInkByte); - -dml_get_var_func(unbounded_request_enabled, bool, mode_lib->mp.UnboundedRequestEnabled); -dml_get_var_func(wm_writeback_urgent, double, mode_lib->mp.Watermark.WritebackUrgentWatermark); -dml_get_var_func(cstate_max_cap_mode, bool, mode_lib->mp.DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); -dml_get_var_func(compbuf_reserved_space_64b, unsigned int, mode_lib->mp.compbuf_reserved_space_64b); -dml_get_var_func(hw_debug5, bool, mode_lib->mp.hw_debug5); -dml_get_var_func(dcfclk_deep_sleep_hysteresis, unsigned int, mode_lib->mp.dcfclk_deep_sleep_hysteresis); - -static void CalculateMaxDETAndMinCompressedBufferSize( - unsigned int ConfigReturnBufferSizeInKByte, - unsigned int ConfigReturnBufferSegmentSizeInKByte, - unsigned int ROBBufferSizeInKByte, - unsigned int MaxNumDPP, - unsigned int nomDETInKByteOverrideEnable, // VBA_DELTA, allow DV to override default DET size - unsigned int nomDETInKByteOverrideValue, // VBA_DELTA - bool is_mrq_present, - - // Output - unsigned int *MaxTotalDETInKByte, - unsigned int *nomDETInKByte, - unsigned int *MinCompressedBufferSizeInKByte) -{ - if (is_mrq_present) - *MaxTotalDETInKByte = (unsigned int) math_ceil2((double)(ConfigReturnBufferSizeInKByte + ROBBufferSizeInKByte)*4/5, 64); - else - *MaxTotalDETInKByte = ConfigReturnBufferSizeInKByte - ConfigReturnBufferSegmentSizeInKByte; - - *nomDETInKByte = (unsigned int)(math_floor2((double)*MaxTotalDETInKByte / (double)MaxNumDPP, ConfigReturnBufferSegmentSizeInKByte)); - *MinCompressedBufferSizeInKByte = ConfigReturnBufferSizeInKByte - *MaxTotalDETInKByte; - - DML_LOG_VERBOSE("DML::%s: is_mrq_present = %u\n", __func__, is_mrq_present); - DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); - DML_LOG_VERBOSE("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); - DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); - DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); - DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); - - if (nomDETInKByteOverrideEnable) { - *nomDETInKByte = nomDETInKByteOverrideValue; - DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); - } -} - -static void PixelClockAdjustmentForProgressiveToInterlaceUnit(const struct dml2_display_cfg *display_cfg, bool ptoi_supported, double *PixelClockBackEnd) -{ - //unsigned int num_active_planes = display_cfg->num_planes; - - //Progressive To Interlace Unit Effect - for (unsigned int k = 0; k < display_cfg->num_planes; ++k) { - PixelClockBackEnd[k] = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced == 1 && ptoi_supported == true) { - // FIXME_STAGE2... can sw pass the pixel rate for interlaced directly - //display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz = 2 * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz; - } - } -} - -static bool dml_is_420(enum dml2_source_format_class source_format) -{ - bool val = false; - - switch (source_format) { - case dml2_444_8: - val = 0; - break; - case dml2_444_16: - val = 0; - break; - case dml2_444_32: - val = 0; - break; - case dml2_444_64: - val = 0; - break; - case dml2_420_8: - val = 1; - break; - case dml2_420_10: - val = 1; - break; - case dml2_420_12: - val = 1; - break; - case dml2_422_planar_8: - val = 0; - break; - case dml2_422_planar_10: - val = 0; - break; - case dml2_422_planar_12: - val = 0; - break; - case dml2_422_packed_8: - val = 0; - break; - case dml2_422_packed_10: - val = 0; - break; - case dml2_422_packed_12: - val = 0; - break; - case dml2_rgbe_alpha: - val = 0; - break; - case dml2_rgbe: - val = 0; - break; - case dml2_mono_8: - val = 0; - break; - case dml2_mono_16: - val = 0; - break; - default: - DML_ASSERT(0); - break; - } - return val; -} - -static unsigned int dml_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode) -{ - if (sw_mode == dml2_sw_linear) - return 256; - else if (sw_mode == dml2_sw_256b_2d) - return 256; - else if (sw_mode == dml2_sw_4kb_2d) - return 4096; - else if (sw_mode == dml2_sw_64kb_2d) - return 65536; - else if (sw_mode == dml2_sw_256kb_2d) - return 262144; - else if (sw_mode == dml2_gfx11_sw_linear) - return 256; - else if (sw_mode == dml2_gfx11_sw_64kb_d) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_d_t) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_d_x) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_r_x) - return 65536; - else if (sw_mode == dml2_gfx11_sw_256kb_d_x) - return 262144; - else if (sw_mode == dml2_gfx11_sw_256kb_r_x) - return 262144; - else { - DML_ASSERT(0); - return 256; - } -} - -static bool dml_is_vertical_rotation(enum dml2_rotation_angle Scan) -{ - bool is_vert = false; - if (Scan == dml2_rotation_90 || Scan == dml2_rotation_270) { - is_vert = true; - } else { - is_vert = false; - } - return is_vert; -} - -static int unsigned dml_get_gfx_version(enum dml2_swizzle_mode sw_mode) -{ - int unsigned version = 0; - - if (sw_mode == dml2_sw_linear || - sw_mode == dml2_sw_256b_2d || - sw_mode == dml2_sw_4kb_2d || - sw_mode == dml2_sw_64kb_2d || - sw_mode == dml2_sw_256kb_2d) { - version = 12; - } else if (sw_mode == dml2_gfx11_sw_linear || - sw_mode == dml2_gfx11_sw_64kb_d || - sw_mode == dml2_gfx11_sw_64kb_d_t || - sw_mode == dml2_gfx11_sw_64kb_d_x || - sw_mode == dml2_gfx11_sw_64kb_r_x || - sw_mode == dml2_gfx11_sw_256kb_d_x || - sw_mode == dml2_gfx11_sw_256kb_r_x) { - version = 11; - } else { - DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); - DML_ASSERT(0); - } - - return version; -} - -static void CalculateBytePerPixelAndBlockSizes( - enum dml2_source_format_class SourcePixelFormat, - enum dml2_swizzle_mode SurfaceTiling, - unsigned int pitch_y, - unsigned int pitch_c, - - // Output - unsigned int *BytePerPixelY, - unsigned int *BytePerPixelC, - double *BytePerPixelDETY, - double *BytePerPixelDETC, - unsigned int *BlockHeight256BytesY, - unsigned int *BlockHeight256BytesC, - unsigned int *BlockWidth256BytesY, - unsigned int *BlockWidth256BytesC, - unsigned int *MacroTileHeightY, - unsigned int *MacroTileHeightC, - unsigned int *MacroTileWidthY, - unsigned int *MacroTileWidthC, - bool *surf_linear128_l, - bool *surf_linear128_c) -{ - *BytePerPixelDETY = 0; - *BytePerPixelDETC = 0; - *BytePerPixelY = 1; - *BytePerPixelC = 1; - - if (SourcePixelFormat == dml2_444_64) { - *BytePerPixelDETY = 8; - *BytePerPixelDETC = 0; - *BytePerPixelY = 8; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml2_444_32 || SourcePixelFormat == dml2_rgbe) { - *BytePerPixelDETY = 4; - *BytePerPixelDETC = 0; - *BytePerPixelY = 4; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml2_444_16 || SourcePixelFormat == dml2_mono_16) { - *BytePerPixelDETY = 2; - *BytePerPixelDETC = 0; - *BytePerPixelY = 2; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml2_444_8 || SourcePixelFormat == dml2_mono_8) { - *BytePerPixelDETY = 1; - *BytePerPixelDETC = 0; - *BytePerPixelY = 1; - *BytePerPixelC = 0; - } else if (SourcePixelFormat == dml2_rgbe_alpha) { - *BytePerPixelDETY = 4; - *BytePerPixelDETC = 1; - *BytePerPixelY = 4; - *BytePerPixelC = 1; - } else if (SourcePixelFormat == dml2_420_8) { - *BytePerPixelDETY = 1; - *BytePerPixelDETC = 2; - *BytePerPixelY = 1; - *BytePerPixelC = 2; - } else if (SourcePixelFormat == dml2_420_12) { - *BytePerPixelDETY = 2; - *BytePerPixelDETC = 4; - *BytePerPixelY = 2; - *BytePerPixelC = 4; - } else if (SourcePixelFormat == dml2_420_10) { - *BytePerPixelDETY = (double)(4.0 / 3); - *BytePerPixelDETC = (double)(8.0 / 3); - *BytePerPixelY = 2; - *BytePerPixelC = 4; - } else { - DML_LOG_VERBOSE("ERROR: DML::%s: SourcePixelFormat = %u not supported!\n", __func__, SourcePixelFormat); - DML_ASSERT(0); - } - - DML_LOG_VERBOSE("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); - DML_LOG_VERBOSE("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); - DML_LOG_VERBOSE("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); - DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); - DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); - DML_LOG_VERBOSE("DML::%s: pitch_y = %u\n", __func__, pitch_y); - DML_LOG_VERBOSE("DML::%s: pitch_c = %u\n", __func__, pitch_c); - DML_LOG_VERBOSE("DML::%s: surf_linear128_l = %u\n", __func__, *surf_linear128_l); - DML_LOG_VERBOSE("DML::%s: surf_linear128_c = %u\n", __func__, *surf_linear128_c); - - if (dml_get_gfx_version(SurfaceTiling) == 11) { - *surf_linear128_l = 0; - *surf_linear128_c = 0; - } else { - if (SurfaceTiling == dml2_sw_linear) { - *surf_linear128_l = (((pitch_y * *BytePerPixelY) % 256) != 0); - - if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) - *surf_linear128_c = (((pitch_c * *BytePerPixelC) % 256) != 0); - } - } - - if (!(dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha)) { - if (SurfaceTiling == dml2_sw_linear) { - *BlockHeight256BytesY = 1; - } else if (SourcePixelFormat == dml2_444_64) { - *BlockHeight256BytesY = 4; - } else if (SourcePixelFormat == dml2_444_8) { - *BlockHeight256BytesY = 16; - } else { - *BlockHeight256BytesY = 8; - } - *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; - *BlockHeight256BytesC = 0; - *BlockWidth256BytesC = 0; - } else { // dual plane - if (SurfaceTiling == dml2_sw_linear) { - *BlockHeight256BytesY = 1; - *BlockHeight256BytesC = 1; - } else if (SourcePixelFormat == dml2_rgbe_alpha) { - *BlockHeight256BytesY = 8; - *BlockHeight256BytesC = 16; - } else if (SourcePixelFormat == dml2_420_8) { - *BlockHeight256BytesY = 16; - *BlockHeight256BytesC = 8; - } else { - *BlockHeight256BytesY = 8; - *BlockHeight256BytesC = 8; - } - *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; - *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; - } - DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); - DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); - DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); - DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); - - if (dml_get_gfx_version(SurfaceTiling) == 11) { - if (SurfaceTiling == dml2_gfx11_sw_linear) { - *MacroTileHeightY = *BlockHeight256BytesY; - *MacroTileWidthY = 256 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 256 / *BytePerPixelC / *MacroTileHeightC; - } - } else if (SurfaceTiling == dml2_gfx11_sw_64kb_d || SurfaceTiling == dml2_gfx11_sw_64kb_d_t || SurfaceTiling == dml2_gfx11_sw_64kb_d_x || SurfaceTiling == dml2_gfx11_sw_64kb_r_x) { - *MacroTileHeightY = 16 * *BlockHeight256BytesY; - *MacroTileWidthY = 65536 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = 16 * *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 65536 / *BytePerPixelC / *MacroTileHeightC; - } - } else { - *MacroTileHeightY = 32 * *BlockHeight256BytesY; - *MacroTileWidthY = 65536 * 4 / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = 32 * *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = 65536 * 4 / *BytePerPixelC / *MacroTileHeightC; - } - } - } else { - unsigned int macro_tile_size_bytes = dml_get_tile_block_size_bytes(SurfaceTiling); - unsigned int macro_tile_scale = 1; // macro tile to 256B req scaling - - if (SurfaceTiling == dml2_sw_linear) { - macro_tile_scale = 1; - } else if (SurfaceTiling == dml2_sw_4kb_2d) { - macro_tile_scale = 4; - } else if (SurfaceTiling == dml2_sw_64kb_2d) { - macro_tile_scale = 16; - } else if (SurfaceTiling == dml2_sw_256kb_2d) { - macro_tile_scale = 32; - } else { - DML_LOG_VERBOSE("ERROR: Invalid SurfaceTiling setting! val=%u\n", SurfaceTiling); - DML_ASSERT(0); - } - - *MacroTileHeightY = macro_tile_scale * *BlockHeight256BytesY; - *MacroTileWidthY = macro_tile_size_bytes / *BytePerPixelY / *MacroTileHeightY; - *MacroTileHeightC = macro_tile_scale * *BlockHeight256BytesC; - if (*MacroTileHeightC == 0) { - *MacroTileWidthC = 0; - } else { - *MacroTileWidthC = macro_tile_size_bytes / *BytePerPixelC / *MacroTileHeightC; - } - } - - DML_LOG_VERBOSE("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); - DML_LOG_VERBOSE("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); - DML_LOG_VERBOSE("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); - DML_LOG_VERBOSE("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); -} - -static void CalculateSinglePipeDPPCLKAndSCLThroughput( - double HRatio, - double HRatioChroma, - double VRatio, - double VRatioChroma, - double MaxDCHUBToPSCLThroughput, - double MaxPSCLToLBThroughput, - double PixelClock, - enum dml2_source_format_class SourcePixelFormat, - unsigned int HTaps, - unsigned int HTapsChroma, - unsigned int VTaps, - unsigned int VTapsChroma, - - // Output - double *PSCL_THROUGHPUT, - double *PSCL_THROUGHPUT_CHROMA, - double *DPPCLKUsingSingleDPP) -{ - double DPPCLKUsingSingleDPPLuma; - double DPPCLKUsingSingleDPPChroma; - - if (HRatio > 1) { - *PSCL_THROUGHPUT = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatio / math_ceil2((double)HTaps / 6.0, 1.0)); - } else { - *PSCL_THROUGHPUT = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); - } - - DPPCLKUsingSingleDPPLuma = PixelClock * math_max3(VTaps / 6 * math_min2(1, HRatio), HRatio * VRatio / *PSCL_THROUGHPUT, 1); - - if ((HTaps > 6 || VTaps > 6) && DPPCLKUsingSingleDPPLuma < 2 * PixelClock) - DPPCLKUsingSingleDPPLuma = 2 * PixelClock; - - if (!dml_is_420(SourcePixelFormat) && SourcePixelFormat != dml2_rgbe_alpha) { - *PSCL_THROUGHPUT_CHROMA = 0; - *DPPCLKUsingSingleDPP = DPPCLKUsingSingleDPPLuma; - } else { - if (HRatioChroma > 1) { - *PSCL_THROUGHPUT_CHROMA = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatioChroma / math_ceil2((double)HTapsChroma / 6.0, 1.0)); - } else { - *PSCL_THROUGHPUT_CHROMA = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); - } - DPPCLKUsingSingleDPPChroma = PixelClock * math_max3(VTapsChroma / 6 * math_min2(1, HRatioChroma), - HRatioChroma * VRatioChroma / *PSCL_THROUGHPUT_CHROMA, 1); - if ((HTapsChroma > 6 || VTapsChroma > 6) && DPPCLKUsingSingleDPPChroma < 2 * PixelClock) - DPPCLKUsingSingleDPPChroma = 2 * PixelClock; - *DPPCLKUsingSingleDPP = math_max2(DPPCLKUsingSingleDPPLuma, DPPCLKUsingSingleDPPChroma); - } -} - -static void CalculateSwathWidth( - const struct dml2_display_cfg *display_cfg, - bool ForceSingleDPP, - unsigned int NumberOfActiveSurfaces, - enum dml2_odm_mode ODMMode[], - unsigned int BytePerPixY[], - unsigned int BytePerPixC[], - unsigned int Read256BytesBlockHeightY[], - unsigned int Read256BytesBlockHeightC[], - unsigned int Read256BytesBlockWidthY[], - unsigned int Read256BytesBlockWidthC[], - bool surf_linear128_l[], - bool surf_linear128_c[], - unsigned int DPPPerSurface[], - - // Output - unsigned int req_per_swath_ub_l[], - unsigned int req_per_swath_ub_c[], - unsigned int SwathWidthSingleDPPY[], // post-rotated plane width - unsigned int SwathWidthSingleDPPC[], - unsigned int SwathWidthY[], // per-pipe - unsigned int SwathWidthC[], // per-pipe - unsigned int MaximumSwathHeightY[], - unsigned int MaximumSwathHeightC[], - unsigned int swath_width_luma_ub[], // per-pipe - unsigned int swath_width_chroma_ub[]) // per-pipe -{ - enum dml2_odm_mode MainSurfaceODMMode; - double odm_hactive_factor = 1.0; - unsigned int req_width_horz_y; - unsigned int req_width_horz_c; - unsigned int surface_width_ub_l; - unsigned int surface_height_ub_l; - unsigned int surface_width_ub_c; - unsigned int surface_height_ub_c; - - DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { - SwathWidthSingleDPPY[k] = (unsigned int)display_cfg->plane_descriptors[k].composition.viewport.plane0.width; - } else { - SwathWidthSingleDPPY[k] = (unsigned int)display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - } - - DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); - DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); - DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); - - MainSurfaceODMMode = ODMMode[k]; - - if (ForceSingleDPP) { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - } else { - if (MainSurfaceODMMode == dml2_odm_mode_combine_4to1) - odm_hactive_factor = 4.0; - else if (MainSurfaceODMMode == dml2_odm_mode_combine_3to1) - odm_hactive_factor = 3.0; - else if (MainSurfaceODMMode == dml2_odm_mode_combine_2to1) - odm_hactive_factor = 2.0; - - if (MainSurfaceODMMode == dml2_odm_mode_combine_4to1 || MainSurfaceODMMode == dml2_odm_mode_combine_3to1 || MainSurfaceODMMode == dml2_odm_mode_combine_2to1) { - SwathWidthY[k] = (unsigned int)(math_min2((double)SwathWidthSingleDPPY[k], math_round((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active / odm_hactive_factor * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio))); - } else if (DPPPerSurface[k] == 2) { - SwathWidthY[k] = SwathWidthSingleDPPY[k] / 2; - } else { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - } - } - - DML_LOG_VERBOSE("DML::%s: k=%u HActive=%lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active); - DML_LOG_VERBOSE("DML::%s: k=%u HRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); - DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); - - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { - SwathWidthC[k] = SwathWidthY[k] / 2; - SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k] / 2; - } else { - SwathWidthC[k] = SwathWidthY[k]; - SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k]; - } - - if (ForceSingleDPP == true) { - SwathWidthY[k] = SwathWidthSingleDPPY[k]; - SwathWidthC[k] = SwathWidthSingleDPPC[k]; - } - - req_width_horz_y = Read256BytesBlockWidthY[k]; - req_width_horz_c = Read256BytesBlockWidthC[k]; - - if (surf_linear128_l[k]) - req_width_horz_y = req_width_horz_y / 2; - - if (surf_linear128_c[k]) - req_width_horz_c = req_width_horz_c / 2; - - surface_width_ub_l = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane0.width, req_width_horz_y); - surface_height_ub_l = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane0.height, Read256BytesBlockHeightY[k]); - surface_width_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.width, req_width_horz_c); - surface_height_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.height, Read256BytesBlockHeightC[k]); - - DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); - DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); - DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); - DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); - DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); - DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); - DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); - DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); - DML_LOG_VERBOSE("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.stationary); - DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); - - req_per_swath_ub_l[k] = 0; - req_per_swath_ub_c[k] = 0; - if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { - MaximumSwathHeightY[k] = Read256BytesBlockHeightY[k]; - MaximumSwathHeightC[k] = Read256BytesBlockHeightC[k]; - if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { - swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_width_ub_l, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start + SwathWidthY[k] + req_width_horz_y - 1, req_width_horz_y) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start, req_width_horz_y))); - } else { - swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_width_ub_l, math_ceil2((double)SwathWidthY[k] - 1, req_width_horz_y) + req_width_horz_y)); - } - req_per_swath_ub_l[k] = swath_width_luma_ub[k] / req_width_horz_y; - - if (BytePerPixC[k] > 0) { - if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { - swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_width_ub_c, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start + SwathWidthC[k] + req_width_horz_c - 1, req_width_horz_c) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start, req_width_horz_c))); - } else { - swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_width_ub_c, math_ceil2((double)SwathWidthC[k] - 1, req_width_horz_c) + req_width_horz_c)); - } - req_per_swath_ub_c[k] = swath_width_chroma_ub[k] / req_width_horz_c; - } else { - swath_width_chroma_ub[k] = 0; - } - } else { - MaximumSwathHeightY[k] = Read256BytesBlockWidthY[k]; - MaximumSwathHeightC[k] = Read256BytesBlockWidthC[k]; - - if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { - swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_height_ub_l, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start + SwathWidthY[k] + Read256BytesBlockHeightY[k] - 1, Read256BytesBlockHeightY[k]) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start, Read256BytesBlockHeightY[k]))); - } else { - swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_height_ub_l, math_ceil2((double)SwathWidthY[k] - 1, Read256BytesBlockHeightY[k]) + Read256BytesBlockHeightY[k])); - } - req_per_swath_ub_l[k] = swath_width_luma_ub[k] / Read256BytesBlockHeightY[k]; - if (BytePerPixC[k] > 0) { - if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { - swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_height_ub_c, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start + SwathWidthC[k] + Read256BytesBlockHeightC[k] - 1, Read256BytesBlockHeightC[k]) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start, Read256BytesBlockHeightC[k]))); - } else { - swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_height_ub_c, math_ceil2((double)SwathWidthC[k] - 1, Read256BytesBlockHeightC[k]) + Read256BytesBlockHeightC[k])); - } - req_per_swath_ub_c[k] = swath_width_chroma_ub[k] / Read256BytesBlockHeightC[k]; - } else { - swath_width_chroma_ub[k] = 0; - } - } - - DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_l=%u\n", __func__, k, req_per_swath_ub_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_c=%u\n", __func__, k, req_per_swath_ub_c[k]); - } -} - -static bool UnboundedRequest(bool unb_req_force_en, bool unb_req_force_val, unsigned int TotalNumberOfActiveDPP, bool NoChromaOrLinear) -{ - bool unb_req_ok = false; - bool unb_req_en = false; - - unb_req_ok = (TotalNumberOfActiveDPP == 1 && NoChromaOrLinear); - unb_req_en = unb_req_ok; - - if (unb_req_force_en) { - unb_req_en = unb_req_force_val && unb_req_ok; - } - DML_LOG_VERBOSE("DML::%s: unb_req_force_en = %u\n", __func__, unb_req_force_en); - DML_LOG_VERBOSE("DML::%s: unb_req_force_val = %u\n", __func__, unb_req_force_val); - DML_LOG_VERBOSE("DML::%s: unb_req_ok = %u\n", __func__, unb_req_ok); - DML_LOG_VERBOSE("DML::%s: unb_req_en = %u\n", __func__, unb_req_en); - return unb_req_en; -} - -static void CalculateDETBufferSize( - struct dml2_core_shared_CalculateDETBufferSize_locals *l, - const struct dml2_display_cfg *display_cfg, - bool ForceSingleDPP, - unsigned int NumberOfActiveSurfaces, - bool UnboundedRequestEnabled, - unsigned int nomDETInKByte, - unsigned int MaxTotalDETInKByte, - unsigned int ConfigReturnBufferSizeInKByte, - unsigned int MinCompressedBufferSizeInKByte, - unsigned int ConfigReturnBufferSegmentSizeInkByte, - unsigned int CompressedBufferSegmentSizeInkByte, - double ReadBandwidthLuma[], - double ReadBandwidthChroma[], - unsigned int full_swath_bytes_l[], - unsigned int full_swath_bytes_c[], - unsigned int DPPPerSurface[], - // Output - unsigned int DETBufferSizeInKByte[], - unsigned int *CompressedBufferSizeInkByte) -{ - memset(l, 0, sizeof(struct dml2_core_shared_CalculateDETBufferSize_locals)); - - bool DETPieceAssignedToThisSurfaceAlready[DML2_MAX_PLANES]; - bool NextPotentialSurfaceToAssignDETPieceFound; - bool MinimizeReallocationSuccess = false; - - DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); - DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); - DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); - DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); - DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); - DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); - DML_LOG_VERBOSE("DML::%s: CompressedBufferSegmentSizeInkByte = %u\n", __func__, CompressedBufferSegmentSizeInkByte); - - // Note: Will use default det size if that fits 2 swaths - if (UnboundedRequestEnabled) { - if (display_cfg->plane_descriptors[0].overrides.det_size_override_kb > 0) { - DETBufferSizeInKByte[0] = display_cfg->plane_descriptors[0].overrides.det_size_override_kb; - } else { - DETBufferSizeInKByte[0] = (unsigned int)math_max2(128.0, math_ceil2(2.0 * ((double)full_swath_bytes_l[0] + (double)full_swath_bytes_c[0]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte)); - } - *CompressedBufferSizeInkByte = ConfigReturnBufferSizeInKByte - DETBufferSizeInKByte[0]; - } else { - l->DETBufferSizePoolInKByte = MaxTotalDETInKByte; - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - DETBufferSizeInKByte[k] = 0; - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { - l->max_minDET = nomDETInKByte - ConfigReturnBufferSegmentSizeInkByte; - } else { - l->max_minDET = nomDETInKByte; - } - l->minDET = 128; - l->minDET_pipe = 0; - - // add DET resource until can hold 2 full swaths - while (l->minDET <= l->max_minDET && l->minDET_pipe == 0) { - if (2.0 * ((double)full_swath_bytes_l[k] + (double)full_swath_bytes_c[k]) / 1024.0 <= l->minDET) - l->minDET_pipe = l->minDET; - l->minDET = l->minDET + ConfigReturnBufferSegmentSizeInkByte; - } - - DML_LOG_VERBOSE("DML::%s: k=%u minDET = %u\n", __func__, k, l->minDET); - DML_LOG_VERBOSE("DML::%s: k=%u max_minDET = %u\n", __func__, k, l->max_minDET); - DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, l->minDET_pipe); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, full_swath_bytes_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, full_swath_bytes_c[k]); - - if (l->minDET_pipe == 0) { - l->minDET_pipe = (unsigned int)(math_max2(128, math_ceil2(((double)full_swath_bytes_l[k] + (double)full_swath_bytes_c[k]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte))); - DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, l->minDET_pipe); - } - - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - DETBufferSizeInKByte[k] = 0; - } else if (display_cfg->plane_descriptors[k].overrides.det_size_override_kb > 0) { - DETBufferSizeInKByte[k] = display_cfg->plane_descriptors[k].overrides.det_size_override_kb; - l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * display_cfg->plane_descriptors[k].overrides.det_size_override_kb; - } else if ((ForceSingleDPP ? 1 : DPPPerSurface[k]) * l->minDET_pipe <= l->DETBufferSizePoolInKByte) { - DETBufferSizeInKByte[k] = l->minDET_pipe; - l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * l->minDET_pipe; - } - - DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); - DML_LOG_VERBOSE("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.det_size_override_kb); - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); - DML_LOG_VERBOSE("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, l->DETBufferSizePoolInKByte); - } - - if (display_cfg->minimize_det_reallocation) { - MinimizeReallocationSuccess = true; - // To minimize det reallocation, we don't distribute based on each surfaces bandwidth proportional to the global - // but rather distribute DET across streams proportionally based on pixel rate, and only distribute based on - // bandwidth between the planes on the same stream. This ensures that large scale re-distribution only on a - // stream count and/or pixel rate change, which is must less likely then general bandwidth changes per plane. - - // Calculate total pixel rate - for (unsigned int k = 0; k < display_cfg->num_streams; ++k) { - l->TotalPixelRate += display_cfg->stream_descriptors[k].timing.pixel_clock_khz; - } - - // Calculate per stream DET budget - for (unsigned int k = 0; k < display_cfg->num_streams; ++k) { - l->DETBudgetPerStream[k] = (unsigned int)((double) display_cfg->stream_descriptors[k].timing.pixel_clock_khz * MaxTotalDETInKByte / l->TotalPixelRate); - l->RemainingDETBudgetPerStream[k] = l->DETBudgetPerStream[k]; - } - - // Calculate the per stream total bandwidth - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - l->TotalBandwidthPerStream[display_cfg->plane_descriptors[k].stream_index] += (unsigned int)(ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); - - // Check the minimum can be satisfied by budget - if (l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] >= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k])) { - l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] -= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k]); - } else { - MinimizeReallocationSuccess = false; - break; - } - } - } - - if (MinimizeReallocationSuccess) { - // Since a fixed budget per stream is sufficient to satisfy the minimums, just re-distribute each streams - // budget proportionally across its planes - l->ResidualDETAfterRounding = MaxTotalDETInKByte; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - l->IdealDETBudget = (unsigned int)(((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / l->TotalBandwidthPerStream[display_cfg->plane_descriptors[k].stream_index]) - * l->DETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]); - - if (l->IdealDETBudget > DETBufferSizeInKByte[k]) { - l->DeltaDETBudget = l->IdealDETBudget - DETBufferSizeInKByte[k]; - if (l->DeltaDETBudget > l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]) - l->DeltaDETBudget = l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]; - - /* split the additional budgeted DET among the pipes per plane */ - DETBufferSizeInKByte[k] += (unsigned int)((double)l->DeltaDETBudget / (ForceSingleDPP ? 1 : DPPPerSurface[k])); - l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] -= l->DeltaDETBudget; - } - - // Round down to segment size - DETBufferSizeInKByte[k] = (DETBufferSizeInKByte[k] / ConfigReturnBufferSegmentSizeInkByte) * ConfigReturnBufferSegmentSizeInkByte; - - l->ResidualDETAfterRounding -= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k]); - } - } - } - } - - if (!MinimizeReallocationSuccess) { - l->TotalBandwidth = 0; - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - l->TotalBandwidth = l->TotalBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; - } - } - DML_LOG_VERBOSE("DML::%s: --- Before bandwidth adjustment ---\n", __func__); - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); - } - DML_LOG_VERBOSE("DML::%s: --- DET allocation with bandwidth ---\n", __func__); - DML_LOG_VERBOSE("DML::%s: TotalBandwidth = %f\n", __func__, l->TotalBandwidth); - l->BandwidthOfSurfacesNotAssignedDETPiece = l->TotalBandwidth; - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - DETPieceAssignedToThisSurfaceAlready[k] = true; - } else if (display_cfg->plane_descriptors[k].overrides.det_size_override_kb > 0 || (((double)(ForceSingleDPP ? 1 : DPPPerSurface[k]) * (double)DETBufferSizeInKByte[k] / (double)MaxTotalDETInKByte) >= ((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / l->TotalBandwidth))) { - DETPieceAssignedToThisSurfaceAlready[k] = true; - l->BandwidthOfSurfacesNotAssignedDETPiece = l->BandwidthOfSurfacesNotAssignedDETPiece - ReadBandwidthLuma[k] - ReadBandwidthChroma[k]; - } else { - DETPieceAssignedToThisSurfaceAlready[k] = false; - } - DML_LOG_VERBOSE("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); - DML_LOG_VERBOSE("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, l->BandwidthOfSurfacesNotAssignedDETPiece); - } - - for (unsigned int j = 0; j < NumberOfActiveSurfaces; ++j) { - NextPotentialSurfaceToAssignDETPieceFound = false; - l->NextSurfaceToAssignDETPiece = 0; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, l->NextSurfaceToAssignDETPiece); - if (!DETPieceAssignedToThisSurfaceAlready[k] && (!NextPotentialSurfaceToAssignDETPieceFound || - ReadBandwidthLuma[k] + ReadBandwidthChroma[k] < ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece])) { - l->NextSurfaceToAssignDETPiece = k; - NextPotentialSurfaceToAssignDETPieceFound = true; - } - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); - DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); - } - - if (NextPotentialSurfaceToAssignDETPieceFound) { - l->NextDETBufferPieceInKByte = (unsigned int)(math_min2( - math_round((double)l->DETBufferSizePoolInKByte * (ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]) / l->BandwidthOfSurfacesNotAssignedDETPiece / - ((ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte)) - * (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte, - math_floor2((double)l->DETBufferSizePoolInKByte, (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte))); - - DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, l->DETBufferSizePoolInKByte); - DML_LOG_VERBOSE("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, l->NextSurfaceToAssignDETPiece); - DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); - DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); - DML_LOG_VERBOSE("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, l->BandwidthOfSurfacesNotAssignedDETPiece); - DML_LOG_VERBOSE("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, l->NextDETBufferPieceInKByte); - DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, l->NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); - - DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] = DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] + l->NextDETBufferPieceInKByte / (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]); - DML_LOG_VERBOSE("to %u\n", DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); - - l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - l->NextDETBufferPieceInKByte; - DETPieceAssignedToThisSurfaceAlready[l->NextSurfaceToAssignDETPiece] = true; - l->BandwidthOfSurfacesNotAssignedDETPiece = l->BandwidthOfSurfacesNotAssignedDETPiece - (ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); - } - } - } - *CompressedBufferSizeInkByte = MinCompressedBufferSizeInKByte; - } - *CompressedBufferSizeInkByte = *CompressedBufferSizeInkByte * CompressedBufferSegmentSizeInkByte / ConfigReturnBufferSegmentSizeInkByte; - - DML_LOG_VERBOSE("DML::%s: --- After bandwidth adjustment ---\n", __func__); - DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); - } -} - -static double CalculateRequiredDispclk( - enum dml2_odm_mode ODMMode, - double PixelClock, - bool isTMDS420) -{ - double DispClk; - - if (ODMMode == dml2_odm_mode_combine_4to1) { - DispClk = PixelClock / 4.0; - } else if (ODMMode == dml2_odm_mode_combine_3to1) { - DispClk = PixelClock / 3.0; - } else if (ODMMode == dml2_odm_mode_combine_2to1) { - DispClk = PixelClock / 2.0; - } else { - DispClk = PixelClock; - } - - if (isTMDS420) { - double TMDS420MinPixClock = PixelClock / 2.0; - DispClk = math_max2(DispClk, TMDS420MinPixClock); - } - - return DispClk; -} - -static double TruncToValidBPP( - struct dml2_core_shared_TruncToValidBPP_locals *l, - double LinkBitRate, - unsigned int Lanes, - unsigned int HTotal, - unsigned int HActive, - double PixelClock, - double DesiredBPP, - bool DSCEnable, - enum dml2_output_encoder_class Output, - enum dml2_output_format_class Format, - unsigned int DSCInputBitPerComponent, - unsigned int DSCSlices, - unsigned int AudioRate, - unsigned int AudioLayout, - enum dml2_odm_mode ODMModeNoDSC, - enum dml2_odm_mode ODMModeDSC, - - // Output - unsigned int *RequiredSlots) -{ - double MaxLinkBPP; - unsigned int MinDSCBPP; - double MaxDSCBPP; - unsigned int NonDSCBPP0; - unsigned int NonDSCBPP1; - unsigned int NonDSCBPP2; - enum dml2_odm_mode ODMMode; - - if (Format == dml2_420) { - NonDSCBPP0 = 12; - NonDSCBPP1 = 15; - NonDSCBPP2 = 18; - MinDSCBPP = 6; - MaxDSCBPP = 16; - } else if (Format == dml2_444) { - NonDSCBPP0 = 24; - NonDSCBPP1 = 30; - NonDSCBPP2 = 36; - MinDSCBPP = 8; - MaxDSCBPP = 16; - } else { - if (Output == dml2_hdmi || Output == dml2_hdmifrl) { - NonDSCBPP0 = 24; - NonDSCBPP1 = 24; - NonDSCBPP2 = 24; - } else { - NonDSCBPP0 = 16; - NonDSCBPP1 = 20; - NonDSCBPP2 = 24; - } - if (Format == dml2_n422 || Output == dml2_hdmifrl) { - MinDSCBPP = 7; - MaxDSCBPP = 16; - } else { - MinDSCBPP = 8; - MaxDSCBPP = 16; - } - } - if (Output == dml2_dp2p0) { - MaxLinkBPP = LinkBitRate * Lanes / PixelClock * 128.0 / 132.0 * 383.0 / 384.0 * 65536.0 / 65540.0; - } else if (DSCEnable && Output == dml2_dp) { - MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock * (1 - 2.4 / 100); - } else { - MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock; - } - - ODMMode = DSCEnable ? ODMModeDSC : ODMModeNoDSC; - - if (ODMMode == dml2_odm_mode_split_1to2) { - MaxLinkBPP = 2 * MaxLinkBPP; - } - - if (DesiredBPP == 0) { - if (DSCEnable) { - if (MaxLinkBPP < MinDSCBPP) { - return __DML2_CALCS_DPP_INVALID__; - } else if (MaxLinkBPP >= MaxDSCBPP) { - return MaxDSCBPP; - } else { - return math_floor2(16.0 * MaxLinkBPP, 1.0) / 16.0; - } - } else { - if (MaxLinkBPP >= NonDSCBPP2) { - return NonDSCBPP2; - } else if (MaxLinkBPP >= NonDSCBPP1) { - return NonDSCBPP1; - } else if (MaxLinkBPP >= NonDSCBPP0) { - return NonDSCBPP0; - } else { - return __DML2_CALCS_DPP_INVALID__; - } - } - } else { - if (!((DSCEnable == false && (DesiredBPP == NonDSCBPP2 || DesiredBPP == NonDSCBPP1 || DesiredBPP == NonDSCBPP0)) || - (DSCEnable && DesiredBPP >= MinDSCBPP && DesiredBPP <= MaxDSCBPP))) { - return __DML2_CALCS_DPP_INVALID__; - } else { - return DesiredBPP; - } - } -} - -// updated for dcn4 -static unsigned int dscceComputeDelay( - unsigned int bpc, - double BPP, - unsigned int sliceWidth, - unsigned int numSlices, - enum dml2_output_format_class pixelFormat, - enum dml2_output_encoder_class Output) -{ - // valid bpc = source bits per component in the set of {8, 10, 12} - // valid bpp = increments of 1/16 of a bit - // min = 6/7/8 in N420/N422/444, respectively - // max = such that compression is 1:1 - //valid sliceWidth = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode) - //valid numSlices = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4} - //valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420} - - // fixed value - unsigned int rcModelSize = 8192; - - // N422/N420 operate at 2 pixels per clock - unsigned int pixelsPerClock, padding_pixels, ssm_group_priming_delay, ssm_pipeline_delay, obsm_pipeline_delay, slice_padded_pixels, ixd_plus_padding, ixd_plus_padding_groups, cycles_per_group, group_delay, pipeline_delay, pixels, additional_group_delay, lines_to_reach_ixd, groups_to_reach_ixd, slice_width_groups, initial_xmit_delay, number_of_lines_to_reach_ixd, slice_width_modified; - - if (pixelFormat == dml2_420) - pixelsPerClock = 2; - // #all other modes operate at 1 pixel per clock - else if (pixelFormat == dml2_444) - pixelsPerClock = 1; - else if (pixelFormat == dml2_n422 || Output == dml2_hdmifrl) - pixelsPerClock = 2; - else - pixelsPerClock = 1; - - //initial transmit delay as per PPS - initial_xmit_delay = (unsigned int)(math_round(rcModelSize / 2.0 / BPP / pixelsPerClock)); - - //slice width as seen by dscc_bcl in pixels or pixels pairs (depending on number of pixels per pixel container based on pixel format) - slice_width_modified = (pixelFormat == dml2_444 || pixelFormat == dml2_420 || Output == dml2_hdmifrl) ? sliceWidth / 2 : sliceWidth; - - padding_pixels = ((slice_width_modified % 3) != 0) ? (3 - (slice_width_modified % 3)) * (initial_xmit_delay / slice_width_modified) : 0; - - if ((3.0 * pixelsPerClock * BPP) >= ((double)((initial_xmit_delay + 2) / 3) * (double)(3 + (pixelFormat == dml2_n422)))) { - if ((initial_xmit_delay + padding_pixels) % 3 == 1) { - initial_xmit_delay++; - } - } - - //sub-stream multiplexer balance fifo priming delay in groups as per dsc standard - if (bpc == 8) - ssm_group_priming_delay = 83; - else if (bpc == 10) - ssm_group_priming_delay = 91; - else if (bpc == 12) - ssm_group_priming_delay = 115; - else if (bpc == 14) - ssm_group_priming_delay = 123; - else - ssm_group_priming_delay = 128; - - //slice width in groups is rounded up to the nearest group as DSC adds padded pixels such that there are an integer number of groups per slice - slice_width_groups = (slice_width_modified + 2) / 3; - - //determine number of padded pixels in the last group of a slice line, computed as - slice_padded_pixels = 3 * slice_width_groups - slice_width_modified; - - //determine integer number of complete slice lines required to reach initial transmit delay without ssm delay considered - number_of_lines_to_reach_ixd = initial_xmit_delay / slice_width_modified; - - //increase initial transmit delay by the number of padded pixels added to a slice line multipled by the integer number of complete lines to reach initial transmit delay - //this step is necessary as each padded pixel added takes up a clock cycle and, therefore, adds to the overall delay - ixd_plus_padding = initial_xmit_delay + slice_padded_pixels * number_of_lines_to_reach_ixd; - - //convert the padded initial transmit delay from pixels to groups by rounding up to the nearest group as DSC processes in groups of pixels - ixd_plus_padding_groups = (ixd_plus_padding + 2) / 3; - - //number of groups required for a slice to reach initial transmit delay is the sum of the padded initial transmit delay plus the ssm group priming delay - groups_to_reach_ixd = ixd_plus_padding_groups + ssm_group_priming_delay; - - //number of lines required to reach padded initial transmit delay in groups in slices to the left of the last horizontal slice - //needs to be rounded up as a complete slice lines are buffered prior to initial transmit delay being reached in the last horizontal slice - lines_to_reach_ixd = (groups_to_reach_ixd + slice_width_groups - 1) / slice_width_groups; //round up lines to reach ixd to next - - //determine if there are non-zero number of pixels reached in the group where initial transmit delay is reached - //an additional group time (i.e., 3 pixel times) is required before the first output if there are no additional pixels beyond initial transmit delay - additional_group_delay = ((initial_xmit_delay - number_of_lines_to_reach_ixd * slice_width_modified) % 3) == 0 ? 1 : 0; - - //number of pipeline delay cycles in the ssm block (can be determined empirically or analytically by inspecting the ssm block) - ssm_pipeline_delay = 2; - - //number of pipe delay cycles in the obsm block (can be determined empirically or analytically by inspecting the obsm block) - obsm_pipeline_delay = 1; - - //a group of pixels is worth 6 pixels in N422/N420 mode or 3 pixels in all other modes - if (pixelFormat == dml2_420 || pixelFormat == dml2_444 || pixelFormat == dml2_n422 || Output == dml2_hdmifrl) - cycles_per_group = 6; - else - cycles_per_group = 3; - //delay of the bit stream contruction layer in pixels is the sum of: - //1. number of pixel containers in a slice line multipled by the number of lines required to reach initial transmit delay multipled by number of slices to the left of the last horizontal slice - //2. number of pixel containers required to reach initial transmit delay (specifically, in the last horizontal slice) - //3. additional group of delay if initial transmit delay is reached exactly in a group - //4. ssm and obsm pipeline delay (i.e., clock cycles of delay) - group_delay = (lines_to_reach_ixd * slice_width_groups * (numSlices - 1)) + groups_to_reach_ixd + additional_group_delay; - pipeline_delay = ssm_pipeline_delay + obsm_pipeline_delay; - - //pixel delay is group_delay (converted to pixels) + pipeline, however, first group is a special case since it is processed as soon as it arrives (i.e., in 3 cycles regardless of pixel format) - pixels = (group_delay - 1) * cycles_per_group + 3 + pipeline_delay; - - DML_LOG_VERBOSE("DML::%s: bpc: %u\n", __func__, bpc); - DML_LOG_VERBOSE("DML::%s: BPP: %f\n", __func__, BPP); - DML_LOG_VERBOSE("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); - DML_LOG_VERBOSE("DML::%s: numSlices: %u\n", __func__, numSlices); - DML_LOG_VERBOSE("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); - DML_LOG_VERBOSE("DML::%s: Output: %u\n", __func__, Output); - DML_LOG_VERBOSE("DML::%s: pixels: %u\n", __func__, pixels); - return pixels; -} - -//updated in dcn4 -static unsigned int dscComputeDelay(enum dml2_output_format_class pixelFormat, enum dml2_output_encoder_class Output) -{ - unsigned int Delay = 0; - unsigned int dispclk_per_dscclk = 3; - - // sfr - Delay = Delay + 2; - - if (pixelFormat == dml2_420 || pixelFormat == dml2_n422 || (Output == dml2_hdmifrl && pixelFormat != dml2_444)) { - dispclk_per_dscclk = 3 * 2; - } - - if (pixelFormat == dml2_420) { - //dscc top delay for pixel compression layer - Delay = Delay + 16 * dispclk_per_dscclk; - - // dscc - input deserializer - Delay = Delay + 5; - - // dscc - input cdc fifo - Delay = Delay + 1 + 4 * dispclk_per_dscclk; - - // dscc - output cdc fifo - Delay = Delay + 3 + 1 * dispclk_per_dscclk; - - // dscc - cdc uncertainty - Delay = Delay + 3 + 3 * dispclk_per_dscclk; - } else if (pixelFormat == dml2_n422 || (Output == dml2_hdmifrl && pixelFormat != dml2_444)) { - //dscc top delay for pixel compression layer - Delay = Delay + 16 * dispclk_per_dscclk; - // dsccif - Delay = Delay + 1; - // dscc - input deserializer - Delay = Delay + 5; - // dscc - input cdc fifo - Delay = Delay + 1 + 4 * dispclk_per_dscclk; - - - // dscc - output cdc fifo - Delay = Delay + 3 + 1 * dispclk_per_dscclk; - // dscc - cdc uncertainty - Delay = Delay + 3 + 3 * dispclk_per_dscclk; - } else if (pixelFormat == dml2_s422) { - //dscc top delay for pixel compression layer - Delay = Delay + 17 * dispclk_per_dscclk; - - // dscc - input deserializer - Delay = Delay + 3; - // dscc - input cdc fifo - Delay = Delay + 1 + 4 * dispclk_per_dscclk; - // dscc - output cdc fifo - Delay = Delay + 3 + 1 * dispclk_per_dscclk; - // dscc - cdc uncertainty - Delay = Delay + 3 + 3 * dispclk_per_dscclk; - } else { - //dscc top delay for pixel compression layer - Delay = Delay + 16 * dispclk_per_dscclk; - // dscc - input deserializer - Delay = Delay + 3; - // dscc - input cdc fifo - Delay = Delay + 1 + 4 * dispclk_per_dscclk; - // dscc - output cdc fifo - Delay = Delay + 3 + 1 * dispclk_per_dscclk; - - // dscc - cdc uncertainty - Delay = Delay + 3 + 3 * dispclk_per_dscclk; - } - - // sft - Delay = Delay + 1; - DML_LOG_VERBOSE("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); - DML_LOG_VERBOSE("DML::%s: Delay = %u\n", __func__, Delay); - - return Delay; -} - -static unsigned int CalculateHostVMDynamicLevels( - bool GPUVMEnable, - bool HostVMEnable, - unsigned int HostVMMinPageSize, - unsigned int HostVMMaxNonCachedPageTableLevels) -{ - unsigned int HostVMDynamicLevels = 0; - - if (GPUVMEnable && HostVMEnable) { - if (HostVMMinPageSize < 2048) - HostVMDynamicLevels = HostVMMaxNonCachedPageTableLevels; - else if (HostVMMinPageSize >= 2048 && HostVMMinPageSize < 1048576) - HostVMDynamicLevels = (unsigned int)math_max2(0, (double)HostVMMaxNonCachedPageTableLevels - 1); - else - HostVMDynamicLevels = (unsigned int)math_max2(0, (double)HostVMMaxNonCachedPageTableLevels - 2); - } else { - HostVMDynamicLevels = 0; - } - return HostVMDynamicLevels; -} - -static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_and_row_bytes_params *p) -{ - unsigned int extra_dpde_bytes; - unsigned int extra_mpde_bytes; - unsigned int MacroTileSizeBytes; - unsigned int vp_height_dpte_ub; - - unsigned int meta_surface_bytes; - unsigned int vm_bytes; - unsigned int vp_height_meta_ub; - unsigned int PixelPTEReqWidth_linear = 0; // VBA_DELTA. VBA doesn't calculate this - - *p->MetaRequestHeight = 8 * p->BlockHeight256Bytes; - *p->MetaRequestWidth = 8 * p->BlockWidth256Bytes; - if (p->SurfaceTiling == dml2_sw_linear) { - *p->meta_row_height = 32; - *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->MetaRequestWidth - 1, *p->MetaRequestWidth) - math_floor2(p->ViewportXStart, *p->MetaRequestWidth)); - *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestHeight * p->BytePerPixel / 256.0); // FIXME_DCN4SW missing in old code but no dcc for linear anyways? - } else if (!dml_is_vertical_rotation(p->RotationAngle)) { - *p->meta_row_height = *p->MetaRequestHeight; - if (p->ViewportStationary && p->NumberOfDPPs == 1) { - *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->MetaRequestWidth - 1, *p->MetaRequestWidth) - math_floor2(p->ViewportXStart, *p->MetaRequestWidth)); - } else { - *p->meta_row_width = (unsigned int)(math_ceil2(p->SwathWidth - 1, *p->MetaRequestWidth) + *p->MetaRequestWidth); - } - *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestHeight * p->BytePerPixel / 256.0); - } else { - *p->meta_row_height = *p->MetaRequestWidth; - if (p->ViewportStationary && p->NumberOfDPPs == 1) { - *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + *p->MetaRequestHeight - 1, *p->MetaRequestHeight) - math_floor2(p->ViewportYStart, *p->MetaRequestHeight)); - } else { - *p->meta_row_width = (unsigned int)(math_ceil2(p->SwathWidth - 1, *p->MetaRequestHeight) + *p->MetaRequestHeight); - } - *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestWidth * p->BytePerPixel / 256.0); - } - - if (p->ViewportStationary && p->is_phantom && (p->NumberOfDPPs == 1 || !dml_is_vertical_rotation(p->RotationAngle))) { - vp_height_meta_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + 64 * p->BlockHeight256Bytes - 1, 64 * p->BlockHeight256Bytes) - math_floor2(p->ViewportYStart, 64 * p->BlockHeight256Bytes)); - } else if (!dml_is_vertical_rotation(p->RotationAngle)) { - vp_height_meta_ub = (unsigned int)(math_ceil2(p->ViewportHeight - 1, 64 * p->BlockHeight256Bytes) + 64 * p->BlockHeight256Bytes); - } else { - vp_height_meta_ub = (unsigned int)(math_ceil2(p->SwathWidth - 1, 64 * p->BlockHeight256Bytes) + 64 * p->BlockHeight256Bytes); - } - - meta_surface_bytes = (unsigned int)(p->DCCMetaPitch * vp_height_meta_ub * p->BytePerPixel / 256.0); - DML_LOG_VERBOSE("DML::%s: DCCMetaPitch = %u\n", __func__, p->DCCMetaPitch); - DML_LOG_VERBOSE("DML::%s: meta_surface_bytes = %u\n", __func__, meta_surface_bytes); - if (p->GPUVMEnable == true) { - double meta_vmpg_bytes = 4.0 * 1024.0; - *p->meta_pte_bytes_per_frame_ub = (unsigned int)((math_ceil2((double) (meta_surface_bytes - meta_vmpg_bytes) / (8 * meta_vmpg_bytes), 1) + 1) * 64); - extra_mpde_bytes = 128 * (p->GPUVMMaxPageTableLevels - 1); - } else { - *p->meta_pte_bytes_per_frame_ub = 0; - extra_mpde_bytes = 0; - } - - if (!p->DCCEnable || !p->mrq_present) { - *p->meta_pte_bytes_per_frame_ub = 0; - extra_mpde_bytes = 0; - *p->meta_row_bytes = 0; - } - - if (!p->GPUVMEnable) { - *p->PixelPTEBytesPerRow = 0; - *p->PixelPTEBytesPerRowStorage = 0; - *p->dpte_row_width_ub = 0; - *p->dpte_row_height = 0; - *p->dpte_row_height_linear = 0; - *p->PixelPTEBytesPerRow_one_row_per_frame = 0; - *p->dpte_row_width_ub_one_row_per_frame = 0; - *p->dpte_row_height_one_row_per_frame = 0; - *p->vmpg_width = 0; - *p->vmpg_height = 0; - *p->PixelPTEReqWidth = 0; - *p->PixelPTEReqHeight = 0; - *p->PTERequestSize = 0; - *p->dpde0_bytes_per_frame_ub = 0; - return 0; - } - - MacroTileSizeBytes = p->MacroTileWidth * p->BytePerPixel * p->MacroTileHeight; - - if (p->ViewportStationary && p->is_phantom && (p->NumberOfDPPs == 1 || !dml_is_vertical_rotation(p->RotationAngle))) { - vp_height_dpte_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + p->MacroTileHeight - 1, p->MacroTileHeight) - math_floor2(p->ViewportYStart, p->MacroTileHeight)); - } else if (!dml_is_vertical_rotation(p->RotationAngle)) { - vp_height_dpte_ub = (unsigned int)(math_ceil2((double)p->ViewportHeight - 1, p->MacroTileHeight) + p->MacroTileHeight); - } else { - vp_height_dpte_ub = (unsigned int)(math_ceil2((double)p->SwathWidth - 1, p->MacroTileHeight) + p->MacroTileHeight); - } - - if (p->GPUVMEnable == true && p->GPUVMMaxPageTableLevels > 1) { - *p->dpde0_bytes_per_frame_ub = (unsigned int)(64 * (math_ceil2((double)(p->Pitch * vp_height_dpte_ub * p->BytePerPixel - MacroTileSizeBytes) / (double)(8 * 2097152), 1) + 1)); - extra_dpde_bytes = 128 * (p->GPUVMMaxPageTableLevels - 2); - } else { - *p->dpde0_bytes_per_frame_ub = 0; - extra_dpde_bytes = 0; - } - - vm_bytes = *p->meta_pte_bytes_per_frame_ub + extra_mpde_bytes + *p->dpde0_bytes_per_frame_ub + extra_dpde_bytes; - - DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->DCCEnable); - DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); - DML_LOG_VERBOSE("DML::%s: SwModeLinear = %u\n", __func__, p->SurfaceTiling == dml2_sw_linear); - DML_LOG_VERBOSE("DML::%s: BytePerPixel = %u\n", __func__, p->BytePerPixel); - DML_LOG_VERBOSE("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, p->GPUVMMaxPageTableLevels); - DML_LOG_VERBOSE("DML::%s: BlockHeight256Bytes = %u\n", __func__, p->BlockHeight256Bytes); - DML_LOG_VERBOSE("DML::%s: BlockWidth256Bytes = %u\n", __func__, p->BlockWidth256Bytes); - DML_LOG_VERBOSE("DML::%s: MacroTileHeight = %u\n", __func__, p->MacroTileHeight); - DML_LOG_VERBOSE("DML::%s: MacroTileWidth = %u\n", __func__, p->MacroTileWidth); - DML_LOG_VERBOSE("DML::%s: meta_pte_bytes_per_frame_ub = %u\n", __func__, *p->meta_pte_bytes_per_frame_ub); - DML_LOG_VERBOSE("DML::%s: dpde0_bytes_per_frame_ub = %u\n", __func__, *p->dpde0_bytes_per_frame_ub); - DML_LOG_VERBOSE("DML::%s: extra_mpde_bytes = %u\n", __func__, extra_mpde_bytes); - DML_LOG_VERBOSE("DML::%s: extra_dpde_bytes = %u\n", __func__, extra_dpde_bytes); - DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); - DML_LOG_VERBOSE("DML::%s: ViewportHeight = %u\n", __func__, p->ViewportHeight); - DML_LOG_VERBOSE("DML::%s: SwathWidth = %u\n", __func__, p->SwathWidth); - DML_LOG_VERBOSE("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); - - if (p->SurfaceTiling == dml2_sw_linear) { - *p->PixelPTEReqHeight = 1; - *p->PixelPTEReqWidth = p->GPUVMMinPageSizeKBytes * 1024 * 8 / p->BytePerPixel; - PixelPTEReqWidth_linear = p->GPUVMMinPageSizeKBytes * 1024 * 8 / p->BytePerPixel; - *p->PTERequestSize = 64; - - *p->vmpg_height = 1; - *p->vmpg_width = p->GPUVMMinPageSizeKBytes * 1024 / p->BytePerPixel; - } else if (p->GPUVMMinPageSizeKBytes * 1024 >= dml_get_tile_block_size_bytes(p->SurfaceTiling)) { // 1 64B 8x1 PTE - *p->PixelPTEReqHeight = p->MacroTileHeight; - *p->PixelPTEReqWidth = 8 * 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); - *p->PTERequestSize = 64; - - *p->vmpg_height = p->MacroTileHeight; - *p->vmpg_width = 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); - - } else if (p->GPUVMMinPageSizeKBytes == 4 && dml_get_tile_block_size_bytes(p->SurfaceTiling) == 65536) { // 2 64B PTE requests to get 16 PTEs to cover the 64K tile - // one 64KB tile, is 16x16x256B req - *p->PixelPTEReqHeight = 16 * p->BlockHeight256Bytes; - *p->PixelPTEReqWidth = 16 * p->BlockWidth256Bytes; - *p->PTERequestSize = 128; - - *p->vmpg_height = *p->PixelPTEReqHeight; - *p->vmpg_width = *p->PixelPTEReqWidth; - } else { - // default for rest of calculation to go through, when vm is disable, the calulated pte related values shouldnt be used anyways - *p->PixelPTEReqHeight = p->MacroTileHeight; - *p->PixelPTEReqWidth = 8 * 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); - *p->PTERequestSize = 64; - - *p->vmpg_height = p->MacroTileHeight; - *p->vmpg_width = 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); - - if (p->GPUVMEnable == true) { - DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes=%u and sw_mode=%u (tile_size=%d) not supported!\n", - __func__, p->GPUVMMinPageSizeKBytes, p->SurfaceTiling, dml_get_tile_block_size_bytes(p->SurfaceTiling)); - DML_ASSERT(0); - } - } - - DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); - DML_LOG_VERBOSE("DML::%s: PixelPTEReqHeight = %u\n", __func__, *p->PixelPTEReqHeight); - DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth = %u\n", __func__, *p->PixelPTEReqWidth); - DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); - DML_LOG_VERBOSE("DML::%s: PTERequestSize = %u\n", __func__, *p->PTERequestSize); - DML_LOG_VERBOSE("DML::%s: Pitch = %u\n", __func__, p->Pitch); - DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, *p->vmpg_width); - DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, *p->vmpg_height); - - *p->dpte_row_height_one_row_per_frame = vp_height_dpte_ub; - *p->dpte_row_width_ub_one_row_per_frame = (unsigned int)((math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height_one_row_per_frame / (double)*p->PixelPTEReqHeight - 1) / (double)*p->PixelPTEReqWidth, 1) + 1) * (double)*p->PixelPTEReqWidth); - *p->PixelPTEBytesPerRow_one_row_per_frame = (unsigned int)((double)*p->dpte_row_width_ub_one_row_per_frame / (double)*p->PixelPTEReqWidth * *p->PTERequestSize); - *p->dpte_row_height_linear = 0; - - if (p->SurfaceTiling == dml2_sw_linear) { - *p->dpte_row_height = (unsigned int)(math_min2(128, (double)(1ULL << (unsigned int)math_floor2(math_log((float)(p->PTEBufferSizeInRequests * *p->PixelPTEReqWidth / p->Pitch), 2.0), 1)))); - *p->dpte_row_width_ub = (unsigned int)(math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height - 1), (double)*p->PixelPTEReqWidth) + *p->PixelPTEReqWidth); - *p->PixelPTEBytesPerRow = (unsigned int)((double)*p->dpte_row_width_ub / (double)*p->PixelPTEReqWidth * *p->PTERequestSize); - - // VBA_DELTA, VBA doesn't have programming value for pte row height linear. - *p->dpte_row_height_linear = (unsigned int)1 << (unsigned int)math_floor2(math_log((float)(p->PTEBufferSizeInRequests * PixelPTEReqWidth_linear / p->Pitch), 2.0), 1); - if (*p->dpte_row_height_linear > 128) - *p->dpte_row_height_linear = 128; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *p->dpte_row_width_ub); -#endif - - } else if (!dml_is_vertical_rotation(p->RotationAngle)) { - *p->dpte_row_height = *p->PixelPTEReqHeight; - - if (p->GPUVMMinPageSizeKBytes > 64) { - *p->dpte_row_width_ub = (unsigned int)((math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height / (double)*p->PixelPTEReqHeight - 1) / (double)*p->PixelPTEReqWidth, 1) + 1) * *p->PixelPTEReqWidth); - } else if (p->ViewportStationary && (p->NumberOfDPPs == 1)) { - *p->dpte_row_width_ub = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->PixelPTEReqWidth - 1, *p->PixelPTEReqWidth) - math_floor2(p->ViewportXStart, *p->PixelPTEReqWidth)); - } else { - *p->dpte_row_width_ub = (unsigned int)((math_ceil2((double)(p->SwathWidth - 1) / (double)*p->PixelPTEReqWidth, 1) + 1.0) * *p->PixelPTEReqWidth); - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *p->dpte_row_width_ub); -#endif - - *p->PixelPTEBytesPerRow = *p->dpte_row_width_ub / *p->PixelPTEReqWidth * *p->PTERequestSize; - } else { - *p->dpte_row_height = (unsigned int)(math_min2(*p->PixelPTEReqWidth, p->MacroTileWidth)); - - if (p->ViewportStationary && (p->NumberOfDPPs == 1)) { - *p->dpte_row_width_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + *p->PixelPTEReqHeight - 1, *p->PixelPTEReqHeight) - math_floor2(p->ViewportYStart, *p->PixelPTEReqHeight)); - } else { - *p->dpte_row_width_ub = (unsigned int)((math_ceil2((double)(p->SwathWidth - 1) / (double)*p->PixelPTEReqHeight, 1) + 1) * *p->PixelPTEReqHeight); - } - - *p->PixelPTEBytesPerRow = (unsigned int)((double)*p->dpte_row_width_ub / (double)*p->PixelPTEReqHeight * *p->PTERequestSize); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *p->dpte_row_width_ub); -#endif - } - - if (p->GPUVMEnable != true) { - *p->PixelPTEBytesPerRow = 0; - *p->PixelPTEBytesPerRow_one_row_per_frame = 0; - } - - *p->PixelPTEBytesPerRowStorage = *p->PixelPTEBytesPerRow; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); - DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); - DML_LOG_VERBOSE("DML::%s: meta_row_height = %u\n", __func__, *p->meta_row_height); - DML_LOG_VERBOSE("DML::%s: dpte_row_height = %u\n", __func__, *p->dpte_row_height); - DML_LOG_VERBOSE("DML::%s: dpte_row_height_linear = %u\n", __func__, *p->dpte_row_height_linear); - DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u\n", __func__, *p->dpte_row_width_ub); - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *p->PixelPTEBytesPerRow); - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *p->PixelPTEBytesPerRowStorage); - DML_LOG_VERBOSE("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, p->PTEBufferSizeInRequests); - DML_LOG_VERBOSE("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *p->dpte_row_height_one_row_per_frame); - DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *p->dpte_row_width_ub_one_row_per_frame); - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *p->PixelPTEBytesPerRow_one_row_per_frame); -#endif - - return vm_bytes; -} // CalculateVMAndRowBytes - -static unsigned int CalculatePrefetchSourceLines( - double VRatio, - unsigned int VTaps, - bool Interlace, - bool ProgressiveToInterlaceUnitInOPP, - unsigned int SwathHeight, - enum dml2_rotation_angle RotationAngle, - bool mirrored, - bool ViewportStationary, - unsigned int SwathWidth, - unsigned int ViewportHeight, - unsigned int ViewportXStart, - unsigned int ViewportYStart, - - // Output - unsigned int *VInitPreFill, - unsigned int *MaxNumSwath) -{ - - unsigned int vp_start_rot = 0; - unsigned int sw0_tmp = 0; - unsigned int MaxPartialSwath = 0; - double numLines = 0; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); - DML_LOG_VERBOSE("DML::%s: VTaps = %u\n", __func__, VTaps); - DML_LOG_VERBOSE("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); - DML_LOG_VERBOSE("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); - DML_LOG_VERBOSE("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); - DML_LOG_VERBOSE("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); -#endif - if (ProgressiveToInterlaceUnitInOPP) - *VInitPreFill = (unsigned int)(math_floor2((VRatio + (double)VTaps + 1) / 2.0, 1)); - else - *VInitPreFill = (unsigned int)(math_floor2((VRatio + (double)VTaps + 1 + (Interlace ? 1 : 0) * 0.5 * VRatio) / 2.0, 1)); - - if (ViewportStationary) { - if (RotationAngle == dml2_rotation_180) { - vp_start_rot = SwathHeight - (((unsigned int)(ViewportYStart + ViewportHeight - 1) % SwathHeight) + 1); - } else if ((RotationAngle == dml2_rotation_270 && !mirrored) || (RotationAngle == dml2_rotation_90 && mirrored)) { - vp_start_rot = ViewportXStart; - } else if ((RotationAngle == dml2_rotation_90 && !mirrored) || (RotationAngle == dml2_rotation_270 && mirrored)) { - vp_start_rot = SwathHeight - (((unsigned int)(ViewportYStart + SwathWidth - 1) % SwathHeight) + 1); - } else { - vp_start_rot = ViewportYStart; - } - sw0_tmp = SwathHeight - (vp_start_rot % SwathHeight); - if (sw0_tmp < *VInitPreFill) { - *MaxNumSwath = (unsigned int)(math_ceil2((*VInitPreFill - sw0_tmp) / (double)SwathHeight, 1) + 1); - } else { - *MaxNumSwath = 1; - } - MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(vp_start_rot + *VInitPreFill - 1) % SwathHeight)); - } else { - *MaxNumSwath = (unsigned int)(math_ceil2((*VInitPreFill - 1.0) / (double)SwathHeight, 1) + 1); - if (*VInitPreFill > 1) { - MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(*VInitPreFill - 2) % SwathHeight)); - } else { - MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(*VInitPreFill + SwathHeight - 2) % SwathHeight)); - } - } - numLines = *MaxNumSwath * SwathHeight + MaxPartialSwath; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); - DML_LOG_VERBOSE("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); - DML_LOG_VERBOSE("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); - DML_LOG_VERBOSE("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); - DML_LOG_VERBOSE("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); -#endif - return (unsigned int)(numLines); - -} - -static void CalculateRowBandwidth( - bool GPUVMEnable, - bool use_one_row_for_frame, - enum dml2_source_format_class SourcePixelFormat, - double VRatio, - double VRatioChroma, - bool DCCEnable, - double LineTime, - unsigned int PixelPTEBytesPerRowLuma, - unsigned int PixelPTEBytesPerRowChroma, - unsigned int dpte_row_height_luma, - unsigned int dpte_row_height_chroma, - - bool mrq_present, - unsigned int meta_row_bytes_per_row_ub_l, - unsigned int meta_row_bytes_per_row_ub_c, - unsigned int meta_row_height_luma, - unsigned int meta_row_height_chroma, - - // Output - double *dpte_row_bw, - double *meta_row_bw) -{ - if (!DCCEnable || !mrq_present) { - *meta_row_bw = 0; - } else if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) { - *meta_row_bw = VRatio * meta_row_bytes_per_row_ub_l / (meta_row_height_luma * LineTime) - + VRatioChroma * meta_row_bytes_per_row_ub_c / (meta_row_height_chroma * LineTime); - } else { - *meta_row_bw = VRatio * meta_row_bytes_per_row_ub_l / (meta_row_height_luma * LineTime); - } - - if (GPUVMEnable != true) { - *dpte_row_bw = 0; - } else if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) { - *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime) - + VRatioChroma * PixelPTEBytesPerRowChroma / (dpte_row_height_chroma * LineTime); - } else { - *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime); - } -} - -static void CalculateMALLUseForStaticScreen( - const struct dml2_display_cfg *display_cfg, - unsigned int NumberOfActiveSurfaces, - unsigned int MALLAllocatedForDCN, - unsigned int SurfaceSizeInMALL[], - bool one_row_per_frame_fits_in_buffer[], - - // Output - bool is_using_mall_for_ss[]) -{ - - unsigned int SurfaceToAddToMALL; - bool CanAddAnotherSurfaceToMALL; - unsigned int TotalSurfaceSizeInMALL; - - TotalSurfaceSizeInMALL = 0; - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - is_using_mall_for_ss[k] = (display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable); - if (is_using_mall_for_ss[k]) - TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, is_using_mall_for_ss[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); -#endif - } - - SurfaceToAddToMALL = 0; - CanAddAnotherSurfaceToMALL = true; - while (CanAddAnotherSurfaceToMALL) { - CanAddAnotherSurfaceToMALL = false; - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k] <= MALLAllocatedForDCN * 1024 * 1024 && - !is_using_mall_for_ss[k] && display_cfg->plane_descriptors[k].overrides.refresh_from_mall != dml2_refresh_from_mall_mode_override_force_disable && one_row_per_frame_fits_in_buffer[k] && - (!CanAddAnotherSurfaceToMALL || SurfaceSizeInMALL[k] < SurfaceSizeInMALL[SurfaceToAddToMALL])) { - CanAddAnotherSurfaceToMALL = true; - SurfaceToAddToMALL = k; - DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, display_cfg->plane_descriptors[k].overrides.refresh_from_mall); - } - } - if (CanAddAnotherSurfaceToMALL) { - is_using_mall_for_ss[SurfaceToAddToMALL] = true; - TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[SurfaceToAddToMALL]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); - DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); -#endif - } - } -} - -static void CalculateDCCConfiguration( - bool DCCEnabled, - bool DCCProgrammingAssumesScanDirectionUnknown, - enum dml2_source_format_class SourcePixelFormat, - unsigned int SurfaceWidthLuma, - unsigned int SurfaceWidthChroma, - unsigned int SurfaceHeightLuma, - unsigned int SurfaceHeightChroma, - unsigned int nomDETInKByte, - unsigned int RequestHeight256ByteLuma, - unsigned int RequestHeight256ByteChroma, - enum dml2_swizzle_mode TilingFormat, - unsigned int BytePerPixelY, - unsigned int BytePerPixelC, - double BytePerPixelDETY, - double BytePerPixelDETC, - enum dml2_rotation_angle RotationAngle, - - // Output - enum dml2_core_internal_request_type *RequestLuma, - enum dml2_core_internal_request_type *RequestChroma, - unsigned int *MaxUncompressedBlockLuma, - unsigned int *MaxUncompressedBlockChroma, - unsigned int *MaxCompressedBlockLuma, - unsigned int *MaxCompressedBlockChroma, - unsigned int *IndependentBlockLuma, - unsigned int *IndependentBlockChroma) -{ - unsigned int DETBufferSizeForDCC = nomDETInKByte * 1024; - - unsigned int segment_order_horz_contiguous_luma; - unsigned int segment_order_horz_contiguous_chroma; - unsigned int segment_order_vert_contiguous_luma; - unsigned int segment_order_vert_contiguous_chroma; - - unsigned int req128_horz_wc_l; - unsigned int req128_horz_wc_c; - unsigned int req128_vert_wc_l; - unsigned int req128_vert_wc_c; - - unsigned int yuv420; - unsigned int horz_div_l; - unsigned int horz_div_c; - unsigned int vert_div_l; - unsigned int vert_div_c; - - unsigned int swath_buf_size; - double detile_buf_vp_horz_limit; - double detile_buf_vp_vert_limit; - - unsigned int MAS_vp_horz_limit; - unsigned int MAS_vp_vert_limit; - unsigned int max_vp_horz_width; - unsigned int max_vp_vert_height; - unsigned int eff_surf_width_l; - unsigned int eff_surf_width_c; - unsigned int eff_surf_height_l; - unsigned int eff_surf_height_c; - - unsigned int full_swath_bytes_horz_wc_l; - unsigned int full_swath_bytes_horz_wc_c; - unsigned int full_swath_bytes_vert_wc_l; - unsigned int full_swath_bytes_vert_wc_c; - - if (dml_is_420(SourcePixelFormat)) - yuv420 = 1; - else - yuv420 = 0; - horz_div_l = 1; - horz_div_c = 1; - vert_div_l = 1; - vert_div_c = 1; - - if (BytePerPixelY == 1) - vert_div_l = 0; - if (BytePerPixelC == 1) - vert_div_c = 0; - - if (BytePerPixelC == 0) { - swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 256; - detile_buf_vp_horz_limit = (double)swath_buf_size / ((double)RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l)); - detile_buf_vp_vert_limit = (double)swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l)); - } else { - swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 2 * 256; - detile_buf_vp_horz_limit = (double)swath_buf_size / ((double)RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l) + (double)RequestHeight256ByteChroma * BytePerPixelC / (1 + horz_div_c) / (1 + yuv420)); - detile_buf_vp_vert_limit = (double)swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l) + 256.0 / RequestHeight256ByteChroma / (1 + vert_div_c) / (1 + yuv420)); - } - - if (SourcePixelFormat == dml2_420_10) { - detile_buf_vp_horz_limit = 1.5 * detile_buf_vp_horz_limit; - detile_buf_vp_vert_limit = 1.5 * detile_buf_vp_vert_limit; - } - - detile_buf_vp_horz_limit = math_floor2(detile_buf_vp_horz_limit - 1, 16); - detile_buf_vp_vert_limit = math_floor2(detile_buf_vp_vert_limit - 1, 16); - - MAS_vp_horz_limit = SourcePixelFormat == dml2_rgbe_alpha ? 3840 : 6144; - MAS_vp_vert_limit = SourcePixelFormat == dml2_rgbe_alpha ? 3840 : (BytePerPixelY == 8 ? 3072 : 6144); - max_vp_horz_width = (unsigned int)(math_min2((double)MAS_vp_horz_limit, detile_buf_vp_horz_limit)); - max_vp_vert_height = (unsigned int)(math_min2((double)MAS_vp_vert_limit, detile_buf_vp_vert_limit)); - eff_surf_width_l = (SurfaceWidthLuma > max_vp_horz_width ? max_vp_horz_width : SurfaceWidthLuma); - eff_surf_width_c = eff_surf_width_l / (1 + yuv420); - eff_surf_height_l = (SurfaceHeightLuma > max_vp_vert_height ? max_vp_vert_height : SurfaceHeightLuma); - eff_surf_height_c = eff_surf_height_l / (1 + yuv420); - - full_swath_bytes_horz_wc_l = eff_surf_width_l * RequestHeight256ByteLuma * BytePerPixelY; - full_swath_bytes_vert_wc_l = eff_surf_height_l * 256 / RequestHeight256ByteLuma; - if (BytePerPixelC > 0) { - full_swath_bytes_horz_wc_c = eff_surf_width_c * RequestHeight256ByteChroma * BytePerPixelC; - full_swath_bytes_vert_wc_c = eff_surf_height_c * 256 / RequestHeight256ByteChroma; - } else { - full_swath_bytes_horz_wc_c = 0; - full_swath_bytes_vert_wc_c = 0; - } - - if (SourcePixelFormat == dml2_420_10) { - full_swath_bytes_horz_wc_l = (unsigned int)(math_ceil2((double)full_swath_bytes_horz_wc_l * 2.0 / 3.0, 256.0)); - full_swath_bytes_horz_wc_c = (unsigned int)(math_ceil2((double)full_swath_bytes_horz_wc_c * 2.0 / 3.0, 256.0)); - full_swath_bytes_vert_wc_l = (unsigned int)(math_ceil2((double)full_swath_bytes_vert_wc_l * 2.0 / 3.0, 256.0)); - full_swath_bytes_vert_wc_c = (unsigned int)(math_ceil2((double)full_swath_bytes_vert_wc_c * 2.0 / 3.0, 256.0)); - } - - if (2 * full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 0; - req128_horz_wc_c = 0; - } else if (full_swath_bytes_horz_wc_l < 1.5 * full_swath_bytes_horz_wc_c && 2 * full_swath_bytes_horz_wc_l + full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 0; - req128_horz_wc_c = 1; - } else if (full_swath_bytes_horz_wc_l >= 1.5 * full_swath_bytes_horz_wc_c && full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { - req128_horz_wc_l = 1; - req128_horz_wc_c = 0; - } else { - req128_horz_wc_l = 1; - req128_horz_wc_c = 1; - } - - if (2 * full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 0; - req128_vert_wc_c = 0; - } else if (full_swath_bytes_vert_wc_l < 1.5 * full_swath_bytes_vert_wc_c && 2 * full_swath_bytes_vert_wc_l + full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 0; - req128_vert_wc_c = 1; - } else if (full_swath_bytes_vert_wc_l >= 1.5 * full_swath_bytes_vert_wc_c && full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { - req128_vert_wc_l = 1; - req128_vert_wc_c = 0; - } else { - req128_vert_wc_l = 1; - req128_vert_wc_c = 1; - } - - if (BytePerPixelY == 2) { - segment_order_horz_contiguous_luma = 0; - segment_order_vert_contiguous_luma = 1; - } else { - segment_order_horz_contiguous_luma = 1; - segment_order_vert_contiguous_luma = 0; - } - - if (BytePerPixelC == 2) { - segment_order_horz_contiguous_chroma = 0; - segment_order_vert_contiguous_chroma = 1; - } else { - segment_order_horz_contiguous_chroma = 1; - segment_order_vert_contiguous_chroma = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); - DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); - DML_LOG_VERBOSE("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); - DML_LOG_VERBOSE("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); - DML_LOG_VERBOSE("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); - DML_LOG_VERBOSE("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); - DML_LOG_VERBOSE("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); - DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); - DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); -#endif - if (DCCProgrammingAssumesScanDirectionUnknown == true) { - if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { - *RequestLuma = dml2_core_internal_request_type_256_bytes; - } else if ((req128_horz_wc_l == 1 && segment_order_horz_contiguous_luma == 0) || (req128_vert_wc_l == 1 && segment_order_vert_contiguous_luma == 0)) { - *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; - } - if (req128_horz_wc_c == 0 && req128_vert_wc_c == 0) { - *RequestChroma = dml2_core_internal_request_type_256_bytes; - } else if ((req128_horz_wc_c == 1 && segment_order_horz_contiguous_chroma == 0) || (req128_vert_wc_c == 1 && segment_order_vert_contiguous_chroma == 0)) { - *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; - } - } else if (!dml_is_vertical_rotation(RotationAngle)) { - if (req128_horz_wc_l == 0) { - *RequestLuma = dml2_core_internal_request_type_256_bytes; - } else if (segment_order_horz_contiguous_luma == 0) { - *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; - } - if (req128_horz_wc_c == 0) { - *RequestChroma = dml2_core_internal_request_type_256_bytes; - } else if (segment_order_horz_contiguous_chroma == 0) { - *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; - } - } else { - if (req128_vert_wc_l == 0) { - *RequestLuma = dml2_core_internal_request_type_256_bytes; - } else if (segment_order_vert_contiguous_luma == 0) { - *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; - } - if (req128_vert_wc_c == 0) { - *RequestChroma = dml2_core_internal_request_type_256_bytes; - } else if (segment_order_vert_contiguous_chroma == 0) { - *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; - } else { - *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; - } - } - - if (*RequestLuma == dml2_core_internal_request_type_256_bytes) { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 256; - *IndependentBlockLuma = 0; - } else if (*RequestLuma == dml2_core_internal_request_type_128_bytes_contiguous) { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 128; - *IndependentBlockLuma = 128; - } else { - *MaxUncompressedBlockLuma = 256; - *MaxCompressedBlockLuma = 64; - *IndependentBlockLuma = 64; - } - - if (*RequestChroma == dml2_core_internal_request_type_256_bytes) { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 256; - *IndependentBlockChroma = 0; - } else if (*RequestChroma == dml2_core_internal_request_type_128_bytes_contiguous) { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 128; - *IndependentBlockChroma = 128; - } else { - *MaxUncompressedBlockChroma = 256; - *MaxCompressedBlockChroma = 64; - *IndependentBlockChroma = 64; - } - - if (DCCEnabled != true || BytePerPixelC == 0) { - *MaxUncompressedBlockChroma = 0; - *MaxCompressedBlockChroma = 0; - *IndependentBlockChroma = 0; - } - - if (DCCEnabled != true) { - *MaxUncompressedBlockLuma = 0; - *MaxCompressedBlockLuma = 0; - *IndependentBlockLuma = 0; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); - DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); - DML_LOG_VERBOSE("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); - DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); - DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); - DML_LOG_VERBOSE("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); -#endif - -} - -static void calculate_mcache_row_bytes( - struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_calculate_mcache_row_bytes_params *p) -{ - unsigned int vmpg_bytes = 0; - unsigned int blk_bytes = 0; - float meta_per_mvmpg_per_channel = 0; - unsigned int est_blk_per_vmpg = 2; - unsigned int mvmpg_per_row_ub = 0; - unsigned int full_vp_width_mvmpg_aligned = 0; - unsigned int full_vp_height_mvmpg_aligned = 0; - unsigned int meta_per_mvmpg_per_channel_ub = 0; - unsigned int mvmpg_per_mcache; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: num_chans = %u\n", __func__, p->num_chans); - DML_LOG_VERBOSE("DML::%s: mem_word_bytes = %u\n", __func__, p->mem_word_bytes); - DML_LOG_VERBOSE("DML::%s: mcache_line_size_bytes = %u\n", __func__, p->mcache_line_size_bytes); - DML_LOG_VERBOSE("DML::%s: mcache_size_bytes = %u\n", __func__, p->mcache_size_bytes); - DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); - DML_LOG_VERBOSE("DML::%s: gpuvm_page_size_kbytes = %u\n", __func__, p->gpuvm_page_size_kbytes); - DML_LOG_VERBOSE("DML::%s: vp_stationary = %u\n", __func__, p->vp_stationary); - DML_LOG_VERBOSE("DML::%s: tiling_mode = %u\n", __func__, p->tiling_mode); - DML_LOG_VERBOSE("DML::%s: vp_start_x = %u\n", __func__, p->vp_start_x); - DML_LOG_VERBOSE("DML::%s: vp_start_y = %u\n", __func__, p->vp_start_y); - DML_LOG_VERBOSE("DML::%s: full_vp_width = %u\n", __func__, p->full_vp_width); - DML_LOG_VERBOSE("DML::%s: full_vp_height = %u\n", __func__, p->full_vp_height); - DML_LOG_VERBOSE("DML::%s: blk_width = %u\n", __func__, p->blk_width); - DML_LOG_VERBOSE("DML::%s: blk_height = %u\n", __func__, p->blk_height); - DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, p->vmpg_width); - DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, p->vmpg_height); - DML_LOG_VERBOSE("DML::%s: full_swath_bytes = %u\n", __func__, p->full_swath_bytes); -#endif - DML_ASSERT(p->mcache_line_size_bytes != 0); - DML_ASSERT(p->mcache_size_bytes != 0); - - *p->mvmpg_width = 0; - *p->mvmpg_height = 0; - - if (p->full_vp_height == 0 && p->full_vp_width == 0) { - *p->num_mcaches = 0; - *p->mcache_row_bytes = 0; - *p->mcache_row_bytes_per_channel = 0; - } else { - blk_bytes = dml_get_tile_block_size_bytes(p->tiling_mode); - - // if gpuvm is not enable, the alignment boundary should be in terms of tiling block size - vmpg_bytes = p->gpuvm_page_size_kbytes * 1024; - - //With vmpg_bytes >= tile blk_bytes, the meta_row_width alignment equations are relative to the vmpg_width/height. - // But for 4KB page with 64KB tile block, we need the meta for all pages in the tile block. - // Therefore, the alignment is relative to the blk_width/height. The factor of 16 vmpg per 64KB tile block is applied at the end. - *p->mvmpg_width = p->blk_width; - *p->mvmpg_height = p->blk_height; - if (p->gpuvm_enable) { - if (vmpg_bytes >= blk_bytes) { - *p->mvmpg_width = p->vmpg_width; - *p->mvmpg_height = p->vmpg_height; - } else if (!((blk_bytes == 65536) && (vmpg_bytes == 4096))) { - DML_LOG_VERBOSE("ERROR: DML::%s: Tiling size and vm page size combination not supported\n", __func__); - DML_ASSERT(0); - } - } - - //For plane0 & 1, first calculate full_vp_width/height_l/c aligned to vmpg_width/height_l/c - full_vp_width_mvmpg_aligned = (unsigned int)(math_floor2((p->vp_start_x + p->full_vp_width) + *p->mvmpg_width - 1, *p->mvmpg_width) - math_floor2(p->vp_start_x, *p->mvmpg_width)); - full_vp_height_mvmpg_aligned = (unsigned int)(math_floor2((p->vp_start_y + p->full_vp_height) + *p->mvmpg_height - 1, *p->mvmpg_height) - math_floor2(p->vp_start_y, *p->mvmpg_height)); - - *p->full_vp_access_width_mvmpg_aligned = p->surf_vert ? full_vp_height_mvmpg_aligned : full_vp_width_mvmpg_aligned; - - //Use the equation for the exact alignment when possible. Note that the exact alignment cannot be used for horizontal access if vmpg_bytes > blk_bytes. - if (!p->surf_vert) { //horizontal access - if (p->vp_stationary == 1 && vmpg_bytes <= blk_bytes) - *p->meta_row_width_ub = full_vp_width_mvmpg_aligned; - else - *p->meta_row_width_ub = (unsigned int)math_ceil2((double)p->full_vp_width - 1, *p->mvmpg_width) + *p->mvmpg_width; - mvmpg_per_row_ub = *p->meta_row_width_ub / *p->mvmpg_width; - } else { //vertical access - if (p->vp_stationary == 1) - *p->meta_row_width_ub = full_vp_height_mvmpg_aligned; - else - *p->meta_row_width_ub = (unsigned int)math_ceil2((double)p->full_vp_height - 1, *p->mvmpg_height) + *p->mvmpg_height; - mvmpg_per_row_ub = *p->meta_row_width_ub / *p->mvmpg_height; - } - - if (p->gpuvm_enable) { - meta_per_mvmpg_per_channel = (float)vmpg_bytes / (float)256 / p->num_chans; - - //but using the est_blk_per_vmpg between 2 and 4, to be not as pessimestic - if (p->surf_vert && vmpg_bytes > blk_bytes) { - meta_per_mvmpg_per_channel = (float)est_blk_per_vmpg * blk_bytes / (float)256 / p->num_chans; - } - - *p->dcc_dram_bw_nom_overhead_factor = 1 + math_max2(1.0 / 256.0, math_ceil2(meta_per_mvmpg_per_channel, p->mem_word_bytes) / (256 * meta_per_mvmpg_per_channel)); // dcc_dr_oh_nom - } else { - meta_per_mvmpg_per_channel = (float) blk_bytes / (float)256 / p->num_chans; - - if (!p->surf_vert) - *p->dcc_dram_bw_nom_overhead_factor = 1 + 1.0 / 256.0; - else - *p->dcc_dram_bw_nom_overhead_factor = 1 + math_max2(1.0 / 256.0, math_ceil2(meta_per_mvmpg_per_channel, p->mem_word_bytes) / (256 * meta_per_mvmpg_per_channel)); - } - - meta_per_mvmpg_per_channel_ub = (unsigned int)math_ceil2((double)meta_per_mvmpg_per_channel, p->mcache_line_size_bytes); - - //but for 4KB vmpg with 64KB tile blk - if (p->gpuvm_enable && (blk_bytes == 65536) && (vmpg_bytes == 4096)) - meta_per_mvmpg_per_channel_ub = 16 * meta_per_mvmpg_per_channel_ub; - - // If this mcache_row_bytes for the full viewport of the surface is less than or equal to mcache_bytes, - // then one mcache can be used for this request stream. If not, it is useful to know the width of the viewport that can be supported in the mcache_bytes. - if (p->gpuvm_enable || p->surf_vert) { - *p->mcache_row_bytes_per_channel = mvmpg_per_row_ub * meta_per_mvmpg_per_channel_ub; - *p->mcache_row_bytes = *p->mcache_row_bytes_per_channel * p->num_chans; - } else { // horizontal and gpuvm disable - *p->mcache_row_bytes = *p->meta_row_width_ub * p->blk_height * p->bytes_per_pixel / 256; - if (p->mcache_line_size_bytes != 0) - *p->mcache_row_bytes_per_channel = (unsigned int)math_ceil2((double)*p->mcache_row_bytes / p->num_chans, p->mcache_line_size_bytes); - } - - *p->dcc_dram_bw_pref_overhead_factor = 1 + math_max2(1.0 / 256.0, *p->mcache_row_bytes / p->full_swath_bytes); // dcc_dr_oh_pref - if (p->mcache_size_bytes != 0) - *p->num_mcaches = (unsigned int)math_ceil2((double)*p->mcache_row_bytes_per_channel / p->mcache_size_bytes, 1); - - mvmpg_per_mcache = p->mcache_size_bytes / meta_per_mvmpg_per_channel_ub; - *p->mvmpg_per_mcache_lb = (unsigned int)math_floor2(mvmpg_per_mcache, 1); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); - DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %u\n", __func__, vmpg_bytes); - DML_LOG_VERBOSE("DML::%s: blk_bytes = %u\n", __func__, blk_bytes); - DML_LOG_VERBOSE("DML::%s: meta_per_mvmpg_per_channel = %f\n", __func__, meta_per_mvmpg_per_channel); - DML_LOG_VERBOSE("DML::%s: mvmpg_per_row_ub = %u\n", __func__, mvmpg_per_row_ub); - DML_LOG_VERBOSE("DML::%s: meta_row_width_ub = %u\n", __func__, *p->meta_row_width_ub); - DML_LOG_VERBOSE("DML::%s: mvmpg_width = %u\n", __func__, *p->mvmpg_width); - DML_LOG_VERBOSE("DML::%s: mvmpg_height = %u\n", __func__, *p->mvmpg_height); - DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_nom_overhead_factor); - DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_pref_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_pref_overhead_factor); -#endif - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: mcache_row_bytes = %u\n", __func__, *p->mcache_row_bytes); - DML_LOG_VERBOSE("DML::%s: mcache_row_bytes_per_channel = %u\n", __func__, *p->mcache_row_bytes_per_channel); - DML_LOG_VERBOSE("DML::%s: num_mcaches = %u\n", __func__, *p->num_mcaches); -#endif - DML_ASSERT(*p->num_mcaches > 0); -} - -static void calculate_mcache_setting( - struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_calculate_mcache_setting_params *p) -{ - unsigned int n; - - struct dml2_core_shared_calculate_mcache_setting_locals *l = &scratch->calculate_mcache_setting_locals; - memset(l, 0, sizeof(struct dml2_core_shared_calculate_mcache_setting_locals)); - - *p->num_mcaches_l = 0; - *p->mcache_row_bytes_l = 0; - *p->mcache_row_bytes_per_channel_l = 0; - *p->dcc_dram_bw_nom_overhead_factor_l = 1.0; - *p->dcc_dram_bw_pref_overhead_factor_l = 1.0; - - *p->num_mcaches_c = 0; - *p->mcache_row_bytes_c = 0; - *p->mcache_row_bytes_per_channel_c = 0; - *p->dcc_dram_bw_nom_overhead_factor_c = 1.0; - *p->dcc_dram_bw_pref_overhead_factor_c = 1.0; - - *p->mall_comb_mcache_l = 0; - *p->mall_comb_mcache_c = 0; - *p->lc_comb_mcache = 0; - - if (!p->dcc_enable) - return; - - l->is_dual_plane = dml_is_420(p->source_format) || p->source_format == dml2_rgbe_alpha; - - l->l_p.num_chans = p->num_chans; - l->l_p.mem_word_bytes = p->mem_word_bytes; - l->l_p.mcache_size_bytes = p->mcache_size_bytes; - l->l_p.mcache_line_size_bytes = p->mcache_line_size_bytes; - l->l_p.gpuvm_enable = p->gpuvm_enable; - l->l_p.gpuvm_page_size_kbytes = p->gpuvm_page_size_kbytes; - l->l_p.surf_vert = p->surf_vert; - l->l_p.vp_stationary = p->vp_stationary; - l->l_p.tiling_mode = p->tiling_mode; - l->l_p.vp_start_x = p->vp_start_x_l; - l->l_p.vp_start_y = p->vp_start_y_l; - l->l_p.full_vp_width = p->full_vp_width_l; - l->l_p.full_vp_height = p->full_vp_height_l; - l->l_p.blk_width = p->blk_width_l; - l->l_p.blk_height = p->blk_height_l; - l->l_p.vmpg_width = p->vmpg_width_l; - l->l_p.vmpg_height = p->vmpg_height_l; - l->l_p.full_swath_bytes = p->full_swath_bytes_l; - l->l_p.bytes_per_pixel = p->bytes_per_pixel_l; - - // output - l->l_p.num_mcaches = p->num_mcaches_l; - l->l_p.mcache_row_bytes = p->mcache_row_bytes_l; - l->l_p.mcache_row_bytes_per_channel = p->mcache_row_bytes_per_channel_l; - l->l_p.dcc_dram_bw_nom_overhead_factor = p->dcc_dram_bw_nom_overhead_factor_l; - l->l_p.dcc_dram_bw_pref_overhead_factor = p->dcc_dram_bw_pref_overhead_factor_l; - l->l_p.mvmpg_width = &l->mvmpg_width_l; - l->l_p.mvmpg_height = &l->mvmpg_height_l; - l->l_p.full_vp_access_width_mvmpg_aligned = &l->full_vp_access_width_mvmpg_aligned_l; - l->l_p.meta_row_width_ub = &l->meta_row_width_l; - l->l_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_l; - - calculate_mcache_row_bytes(scratch, &l->l_p); - DML_ASSERT(*p->num_mcaches_l > 0); - - if (l->is_dual_plane) { - l->c_p.num_chans = p->num_chans; - l->c_p.mem_word_bytes = p->mem_word_bytes; - l->c_p.mcache_size_bytes = p->mcache_size_bytes; - l->c_p.mcache_line_size_bytes = p->mcache_line_size_bytes; - l->c_p.gpuvm_enable = p->gpuvm_enable; - l->c_p.gpuvm_page_size_kbytes = p->gpuvm_page_size_kbytes; - l->c_p.surf_vert = p->surf_vert; - l->c_p.vp_stationary = p->vp_stationary; - l->c_p.tiling_mode = p->tiling_mode; - l->c_p.vp_start_x = p->vp_start_x_c; - l->c_p.vp_start_y = p->vp_start_y_c; - l->c_p.full_vp_width = p->full_vp_width_c; - l->c_p.full_vp_height = p->full_vp_height_c; - l->c_p.blk_width = p->blk_width_c; - l->c_p.blk_height = p->blk_height_c; - l->c_p.vmpg_width = p->vmpg_width_c; - l->c_p.vmpg_height = p->vmpg_height_c; - l->c_p.full_swath_bytes = p->full_swath_bytes_c; - l->c_p.bytes_per_pixel = p->bytes_per_pixel_c; - - // output - l->c_p.num_mcaches = p->num_mcaches_c; - l->c_p.mcache_row_bytes = p->mcache_row_bytes_c; - l->c_p.mcache_row_bytes_per_channel = p->mcache_row_bytes_per_channel_c; - l->c_p.dcc_dram_bw_nom_overhead_factor = p->dcc_dram_bw_nom_overhead_factor_c; - l->c_p.dcc_dram_bw_pref_overhead_factor = p->dcc_dram_bw_pref_overhead_factor_c; - l->c_p.mvmpg_width = &l->mvmpg_width_c; - l->c_p.mvmpg_height = &l->mvmpg_height_c; - l->c_p.full_vp_access_width_mvmpg_aligned = &l->full_vp_access_width_mvmpg_aligned_c; - l->c_p.meta_row_width_ub = &l->meta_row_width_c; - l->c_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_c; - - calculate_mcache_row_bytes(scratch, &l->c_p); - DML_ASSERT(*p->num_mcaches_c > 0); - } - - // Sharing for iMALL access - l->mcache_remainder_l = *p->mcache_row_bytes_per_channel_l % p->mcache_size_bytes; - l->mcache_remainder_c = *p->mcache_row_bytes_per_channel_c % p->mcache_size_bytes; - l->mvmpg_access_width_l = p->surf_vert ? l->mvmpg_height_l : l->mvmpg_width_l; - l->mvmpg_access_width_c = p->surf_vert ? l->mvmpg_height_c : l->mvmpg_width_c; - - if (p->imall_enable) { - *p->mall_comb_mcache_l = (2 * l->mcache_remainder_l <= p->mcache_size_bytes); - - if (l->is_dual_plane) - *p->mall_comb_mcache_c = (2 * l->mcache_remainder_c <= p->mcache_size_bytes); - } - - if (!p->surf_vert) // horizonatal access - l->luma_time_factor = (double)l->mvmpg_height_c / l->mvmpg_height_l * 2; - else // vertical access - l->luma_time_factor = (double)l->mvmpg_width_c / l->mvmpg_width_l * 2; - - // The algorithm starts with computing a non-integer, avg_mcache_element_size_l/c: - if (*p->num_mcaches_l) { - l->avg_mcache_element_size_l = l->meta_row_width_l / *p->num_mcaches_l; - } - if (l->is_dual_plane) { - l->avg_mcache_element_size_c = l->meta_row_width_c / *p->num_mcaches_c; - - /* if either remainder is 0, then mcache sharing is not needed or not possible due to full utilization */ - if (l->mcache_remainder_l && l->mcache_remainder_c) { - if (!p->imall_enable || (*p->mall_comb_mcache_l == *p->mall_comb_mcache_c)) { - l->lc_comb_last_mcache_size = (unsigned int)((l->mcache_remainder_l * (*p->mall_comb_mcache_l ? 2 : 1) * l->luma_time_factor) + - (l->mcache_remainder_c * (*p->mall_comb_mcache_c ? 2 : 1))); - } - *p->lc_comb_mcache = (l->lc_comb_last_mcache_size <= p->mcache_size_bytes) && (*p->mall_comb_mcache_l == *p->mall_comb_mcache_c); - } - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: imall_enable = %u\n", __func__, p->imall_enable); - DML_LOG_VERBOSE("DML::%s: is_dual_plane = %u\n", __func__, l->is_dual_plane); - DML_LOG_VERBOSE("DML::%s: surf_vert = %u\n", __func__, p->surf_vert); - DML_LOG_VERBOSE("DML::%s: mvmpg_width_l = %u\n", __func__, l->mvmpg_width_l); - DML_LOG_VERBOSE("DML::%s: mvmpg_height_l = %u\n", __func__, l->mvmpg_height_l); - DML_LOG_VERBOSE("DML::%s: mcache_remainder_l = %f\n", __func__, l->mcache_remainder_l); - DML_LOG_VERBOSE("DML::%s: num_mcaches_l = %u\n", __func__, *p->num_mcaches_l); - DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_l = %u\n", __func__, l->avg_mcache_element_size_l); - DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_l = %u\n", __func__, l->mvmpg_access_width_l); - DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_l = %u\n", __func__, *p->mall_comb_mcache_l); - - if (l->is_dual_plane) { - DML_LOG_VERBOSE("DML::%s: mvmpg_width_c = %u\n", __func__, l->mvmpg_width_c); - DML_LOG_VERBOSE("DML::%s: mvmpg_height_c = %u\n", __func__, l->mvmpg_height_c); - DML_LOG_VERBOSE("DML::%s: mcache_remainder_c = %f\n", __func__, l->mcache_remainder_c); - DML_LOG_VERBOSE("DML::%s: luma_time_factor = %f\n", __func__, l->luma_time_factor); - DML_LOG_VERBOSE("DML::%s: num_mcaches_c = %u\n", __func__, *p->num_mcaches_c); - DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_c = %u\n", __func__, l->avg_mcache_element_size_c); - DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_c = %u\n", __func__, l->mvmpg_access_width_c); - DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_c = %u\n", __func__, *p->mall_comb_mcache_c); - DML_LOG_VERBOSE("DML::%s: lc_comb_last_mcache_size = %u\n", __func__, l->lc_comb_last_mcache_size); - DML_LOG_VERBOSE("DML::%s: lc_comb_mcache = %u\n", __func__, *p->lc_comb_mcache); - } -#endif - // calculate split_coordinate - l->full_vp_access_width_l = p->surf_vert ? p->full_vp_height_l : p->full_vp_width_l; - l->full_vp_access_width_c = p->surf_vert ? p->full_vp_height_c : p->full_vp_width_c; - - for (n = 0; n < *p->num_mcaches_l - 1; n++) { - p->mcache_offsets_l[n] = (unsigned int)(math_floor2((n + 1) * l->avg_mcache_element_size_l / l->mvmpg_access_width_l, 1)) * l->mvmpg_access_width_l; - } - p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l; - - if (l->is_dual_plane) { - for (n = 0; n < *p->num_mcaches_c - 1; n++) { - p->mcache_offsets_c[n] = (unsigned int)(math_floor2((n + 1) * l->avg_mcache_element_size_c / l->mvmpg_access_width_c, 1)) * l->mvmpg_access_width_c; - } - p->mcache_offsets_c[*p->num_mcaches_c - 1] = l->full_vp_access_width_c; - } -#ifdef __DML_VBA_DEBUG__ - for (n = 0; n < *p->num_mcaches_l; n++) - DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); - - if (l->is_dual_plane) { - for (n = 0; n < *p->num_mcaches_c; n++) - DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); - } -#endif - - // Luma/Chroma combine in the last mcache - // In the case of Luma/Chroma combine-mCache (with lc_comb_mcache==1), all mCaches except the last segment are filled as much as possible, when stay aligned to mvmpg boundary - if (*p->lc_comb_mcache && l->is_dual_plane) { - for (n = 0; n < *p->num_mcaches_l - 1; n++) - p->mcache_offsets_l[n] = (n + 1) * l->mvmpg_per_mcache_lb_l * l->mvmpg_access_width_l; - p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l; - - for (n = 0; n < *p->num_mcaches_c - 1; n++) - p->mcache_offsets_c[n] = (n + 1) * l->mvmpg_per_mcache_lb_c * l->mvmpg_access_width_c; - p->mcache_offsets_c[*p->num_mcaches_c - 1] = l->full_vp_access_width_c; - -#ifdef __DML_VBA_DEBUG__ - for (n = 0; n < *p->num_mcaches_l; n++) - DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); - - for (n = 0; n < *p->num_mcaches_c; n++) - DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); -#endif - } - - *p->mcache_shift_granularity_l = l->mvmpg_access_width_l; - *p->mcache_shift_granularity_c = l->mvmpg_access_width_c; -} - -static void calculate_mall_bw_overhead_factor( - double mall_prefetch_sdp_overhead_factor[], //mall_sdp_oh_nom/pref - double mall_prefetch_dram_overhead_factor[], //mall_dram_oh_nom/pref - - // input - const struct dml2_display_cfg *display_cfg, - unsigned int num_active_planes) -{ - for (unsigned int k = 0; k < num_active_planes; ++k) { - mall_prefetch_sdp_overhead_factor[k] = 1.0; - mall_prefetch_dram_overhead_factor[k] = 1.0; - - // SDP - on the return side - if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) // always no data return - mall_prefetch_sdp_overhead_factor[k] = 1.25; - else if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) - mall_prefetch_sdp_overhead_factor[k] = 0.25; - - // DRAM - if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) - mall_prefetch_dram_overhead_factor[k] = 2.0; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_sdp_overhead_factor = %f\n", __func__, k, mall_prefetch_sdp_overhead_factor[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_dram_overhead_factor = %f\n", __func__, k, mall_prefetch_dram_overhead_factor[k]); -#endif - } -} - -static double dml_get_return_bandwidth_available( - const struct dml2_soc_bb *soc, - enum dml2_core_internal_soc_state_type state_type, - enum dml2_core_internal_bw_type bw_type, - bool is_avg_bw, - bool is_hvm_en, - bool is_hvm_only, - double dcfclk_mhz, - double fclk_mhz, - double dram_bw_mbps) -{ - double return_bw_mbps = 0.; - double ideal_sdp_bandwidth = (double)soc->return_bus_width_bytes * dcfclk_mhz; - double ideal_fabric_bandwidth = fclk_mhz * (double)soc->fabric_datapath_to_dcn_data_return_bytes; - double ideal_dram_bandwidth = dram_bw_mbps; //dram_speed_mts * soc->clk_table.dram_config.channel_count * soc->clk_table.dram_config.channel_width_bytes; - - double derate_sdp_factor; - double derate_fabric_factor; - double derate_dram_factor; - - double derate_sdp_bandwidth; - double derate_fabric_bandwidth; - double derate_dram_bandwidth; - - if (is_avg_bw) { - if (state_type == dml2_core_internal_soc_state_svp_prefetch) { - derate_sdp_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.dcfclk_derate_percent / 100.0; - derate_fabric_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.fclk_derate_percent / 100.0; - derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.dram_derate_percent_pixel / 100.0; - } else { // just assume sys_active - derate_sdp_factor = soc->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100.0; - derate_fabric_factor = soc->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100.0; - derate_dram_factor = soc->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100.0; - } - } else { // urgent bw - if (state_type == dml2_core_internal_soc_state_svp_prefetch) { - derate_sdp_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dcfclk_derate_percent / 100.0; - derate_fabric_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.fclk_derate_percent / 100.0; - derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100.0; - - if (is_hvm_en) { - if (is_hvm_only) - derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_vm / 100.0; - else - derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel_and_vm / 100.0; - } else { - derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100.0; - } - } else { // just assume sys_active - derate_sdp_factor = soc->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0; - derate_fabric_factor = soc->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100.0; - - if (is_hvm_en) { - if (is_hvm_only) - derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_vm / 100.0; - else - derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel_and_vm / 100.0; - } else { - derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100.0; - } - } - } - - derate_sdp_bandwidth = ideal_sdp_bandwidth * derate_sdp_factor; - derate_fabric_bandwidth = ideal_fabric_bandwidth * derate_fabric_factor; - derate_dram_bandwidth = ideal_dram_bandwidth * derate_dram_factor; - - if (bw_type == dml2_core_internal_bw_sdp) - return_bw_mbps = math_min2(derate_sdp_bandwidth, derate_fabric_bandwidth); - else // dml2_core_internal_bw_dram - return_bw_mbps = derate_dram_bandwidth; - - DML_LOG_VERBOSE("DML::%s: is_avg_bw = %u\n", __func__, is_avg_bw); - DML_LOG_VERBOSE("DML::%s: is_hvm_en = %u\n", __func__, is_hvm_en); - DML_LOG_VERBOSE("DML::%s: is_hvm_only = %u\n", __func__, is_hvm_only); - DML_LOG_VERBOSE("DML::%s: state_type = %s\n", __func__, dml2_core_internal_soc_state_type_str(state_type)); - DML_LOG_VERBOSE("DML::%s: bw_type = %s\n", __func__, dml2_core_internal_bw_type_str(bw_type)); - DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); - DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); - DML_LOG_VERBOSE("DML::%s: ideal_sdp_bandwidth = %f\n", __func__, ideal_sdp_bandwidth); - DML_LOG_VERBOSE("DML::%s: ideal_fabric_bandwidth = %f\n", __func__, ideal_fabric_bandwidth); - DML_LOG_VERBOSE("DML::%s: ideal_dram_bandwidth = %f\n", __func__, ideal_dram_bandwidth); - DML_LOG_VERBOSE("DML::%s: derate_sdp_bandwidth = %f (derate %f)\n", __func__, derate_sdp_bandwidth, derate_sdp_factor); - DML_LOG_VERBOSE("DML::%s: derate_fabric_bandwidth = %f (derate %f)\n", __func__, derate_fabric_bandwidth, derate_fabric_factor); - DML_LOG_VERBOSE("DML::%s: derate_dram_bandwidth = %f (derate %f)\n", __func__, derate_dram_bandwidth, derate_dram_factor); - DML_LOG_VERBOSE("DML::%s: return_bw_mbps = %f\n", __func__, return_bw_mbps); - return return_bw_mbps; -} - -static noinline_for_stack void calculate_bandwidth_available( - double avg_bandwidth_available_min[dml2_core_internal_soc_state_max], - double avg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_bandwidth_available_min[dml2_core_internal_soc_state_max], // min between SDP and DRAM - double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_max], - double urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_max], - - const struct dml2_soc_bb *soc, - bool HostVMEnable, - double dcfclk_mhz, - double fclk_mhz, - double dram_bw_mbps) -{ - unsigned int n, m; - - DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); - DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); - DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, dram_bw_mbps); - - // Calculate all the bandwidth availabe - for (m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (n = 0; n < dml2_core_internal_bw_max; n++) { - avg_bandwidth_available[m][n] = dml_get_return_bandwidth_available(soc, - m, // soc_state - n, // bw_type - 1, // avg_bw - HostVMEnable, - 0, // hvm_only - dcfclk_mhz, - fclk_mhz, - dram_bw_mbps); - - urg_bandwidth_available[m][n] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 0, dcfclk_mhz, fclk_mhz, dram_bw_mbps); - - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), avg_bandwidth_available[m][n]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n]); -#endif - - // urg_bandwidth_available_vm_only is indexed by soc_state - if (n == dml2_core_internal_bw_dram) { - urg_bandwidth_available_vm_only[m] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 1, dcfclk_mhz, fclk_mhz, dram_bw_mbps); - urg_bandwidth_available_pixel_and_vm[m] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 0, dcfclk_mhz, fclk_mhz, dram_bw_mbps); - } - } - - avg_bandwidth_available_min[m] = math_min2(avg_bandwidth_available[m][dml2_core_internal_bw_dram], avg_bandwidth_available[m][dml2_core_internal_bw_sdp]); - urg_bandwidth_available_min[m] = math_min2(urg_bandwidth_available[m][dml2_core_internal_bw_dram], urg_bandwidth_available[m][dml2_core_internal_bw_sdp]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), avg_bandwidth_available_min[m]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_min[m]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_vm_only[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_vm_only[n]); -#endif - } -} - -static void calculate_avg_bandwidth_required( - double avg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - - // input - const struct dml2_display_cfg *display_cfg, - unsigned int num_active_planes, - double ReadBandwidthLuma[], - double ReadBandwidthChroma[], - double cursor_bw[], - double dcc_dram_bw_nom_overhead_factor_p0[], - double dcc_dram_bw_nom_overhead_factor_p1[], - double mall_prefetch_dram_overhead_factor[], - double mall_prefetch_sdp_overhead_factor[]) -{ - unsigned int n, m, k; - double sdp_overhead_factor; - double dram_overhead_factor_p0; - double dram_overhead_factor_p1; - - // Average BW support check - for (m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (n = 0; n < dml2_core_internal_bw_max; n++) { // sdp, dram - avg_bandwidth_required[m][n] = 0; - } - } - - // SysActive and SVP Prefetch AVG bandwidth Check - for (k = 0; k < num_active_planes; ++k) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: plane %0d\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: ReadBandwidthLuma=%f\n", __func__, ReadBandwidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: ReadBandwidthChroma=%f\n", __func__, ReadBandwidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p0=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p0[k]); - DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p1=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p1[k]); - DML_LOG_VERBOSE("DML::%s: mall_prefetch_dram_overhead_factor=%f\n", __func__, mall_prefetch_dram_overhead_factor[k]); - DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor=%f\n", __func__, mall_prefetch_sdp_overhead_factor[k]); -#endif - - sdp_overhead_factor = mall_prefetch_sdp_overhead_factor[k]; - dram_overhead_factor_p0 = dcc_dram_bw_nom_overhead_factor_p0[k] * mall_prefetch_dram_overhead_factor[k]; - dram_overhead_factor_p1 = dcc_dram_bw_nom_overhead_factor_p1[k] * mall_prefetch_dram_overhead_factor[k]; - - // FIXME_DCN4, was missing cursor_bw in here, but do I actually need that and tdlut bw for average bandwidth calculation? - // active avg bw not include phantom, but svp_prefetch avg bw should include phantom pipes - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] += sdp_overhead_factor * (ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) + cursor_bw[k]; - avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] += dram_overhead_factor_p0 * ReadBandwidthLuma[k] + dram_overhead_factor_p1 * ReadBandwidthChroma[k] + cursor_bw[k]; - } - avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] += sdp_overhead_factor * (ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) + cursor_bw[k]; - avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] += dram_overhead_factor_p0 * ReadBandwidthLuma[k] + dram_overhead_factor_p1 * ReadBandwidthChroma[k] + cursor_bw[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); -#endif - } -} - -static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_CalculateVMRowAndSwath_params *p) -{ - struct dml2_core_calcs_CalculateVMRowAndSwath_locals *s = &scratch->CalculateVMRowAndSwath_locals; - - s->HostVMDynamicLevels = CalculateHostVMDynamicLevels(p->display_cfg->gpuvm_enable, p->display_cfg->hostvm_enable, p->HostVMMinPageSize, p->display_cfg->hostvm_max_non_cached_page_table_levels); - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->display_cfg->gpuvm_enable == true) { - p->vm_group_bytes[k] = 512; - p->dpte_group_bytes[k] = 512; - } else { - p->vm_group_bytes[k] = 0; - p->dpte_group_bytes[k] = 0; - } - - if (dml_is_420(p->myPipe[k].SourcePixelFormat) || p->myPipe[k].SourcePixelFormat == dml2_rgbe_alpha) { - if ((p->myPipe[k].SourcePixelFormat == dml2_420_10 || p->myPipe[k].SourcePixelFormat == dml2_420_12) && !dml_is_vertical_rotation(p->myPipe[k].RotationAngle)) { - s->PTEBufferSizeInRequestsForLuma[k] = (p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma) / 2; - s->PTEBufferSizeInRequestsForChroma[k] = s->PTEBufferSizeInRequestsForLuma[k]; - } else { - s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma; - s->PTEBufferSizeInRequestsForChroma[k] = p->PTEBufferSizeInRequestsChroma; - } - - scratch->calculate_vm_and_row_bytes_params.ViewportStationary = p->myPipe[k].ViewportStationary; - scratch->calculate_vm_and_row_bytes_params.DCCEnable = p->myPipe[k].DCCEnable; - scratch->calculate_vm_and_row_bytes_params.NumberOfDPPs = p->myPipe[k].DPPPerSurface; - scratch->calculate_vm_and_row_bytes_params.BlockHeight256Bytes = p->myPipe[k].BlockHeight256BytesC; - scratch->calculate_vm_and_row_bytes_params.BlockWidth256Bytes = p->myPipe[k].BlockWidth256BytesC; - scratch->calculate_vm_and_row_bytes_params.SourcePixelFormat = p->myPipe[k].SourcePixelFormat; - scratch->calculate_vm_and_row_bytes_params.SurfaceTiling = p->myPipe[k].SurfaceTiling; - scratch->calculate_vm_and_row_bytes_params.BytePerPixel = p->myPipe[k].BytePerPixelC; - scratch->calculate_vm_and_row_bytes_params.RotationAngle = p->myPipe[k].RotationAngle; - scratch->calculate_vm_and_row_bytes_params.SwathWidth = p->SwathWidthC[k]; - scratch->calculate_vm_and_row_bytes_params.ViewportHeight = p->myPipe[k].ViewportHeightC; - scratch->calculate_vm_and_row_bytes_params.ViewportXStart = p->myPipe[k].ViewportXStartC; - scratch->calculate_vm_and_row_bytes_params.ViewportYStart = p->myPipe[k].ViewportYStartC; - scratch->calculate_vm_and_row_bytes_params.GPUVMEnable = p->display_cfg->gpuvm_enable; - scratch->calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = p->display_cfg->gpuvm_max_page_table_levels; - scratch->calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - scratch->calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = s->PTEBufferSizeInRequestsForChroma[k]; - scratch->calculate_vm_and_row_bytes_params.Pitch = p->myPipe[k].PitchC; - scratch->calculate_vm_and_row_bytes_params.MacroTileWidth = p->myPipe[k].BlockWidthC; - scratch->calculate_vm_and_row_bytes_params.MacroTileHeight = p->myPipe[k].BlockHeightC; - scratch->calculate_vm_and_row_bytes_params.is_phantom = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]); - scratch->calculate_vm_and_row_bytes_params.DCCMetaPitch = p->myPipe[k].DCCMetaPitchC; - scratch->calculate_vm_and_row_bytes_params.mrq_present = p->mrq_present; - - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &s->PixelPTEBytesPerRowC[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &s->PixelPTEBytesPerRowStorageC[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub = &p->dpte_row_width_chroma_ub[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height = &p->dpte_row_height_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height_linear = &p->dpte_row_height_linear_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &s->PixelPTEBytesPerRowC_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &s->dpte_row_width_chroma_ub_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &s->dpte_row_height_chroma_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.vmpg_width = &p->vmpg_width_c[k]; - scratch->calculate_vm_and_row_bytes_params.vmpg_height = &p->vmpg_height_c[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &p->PixelPTEReqWidthC[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &p->PixelPTEReqHeightC[k]; - scratch->calculate_vm_and_row_bytes_params.PTERequestSize = &p->PTERequestSizeC[k]; - scratch->calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &p->dpde0_bytes_per_frame_ub_c[k]; - - scratch->calculate_vm_and_row_bytes_params.meta_row_bytes = &s->meta_row_bytes_per_row_ub_c[k]; - scratch->calculate_vm_and_row_bytes_params.MetaRequestWidth = &p->meta_req_width_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.MetaRequestHeight = &p->meta_req_height_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_row_width = &p->meta_row_width_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_row_height = &p->meta_row_height_chroma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &p->meta_pte_bytes_per_frame_ub_c[k]; - - s->vm_bytes_c = CalculateVMAndRowBytes(&scratch->calculate_vm_and_row_bytes_params); - - p->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines( - p->myPipe[k].VRatioChroma, - p->myPipe[k].VTapsChroma, - p->myPipe[k].InterlaceEnable, - p->myPipe[k].ProgressiveToInterlaceUnitInOPP, - p->myPipe[k].SwathHeightC, - p->myPipe[k].RotationAngle, - p->myPipe[k].mirrored, - p->myPipe[k].ViewportStationary, - p->SwathWidthC[k], - p->myPipe[k].ViewportHeightC, - p->myPipe[k].ViewportXStartC, - p->myPipe[k].ViewportYStartC, - - // Output - &p->VInitPreFillC[k], - &p->MaxNumSwathC[k]); - } else { - s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma; - s->PTEBufferSizeInRequestsForChroma[k] = 0; - s->PixelPTEBytesPerRowC[k] = 0; - s->PixelPTEBytesPerRowStorageC[k] = 0; - s->vm_bytes_c = 0; - p->MaxNumSwathC[k] = 0; - p->PrefetchSourceLinesC[k] = 0; - s->dpte_row_height_chroma_one_row_per_frame[k] = 0; - s->dpte_row_width_chroma_ub_one_row_per_frame[k] = 0; - s->PixelPTEBytesPerRowC_one_row_per_frame[k] = 0; - } - - scratch->calculate_vm_and_row_bytes_params.ViewportStationary = p->myPipe[k].ViewportStationary; - scratch->calculate_vm_and_row_bytes_params.DCCEnable = p->myPipe[k].DCCEnable; - scratch->calculate_vm_and_row_bytes_params.NumberOfDPPs = p->myPipe[k].DPPPerSurface; - scratch->calculate_vm_and_row_bytes_params.BlockHeight256Bytes = p->myPipe[k].BlockHeight256BytesY; - scratch->calculate_vm_and_row_bytes_params.BlockWidth256Bytes = p->myPipe[k].BlockWidth256BytesY; - scratch->calculate_vm_and_row_bytes_params.SourcePixelFormat = p->myPipe[k].SourcePixelFormat; - scratch->calculate_vm_and_row_bytes_params.SurfaceTiling = p->myPipe[k].SurfaceTiling; - scratch->calculate_vm_and_row_bytes_params.BytePerPixel = p->myPipe[k].BytePerPixelY; - scratch->calculate_vm_and_row_bytes_params.RotationAngle = p->myPipe[k].RotationAngle; - scratch->calculate_vm_and_row_bytes_params.SwathWidth = p->SwathWidthY[k]; - scratch->calculate_vm_and_row_bytes_params.ViewportHeight = p->myPipe[k].ViewportHeight; - scratch->calculate_vm_and_row_bytes_params.ViewportXStart = p->myPipe[k].ViewportXStart; - scratch->calculate_vm_and_row_bytes_params.ViewportYStart = p->myPipe[k].ViewportYStart; - scratch->calculate_vm_and_row_bytes_params.GPUVMEnable = p->display_cfg->gpuvm_enable; - scratch->calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = p->display_cfg->gpuvm_max_page_table_levels; - scratch->calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - scratch->calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = s->PTEBufferSizeInRequestsForLuma[k]; - scratch->calculate_vm_and_row_bytes_params.Pitch = p->myPipe[k].PitchY; - scratch->calculate_vm_and_row_bytes_params.MacroTileWidth = p->myPipe[k].BlockWidthY; - scratch->calculate_vm_and_row_bytes_params.MacroTileHeight = p->myPipe[k].BlockHeightY; - scratch->calculate_vm_and_row_bytes_params.is_phantom = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]); - scratch->calculate_vm_and_row_bytes_params.DCCMetaPitch = p->myPipe[k].DCCMetaPitchY; - scratch->calculate_vm_and_row_bytes_params.mrq_present = p->mrq_present; - - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &s->PixelPTEBytesPerRowY[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &s->PixelPTEBytesPerRowStorageY[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub = &p->dpte_row_width_luma_ub[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height = &p->dpte_row_height_luma[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height_linear = &p->dpte_row_height_linear_luma[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &s->PixelPTEBytesPerRowY_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &s->dpte_row_width_luma_ub_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &s->dpte_row_height_luma_one_row_per_frame[k]; - scratch->calculate_vm_and_row_bytes_params.vmpg_width = &p->vmpg_width_y[k]; - scratch->calculate_vm_and_row_bytes_params.vmpg_height = &p->vmpg_height_y[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &p->PixelPTEReqWidthY[k]; - scratch->calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &p->PixelPTEReqHeightY[k]; - scratch->calculate_vm_and_row_bytes_params.PTERequestSize = &p->PTERequestSizeY[k]; - scratch->calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &p->dpde0_bytes_per_frame_ub_l[k]; - - scratch->calculate_vm_and_row_bytes_params.meta_row_bytes = &s->meta_row_bytes_per_row_ub_l[k]; - scratch->calculate_vm_and_row_bytes_params.MetaRequestWidth = &p->meta_req_width_luma[k]; - scratch->calculate_vm_and_row_bytes_params.MetaRequestHeight = &p->meta_req_height_luma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_row_width = &p->meta_row_width_luma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_row_height = &p->meta_row_height_luma[k]; - scratch->calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &p->meta_pte_bytes_per_frame_ub_l[k]; - - s->vm_bytes_l = CalculateVMAndRowBytes(&scratch->calculate_vm_and_row_bytes_params); - - p->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines( - p->myPipe[k].VRatio, - p->myPipe[k].VTaps, - p->myPipe[k].InterlaceEnable, - p->myPipe[k].ProgressiveToInterlaceUnitInOPP, - p->myPipe[k].SwathHeightY, - p->myPipe[k].RotationAngle, - p->myPipe[k].mirrored, - p->myPipe[k].ViewportStationary, - p->SwathWidthY[k], - p->myPipe[k].ViewportHeight, - p->myPipe[k].ViewportXStart, - p->myPipe[k].ViewportYStart, - - // Output - &p->VInitPreFillY[k], - &p->MaxNumSwathY[k]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_l = %u (before hvm level)\n", __func__, k, s->vm_bytes_l); - DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_c = %u (before hvm level)\n", __func__, k, s->vm_bytes_c); - DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_l = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_c = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_c[k]); -#endif - p->vm_bytes[k] = (s->vm_bytes_l + s->vm_bytes_c) * (1 + 8 * s->HostVMDynamicLevels); - p->meta_row_bytes[k] = s->meta_row_bytes_per_row_ub_l[k] + s->meta_row_bytes_per_row_ub_c[k]; - p->meta_row_bytes_per_row_ub_l[k] = s->meta_row_bytes_per_row_ub_l[k]; - p->meta_row_bytes_per_row_ub_c[k] = s->meta_row_bytes_per_row_ub_c[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes = %u\n", __func__, k, p->meta_row_bytes[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes = %u (after hvm level)\n", __func__, k, p->vm_bytes[k]); -#endif - if (s->PixelPTEBytesPerRowStorageY[k] <= 64 * s->PTEBufferSizeInRequestsForLuma[k] && s->PixelPTEBytesPerRowStorageC[k] <= 64 * s->PTEBufferSizeInRequestsForChroma[k]) { - p->PTEBufferSizeNotExceeded[k] = true; - } else { - p->PTEBufferSizeNotExceeded[k] = false; - } - - s->one_row_per_frame_fits_in_buffer[k] = (s->PixelPTEBytesPerRowY_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForLuma[k] && - s->PixelPTEBytesPerRowC_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForChroma[k]); -#ifdef __DML_VBA_DEBUG__ - if (p->PTEBufferSizeNotExceeded[k] == 0 || s->one_row_per_frame_fits_in_buffer[k] == 0) { - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded (not one_row_per_frame) = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); - - DML_LOG_VERBOSE("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); - } -#endif - } - - CalculateMALLUseForStaticScreen( - p->display_cfg, - p->NumberOfActiveSurfaces, - p->MALLAllocatedForDCN, - p->SurfaceSizeInMALL, - s->one_row_per_frame_fits_in_buffer, - // Output - p->is_using_mall_for_ss); - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->display_cfg->gpuvm_enable) { - if (p->display_cfg->plane_descriptors[k].overrides.hw.force_pte_buffer_mode.enable == 1) { - p->PTE_BUFFER_MODE[k] = p->display_cfg->plane_descriptors[k].overrides.hw.force_pte_buffer_mode.value; - } - p->PTE_BUFFER_MODE[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->is_using_mall_for_ss[k] || (p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) || - dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) || (p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes > 64); - p->BIGK_FRAGMENT_SIZE[k] = (unsigned int)(math_log((float)p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes * 1024, 2) - 12); - } else { - p->PTE_BUFFER_MODE[k] = 0; - p->BIGK_FRAGMENT_SIZE[k] = 0; - } - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - p->DCCMetaBufferSizeNotExceeded[k] = true; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, p->is_using_mall_for_ss[k]); -#endif - p->use_one_row_for_frame[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->is_using_mall_for_ss[k] || (p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) || - (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) || (p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes > 64 && dml_is_vertical_rotation(p->myPipe[k].RotationAngle)); - - p->use_one_row_for_frame_flip[k] = p->use_one_row_for_frame[k] && !(p->display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame); - - if (p->use_one_row_for_frame[k]) { - p->dpte_row_height_luma[k] = s->dpte_row_height_luma_one_row_per_frame[k]; - p->dpte_row_width_luma_ub[k] = s->dpte_row_width_luma_ub_one_row_per_frame[k]; - s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY_one_row_per_frame[k]; - p->dpte_row_height_chroma[k] = s->dpte_row_height_chroma_one_row_per_frame[k]; - p->dpte_row_width_chroma_ub[k] = s->dpte_row_width_chroma_ub_one_row_per_frame[k]; - s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC_one_row_per_frame[k]; - p->PTEBufferSizeNotExceeded[k] = s->one_row_per_frame_fits_in_buffer[k]; - } - - if (p->meta_row_bytes[k] <= p->DCCMetaBufferSizeBytes) { - p->DCCMetaBufferSizeNotExceeded[k] = true; - } else { - p->DCCMetaBufferSizeNotExceeded[k] = false; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bytes = %d\n", __func__, k, p->meta_row_bytes[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeBytes = %d\n", __func__, k, p->DCCMetaBufferSizeBytes); - DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeNotExceeded = %d\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); -#endif - } - - s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY[k] * (1 + 8 * s->HostVMDynamicLevels); - s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC[k] * (1 + 8 * s->HostVMDynamicLevels); - p->PixelPTEBytesPerRow[k] = s->PixelPTEBytesPerRowY[k] + s->PixelPTEBytesPerRowC[k]; - p->dpte_row_bytes_per_row_l[k] = s->PixelPTEBytesPerRowY[k]; - p->dpte_row_bytes_per_row_c[k] = s->PixelPTEBytesPerRowC[k]; - - // if one row of dPTEs is meant to span the entire frame, then for these calculations, we will pretend like that one big row is fetched in two halfs - if (p->use_one_row_for_frame[k]) - p->PixelPTEBytesPerRow[k] = p->PixelPTEBytesPerRow[k] / 2; - - CalculateRowBandwidth( - p->display_cfg->gpuvm_enable, - p->use_one_row_for_frame[k], - p->myPipe[k].SourcePixelFormat, - p->myPipe[k].VRatio, - p->myPipe[k].VRatioChroma, - p->myPipe[k].DCCEnable, - p->myPipe[k].HTotal / p->myPipe[k].PixelClock, - s->PixelPTEBytesPerRowY[k], - s->PixelPTEBytesPerRowC[k], - p->dpte_row_height_luma[k], - p->dpte_row_height_chroma[k], - - p->mrq_present, - p->meta_row_bytes_per_row_ub_l[k], - p->meta_row_bytes_per_row_ub_c[k], - p->meta_row_height_luma[k], - p->meta_row_height_chroma[k], - - // Output - &p->dpte_row_bw[k], - &p->meta_row_bw[k]); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, gpuvm_enable = %u\n", __func__, k, p->display_cfg->gpuvm_enable); - DML_LOG_VERBOSE("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); -#endif - } -} - -static double CalculateUrgentLatency( - double UrgentLatencyPixelDataOnly, - double UrgentLatencyPixelMixedWithVMData, - double UrgentLatencyVMDataOnly, - bool DoUrgentLatencyAdjustment, - double UrgentLatencyAdjustmentFabricClockComponent, - double UrgentLatencyAdjustmentFabricClockReference, - double FabricClock, - double uclk_freq_mhz, - enum dml2_qos_param_type qos_type, - unsigned int urgent_ramp_uclk_cycles, - unsigned int df_qos_response_time_fclk_cycles, - unsigned int max_round_trip_to_furthest_cs_fclk_cycles, - unsigned int mall_overhead_fclk_cycles, - double umc_urgent_ramp_latency_margin, - double fabric_max_transport_latency_margin) -{ - double urgent_latency = 0; - if (qos_type == dml2_qos_param_type_dcn4x) { - urgent_latency = (df_qos_response_time_fclk_cycles + mall_overhead_fclk_cycles) / FabricClock - + max_round_trip_to_furthest_cs_fclk_cycles / FabricClock * (1 + fabric_max_transport_latency_margin / 100.0) - + urgent_ramp_uclk_cycles / uclk_freq_mhz * (1 + umc_urgent_ramp_latency_margin / 100.0); - } else { - urgent_latency = math_max3(UrgentLatencyPixelDataOnly, UrgentLatencyPixelMixedWithVMData, UrgentLatencyVMDataOnly); - if (DoUrgentLatencyAdjustment == true) { - urgent_latency = urgent_latency + UrgentLatencyAdjustmentFabricClockComponent * (UrgentLatencyAdjustmentFabricClockReference / FabricClock - 1); - } - } -#ifdef __DML_VBA_DEBUG__ - if (qos_type == dml2_qos_param_type_dcn4x) { - DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); - DML_LOG_VERBOSE("DML::%s: urgent_ramp_uclk_cycles = %d\n", __func__, urgent_ramp_uclk_cycles); - DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: umc_urgent_ramp_latency_margin = %f\n", __func__, umc_urgent_ramp_latency_margin); - } else { - DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelDataOnly = %f\n", __func__, UrgentLatencyPixelDataOnly); - DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelMixedWithVMData = %f\n", __func__, UrgentLatencyPixelMixedWithVMData); - DML_LOG_VERBOSE("DML::%s: UrgentLatencyVMDataOnly = %f\n", __func__, UrgentLatencyVMDataOnly); - DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockComponent = %f\n", __func__, UrgentLatencyAdjustmentFabricClockComponent); - DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockReference = %f\n", __func__, UrgentLatencyAdjustmentFabricClockReference); - } - DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); - DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, urgent_latency); -#endif - return urgent_latency; -} - -static double CalculateTripToMemory( - double UrgLatency, - double FabricClock, - double uclk_freq_mhz, - enum dml2_qos_param_type qos_type, - unsigned int trip_to_memory_uclk_cycles, - unsigned int max_round_trip_to_furthest_cs_fclk_cycles, - unsigned int mall_overhead_fclk_cycles, - double umc_max_latency_margin, - double fabric_max_transport_latency_margin) -{ - double trip_to_memory_us; - if (qos_type == dml2_qos_param_type_dcn4x) { - trip_to_memory_us = mall_overhead_fclk_cycles / FabricClock - + max_round_trip_to_furthest_cs_fclk_cycles / FabricClock * (1.0 + fabric_max_transport_latency_margin / 100.0) - + trip_to_memory_uclk_cycles / uclk_freq_mhz * (1.0 + umc_max_latency_margin / 100.0); - } else { - trip_to_memory_us = UrgLatency; - } - -#ifdef __DML_VBA_DEBUG__ - if (qos_type == dml2_qos_param_type_dcn4x) { - DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); - DML_LOG_VERBOSE("DML::%s: max_round_trip_to_furthest_cs_fclk_cycles = %d\n", __func__, max_round_trip_to_furthest_cs_fclk_cycles); - DML_LOG_VERBOSE("DML::%s: mall_overhead_fclk_cycles = %d\n", __func__, mall_overhead_fclk_cycles); - DML_LOG_VERBOSE("DML::%s: trip_to_memory_uclk_cycles = %d\n", __func__, trip_to_memory_uclk_cycles); - DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); - DML_LOG_VERBOSE("DML::%s: fabric_max_transport_latency_margin = %f\n", __func__, fabric_max_transport_latency_margin); - DML_LOG_VERBOSE("DML::%s: umc_max_latency_margin = %f\n", __func__, umc_max_latency_margin); - } else { - DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); - } - DML_LOG_VERBOSE("DML::%s: trip_to_memory_us = %f\n", __func__, trip_to_memory_us); -#endif - - - return trip_to_memory_us; -} - -static double CalculateMetaTripToMemory( - double UrgLatency, - double FabricClock, - double uclk_freq_mhz, - enum dml2_qos_param_type qos_type, - unsigned int meta_trip_to_memory_uclk_cycles, - unsigned int meta_trip_to_memory_fclk_cycles, - double umc_max_latency_margin, - double fabric_max_transport_latency_margin) -{ - double meta_trip_to_memory_us; - if (qos_type == dml2_qos_param_type_dcn4x) { - meta_trip_to_memory_us = meta_trip_to_memory_fclk_cycles / FabricClock * (1.0 + fabric_max_transport_latency_margin / 100.0) - + meta_trip_to_memory_uclk_cycles / uclk_freq_mhz * (1.0 + umc_max_latency_margin / 100.0); - } else { - meta_trip_to_memory_us = UrgLatency; - } - -#ifdef __DML_VBA_DEBUG__ - if (qos_type == dml2_qos_param_type_dcn4x) { - DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); - DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_fclk_cycles = %d\n", __func__, meta_trip_to_memory_fclk_cycles); - DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_uclk_cycles = %d\n", __func__, meta_trip_to_memory_uclk_cycles); - DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); - } else { - DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); - } - DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_us = %f\n", __func__, meta_trip_to_memory_us); -#endif - - - return meta_trip_to_memory_us; -} - -static void calculate_cursor_req_attributes( - unsigned int cursor_width, - unsigned int cursor_bpp, - - // output - unsigned int *cursor_lines_per_chunk, - unsigned int *cursor_bytes_per_line, - unsigned int *cursor_bytes_per_chunk, - unsigned int *cursor_bytes) -{ - unsigned int cursor_bytes_per_req = 0; - unsigned int cursor_width_bytes = 0; - unsigned int cursor_height = 0; - - //SW determines the cursor pitch to support the maximum cursor_width that will be used but the following restrictions apply. - //- For 2bpp, cursor_pitch = 256 pixels due to min cursor request size of 64B - //- For 32 or 64 bpp, cursor_pitch = 64, 128 or 256 pixels depending on the cursor width - - //The cursor requestor uses a cursor request size of 64B, 128B, or 256B depending on the cursor_width and cursor_bpp as follows. - - cursor_width_bytes = (unsigned int)math_ceil2((double)cursor_width * cursor_bpp / 8, 1); - if (cursor_width_bytes <= 64) - cursor_bytes_per_req = 64; - else if (cursor_width_bytes <= 128) - cursor_bytes_per_req = 128; - else - cursor_bytes_per_req = 256; - - //If cursor_width_bytes is greater than 256B, then multiple 256B requests are issued to fetch the entire cursor line. - *cursor_bytes_per_line = (unsigned int)math_ceil2((double)cursor_width_bytes, cursor_bytes_per_req); - - //Nominally, the cursor chunk is 1KB or 2KB but it is restricted to a power of 2 number of lines with a maximum of 16 lines. - if (cursor_bpp == 2) { - *cursor_lines_per_chunk = 16; - } else if (cursor_bpp == 32) { - if (cursor_width <= 32) - *cursor_lines_per_chunk = 16; - else if (cursor_width <= 64) - *cursor_lines_per_chunk = 8; - else if (cursor_width <= 128) - *cursor_lines_per_chunk = 4; - else - *cursor_lines_per_chunk = 2; - } else if (cursor_bpp == 64) { - if (cursor_width <= 16) - *cursor_lines_per_chunk = 16; - else if (cursor_width <= 32) - *cursor_lines_per_chunk = 8; - else if (cursor_width <= 64) - *cursor_lines_per_chunk = 4; - else if (cursor_width <= 128) - *cursor_lines_per_chunk = 2; - else - *cursor_lines_per_chunk = 1; - } else { - if (cursor_width > 0) { - DML_LOG_VERBOSE("DML::%s: Invalid cursor_bpp = %d\n", __func__, cursor_bpp); - DML_ASSERT(0); - } - } - - *cursor_bytes_per_chunk = *cursor_bytes_per_line * *cursor_lines_per_chunk; - - // For the cursor implementation, all requested data is stored in the return buffer. Given this fact, the cursor_bytes can be directly compared with the CursorBufferSize. - // Only cursor_width is provided for worst case sizing so assume that the cursor is square - cursor_height = cursor_width; - *cursor_bytes = *cursor_bytes_per_line * cursor_height; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: cursor_bpp = %d\n", __func__, cursor_bpp); - DML_LOG_VERBOSE("DML::%s: cursor_width = %d\n", __func__, cursor_width); - DML_LOG_VERBOSE("DML::%s: cursor_width_bytes = %d\n", __func__, cursor_width_bytes); - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_req = %d\n", __func__, cursor_bytes_per_req); - DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %d\n", __func__, *cursor_lines_per_chunk); - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, *cursor_bytes_per_line); - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, *cursor_bytes_per_chunk); - DML_LOG_VERBOSE("DML::%s: cursor_bytes = %d\n", __func__, *cursor_bytes); - DML_LOG_VERBOSE("DML::%s: cursor_pitch = %d\n", __func__, cursor_bpp == 2 ? 256 : (unsigned int)1 << (unsigned int)math_ceil2(math_log((float)cursor_width, 2), 1)); -#endif -} - -static void calculate_cursor_urgent_burst_factor( - unsigned int CursorBufferSize, - unsigned int CursorWidth, - unsigned int cursor_bytes_per_chunk, - unsigned int cursor_lines_per_chunk, - double LineTime, - double UrgentLatency, - - double *UrgentBurstFactorCursor, - bool *NotEnoughUrgentLatencyHiding) -{ - unsigned int LinesInCursorBuffer = 0; - double CursorBufferSizeInTime = 0; - - if (CursorWidth > 0) { - LinesInCursorBuffer = (unsigned int)math_floor2(CursorBufferSize * 1024.0 / (double)cursor_bytes_per_chunk, 1) * cursor_lines_per_chunk; - - CursorBufferSizeInTime = LinesInCursorBuffer * LineTime; - if (CursorBufferSizeInTime - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorCursor = 1; - } else { - *NotEnoughUrgentLatencyHiding = 0; - *UrgentBurstFactorCursor = CursorBufferSizeInTime / (CursorBufferSizeInTime - UrgentLatency); - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: LinesInCursorBuffer = %u\n", __func__, LinesInCursorBuffer); - DML_LOG_VERBOSE("DML::%s: CursorBufferSizeInTime = %f\n", __func__, CursorBufferSizeInTime); - DML_LOG_VERBOSE("DML::%s: CursorBufferSize = %u (kbytes)\n", __func__, CursorBufferSize); - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %u\n", __func__, cursor_bytes_per_chunk); - DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %u\n", __func__, cursor_lines_per_chunk); - DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, *UrgentBurstFactorCursor); - DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); -#endif - - } -} - -static void CalculateUrgentBurstFactor( - const struct dml2_plane_parameters *plane_cfg, - unsigned int swath_width_luma_ub, - unsigned int swath_width_chroma_ub, - unsigned int SwathHeightY, - unsigned int SwathHeightC, - double LineTime, - double UrgentLatency, - double VRatio, - double VRatioC, - double BytePerPixelInDETY, - double BytePerPixelInDETC, - unsigned int DETBufferSizeY, - unsigned int DETBufferSizeC, - // Output - double *UrgentBurstFactorLuma, - double *UrgentBurstFactorChroma, - bool *NotEnoughUrgentLatencyHiding) -{ - double LinesInDETLuma; - double LinesInDETChroma; - double DETBufferSizeInTimeLuma; - double DETBufferSizeInTimeChroma; - - *NotEnoughUrgentLatencyHiding = 0; - *UrgentBurstFactorLuma = 0; - *UrgentBurstFactorChroma = 0; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); - DML_LOG_VERBOSE("DML::%s: VRatioC = %f\n", __func__, VRatioC); - DML_LOG_VERBOSE("DML::%s: DETBufferSizeY = %d\n", __func__, DETBufferSizeY); - DML_LOG_VERBOSE("DML::%s: DETBufferSizeC = %d\n", __func__, DETBufferSizeC); - DML_LOG_VERBOSE("DML::%s: BytePerPixelInDETY = %f\n", __func__, BytePerPixelInDETY); - DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %d\n", __func__, swath_width_luma_ub); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); -#endif - DML_ASSERT(VRatio > 0); - - LinesInDETLuma = (dml_is_phantom_pipe(plane_cfg) ? 1024 * 1024 : DETBufferSizeY) / BytePerPixelInDETY / swath_width_luma_ub; - - DETBufferSizeInTimeLuma = math_floor2(LinesInDETLuma, SwathHeightY) * LineTime / VRatio; - if (DETBufferSizeInTimeLuma - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorLuma = 1; - } else { - *UrgentBurstFactorLuma = DETBufferSizeInTimeLuma / (DETBufferSizeInTimeLuma - UrgentLatency); - } - - if (BytePerPixelInDETC > 0) { - LinesInDETChroma = (dml_is_phantom_pipe(plane_cfg) ? 1024 * 1024 : DETBufferSizeC) / BytePerPixelInDETC / swath_width_chroma_ub; - - DETBufferSizeInTimeChroma = math_floor2(LinesInDETChroma, SwathHeightC) * LineTime / VRatioC; - if (DETBufferSizeInTimeChroma - UrgentLatency <= 0) { - *NotEnoughUrgentLatencyHiding = 1; - *UrgentBurstFactorChroma = 1; - } else { - *UrgentBurstFactorChroma = DETBufferSizeInTimeChroma / (DETBufferSizeInTimeChroma - UrgentLatency); - } - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: LinesInDETLuma = %f\n", __func__, LinesInDETLuma); - DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); - DML_LOG_VERBOSE("DML::%s: DETBufferSizeInTimeLuma = %f\n", __func__, DETBufferSizeInTimeLuma); - DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, *UrgentBurstFactorLuma); - DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, *UrgentBurstFactorChroma); - DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); -#endif -} - -static void CalculateDCFCLKDeepSleepTdlut( - const struct dml2_display_cfg *display_cfg, - unsigned int NumberOfActiveSurfaces, - unsigned int BytePerPixelY[], - unsigned int BytePerPixelC[], - unsigned int SwathWidthY[], - unsigned int SwathWidthC[], - unsigned int DPPPerSurface[], - double PSCL_THROUGHPUT[], - double PSCL_THROUGHPUT_CHROMA[], - double Dppclk[], - double ReadBandwidthLuma[], - double ReadBandwidthChroma[], - unsigned int ReturnBusWidth, - - double dispclk, - unsigned int tdlut_bytes_to_deliver[], - double prefetch_swath_time_us[], - - // Output - double *DCFClkDeepSleep) -{ - double DisplayPipeLineDeliveryTimeLuma; - double DisplayPipeLineDeliveryTimeChroma; - double DCFClkDeepSleepPerSurface[DML2_MAX_PLANES]; - double ReadBandwidth = 0.0; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - double pixel_rate_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - - if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio <= 1) { - DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] * DPPPerSurface[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_rate_mhz; - } else { - DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChroma = 0; - } else { - if (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio <= 1) { - DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] * DPPPerSurface[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_rate_mhz; - } else { - DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } - - if (BytePerPixelC[k] > 0) { - DCFClkDeepSleepPerSurface[k] = math_max2(__DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 32.0 / DisplayPipeLineDeliveryTimeLuma, - __DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthC[k] * BytePerPixelC[k] / 32.0 / DisplayPipeLineDeliveryTimeChroma); - } else { - DCFClkDeepSleepPerSurface[k] = __DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 64.0 / DisplayPipeLineDeliveryTimeLuma; - } - DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], pixel_rate_mhz / 16); - - // adjust for 3dlut delivery time - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && tdlut_bytes_to_deliver[k] > 0) { - double tdlut_required_deepsleep_dcfclk = (double) tdlut_bytes_to_deliver[k] / 64.0 / prefetch_swath_time_us[k]; - - DML_LOG_VERBOSE("DML::%s: k=%d, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_bytes_to_deliver = %d\n", __func__, k, tdlut_bytes_to_deliver[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_swath_time_us = %f\n", __func__, k, prefetch_swath_time_us[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_required_deepsleep_dcfclk = %f\n", __func__, k, tdlut_required_deepsleep_dcfclk); - - // increase the deepsleep dcfclk to match the original dispclk throughput rate - if (tdlut_required_deepsleep_dcfclk > DCFClkDeepSleepPerSurface[k]) { - DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], tdlut_required_deepsleep_dcfclk); - DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], dispclk / 4.0); - } - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, PixelClock = %f\n", __func__, k, pixel_rate_mhz); - DML_LOG_VERBOSE("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); -#endif - } - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - ReadBandwidth = ReadBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; - } - - *DCFClkDeepSleep = math_max2(8.0, __DML2_CALCS_DCFCLK_FACTOR__ * ReadBandwidth / (double)ReturnBusWidth); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: __DML2_CALCS_DCFCLK_FACTOR__ = %f\n", __func__, __DML2_CALCS_DCFCLK_FACTOR__); - DML_LOG_VERBOSE("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); - DML_LOG_VERBOSE("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); - DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); -#endif - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - *DCFClkDeepSleep = math_max2(*DCFClkDeepSleep, DCFClkDeepSleepPerSurface[k]); - } - - DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); -} - -static noinline_for_stack void CalculateDCFCLKDeepSleep( - const struct dml2_display_cfg *display_cfg, - unsigned int NumberOfActiveSurfaces, - unsigned int BytePerPixelY[], - unsigned int BytePerPixelC[], - unsigned int SwathWidthY[], - unsigned int SwathWidthC[], - unsigned int DPPPerSurface[], - double PSCL_THROUGHPUT[], - double PSCL_THROUGHPUT_CHROMA[], - double Dppclk[], - double ReadBandwidthLuma[], - double ReadBandwidthChroma[], - unsigned int ReturnBusWidth, - - // Output - double *DCFClkDeepSleep) -{ - double zero_double[DML2_MAX_PLANES]; - unsigned int zero_integer[DML2_MAX_PLANES]; - - memset(zero_double, 0, DML2_MAX_PLANES * sizeof(double)); - memset(zero_integer, 0, DML2_MAX_PLANES * sizeof(unsigned int)); - - CalculateDCFCLKDeepSleepTdlut( - display_cfg, - NumberOfActiveSurfaces, - BytePerPixelY, - BytePerPixelC, - SwathWidthY, - SwathWidthC, - DPPPerSurface, - PSCL_THROUGHPUT, - PSCL_THROUGHPUT_CHROMA, - Dppclk, - ReadBandwidthLuma, - ReadBandwidthChroma, - ReturnBusWidth, - 0, - zero_integer, //tdlut_bytes_to_deliver, - zero_double, //prefetch_swath_time_us, - - // Output - DCFClkDeepSleep); -} - -static double CalculateWriteBackDelay( - enum dml2_source_format_class WritebackPixelFormat, - double WritebackHRatio, - double WritebackVRatio, - unsigned int WritebackVTaps, - unsigned int WritebackDestinationWidth, - unsigned int WritebackDestinationHeight, - unsigned int WritebackSourceHeight, - unsigned int HTotal) -{ - double CalculateWriteBackDelay; - double Line_length; - double Output_lines_last_notclamped; - double WritebackVInit; - - WritebackVInit = (WritebackVRatio + WritebackVTaps + 1) / 2; - Line_length = math_max2((double)WritebackDestinationWidth, math_ceil2((double)WritebackDestinationWidth / 6.0, 1.0) * WritebackVTaps); - Output_lines_last_notclamped = WritebackDestinationHeight - 1 - math_ceil2(((double)WritebackSourceHeight - (double)WritebackVInit) / (double)WritebackVRatio, 1.0); - if (Output_lines_last_notclamped < 0) { - CalculateWriteBackDelay = 0; - } else { - CalculateWriteBackDelay = Output_lines_last_notclamped * Line_length + (HTotal - WritebackDestinationWidth) + 80; - } - return CalculateWriteBackDelay; -} - -static unsigned int CalculateMaxVStartup( - bool ptoi_supported, - unsigned int vblank_nom_default_us, - const struct dml2_timing_cfg *timing, - double write_back_delay_us) -{ - unsigned int vblank_size = 0; - unsigned int max_vstartup_lines = 0; - - double line_time_us = (double)timing->h_total / ((double)timing->pixel_clock_khz / 1000); - unsigned int vblank_actual = timing->v_total - timing->v_active; - unsigned int vblank_nom_default_in_line = (unsigned int)math_floor2((double)vblank_nom_default_us / line_time_us, 1.0); - unsigned int vblank_avail = (timing->vblank_nom == 0) ? vblank_nom_default_in_line : (unsigned int)timing->vblank_nom; - - vblank_size = (unsigned int)math_min2(vblank_actual, vblank_avail); - - if (timing->interlaced && !ptoi_supported) - max_vstartup_lines = (unsigned int)(math_floor2((vblank_size - 1) / 2.0, 1.0)); - else - max_vstartup_lines = vblank_size - (unsigned int)math_max2(1.0, math_ceil2(write_back_delay_us / line_time_us, 1.0)); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VBlankNom = %lu\n", __func__, timing->vblank_nom); - DML_LOG_VERBOSE("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); - DML_LOG_VERBOSE("DML::%s: line_time_us = %f\n", __func__, line_time_us); - DML_LOG_VERBOSE("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); - DML_LOG_VERBOSE("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); - DML_LOG_VERBOSE("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); -#endif - max_vstartup_lines = (unsigned int)math_min2(max_vstartup_lines, DML_MAX_VSTARTUP_START); - return max_vstartup_lines; -} - -static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *p) -{ - unsigned int MaximumSwathHeightY[DML2_MAX_PLANES] = { 0 }; - unsigned int MaximumSwathHeightC[DML2_MAX_PLANES] = { 0 }; - unsigned int RoundedUpSwathSizeBytesY[DML2_MAX_PLANES] = { 0 }; - unsigned int RoundedUpSwathSizeBytesC[DML2_MAX_PLANES] = { 0 }; - unsigned int SwathWidthSingleDPP[DML2_MAX_PLANES] = { 0 }; - unsigned int SwathWidthSingleDPPChroma[DML2_MAX_PLANES] = { 0 }; - - unsigned int TotalActiveDPP = 0; - bool NoChromaOrLinear = true; - unsigned int SurfaceDoingUnboundedRequest = 0; - unsigned int DETBufferSizeInKByteForSwathCalculation; - - const long TTUFIFODEPTH = 8; - const long MAXIMUMCOMPRESSION = 4; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - DML_LOG_VERBOSE("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); - } -#endif - CalculateSwathWidth( - p->display_cfg, - p->ForceSingleDPP, - p->NumberOfActiveSurfaces, - p->ODMMode, - p->BytePerPixY, - p->BytePerPixC, - p->Read256BytesBlockHeightY, - p->Read256BytesBlockHeightC, - p->Read256BytesBlockWidthY, - p->Read256BytesBlockWidthC, - p->surf_linear128_l, - p->surf_linear128_c, - p->DPPPerSurface, - - // Output - p->req_per_swath_ub_l, - p->req_per_swath_ub_c, - SwathWidthSingleDPP, - SwathWidthSingleDPPChroma, - p->SwathWidth, - p->SwathWidthChroma, - MaximumSwathHeightY, - MaximumSwathHeightC, - p->swath_width_luma_ub, - p->swath_width_chroma_ub); - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - p->full_swath_bytes_l[k] = (unsigned int)(p->swath_width_luma_ub[k] * p->BytePerPixDETY[k] * MaximumSwathHeightY[k]); - p->full_swath_bytes_c[k] = (unsigned int)(p->swath_width_chroma_ub[k] * p->BytePerPixDETC[k] * MaximumSwathHeightC[k]); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); - DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); -#endif - if (p->display_cfg->plane_descriptors[k].pixel_format == dml2_420_10) { - p->full_swath_bytes_l[k] = (unsigned int)(math_ceil2((double)p->full_swath_bytes_l[k], 256)); - p->full_swath_bytes_c[k] = (unsigned int)(math_ceil2((double)p->full_swath_bytes_c[k], 256)); - } - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - TotalActiveDPP = TotalActiveDPP + (p->ForceSingleDPP ? 1 : p->DPPPerSurface[k]); - if (p->DPPPerSurface[k] > 0) - SurfaceDoingUnboundedRequest = k; - if (dml_is_420(p->display_cfg->plane_descriptors[k].pixel_format) || p->display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha - || p->display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { - NoChromaOrLinear = false; - } - } - - *p->UnboundedRequestEnabled = UnboundedRequest(p->display_cfg->overrides.hw.force_unbounded_requesting.enable, p->display_cfg->overrides.hw.force_unbounded_requesting.value, TotalActiveDPP, NoChromaOrLinear); - - CalculateDETBufferSize( - &scratch->CalculateDETBufferSize_locals, - p->display_cfg, - p->ForceSingleDPP, - p->NumberOfActiveSurfaces, - *p->UnboundedRequestEnabled, - p->nomDETInKByte, - p->MaxTotalDETInKByte, - p->ConfigReturnBufferSizeInKByte, - p->MinCompressedBufferSizeInKByte, - p->ConfigReturnBufferSegmentSizeInkByte, - p->CompressedBufferSegmentSizeInkByte, - p->ReadBandwidthLuma, - p->ReadBandwidthChroma, - p->full_swath_bytes_l, - p->full_swath_bytes_c, - p->DPPPerSurface, - - // Output - p->DETBufferSizeInKByte, // per hubp pipe - p->CompressedBufferSizeInkByte); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); - DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); - DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); - DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); - DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); -#endif - - *p->ViewportSizeSupport = true; - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - - DETBufferSizeInKByteForSwathCalculation = (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) ? 1024 : p->DETBufferSizeInKByte[k]); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); -#endif - if (p->display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { - p->SwathHeightY[k] = MaximumSwathHeightY[k]; - p->SwathHeightC[k] = MaximumSwathHeightC[k]; - RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; - RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; - - if (p->surf_linear128_l[k]) - p->request_size_bytes_luma[k] = 128; - else - p->request_size_bytes_luma[k] = 256; - - if (p->surf_linear128_c[k]) - p->request_size_bytes_chroma[k] = 128; - else - p->request_size_bytes_chroma[k] = 256; - - } else if (p->full_swath_bytes_l[k] + p->full_swath_bytes_c[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k]; - p->SwathHeightC[k] = MaximumSwathHeightC[k]; - RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; - RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; - p->request_size_bytes_luma[k] = 256; - p->request_size_bytes_chroma[k] = 256; - - } else if (p->full_swath_bytes_l[k] >= 1.5 * p->full_swath_bytes_c[k] && p->full_swath_bytes_l[k] / 2 + p->full_swath_bytes_c[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; - p->SwathHeightC[k] = MaximumSwathHeightC[k]; - RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k] / 2; - RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; - p->request_size_bytes_luma[k] = ((p->BytePerPixY[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; - p->request_size_bytes_chroma[k] = 256; - - } else if (p->full_swath_bytes_l[k] < 1.5 * p->full_swath_bytes_c[k] && p->full_swath_bytes_l[k] + p->full_swath_bytes_c[k] / 2 <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { - p->SwathHeightY[k] = MaximumSwathHeightY[k]; - p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; - RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; - RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k] / 2; - p->request_size_bytes_luma[k] = 256; - p->request_size_bytes_chroma[k] = ((p->BytePerPixC[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; - - } else { - p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; - p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; - RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k] / 2; - RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k] / 2; - p->request_size_bytes_luma[k] = ((p->BytePerPixY[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; - p->request_size_bytes_chroma[k] = ((p->BytePerPixC[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; - } - - if (p->SwathHeightC[k] == 0) - p->request_size_bytes_chroma[k] = 0; - - if ((p->full_swath_bytes_l[k] / 2 + p->full_swath_bytes_c[k] / 2 > DETBufferSizeInKByteForSwathCalculation * 1024 / 2) || - p->SwathWidth[k] > p->MaximumSwathWidthLuma[k] || (p->SwathHeightC[k] > 0 && p->SwathWidthChroma[k] > p->MaximumSwathWidthChroma[k])) { - *p->ViewportSizeSupport = false; - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l=%u\n", __func__, k, p->full_swath_bytes_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c=%u\n", __func__, k, p->full_swath_bytes_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation=%u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); - DML_LOG_VERBOSE("DML::%s: k=%u SwathWidth=%u\n", __func__, k, p->SwathWidth[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, p->MaximumSwathWidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthChroma=%d\n", __func__, k, p->SwathWidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, p->MaximumSwathWidthChroma[k]); - p->ViewportSizeSupportPerSurface[k] = false; - } else { - p->ViewportSizeSupportPerSurface[k] = true; - } - - if (p->SwathHeightC[k] == 0) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, All DET will be used for plane0\n", __func__, k); -#endif - p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024; - p->DETBufferSizeC[k] = 0; - } else if (RoundedUpSwathSizeBytesY[k] <= 1.5 * RoundedUpSwathSizeBytesC[k]) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, Half DET will be used for plane0, and half for plane1\n", __func__, k); -#endif - p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; - p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; - } else { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, 2/3 DET will be used for plane0, and 1/3 for plane1\n", __func__, k); -#endif - p->DETBufferSizeY[k] = (unsigned int)(math_floor2(p->DETBufferSizeInKByte[k] * 1024 * 2 / 3, 1024)); - p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 - p->DETBufferSizeY[k]; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); -#endif - - } - - *p->compbuf_reserved_space_64b = 2 * p->pixel_chunk_size_kbytes * 1024 / 64; - if (*p->UnboundedRequestEnabled) { - *p->compbuf_reserved_space_64b = (unsigned int)math_ceil2(math_max2(*p->compbuf_reserved_space_64b, - (double)(p->rob_buffer_size_kbytes * 1024 / 64) - (double)(RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest] * TTUFIFODEPTH / (p->mrq_present ? MAXIMUMCOMPRESSION : 1) / 64)), 1.0); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: RoundedUpSwathSizeBytesY[%d] = %u\n", __func__, SurfaceDoingUnboundedRequest, RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest]); - DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %u\n", __func__, p->rob_buffer_size_kbytes); -#endif - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_64b = %u\n", __func__, *p->compbuf_reserved_space_64b); -#endif - - *p->hw_debug5 = false; -#ifdef ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE - if (p->NumberOfActiveSurfaces > 1) - *p->hw_debug5 = true; -#else - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (!(p->mrq_present) && (!(*p->UnboundedRequestEnabled)) && (TotalActiveDPP == 1) - && p->display_cfg->plane_descriptors[k].surface.dcc.enable - && ((p->rob_buffer_size_kbytes * 1024 * (p->mrq_present ? MAXIMUMCOMPRESSION : 1) - + *p->CompressedBufferSizeInkByte * MAXIMUMCOMPRESSION * 1024) > TTUFIFODEPTH * (RoundedUpSwathSizeBytesY[k] + RoundedUpSwathSizeBytesC[k]))) - *p->hw_debug5 = true; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u UnboundedRequestEnabled = %u\n", __func__, k, *p->UnboundedRequestEnabled); - DML_LOG_VERBOSE("DML::%s: k=%u MAXIMUMCOMPRESSION = %lu\n", __func__, k, MAXIMUMCOMPRESSION); - DML_LOG_VERBOSE("DML::%s: k=%u TTUFIFODEPTH = %lu\n", __func__, k, TTUFIFODEPTH); - DML_LOG_VERBOSE("DML::%s: k=%u CompressedBufferSizeInkByte = %u\n", __func__, k, *p->CompressedBufferSizeInkByte); - DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u hw_debug5 = %u\n", __func__, k, *p->hw_debug5); -#endif - } -#endif -} - -static enum dml2_odm_mode DecideODMMode(unsigned int HActive, - double MaxDispclk, - unsigned int MaximumPixelsPerLinePerDSCUnit, - enum dml2_output_format_class OutFormat, - bool UseDSC, - unsigned int NumberOfDSCSlices, - double SurfaceRequiredDISPCLKWithoutODMCombine, - double SurfaceRequiredDISPCLKWithODMCombineTwoToOne, - double SurfaceRequiredDISPCLKWithODMCombineThreeToOne, - double SurfaceRequiredDISPCLKWithODMCombineFourToOne) -{ - enum dml2_odm_mode MinimumRequiredODMModeForMaxDispClock; - enum dml2_odm_mode MinimumRequiredODMModeForMaxDSCHActive; - enum dml2_odm_mode MinimumRequiredODMModeForMax420HActive; - enum dml2_odm_mode ODMMode = dml2_odm_mode_bypass; - - MinimumRequiredODMModeForMaxDispClock = - (SurfaceRequiredDISPCLKWithoutODMCombine <= MaxDispclk) ? dml2_odm_mode_bypass : - (SurfaceRequiredDISPCLKWithODMCombineTwoToOne <= MaxDispclk) ? dml2_odm_mode_combine_2to1 : - (SurfaceRequiredDISPCLKWithODMCombineThreeToOne <= MaxDispclk) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; - if (ODMMode < MinimumRequiredODMModeForMaxDispClock) - ODMMode = MinimumRequiredODMModeForMaxDispClock; - - if (UseDSC) { - MinimumRequiredODMModeForMaxDSCHActive = - (HActive <= 1 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_bypass : - (HActive <= 2 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_combine_2to1 : - (HActive <= 3 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; - if (ODMMode < MinimumRequiredODMModeForMaxDSCHActive) - ODMMode = MinimumRequiredODMModeForMaxDSCHActive; - } - - if (OutFormat == dml2_420) { - MinimumRequiredODMModeForMax420HActive = - (HActive <= 1 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_bypass : - (HActive <= 2 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_combine_2to1 : - (HActive <= 3 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; - if (ODMMode < MinimumRequiredODMModeForMax420HActive) - ODMMode = MinimumRequiredODMModeForMax420HActive; - } - - if (UseDSC) { - if (ODMMode == dml2_odm_mode_bypass && NumberOfDSCSlices > 4) - ODMMode = dml2_odm_mode_combine_2to1; - if (ODMMode == dml2_odm_mode_combine_2to1 && NumberOfDSCSlices > 8) - ODMMode = dml2_odm_mode_combine_3to1; - if (ODMMode == dml2_odm_mode_combine_3to1 && NumberOfDSCSlices != 12) - ODMMode = dml2_odm_mode_combine_4to1; - } - - return ODMMode; -} - -static void CalculateODMConstraints( - enum dml2_odm_mode ODMUse, - double SurfaceRequiredDISPCLKWithoutODMCombine, - double SurfaceRequiredDISPCLKWithODMCombineTwoToOne, - double SurfaceRequiredDISPCLKWithODMCombineThreeToOne, - double SurfaceRequiredDISPCLKWithODMCombineFourToOne, - unsigned int MaximumPixelsPerLinePerDSCUnit, - /* Output */ - double *DISPCLKRequired, - unsigned int *NumberOfDPPRequired, - unsigned int *MaxHActiveForDSC, - unsigned int *MaxDSCSlices, - unsigned int *MaxHActiveFor420) -{ - switch (ODMUse) { - case dml2_odm_mode_combine_2to1: - *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineTwoToOne; - *NumberOfDPPRequired = 2; - break; - case dml2_odm_mode_combine_3to1: - *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineThreeToOne; - *NumberOfDPPRequired = 3; - break; - case dml2_odm_mode_combine_4to1: - *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineFourToOne; - *NumberOfDPPRequired = 4; - break; - case dml2_odm_mode_auto: - case dml2_odm_mode_split_1to2: - case dml2_odm_mode_mso_1to2: - case dml2_odm_mode_mso_1to4: - case dml2_odm_mode_bypass: - default: - *DISPCLKRequired = SurfaceRequiredDISPCLKWithoutODMCombine; - *NumberOfDPPRequired = 1; - break; - } - *MaxHActiveForDSC = *NumberOfDPPRequired * MaximumPixelsPerLinePerDSCUnit; - *MaxDSCSlices = *NumberOfDPPRequired * DML_MAX_NUM_OF_SLICES_PER_DSC; - *MaxHActiveFor420 = *NumberOfDPPRequired * DML2_MAX_FMT_420_BUFFER_WIDTH; -} - -static bool ValidateODMMode(enum dml2_odm_mode ODMMode, - double MaxDispclk, - unsigned int HActive, - enum dml2_output_format_class OutFormat, - bool UseDSC, - unsigned int NumberOfDSCSlices, - unsigned int TotalNumberOfActiveDPP, - unsigned int TotalNumberOfActiveOPP, - unsigned int MaxNumDPP, - unsigned int MaxNumOPP, - double DISPCLKRequired, - unsigned int NumberOfDPPRequired, - unsigned int MaxHActiveForDSC, - unsigned int MaxDSCSlices, - unsigned int MaxHActiveFor420) -{ - bool are_odm_segments_symmetrical = (ODMMode == dml2_odm_mode_combine_3to1) ? UseDSC : true; - bool is_max_dsc_slice_required = (ODMMode == dml2_odm_mode_combine_3to1); - unsigned int pixels_per_clock_cycle = (OutFormat == dml2_420 || OutFormat == dml2_n422) ? 2 : 1; - unsigned int h_timing_div_mode = - (ODMMode == dml2_odm_mode_combine_4to1 || ODMMode == dml2_odm_mode_combine_3to1) ? 4 : - (ODMMode == dml2_odm_mode_combine_2to1) ? 2 : pixels_per_clock_cycle; - - if (DISPCLKRequired > MaxDispclk) - return false; - if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP || (TotalNumberOfActiveOPP + NumberOfDPPRequired) > MaxNumOPP) - return false; - if (are_odm_segments_symmetrical) { - if (HActive % (NumberOfDPPRequired * pixels_per_clock_cycle)) - return false; - } - if (HActive % h_timing_div_mode) - /* - * TODO - OTG_H_TOTAL, OTG_H_BLANK_START/END and - * OTG_H_SYNC_A_START/END all need to be visible by h timing div - * mode. This logic only checks H active. - */ - return false; - - if (UseDSC) { - if (HActive > MaxHActiveForDSC) - return false; - if (NumberOfDSCSlices > MaxDSCSlices) - return false; - if (HActive % NumberOfDSCSlices) - return false; - if (NumberOfDSCSlices % NumberOfDPPRequired) - return false; - if (is_max_dsc_slice_required) { - if (NumberOfDSCSlices != MaxDSCSlices) - return false; - } - } - - if (OutFormat == dml2_420) { - if (HActive > MaxHActiveFor420) - return false; - } - - return true; -} - -static noinline_for_stack void CalculateODMMode( - unsigned int MaximumPixelsPerLinePerDSCUnit, - unsigned int HActive, - enum dml2_output_format_class OutFormat, - enum dml2_output_encoder_class Output, - enum dml2_odm_mode ODMUse, - double MaxDispclk, - bool DSCEnable, - unsigned int TotalNumberOfActiveDPP, - unsigned int TotalNumberOfActiveOPP, - unsigned int MaxNumDPP, - unsigned int MaxNumOPP, - double PixelClock, - unsigned int NumberOfDSCSlices, - - // Output - bool *TotalAvailablePipesSupport, - unsigned int *NumberOfDPP, - enum dml2_odm_mode *ODMMode, - double *RequiredDISPCLKPerSurface) -{ - double SurfaceRequiredDISPCLKWithoutODMCombine; - double SurfaceRequiredDISPCLKWithODMCombineTwoToOne; - double SurfaceRequiredDISPCLKWithODMCombineThreeToOne; - double SurfaceRequiredDISPCLKWithODMCombineFourToOne; - double DISPCLKRequired; - unsigned int NumberOfDPPRequired; - unsigned int MaxHActiveForDSC; - unsigned int MaxDSCSlices; - unsigned int MaxHActiveFor420; - bool success; - bool UseDSC = DSCEnable && (NumberOfDSCSlices > 0); - enum dml2_odm_mode DecidedODMMode; - bool isTMDS420 = (OutFormat == dml2_420 && Output == dml2_hdmi); - - SurfaceRequiredDISPCLKWithoutODMCombine = CalculateRequiredDispclk(dml2_odm_mode_bypass, PixelClock, isTMDS420); - SurfaceRequiredDISPCLKWithODMCombineTwoToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_2to1, PixelClock, isTMDS420); - SurfaceRequiredDISPCLKWithODMCombineThreeToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_3to1, PixelClock, isTMDS420); - SurfaceRequiredDISPCLKWithODMCombineFourToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_4to1, PixelClock, isTMDS420); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: ODMUse = %d\n", __func__, ODMUse); - DML_LOG_VERBOSE("DML::%s: Output = %d\n", __func__, Output); - DML_LOG_VERBOSE("DML::%s: DSCEnable = %d\n", __func__, DSCEnable); - DML_LOG_VERBOSE("DML::%s: MaxDispclk = %f\n", __func__, MaxDispclk); - DML_LOG_VERBOSE("DML::%s: MaximumPixelsPerLinePerDSCUnit = %d\n", __func__, MaximumPixelsPerLinePerDSCUnit); - DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithoutODMCombine = %f\n", __func__, SurfaceRequiredDISPCLKWithoutODMCombine); - DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineTwoToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineTwoToOne); - DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineThreeToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineThreeToOne); - DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineFourToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineFourToOne); -#endif - if (ODMUse == dml2_odm_mode_auto) - DecidedODMMode = DecideODMMode(HActive, - MaxDispclk, - MaximumPixelsPerLinePerDSCUnit, - OutFormat, - UseDSC, - NumberOfDSCSlices, - SurfaceRequiredDISPCLKWithoutODMCombine, - SurfaceRequiredDISPCLKWithODMCombineTwoToOne, - SurfaceRequiredDISPCLKWithODMCombineThreeToOne, - SurfaceRequiredDISPCLKWithODMCombineFourToOne); - else - DecidedODMMode = ODMUse; - CalculateODMConstraints(DecidedODMMode, - SurfaceRequiredDISPCLKWithoutODMCombine, - SurfaceRequiredDISPCLKWithODMCombineTwoToOne, - SurfaceRequiredDISPCLKWithODMCombineThreeToOne, - SurfaceRequiredDISPCLKWithODMCombineFourToOne, - MaximumPixelsPerLinePerDSCUnit, - &DISPCLKRequired, - &NumberOfDPPRequired, - &MaxHActiveForDSC, - &MaxDSCSlices, - &MaxHActiveFor420); - success = ValidateODMMode(DecidedODMMode, - MaxDispclk, - HActive, - OutFormat, - UseDSC, - NumberOfDSCSlices, - TotalNumberOfActiveDPP, - TotalNumberOfActiveOPP, - MaxNumDPP, - MaxNumOPP, - DISPCLKRequired, - NumberOfDPPRequired, - MaxHActiveForDSC, - MaxDSCSlices, - MaxHActiveFor420); - - *ODMMode = DecidedODMMode; - *TotalAvailablePipesSupport = success; - *NumberOfDPP = NumberOfDPPRequired; - *RequiredDISPCLKPerSurface = success ? DISPCLKRequired : 0; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: ODMMode = %d\n", __func__, *ODMMode); - DML_LOG_VERBOSE("DML::%s: NumberOfDPP = %d\n", __func__, *NumberOfDPP); - DML_LOG_VERBOSE("DML::%s: TotalAvailablePipesSupport = %d\n", __func__, *TotalAvailablePipesSupport); - DML_LOG_VERBOSE("DML::%s: RequiredDISPCLKPerSurface = %f\n", __func__, *RequiredDISPCLKPerSurface); -#endif -} - -static noinline_for_stack void CalculateOutputLink( - struct dml2_core_internal_scratch *s, - double PHYCLK, - double PHYCLKD18, - double PHYCLKD32, - double Downspreading, - enum dml2_output_encoder_class Output, - enum dml2_output_format_class OutputFormat, - unsigned int HTotal, - unsigned int HActive, - double PixelClockBackEnd, - double ForcedOutputLinkBPP, - unsigned int DSCInputBitPerComponent, - unsigned int NumberOfDSCSlices, - double AudioSampleRate, - unsigned int AudioSampleLayout, - enum dml2_odm_mode ODMModeNoDSC, - enum dml2_odm_mode ODMModeDSC, - enum dml2_dsc_enable_option DSCEnable, - unsigned int OutputLinkDPLanes, - enum dml2_output_link_dp_rate OutputLinkDPRate, - - // Output - bool *RequiresDSC, - bool *RequiresFEC, - double *OutBpp, - enum dml2_core_internal_output_type *OutputType, - enum dml2_core_internal_output_type_rate *OutputRate, - unsigned int *RequiredSlots) -{ - bool LinkDSCEnable; - unsigned int dummy; - *RequiresDSC = false; - *RequiresFEC = false; - *OutBpp = 0; - - *OutputType = dml2_core_internal_output_type_unknown; - *OutputRate = dml2_core_internal_output_rate_unknown; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DSCEnable = %u (dis, en, en_if_necessary)\n", __func__, DSCEnable); - DML_LOG_VERBOSE("DML::%s: PHYCLK = %f\n", __func__, PHYCLK); - DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); - DML_LOG_VERBOSE("DML::%s: AudioSampleRate = %f\n", __func__, AudioSampleRate); - DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); - DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); - DML_LOG_VERBOSE("DML::%s: ODMModeNoDSC = %u\n", __func__, ODMModeNoDSC); - DML_LOG_VERBOSE("DML::%s: ODMModeDSC = %u\n", __func__, ODMModeDSC); - DML_LOG_VERBOSE("DML::%s: ForcedOutputLinkBPP = %f\n", __func__, ForcedOutputLinkBPP); - DML_LOG_VERBOSE("DML::%s: Output (encoder) = %u\n", __func__, Output); - DML_LOG_VERBOSE("DML::%s: OutputLinkDPRate = %u\n", __func__, OutputLinkDPRate); -#endif - { - if (Output == dml2_hdmi) { - *RequiresDSC = false; - *RequiresFEC = false; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, math_min2(600, PHYCLK) * 10, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, false, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = "HDMI"; - *OutputType = dml2_core_internal_output_type_hdmi; - } else if (Output == dml2_dp || Output == dml2_dp2p0 || Output == dml2_edp) { - if (DSCEnable == dml2_dsc_enable) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml2_dp || Output == dml2_dp2p0) { - *RequiresFEC = true; - } else { - *RequiresFEC = false; - } - } else { - *RequiresDSC = false; - LinkDSCEnable = false; - if (Output == dml2_dp2p0) { - *RequiresFEC = true; - } else { - *RequiresFEC = false; - } - } - if (Output == dml2_dp2p0) { - *OutBpp = 0; - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr10) && PHYCLKD32 >= 10000.0 / 32) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && PHYCLKD32 < 13500.0 / 32 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR10"; - *OutputType = dml2_core_internal_output_type_dp2p0; - *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr10; - } - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr13p5) && *OutBpp == 0 && PHYCLKD32 >= 13500.0 / 32) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && PHYCLKD32 < 20000.0 / 32 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR13p5"; - *OutputType = dml2_core_internal_output_type_dp2p0; - *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr13p5; - } - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr20) && *OutBpp == 0 && PHYCLKD32 >= 20000.0 / 32) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " UHBR20"; - *OutputType = dml2_core_internal_output_type_dp2p0; - *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr20; - } - } else { // output is dp or edp - *OutBpp = 0; - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr) && PHYCLK >= 270) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - if (*OutBpp == 0 && PHYCLK < 540 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml2_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR"; - *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; - *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr; - } - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr2) && *OutBpp == 0 && PHYCLK >= 540) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && PHYCLK < 810 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml2_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR2"; - *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; - *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr2; - } - if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr3) && *OutBpp == 0 && PHYCLK >= 810) { // VBA_ERROR, vba code doesn't have hbr3 check - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - - if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - if (Output == dml2_dp) { - *RequiresFEC = true; - } - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, - OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); - } - //OutputTypeAndRate = Output & " HBR3"; - *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; - *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr3; - } - } - } else if (Output == dml2_hdmifrl) { - if (DSCEnable == dml2_dsc_enable) { - *RequiresDSC = true; - LinkDSCEnable = true; - *RequiresFEC = true; - } else { - *RequiresDSC = false; - LinkDSCEnable = false; - *RequiresFEC = false; - } - *OutBpp = 0; - if (PHYCLKD18 >= 3000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 3000, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = Output & "3x3"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_3x3; - } - if (*OutBpp == 0 && PHYCLKD18 >= 6000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 6000, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = Output & "6x3"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_6x3; - } - if (*OutBpp == 0 && PHYCLKD18 >= 6000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 6000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = Output & "6x4"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_6x4; - } - if (*OutBpp == 0 && PHYCLKD18 >= 8000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 8000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - //OutputTypeAndRate = Output & "8x4"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_8x4; - } - if (*OutBpp == 0 && PHYCLKD18 >= 10000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 10000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0 && PHYCLKD18 < 12000.0 / 18) { - *RequiresDSC = true; - LinkDSCEnable = true; - *RequiresFEC = true; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 10000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - } - //OutputTypeAndRate = Output & "10x4"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_10x4; - } - if (*OutBpp == 0 && PHYCLKD18 >= 12000.0 / 18) { - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 12000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { - *RequiresDSC = true; - LinkDSCEnable = true; - *RequiresFEC = true; - *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 12000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); - } - //OutputTypeAndRate = Output & "12x4"; - *OutputType = dml2_core_internal_output_type_hdmifrl; - *OutputRate = dml2_core_internal_output_rate_hdmi_rate_12x4; - } - } - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: RequiresDSC = %u\n", __func__, *RequiresDSC); - DML_LOG_VERBOSE("DML::%s: RequiresFEC = %u\n", __func__, *RequiresFEC); - DML_LOG_VERBOSE("DML::%s: OutBpp = %f\n", __func__, *OutBpp); -#endif -} - -static double CalculateWriteBackDISPCLK( - enum dml2_source_format_class WritebackPixelFormat, - double PixelClock, - double WritebackHRatio, - double WritebackVRatio, - unsigned int WritebackHTaps, - unsigned int WritebackVTaps, - unsigned int WritebackSourceWidth, - unsigned int WritebackDestinationWidth, - unsigned int HTotal, - unsigned int WritebackLineBufferSize) -{ - double DISPCLK_H, DISPCLK_V, DISPCLK_HB; - - DISPCLK_H = PixelClock * math_ceil2((double)WritebackHTaps / 8.0, 1) / WritebackHRatio; - DISPCLK_V = PixelClock * (WritebackVTaps * math_ceil2((double)WritebackDestinationWidth / 6.0, 1) + 8.0) / (double)HTotal; - DISPCLK_HB = PixelClock * WritebackVTaps * (WritebackDestinationWidth * WritebackVTaps - WritebackLineBufferSize / 57.0) / 6.0 / (double)WritebackSourceWidth; - return math_max3(DISPCLK_H, DISPCLK_V, DISPCLK_HB); -} - -static double RequiredDTBCLK( - bool DSCEnable, - double PixelClock, - enum dml2_output_format_class OutputFormat, - double OutputBpp, - unsigned int DSCSlices, - unsigned int HTotal, - unsigned int HActive, - unsigned int AudioRate, - unsigned int AudioLayout) -{ - if (DSCEnable != true) { - return math_max2(PixelClock / 4.0 * OutputBpp / 24.0, 25.0); - } else { - double PixelWordRate = PixelClock / (OutputFormat == dml2_444 ? 1 : 2); - double HCActive = math_ceil2(DSCSlices * math_ceil2(OutputBpp * math_ceil2(HActive / DSCSlices, 1) / 8.0, 1) / 3.0, 1); - double HCBlank = 64 + 32 * math_ceil2(AudioRate * (AudioLayout == 1 ? 1 : 0.25) * HTotal / (PixelClock * 1000), 1); - double AverageTribyteRate = PixelWordRate * (HCActive + HCBlank) / HTotal; - double HActiveTribyteRate = PixelWordRate * HCActive / HActive; - return math_max4(PixelWordRate / 4.0, AverageTribyteRate / 4.0, HActiveTribyteRate / 4.0, 25.0) * 1.002; - } -} - -static unsigned int DSCDelayRequirement( - bool DSCEnabled, - enum dml2_odm_mode ODMMode, - unsigned int DSCInputBitPerComponent, - double OutputBpp, - unsigned int HActive, - unsigned int HTotal, - unsigned int NumberOfDSCSlices, - enum dml2_output_format_class OutputFormat, - enum dml2_output_encoder_class Output, - double PixelClock, - double PixelClockBackEnd) -{ - unsigned int DSCDelayRequirement_val = 0; - unsigned int NumberOfDSCSlicesFactor = 1; - - if (DSCEnabled == true && OutputBpp != 0) { - - if (ODMMode == dml2_odm_mode_combine_4to1) - NumberOfDSCSlicesFactor = 4; - else if (ODMMode == dml2_odm_mode_combine_3to1) - NumberOfDSCSlicesFactor = 3; - else if (ODMMode == dml2_odm_mode_combine_2to1) - NumberOfDSCSlicesFactor = 2; - - DSCDelayRequirement_val = NumberOfDSCSlicesFactor * (dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (unsigned int)(math_ceil2((double)HActive / (double)NumberOfDSCSlices, 1.0)), - (NumberOfDSCSlices / NumberOfDSCSlicesFactor), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output)); - - DSCDelayRequirement_val = (unsigned int)(DSCDelayRequirement_val + (HTotal - HActive) * math_ceil2((double)DSCDelayRequirement_val / (double)HActive, 1.0)); - DSCDelayRequirement_val = (unsigned int)(DSCDelayRequirement_val * PixelClock / PixelClockBackEnd); - - } else { - DSCDelayRequirement_val = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DSCEnabled= %u\n", __func__, DSCEnabled); - DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, ODMMode); - DML_LOG_VERBOSE("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); - DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); - DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); - DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); - DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); - DML_LOG_VERBOSE("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); - DML_LOG_VERBOSE("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); - DML_LOG_VERBOSE("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); - DML_LOG_VERBOSE("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); -#endif - - return DSCDelayRequirement_val; -} - -static void CalculateSurfaceSizeInMall( - const struct dml2_display_cfg *display_cfg, - unsigned int NumberOfActiveSurfaces, - unsigned int MALLAllocatedForDCN, - unsigned int BytesPerPixelY[], - unsigned int BytesPerPixelC[], - unsigned int Read256BytesBlockWidthY[], - unsigned int Read256BytesBlockWidthC[], - unsigned int Read256BytesBlockHeightY[], - unsigned int Read256BytesBlockHeightC[], - unsigned int ReadBlockWidthY[], - unsigned int ReadBlockWidthC[], - unsigned int ReadBlockHeightY[], - unsigned int ReadBlockHeightC[], - - // Output - unsigned int SurfaceSizeInMALL[], - bool *ExceededMALLSize) -{ - unsigned int TotalSurfaceSizeInMALLForSS = 0; - unsigned int TotalSurfaceSizeInMALLForSubVP = 0; - unsigned int MALLAllocatedForDCNInBytes = MALLAllocatedForDCN * 1024 * 1024; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - const struct dml2_composition_cfg *composition = &display_cfg->plane_descriptors[k].composition; - const struct dml2_surface_cfg *surface = &display_cfg->plane_descriptors[k].surface; - - if (composition->viewport.stationary) { - SurfaceSizeInMALL[k] = (unsigned int)(math_min2(math_ceil2((double)surface->plane0.width, ReadBlockWidthY[k]), - math_floor2(composition->viewport.plane0.x_start + composition->viewport.plane0.width + ReadBlockWidthY[k] - 1, ReadBlockWidthY[k]) - - math_floor2((double)composition->viewport.plane0.x_start, ReadBlockWidthY[k])) * - math_min2(math_ceil2((double)surface->plane0.height, ReadBlockHeightY[k]), - math_floor2((double)composition->viewport.plane0.y_start + composition->viewport.plane0.height + ReadBlockHeightY[k] - 1, ReadBlockHeightY[k]) - - math_floor2((double)composition->viewport.plane0.y_start, ReadBlockHeightY[k])) * BytesPerPixelY[k]); - - if (ReadBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (unsigned int)(SurfaceSizeInMALL[k] + - math_min2(math_ceil2((double)surface->plane1.width, ReadBlockWidthC[k]), - math_floor2((double)composition->viewport.plane1.y_start + composition->viewport.plane1.width + ReadBlockWidthC[k] - 1, ReadBlockWidthC[k]) - - math_floor2((double)composition->viewport.plane1.y_start, ReadBlockWidthC[k])) * - math_min2(math_ceil2((double)surface->plane1.height, ReadBlockHeightC[k]), - math_floor2((double)composition->viewport.plane1.y_start + composition->viewport.plane1.height + ReadBlockHeightC[k] - 1, ReadBlockHeightC[k]) - - math_floor2(composition->viewport.plane1.y_start, ReadBlockHeightC[k])) * BytesPerPixelC[k]); - } - } else { - SurfaceSizeInMALL[k] = (unsigned int)(math_ceil2(math_min2(surface->plane0.width, composition->viewport.plane0.width + ReadBlockWidthY[k] - 1), ReadBlockWidthY[k]) * - math_ceil2(math_min2(surface->plane0.height, composition->viewport.plane0.height + ReadBlockHeightY[k] - 1), ReadBlockHeightY[k]) * BytesPerPixelY[k]); - if (ReadBlockWidthC[k] > 0) { - SurfaceSizeInMALL[k] = (unsigned int)(SurfaceSizeInMALL[k] + - math_ceil2(math_min2(surface->plane1.width, composition->viewport.plane1.width + ReadBlockWidthC[k] - 1), ReadBlockWidthC[k]) * - math_ceil2(math_min2(surface->plane1.height, composition->viewport.plane1.height + ReadBlockHeightC[k] - 1), ReadBlockHeightC[k]) * BytesPerPixelC[k]); - } - } - } - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - /* SS and Subvp counted separate as they are never used at the same time */ - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) - TotalSurfaceSizeInMALLForSubVP += SurfaceSizeInMALL[k]; - else if (display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable) - TotalSurfaceSizeInMALLForSS += SurfaceSizeInMALL[k]; - } - - *ExceededMALLSize = (TotalSurfaceSizeInMALLForSS > MALLAllocatedForDCNInBytes) || - (TotalSurfaceSizeInMALLForSubVP > MALLAllocatedForDCNInBytes); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: MALLAllocatedForDCN = %u\n", __func__, MALLAllocatedForDCN * 1024 * 1024); - DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSubVP = %u\n", __func__, TotalSurfaceSizeInMALLForSubVP); - DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSS = %u\n", __func__, TotalSurfaceSizeInMALLForSS); - DML_LOG_VERBOSE("DML::%s: ExceededMALLSize = %u\n", __func__, *ExceededMALLSize); -#endif -} - -static void calculate_tdlut_setting( - struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_calculate_tdlut_setting_params *p) -{ - // locals - unsigned int tdlut_bpe = 8; - unsigned int tdlut_width; - unsigned int tdlut_pitch_bytes; - unsigned int tdlut_footprint_bytes; - unsigned int vmpg_bytes; - unsigned int tdlut_vmpg_per_frame; - unsigned int tdlut_pte_req_per_frame; - unsigned int tdlut_bytes_per_line; - double tdlut_drain_rate; - unsigned int tdlut_mpc_width; - unsigned int tdlut_bytes_per_group_simple; - - if (!p->setup_for_tdlut) { - *p->tdlut_groups_per_2row_ub = 0; - *p->tdlut_opt_time = 0; - *p->tdlut_drain_time = 0; - *p->tdlut_bytes_to_deliver = 0; - *p->tdlut_bytes_per_group = 0; - *p->tdlut_pte_bytes_per_frame = 0; - *p->tdlut_bytes_per_frame = 0; - return; - } - - if (p->tdlut_mpc_width_flag) { - tdlut_mpc_width = 33; - tdlut_bytes_per_group_simple = 39*256; - } else { - tdlut_mpc_width = 17; - tdlut_bytes_per_group_simple = 10*256; - } - - vmpg_bytes = p->gpuvm_page_size_kbytes * 1024; - - if (p->tdlut_addressing_mode == dml2_tdlut_simple_linear) { - if (p->tdlut_width_mode == dml2_tdlut_width_17_cube) - tdlut_width = 4916; - else - tdlut_width = 35940; - } else { - if (p->tdlut_width_mode == dml2_tdlut_width_17_cube) - tdlut_width = 17; - else // dml2_tdlut_width_33_cube - tdlut_width = 33; - } - - if (p->is_gfx11) - tdlut_pitch_bytes = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 256); //256B alignment - else - tdlut_pitch_bytes = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 128); //128B alignment - - if (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) - tdlut_footprint_bytes = tdlut_pitch_bytes * tdlut_width * tdlut_width; - else - tdlut_footprint_bytes = tdlut_pitch_bytes; - - if (!p->gpuvm_enable) { - tdlut_vmpg_per_frame = 0; - tdlut_pte_req_per_frame = 0; - } else { - tdlut_vmpg_per_frame = (unsigned int)math_ceil2(tdlut_footprint_bytes - 1, vmpg_bytes) / vmpg_bytes + 1; - tdlut_pte_req_per_frame = (unsigned int)math_ceil2(tdlut_vmpg_per_frame - 1, 8) / 8 + 1; - } - tdlut_bytes_per_line = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 64); //64b request - *p->tdlut_pte_bytes_per_frame = tdlut_pte_req_per_frame * 64; - - if (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) { - //the tdlut_width is either 17 or 33 but the 33x33x33 is subsampled every other line/slice - *p->tdlut_bytes_per_frame = tdlut_bytes_per_line * tdlut_mpc_width * tdlut_mpc_width; - *p->tdlut_bytes_per_group = tdlut_bytes_per_line * tdlut_mpc_width; - //the delivery cycles is DispClk cycles per line * number of lines * number of slices - //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width; - tdlut_drain_rate = tdlut_bytes_per_line * p->dispclk_mhz / math_ceil2(tdlut_mpc_width/2.0, 1); - } else { - //tdlut_addressing_mode = tdlut_simple_linear, 3dlut width should be 4*1229=4916 elements - *p->tdlut_bytes_per_frame = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 256); - *p->tdlut_bytes_per_group = tdlut_bytes_per_group_simple; - //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_width/2.0, 1); - tdlut_drain_rate = 2 * tdlut_bpe * p->dispclk_mhz; - } - - //the tdlut is fetched during the 2 row times of prefetch. - if (p->setup_for_tdlut) { - *p->tdlut_groups_per_2row_ub = (unsigned int)math_ceil2((double) *p->tdlut_bytes_per_frame / *p->tdlut_bytes_per_group, 1); - if (*p->tdlut_bytes_per_frame > p->cursor_buffer_size * 1024) - *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; - else - *p->tdlut_opt_time = 0; - *p->tdlut_drain_time = p->cursor_buffer_size * 1024 / tdlut_drain_rate; - *p->tdlut_bytes_to_deliver = (unsigned int) (p->cursor_buffer_size * 1024.0); - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %d\n", __func__, p->gpuvm_enable); - DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %d\n", __func__, vmpg_bytes); - DML_LOG_VERBOSE("DML::%s: tdlut_vmpg_per_frame = %d\n", __func__, tdlut_vmpg_per_frame); - DML_LOG_VERBOSE("DML::%s: tdlut_pte_req_per_frame = %d\n", __func__, tdlut_pte_req_per_frame); - - DML_LOG_VERBOSE("DML::%s: dispclk_mhz = %f\n", __func__, p->dispclk_mhz); - DML_LOG_VERBOSE("DML::%s: tdlut_width = %u\n", __func__, tdlut_width); - DML_LOG_VERBOSE("DML::%s: tdlut_addressing_mode = %s\n", __func__, (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) ? "sw_linear" : "simple_linear"); - DML_LOG_VERBOSE("DML::%s: tdlut_pitch_bytes = %u\n", __func__, tdlut_pitch_bytes); - DML_LOG_VERBOSE("DML::%s: tdlut_footprint_bytes = %u\n", __func__, tdlut_footprint_bytes); - DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_frame = %u\n", __func__, *p->tdlut_bytes_per_frame); - DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_line = %u\n", __func__, tdlut_bytes_per_line); - DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_group = %u\n", __func__, *p->tdlut_bytes_per_group); - DML_LOG_VERBOSE("DML::%s: tdlut_drain_rate = %f\n", __func__, tdlut_drain_rate); - DML_LOG_VERBOSE("DML::%s: tdlut_delivery_cycles = %u\n", __func__, p->tdlut_addressing_mode == dml2_tdlut_sw_linear ? (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width : (unsigned int)math_ceil2(tdlut_width/2.0, 1)); - DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, *p->tdlut_opt_time); - DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, *p->tdlut_drain_time); - DML_LOG_VERBOSE("DML::%s: tdlut_bytes_to_deliver = %d\n", __func__, *p->tdlut_bytes_to_deliver); - DML_LOG_VERBOSE("DML::%s: tdlut_groups_per_2row_ub = %d\n", __func__, *p->tdlut_groups_per_2row_ub); -#endif -} - -static void CalculateTarb( - const struct dml2_display_cfg *display_cfg, - unsigned int PixelChunkSizeInKByte, - unsigned int NumberOfActiveSurfaces, - unsigned int NumberOfDPP[], - unsigned int dpte_group_bytes[], - unsigned int tdlut_bytes_per_group[], - double HostVMInefficiencyFactor, - double HostVMInefficiencyFactorPrefetch, - unsigned int HostVMMinPageSize, - double ReturnBW, - unsigned int MetaChunkSize, - - // output - double *Tarb, - double *Tarb_prefetch) -{ - double extra_bytes = 0; - double extra_bytes_prefetch = 0; - double HostVMDynamicLevels = CalculateHostVMDynamicLevels(display_cfg->gpuvm_enable, display_cfg->hostvm_enable, HostVMMinPageSize, display_cfg->hostvm_max_non_cached_page_table_levels); - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - extra_bytes = extra_bytes + (NumberOfDPP[k] * PixelChunkSizeInKByte * 1024); - - if (display_cfg->plane_descriptors[k].surface.dcc.enable) - extra_bytes = extra_bytes + (MetaChunkSize * 1024); - - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) - extra_bytes = extra_bytes + tdlut_bytes_per_group[k]; - } - - extra_bytes_prefetch = extra_bytes; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (display_cfg->gpuvm_enable == true) { - extra_bytes = extra_bytes + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactor; - extra_bytes_prefetch = extra_bytes_prefetch + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactorPrefetch; - } - } - *Tarb = extra_bytes / ReturnBW; - *Tarb_prefetch = extra_bytes_prefetch / ReturnBW; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %d\n", __func__, PixelChunkSizeInKByte); - DML_LOG_VERBOSE("DML::%s: MetaChunkSize = %d\n", __func__, MetaChunkSize); - DML_LOG_VERBOSE("DML::%s: extra_bytes = %f\n", __func__, extra_bytes); - DML_LOG_VERBOSE("DML::%s: extra_bytes_prefetch = %f\n", __func__, extra_bytes_prefetch); -#endif -} - -static double CalculateTWait( - long reserved_vblank_time_ns, - double UrgentLatency, - double Ttrip, - double g6_temp_read_blackout_us) -{ - double TWait; - double t_urg_trip = math_max2(UrgentLatency, Ttrip); - TWait = math_max2(reserved_vblank_time_ns/1000.0, g6_temp_read_blackout_us) + t_urg_trip; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: reserved_vblank_time_ns = %ld\n", __func__, reserved_vblank_time_ns); - DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); - DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, Ttrip); - DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, TWait); -#endif - return TWait; -} - - -static void CalculateVUpdateAndDynamicMetadataParameters( - unsigned int MaxInterDCNTileRepeaters, - double Dppclk, - double Dispclk, - double DCFClkDeepSleep, - double PixelClock, - unsigned int HTotal, - unsigned int VBlank, - unsigned int DynamicMetadataTransmittedBytes, - unsigned int DynamicMetadataLinesBeforeActiveRequired, - unsigned int InterlaceEnable, - bool ProgressiveToInterlaceUnitInOPP, - - // Output - double *TSetup, - double *Tdmbf, - double *Tdmec, - double *Tdmsks, - unsigned int *VUpdateOffsetPix, - unsigned int *VUpdateWidthPix, - unsigned int *VReadyOffsetPix) -{ - double TotalRepeaterDelayTime; - TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2 / Dppclk + 3 / Dispclk); - *VUpdateWidthPix = (unsigned int)(math_ceil2((14.0 / DCFClkDeepSleep + 12.0 / Dppclk + TotalRepeaterDelayTime) * PixelClock, 1.0)); - *VReadyOffsetPix = (unsigned int)(math_ceil2(math_max2(150.0 / Dppclk, TotalRepeaterDelayTime + 20.0 / DCFClkDeepSleep + 10.0 / Dppclk) * PixelClock, 1.0)); - *VUpdateOffsetPix = (unsigned int)(math_ceil2(HTotal / 4.0, 1.0)); - *TSetup = (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock; - *Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / Dispclk; - *Tdmec = HTotal / PixelClock; - - if (DynamicMetadataLinesBeforeActiveRequired == 0) { - *Tdmsks = VBlank * HTotal / PixelClock / 2.0; - } else { - *Tdmsks = DynamicMetadataLinesBeforeActiveRequired * HTotal / PixelClock; - } - if (InterlaceEnable == 1 && ProgressiveToInterlaceUnitInOPP == false) { - *Tdmsks = *Tdmsks / 2; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); - DML_LOG_VERBOSE("DML::%s: VBlank = %u\n", __func__, VBlank); - DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); - DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); - DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, Dppclk); - DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); - DML_LOG_VERBOSE("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); - DML_LOG_VERBOSE("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); - - DML_LOG_VERBOSE("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); - DML_LOG_VERBOSE("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); - DML_LOG_VERBOSE("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); - - DML_LOG_VERBOSE("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); -#endif -} - -static double get_urgent_bandwidth_required( - struct dml2_core_shared_get_urgent_bandwidth_required_locals *l, - const struct dml2_display_cfg *display_cfg, - enum dml2_core_internal_soc_state_type state_type, - enum dml2_core_internal_bw_type bw_type, - bool inc_flip_bw, // including flip bw - bool use_qual_row_bw, - unsigned int NumberOfActiveSurfaces, - unsigned int NumberOfDPP[], - double dcc_dram_bw_nom_overhead_factor_p0[], - double dcc_dram_bw_nom_overhead_factor_p1[], - double dcc_dram_bw_pref_overhead_factor_p0[], - double dcc_dram_bw_pref_overhead_factor_p1[], - double mall_prefetch_sdp_overhead_factor[], - double mall_prefetch_dram_overhead_factor[], - double ReadBandwidthLuma[], - double ReadBandwidthChroma[], - double PrefetchBandwidthLuma[], - double PrefetchBandwidthChroma[], - double PrefetchBandwidthMax[], - double excess_vactive_fill_bw_l[], - double excess_vactive_fill_bw_c[], - double cursor_bw[], - double dpte_row_bw[], - double meta_row_bw[], - double prefetch_cursor_bw[], - double prefetch_vmrow_bw[], - double flip_bw[], - double UrgentBurstFactorLuma[], - double UrgentBurstFactorChroma[], - double UrgentBurstFactorCursor[], - double UrgentBurstFactorLumaPre[], - double UrgentBurstFactorChromaPre[], - double UrgentBurstFactorCursorPre[], - /* outputs */ - double surface_required_bw[], - double surface_peak_required_bw[]) -{ - // set inc_flip_bw = 0 for total_dchub_urgent_read_bw_noflip calculation, 1 for total_dchub_urgent_read_bw as described in the MAS - // set use_qual_row_bw = 1 to calculate using qualified row bandwidth, used for total_flip_bw calculation - - memset(l, 0, sizeof(struct dml2_core_shared_get_urgent_bandwidth_required_locals)); - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - l->mall_svp_prefetch_factor = (state_type == dml2_core_internal_soc_state_svp_prefetch) ? (bw_type == dml2_core_internal_bw_dram ? mall_prefetch_dram_overhead_factor[k] : mall_prefetch_sdp_overhead_factor[k]) : 1.0; - l->tmp_nom_adj_factor_p0 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p0[k] : 1.0) * l->mall_svp_prefetch_factor; - l->tmp_nom_adj_factor_p1 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p1[k] : 1.0) * l->mall_svp_prefetch_factor; - l->tmp_pref_adj_factor_p0 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_pref_overhead_factor_p0[k] : 1.0) * l->mall_svp_prefetch_factor; - l->tmp_pref_adj_factor_p1 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_pref_overhead_factor_p1[k] : 1.0) * l->mall_svp_prefetch_factor; - - l->adj_factor_p0 = UrgentBurstFactorLuma[k] * l->tmp_nom_adj_factor_p0; - l->adj_factor_p1 = UrgentBurstFactorChroma[k] * l->tmp_nom_adj_factor_p1; - l->adj_factor_cur = UrgentBurstFactorCursor[k]; - l->adj_factor_p0_pre = UrgentBurstFactorLumaPre[k] * l->tmp_pref_adj_factor_p0; - l->adj_factor_p1_pre = UrgentBurstFactorChromaPre[k] * l->tmp_pref_adj_factor_p1; - l->adj_factor_cur_pre = UrgentBurstFactorCursorPre[k]; - - bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]); - bool exclude_this_plane = false; - - // Exclude phantom pipe in bw calculation for non svp prefetch state - if (state_type != dml2_core_internal_soc_state_svp_prefetch && is_phantom) - exclude_this_plane = true; - - // The qualified row bandwidth, qual_row_bw, accounts for the regular non-flip row bandwidth when there is no possible immediate flip or HostVM invalidation flip. - // The qual_row_bw is zero if HostVM is possible and only non-zero and equal to row_bw(i) if immediate flip is not allowed for that pipe. - if (use_qual_row_bw) { - if (display_cfg->hostvm_enable) - l->per_plane_flip_bw[k] = 0; // qual_row_bw - else if (!display_cfg->plane_descriptors[k].immediate_flip) - l->per_plane_flip_bw[k] = NumberOfDPP[k] * (dpte_row_bw[k] + meta_row_bw[k]); - } else { - // the final_flip_bw includes the regular row_bw when immediate flip is disallowed (and no HostVM) - if ((!display_cfg->plane_descriptors[k].immediate_flip && !display_cfg->hostvm_enable) || !inc_flip_bw) - l->per_plane_flip_bw[k] = NumberOfDPP[k] * (dpte_row_bw[k] + meta_row_bw[k]); - else - l->per_plane_flip_bw[k] = NumberOfDPP[k] * flip_bw[k]; - } - - if (!exclude_this_plane) { - l->vm_row_bw = NumberOfDPP[k] * prefetch_vmrow_bw[k]; - l->flip_and_active_bw = l->per_plane_flip_bw[k] + ReadBandwidthLuma[k] * l->adj_factor_p0 + ReadBandwidthChroma[k] * l->adj_factor_p1 + cursor_bw[k] * l->adj_factor_cur; - l->flip_and_prefetch_bw = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; - l->flip_and_prefetch_bw_max = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthMax[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; - l->active_and_excess_bw = (ReadBandwidthLuma[k] + excess_vactive_fill_bw_l[k]) * l->tmp_nom_adj_factor_p0 + (ReadBandwidthChroma[k] + excess_vactive_fill_bw_c[k]) * l->tmp_nom_adj_factor_p1 + dpte_row_bw[k] + meta_row_bw[k]; - surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_max); - - /* export peak required bandwidth for the surface */ - surface_peak_required_bw[k] = math_max2(surface_required_bw[k], surface_peak_required_bw[k]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, max1: vm_row_bw=%f\n", __func__, k, l->vm_row_bw); - DML_LOG_VERBOSE("DML::%s: k=%d, max2: flip_and_active_bw=%f\n", __func__, k, l->flip_and_active_bw); - DML_LOG_VERBOSE("DML::%s: k=%d, max3: flip_and_prefetch_bw=%f\n", __func__, k, l->flip_and_prefetch_bw); - DML_LOG_VERBOSE("DML::%s: k=%d, max4: active_and_excess_bw=%f\n", __func__, k, l->active_and_excess_bw); - DML_LOG_VERBOSE("DML::%s: k=%d, surface_required_bw=%f\n", __func__, k, surface_required_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, surface_peak_required_bw=%f\n", __func__, k, surface_peak_required_bw[k]); -#endif - } else { - surface_required_bw[k] = 0.0; - } - - l->required_bandwidth_mbps += surface_required_bw[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, NumberOfDPP=%d\n", __func__, k, NumberOfDPP[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, use_qual_row_bw=%d\n", __func__, k, use_qual_row_bw); - DML_LOG_VERBOSE("DML::%s: k=%d, immediate_flip=%d\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); - DML_LOG_VERBOSE("DML::%s: k=%d, mall_svp_prefetch_factor=%f\n", __func__, k, l->mall_svp_prefetch_factor); - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0=%f\n", __func__, k, l->adj_factor_p0); - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1=%f\n", __func__, k, l->adj_factor_p1); - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur=%f\n", __func__, k, l->adj_factor_cur); - - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0_pre=%f\n", __func__, k, l->adj_factor_p0_pre); - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1_pre=%f\n", __func__, k, l->adj_factor_p1_pre); - DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur_pre=%f\n", __func__, k, l->adj_factor_cur_pre); - - DML_LOG_VERBOSE("DML::%s: k=%d, per_plane_flip_bw=%f\n", __func__, k, l->per_plane_flip_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_vmrow_bw=%f\n", __func__, k, prefetch_vmrow_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthLuma=%f\n", __func__, k, ReadBandwidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthChroma=%f\n", __func__, k, ReadBandwidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_l=%f\n", __func__, k, excess_vactive_fill_bw_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_c=%f\n", __func__, k, excess_vactive_fill_bw_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, cursor_bw=%f\n", __func__, k, cursor_bw[k]); - - DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bw=%f\n", __func__, k, meta_row_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, dpte_row_bw=%f\n", __func__, k, dpte_row_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthLuma=%f\n", __func__, k, PrefetchBandwidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthChroma=%f\n", __func__, k, PrefetchBandwidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_cursor_bw=%f\n", __func__, k, prefetch_cursor_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); - DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), soc_state=%s, inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, dml2_core_internal_soc_state_type_str(state_type), inc_flip_bw, is_phantom, exclude_this_plane); - DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); -#endif - } - - return l->required_bandwidth_mbps; -} - -static void CalculateExtraLatency( - const struct dml2_display_cfg *display_cfg, - unsigned int ROBBufferSizeInKByte, - unsigned int RoundTripPingLatencyCycles, - unsigned int ReorderingBytes, - double DCFCLK, - double FabricClock, - unsigned int PixelChunkSizeInKByte, - double ReturnBW, - unsigned int NumberOfActiveSurfaces, - unsigned int NumberOfDPP[], - unsigned int dpte_group_bytes[], - unsigned int tdlut_bytes_per_group[], - double HostVMInefficiencyFactor, - double HostVMInefficiencyFactorPrefetch, - unsigned int HostVMMinPageSize, - enum dml2_qos_param_type qos_type, - bool max_outstanding_when_urgent_expected, - unsigned int max_outstanding_requests, - unsigned int request_size_bytes_luma[], - unsigned int request_size_bytes_chroma[], - unsigned int MetaChunkSize, - unsigned int dchub_arb_to_ret_delay, - double Ttrip, - unsigned int hostvm_mode, - - // output - double *ExtraLatency, // Tex - double *ExtraLatency_sr, // Tex_sr - double *ExtraLatencyPrefetch) - -{ - double Tarb; - double Tarb_prefetch; - double Tex_trips; - unsigned int max_request_size_bytes = 0; - - CalculateTarb( - display_cfg, - PixelChunkSizeInKByte, - NumberOfActiveSurfaces, - NumberOfDPP, - dpte_group_bytes, - tdlut_bytes_per_group, - HostVMInefficiencyFactor, - HostVMInefficiencyFactorPrefetch, - HostVMMinPageSize, - ReturnBW, - MetaChunkSize, - // output - &Tarb, - &Tarb_prefetch); - - Tex_trips = (display_cfg->hostvm_enable && hostvm_mode == 1) ? (2.0 * Ttrip) : 0.0; - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - if (request_size_bytes_luma[k] > max_request_size_bytes) - max_request_size_bytes = request_size_bytes_luma[k]; - if (request_size_bytes_chroma[k] > max_request_size_bytes) - max_request_size_bytes = request_size_bytes_chroma[k]; - } - - if (qos_type == dml2_qos_param_type_dcn4x) { - *ExtraLatency_sr = dchub_arb_to_ret_delay / DCFCLK; - *ExtraLatency = *ExtraLatency_sr; - if (max_outstanding_when_urgent_expected) - *ExtraLatency = *ExtraLatency + (ROBBufferSizeInKByte * 1024 - max_outstanding_requests * max_request_size_bytes) / ReturnBW; - } else { - *ExtraLatency_sr = dchub_arb_to_ret_delay / DCFCLK + RoundTripPingLatencyCycles / FabricClock + ReorderingBytes / ReturnBW; - *ExtraLatency = *ExtraLatency_sr; - } - *ExtraLatency = *ExtraLatency + Tex_trips; - *ExtraLatencyPrefetch = *ExtraLatency + Tarb_prefetch; - *ExtraLatency = *ExtraLatency + Tarb; - *ExtraLatency_sr = *ExtraLatency_sr + Tarb; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: qos_type=%u\n", __func__, qos_type); - DML_LOG_VERBOSE("DML::%s: hostvm_mode=%u\n", __func__, hostvm_mode); - DML_LOG_VERBOSE("DML::%s: Tex_trips=%f\n", __func__, Tex_trips); - DML_LOG_VERBOSE("DML::%s: max_outstanding_when_urgent_expected=%u\n", __func__, max_outstanding_when_urgent_expected); - DML_LOG_VERBOSE("DML::%s: FabricClock=%f\n", __func__, FabricClock); - DML_LOG_VERBOSE("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); - DML_LOG_VERBOSE("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); - DML_LOG_VERBOSE("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); - DML_LOG_VERBOSE("DML::%s: ReorderingBytes=%u\n", __func__, ReorderingBytes); - DML_LOG_VERBOSE("DML::%s: Tarb=%f\n", __func__, Tarb); - DML_LOG_VERBOSE("DML::%s: ExtraLatency=%f\n", __func__, *ExtraLatency); - DML_LOG_VERBOSE("DML::%s: ExtraLatency_sr=%f\n", __func__, *ExtraLatency_sr); - DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch=%f\n", __func__, *ExtraLatencyPrefetch); -#endif -} - -static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch, struct dml2_core_calcs_CalculatePrefetchSchedule_params *p) -{ - struct dml2_core_calcs_CalculatePrefetchSchedule_locals *s = &scratch->CalculatePrefetchSchedule_locals; - bool dcc_mrq_enable; - - unsigned int vm_bytes; - unsigned int extra_tdpe_bytes; - unsigned int tdlut_row_bytes; - unsigned int Lo; - - s->NoTimeToPrefetch = false; - s->DPPCycles = 0; - s->DISPCLKCycles = 0; - s->DSTTotalPixelsAfterScaler = 0.0; - s->LineTime = 0.0; - s->dst_y_prefetch_equ = 0.0; - s->prefetch_bw_oto = 0.0; - s->Tvm_oto = 0.0; - s->Tr0_oto = 0.0; - s->Tvm_oto_lines = 0.0; - s->Tr0_oto_lines = 0.0; - s->dst_y_prefetch_oto = 0.0; - s->TimeForFetchingVM = 0.0; - s->TimeForFetchingRowInVBlank = 0.0; - s->LinesToRequestPrefetchPixelData = 0.0; - s->HostVMDynamicLevelsTrips = 0; - s->trip_to_mem = 0.0; - *p->Tvm_trips = 0.0; - *p->Tr0_trips = 0.0; - s->Tvm_trips_rounded = 0.0; - s->Tr0_trips_rounded = 0.0; - s->max_Tsw = 0.0; - s->Lsw_oto = 0.0; - *p->Tpre_rounded = 0.0; - s->prefetch_bw_equ = 0.0; - s->Tvm_equ = 0.0; - s->Tr0_equ = 0.0; - s->Tdmbf = 0.0; - s->Tdmec = 0.0; - s->Tdmsks = 0.0; - *p->prefetch_sw_bytes = 0.0; - s->prefetch_bw_pr = 0.0; - s->bytes_pp = 0.0; - s->dep_bytes = 0.0; - s->min_Lsw_oto = 0.0; - s->min_Lsw_equ = 0.0; - s->Tsw_est1 = 0.0; - s->Tsw_est2 = 0.0; - s->Tsw_est3 = 0.0; - s->cursor_prefetch_bytes = 0; - *p->prefetch_cursor_bw = 0; - *p->RequiredPrefetchBWMax = 0.0; - - dcc_mrq_enable = (p->dcc_enable && p->mrq_present); - - s->TWait_p = p->TWait - p->Ttrip; // TWait includes max(Turg, Ttrip) and Ttrip here is already max(Turg, Ttrip) - - if (p->display_cfg->gpuvm_enable == true && p->display_cfg->hostvm_enable == true) { - s->HostVMDynamicLevelsTrips = p->display_cfg->hostvm_max_non_cached_page_table_levels; - } else { - s->HostVMDynamicLevelsTrips = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: dcc_enable = %u\n", __func__, p->dcc_enable); - DML_LOG_VERBOSE("DML::%s: mrq_present = %u\n", __func__, p->mrq_present); - DML_LOG_VERBOSE("DML::%s: dcc_mrq_enable = %u\n", __func__, dcc_mrq_enable); - DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->display_cfg->gpuvm_enable); - DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); - DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); - DML_LOG_VERBOSE("DML::%s: VStartup = %u\n", __func__, p->VStartup); - DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, p->display_cfg->hostvm_enable); - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); - DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); - DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); - DML_LOG_VERBOSE("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); - DML_LOG_VERBOSE("DML::%s: myPipe->Dispclk = %f\n", __func__, p->myPipe->Dispclk); -#endif - CalculateVUpdateAndDynamicMetadataParameters( - p->MaxInterDCNTileRepeaters, - p->myPipe->Dppclk, - p->myPipe->Dispclk, - p->myPipe->DCFClkDeepSleep, - p->myPipe->PixelClock, - p->myPipe->HTotal, - p->myPipe->VBlank, - p->DynamicMetadataTransmittedBytes, - p->DynamicMetadataLinesBeforeActiveRequired, - p->myPipe->InterlaceEnable, - p->myPipe->ProgressiveToInterlaceUnitInOPP, - p->TSetup, - - // Output - &s->Tdmbf, - &s->Tdmec, - &s->Tdmsks, - p->VUpdateOffsetPix, - p->VUpdateWidthPix, - p->VReadyOffsetPix); - - s->LineTime = p->myPipe->HTotal / p->myPipe->PixelClock; - s->trip_to_mem = p->Ttrip; - *p->Tvm_trips = p->ExtraLatencyPrefetch + math_max2(s->trip_to_mem * (p->display_cfg->gpuvm_max_page_table_levels * (s->HostVMDynamicLevelsTrips + 1)), p->Turg); - if (dcc_mrq_enable) - *p->Tvm_trips_flip = *p->Tvm_trips; - else - *p->Tvm_trips_flip = *p->Tvm_trips - s->trip_to_mem; - - *p->Tr0_trips_flip = s->trip_to_mem * (s->HostVMDynamicLevelsTrips + 1); - *p->Tr0_trips = math_max2(*p->Tr0_trips_flip, p->tdlut_opt_time / 2); - - if (p->DynamicMetadataVMEnabled == true) { - *p->Tdmdl_vm = s->TWait_p + *p->Tvm_trips; - *p->Tdmdl = *p->Tdmdl_vm + p->Ttrip; - } else { - *p->Tdmdl_vm = 0; - *p->Tdmdl = s->TWait_p + p->ExtraLatencyPrefetch + p->Ttrip; // Tex - } - - if (p->DynamicMetadataEnable == true) { - if (p->VStartup * s->LineTime < *p->TSetup + *p->Tdmdl + s->Tdmbf + s->Tdmec + s->Tdmsks) { - *p->NotEnoughTimeForDynamicMetadata = true; - DML_LOG_VERBOSE("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); - DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); - } else { - *p->NotEnoughTimeForDynamicMetadata = false; - } - } else { - *p->NotEnoughTimeForDynamicMetadata = false; - } - - if (p->myPipe->ScalerEnabled) - s->DPPCycles = (unsigned int)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCL); - else - s->DPPCycles = (unsigned int)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCLLBOnly); - - s->DPPCycles = (unsigned int)(s->DPPCycles + p->myPipe->NumberOfCursors * p->DPPCLKDelayCNVCCursor); - - s->DISPCLKCycles = (unsigned int)p->DISPCLKDelaySubtotal; - - if (p->myPipe->Dppclk == 0.0 || p->myPipe->Dispclk == 0.0) - return true; - - *p->DSTXAfterScaler = (unsigned int)math_round(s->DPPCycles * p->myPipe->PixelClock / p->myPipe->Dppclk + s->DISPCLKCycles * p->myPipe->PixelClock / p->myPipe->Dispclk + p->DSCDelay); - *p->DSTXAfterScaler = (unsigned int)math_round(*p->DSTXAfterScaler + (p->myPipe->ODMMode != dml2_odm_mode_bypass ? 18 : 0) + (p->myPipe->DPPPerSurface - 1) * p->DPP_RECOUT_WIDTH + - ((p->myPipe->ODMMode == dml2_odm_mode_split_1to2 || p->myPipe->ODMMode == dml2_odm_mode_mso_1to2) ? (double)p->myPipe->HActive / 2.0 : 0) + - ((p->myPipe->ODMMode == dml2_odm_mode_mso_1to4) ? (double)p->myPipe->HActive * 3.0 / 4.0 : 0)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DynamicMetadataVMEnabled = %u\n", __func__, p->DynamicMetadataVMEnabled); - DML_LOG_VERBOSE("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); - DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); - DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); - DML_LOG_VERBOSE("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); - DML_LOG_VERBOSE("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); - DML_LOG_VERBOSE("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); - DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); - DML_LOG_VERBOSE("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); - DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); - - DML_LOG_VERBOSE("DML::%s: setup_for_tdlut = %u\n", __func__, p->setup_for_tdlut); - DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, p->tdlut_opt_time); - DML_LOG_VERBOSE("DML::%s: tdlut_pte_bytes_per_frame = %u\n", __func__, p->tdlut_pte_bytes_per_frame); - DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, p->tdlut_drain_time); -#endif - - if (p->OutputFormat == dml2_420 || (p->myPipe->InterlaceEnable && p->myPipe->ProgressiveToInterlaceUnitInOPP)) - *p->DSTYAfterScaler = 1; - else - *p->DSTYAfterScaler = 0; - - s->DSTTotalPixelsAfterScaler = *p->DSTYAfterScaler * p->myPipe->HTotal + *p->DSTXAfterScaler; - *p->DSTYAfterScaler = (unsigned int)(math_floor2(s->DSTTotalPixelsAfterScaler / p->myPipe->HTotal, 1)); - *p->DSTXAfterScaler = (unsigned int)(s->DSTTotalPixelsAfterScaler - ((double)(*p->DSTYAfterScaler * p->myPipe->HTotal))); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); - DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); -#endif - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); - DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); - DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); - DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); - DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); - DML_LOG_VERBOSE("DML::%s: HostVMDynamicLevelsTrips = %u\n", __func__, s->HostVMDynamicLevelsTrips); -#endif - if (p->display_cfg->gpuvm_enable) { - s->Tvm_trips_rounded = math_ceil2(4.0 * *p->Tvm_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; - *p->Tvm_trips_flip_rounded = math_ceil2(4.0 * *p->Tvm_trips_flip / s->LineTime, 1.0) / 4.0 * s->LineTime; - } else { - if (p->DynamicMetadataEnable || dcc_mrq_enable || p->setup_for_tdlut) - s->Tvm_trips_rounded = math_max2(s->LineTime * math_ceil2(4.0*math_max3(p->ExtraLatencyPrefetch, p->Turg, s->trip_to_mem)/s->LineTime, 1)/4, s->LineTime/4.0); - else - s->Tvm_trips_rounded = s->LineTime / 4.0; - *p->Tvm_trips_flip_rounded = s->LineTime / 4.0; - } - - s->Tvm_trips_rounded = math_max2(s->Tvm_trips_rounded, s->LineTime / 4.0); - *p->Tvm_trips_flip_rounded = math_max2(*p->Tvm_trips_flip_rounded, s->LineTime / 4.0); - - if (p->display_cfg->gpuvm_enable == true || p->setup_for_tdlut || dcc_mrq_enable) { - s->Tr0_trips_rounded = math_ceil2(4.0 * *p->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; - *p->Tr0_trips_flip_rounded = math_ceil2(4.0 * *p->Tr0_trips_flip / s->LineTime, 1.0) / 4.0 * s->LineTime; - } else { - s->Tr0_trips_rounded = s->LineTime / 4.0; - *p->Tr0_trips_flip_rounded = s->LineTime / 4.0; - } - s->Tr0_trips_rounded = math_max2(s->Tr0_trips_rounded, s->LineTime / 4.0); - *p->Tr0_trips_flip_rounded = math_max2(*p->Tr0_trips_flip_rounded, s->LineTime / 4.0); - - if (p->display_cfg->gpuvm_enable == true) { - if (p->display_cfg->gpuvm_max_page_table_levels >= 3) { - *p->Tno_bw = p->ExtraLatencyPrefetch + s->trip_to_mem * (double)((p->display_cfg->gpuvm_max_page_table_levels - 2) * (s->HostVMDynamicLevelsTrips + 1)); - } else if (p->display_cfg->gpuvm_max_page_table_levels == 1 && !dcc_mrq_enable && !p->setup_for_tdlut) { - *p->Tno_bw = p->ExtraLatencyPrefetch; - } else { - *p->Tno_bw = 0; - } - } else { - *p->Tno_bw = 0; - } - - if (p->mrq_present || p->display_cfg->gpuvm_max_page_table_levels >= 3) - *p->Tno_bw_flip = *p->Tno_bw; - else - *p->Tno_bw_flip = 0; //because there is no 3DLUT for iFlip - - if (dml_is_420(p->myPipe->SourcePixelFormat)) { - s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC / 4.0; - } else { - s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC; - } - - *p->prefetch_sw_bytes = p->PrefetchSourceLinesY * p->swath_width_luma_ub * p->myPipe->BytePerPixelY + p->PrefetchSourceLinesC * p->swath_width_chroma_ub * p->myPipe->BytePerPixelC; - *p->prefetch_sw_bytes = *p->prefetch_sw_bytes * p->mall_prefetch_sdp_overhead_factor; - - vm_bytes = p->vm_bytes; // vm_bytes is dpde0_bytes_per_frame_ub_l + dpde0_bytes_per_frame_ub_c + 2*extra_dpde_bytes; - extra_tdpe_bytes = (unsigned int)math_max2(0, (p->display_cfg->gpuvm_max_page_table_levels - 1) * 128); - - if (p->setup_for_tdlut) - vm_bytes = vm_bytes + p->tdlut_pte_bytes_per_frame + (p->display_cfg->gpuvm_enable ? extra_tdpe_bytes : 0); - - tdlut_row_bytes = (unsigned long) math_ceil2(p->tdlut_bytes_per_frame/2.0, 1.0); - - s->min_Lsw_oto = math_max2(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML2_CALCS_MAX_VRATIO_PRE_OTO__; - s->min_Lsw_oto = math_max2(s->min_Lsw_oto, p->tdlut_drain_time / s->LineTime); - s->min_Lsw_oto = math_max2(s->min_Lsw_oto, 2.0); - - // use vactive swath bw for prefetch oto and also cap prefetch_bw_oto to max_vratio_oto - // Note: in prefetch calculation, acounting is done mostly per-pipe. - // vactive swath bw represents the per-surface (aka per dml plane) bw to move vratio_l/c lines of bytes_l/c per line time - s->per_pipe_vactive_sw_bw = p->vactive_sw_bw_l / (double)p->myPipe->DPPPerSurface; - - // one-to-one prefetch bw as one line of bytes per line time (as per vratio_pre_l/c = 1) - s->prefetch_bw_oto = (p->swath_width_luma_ub * p->myPipe->BytePerPixelY) / s->LineTime; - - if (p->myPipe->BytePerPixelC > 0) { - s->per_pipe_vactive_sw_bw += p->vactive_sw_bw_c / (double)p->myPipe->DPPPerSurface; - s->prefetch_bw_oto += (p->swath_width_chroma_ub * p->myPipe->BytePerPixelC) / s->LineTime; - } - - /* oto prefetch bw should be always be less than total vactive bw */ - //DML_ASSERT(s->prefetch_bw_oto < s->per_pipe_vactive_sw_bw * p->myPipe->DPPPerSurface); - - s->prefetch_bw_oto = math_max2(s->per_pipe_vactive_sw_bw, s->prefetch_bw_oto) * p->mall_prefetch_sdp_overhead_factor; - - s->prefetch_bw_oto = math_min2(s->prefetch_bw_oto, *p->prefetch_sw_bytes/(s->min_Lsw_oto*s->LineTime)); - - s->Lsw_oto = math_ceil2(4.0 * *p->prefetch_sw_bytes / s->prefetch_bw_oto / s->LineTime, 1.0) / 4.0; - - s->prefetch_bw_oto = math_max3(s->prefetch_bw_oto, - p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, - (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); - - /* oto bw needs to be outputted even if the oto schedule isn't being used to avoid ms/mp mismatch. - * mp will fail if ms decides to use equ schedule and mp decides to use oto schedule - * and the required bandwidth increases when going from ms to mp - */ - *p->RequiredPrefetchBWMax = s->prefetch_bw_oto; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); - DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c = %f\n", __func__, p->vactive_sw_bw_c); - DML_LOG_VERBOSE("DML::%s: per_pipe_vactive_sw_bw = %f\n", __func__, s->per_pipe_vactive_sw_bw); -#endif - - if (p->display_cfg->gpuvm_enable == true) { - s->Tvm_oto = math_max3( - *p->Tvm_trips, - *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto, - s->LineTime / 4.0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Tvm_oto max0 = %f\n", __func__, *p->Tvm_trips); - DML_LOG_VERBOSE("DML::%s: Tvm_oto max1 = %f\n", __func__, *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto); - DML_LOG_VERBOSE("DML::%s: Tvm_oto max2 = %f\n", __func__, s->LineTime / 4.0); -#endif - } else { - s->Tvm_oto = s->Tvm_trips_rounded; - } - - if ((p->display_cfg->gpuvm_enable == true || p->setup_for_tdlut || dcc_mrq_enable)) { - s->Tr0_oto = math_max3( - *p->Tr0_trips, - (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto, - s->LineTime / 4.0); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Tr0_oto max0 = %f\n", __func__, *p->Tr0_trips); - DML_LOG_VERBOSE("DML::%s: Tr0_oto max1 = %f\n", __func__, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto); - DML_LOG_VERBOSE("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime / 4); -#endif - } else - s->Tr0_oto = s->LineTime / 4.0; - - s->Tvm_oto_lines = math_ceil2(4.0 * s->Tvm_oto / s->LineTime, 1) / 4.0; - s->Tr0_oto_lines = math_ceil2(4.0 * s->Tr0_oto / s->LineTime, 1) / 4.0; - s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto; - -#ifdef DML_GLOBAL_PREFETCH_CHECK - DML_LOG_VERBOSE("DML::%s: impacted_Tpre = %f\n", __func__, p->impacted_dst_y_pre); - if (p->impacted_dst_y_pre > 0) { - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); - s->dst_y_prefetch_oto = math_max2(s->dst_y_prefetch_oto, p->impacted_dst_y_pre); - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f (impacted)\n", __func__, s->dst_y_prefetch_oto); - } -#endif - *p->Tpre_oto = s->dst_y_prefetch_oto * s->LineTime; - - //To (time for delay after scaler) in line time - Lo = (unsigned int)(*p->DSTYAfterScaler + (double)*p->DSTXAfterScaler / (double)p->myPipe->HTotal); - - s->min_Lsw_equ = math_max2(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML2_CALCS_MAX_VRATIO_PRE_EQU__; - s->min_Lsw_equ = math_max2(s->min_Lsw_equ, p->tdlut_drain_time / s->LineTime); - s->min_Lsw_equ = math_max2(s->min_Lsw_equ, 2.0); - //Tpre_equ in line time - if (p->DynamicMetadataVMEnabled && p->DynamicMetadataEnable) - s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + math_max2(p->TCalc, *p->Tvm_trips) + s->TWait_p) / s->LineTime - Lo; - else - s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + math_max2(p->TCalc, p->ExtraLatencyPrefetch) + s->TWait_p) / s->LineTime - Lo; - -#ifdef DML_GLOBAL_PREFETCH_CHECK - s->dst_y_prefetch_equ_impacted = math_max2(p->impacted_dst_y_pre, s->dst_y_prefetch_equ); - - s->dst_y_prefetch_equ_impacted = math_min2(s->dst_y_prefetch_equ_impacted, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH - - if (s->dst_y_prefetch_equ_impacted > s->dst_y_prefetch_equ) - s->dst_y_prefetch_equ -= s->dst_y_prefetch_equ_impacted - s->dst_y_prefetch_equ; -#endif - - s->dst_y_prefetch_equ = math_min2(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); - DML_LOG_VERBOSE("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); - DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); - DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); - DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, *p->Tno_bw_flip); - DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); - DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); - DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor = %f\n", __func__, p->mall_prefetch_sdp_overhead_factor); - DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); - DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - DML_LOG_VERBOSE("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); - DML_LOG_VERBOSE("DML::%s: prefetch_sw_bytes = %f\n", __func__, *p->prefetch_sw_bytes); - DML_LOG_VERBOSE("DML::%s: max_Tsw = %f\n", __func__, s->max_Tsw); - DML_LOG_VERBOSE("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); - DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); - DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); - DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, *p->Tvm_trips_flip); - DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, *p->Tr0_trips_flip); - DML_LOG_VERBOSE("DML::%s: prefetch_bw_pr = %f\n", __func__, s->prefetch_bw_pr); - DML_LOG_VERBOSE("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); - DML_LOG_VERBOSE("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); - DML_LOG_VERBOSE("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); - DML_LOG_VERBOSE("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); - DML_LOG_VERBOSE("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); - DML_LOG_VERBOSE("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); - DML_LOG_VERBOSE("DML::%s: tdlut_row_bytes = %d\n", __func__, tdlut_row_bytes); - DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, p->meta_row_bytes); -#endif - s->dst_y_prefetch_equ = math_floor2(4.0 * (s->dst_y_prefetch_equ + 0.125), 1) / 4.0; - *p->Tpre_rounded = s->dst_y_prefetch_equ * s->LineTime; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); - DML_LOG_VERBOSE("DML::%s: LineTime: %f\n", __func__, s->LineTime); - DML_LOG_VERBOSE("DML::%s: VStartup: %u\n", __func__, p->VStartup); - DML_LOG_VERBOSE("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); - DML_LOG_VERBOSE("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); - DML_LOG_VERBOSE("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); - DML_LOG_VERBOSE("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); - DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); - DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); - DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); - DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); - DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); - DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); - DML_LOG_VERBOSE("DML::%s: Tex = %f\n", __func__, p->ExtraLatencyPrefetch); - DML_LOG_VERBOSE("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); - DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); - DML_LOG_VERBOSE("DML::%s: TWait_p: %fus\n", __func__, s->TWait_p); - DML_LOG_VERBOSE("DML::%s: Ttrip: %fus\n", __func__, p->Ttrip); - DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); - DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); - DML_LOG_VERBOSE("DML::%s: vm_bytes: %f (hvm inefficiency scaled)\n", __func__, vm_bytes*p->HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: row_bytes: %f (hvm inefficiency scaled, 1 row)\n", __func__, p->PixelPTEBytesPerRow*p->HostVMInefficiencyFactor+p->meta_row_bytes+tdlut_row_bytes); - DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); - DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, (s->dst_y_prefetch_equ * s->LineTime), *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); - DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); -#endif - - *p->dst_y_per_vm_vblank = 0; - *p->dst_y_per_row_vblank = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixelDataBWLuma = 0; - - // Derive bandwidth by finding how much data to move within the time constraint - // Tpre_rounded is Tpre rounding to 2-bit fraction - // Tvm_trips_rounded is Tvm_trips ceiling to 1/4 line time - // Tr0_trips_rounded is Tr0_trips ceiling to 1/4 line time - // So that means prefetch bw calculated can be higher since the total time available for prefetch is less - bool min_Lsw_equ_ok = *p->Tpre_rounded >= s->Tvm_trips_rounded + 2.0*s->Tr0_trips_rounded + s->min_Lsw_equ*s->LineTime; - bool tpre_gt_req_latency = true; -#if 0 - // Check that Tpre_rounded is big enough if all of the stages of the prefetch are time constrained. - // The terms Tvm_trips_rounded and Tr0_trips_rounded represent the min time constraints for the VM and row stages. - // Normally, these terms cover the overall time constraint for Tpre >= (Tex + max{Ttrip, Turg}), but if these terms are at their minimum, an explicit check is necessary. - tpre_gt_req_latency = *p->Tpre_rounded > (math_max2(p->Turg, s->trip_to_mem) + p->ExtraLatencyPrefetch); -#endif - - if (s->dst_y_prefetch_equ > 1 && min_Lsw_equ_ok && tpre_gt_req_latency) { - s->prefetch_bw1 = 0.; - s->prefetch_bw2 = 0.; - s->prefetch_bw3 = 0.; - s->prefetch_bw4 = 0.; - - // prefetch_bw1: VM + 2*R0 + SW - if (*p->Tpre_rounded - *p->Tno_bw > 0) { - s->prefetch_bw1 = (vm_bytes * p->HostVMInefficiencyFactor - + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) - + *p->prefetch_sw_bytes) - / (*p->Tpre_rounded - *p->Tno_bw); - s->Tsw_est1 = *p->prefetch_sw_bytes / s->prefetch_bw1; - } else - s->prefetch_bw1 = 0; - - DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f\n", __func__, s->prefetch_bw1); - if ((s->Tsw_est1 < s->min_Lsw_equ * s->LineTime) && (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw > 0)) { - s->prefetch_bw1 = (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / - (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: vm and 2 rows bytes = %f\n", __func__, (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes))); - DML_LOG_VERBOSE("DML::%s: Tpre_rounded = %f\n", __func__, *p->Tpre_rounded); - DML_LOG_VERBOSE("DML::%s: minus term = %f\n", __func__, s->min_Lsw_equ * s->LineTime + 0.75 * s->LineTime + *p->Tno_bw); - DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); - DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); - DML_LOG_VERBOSE("DML::%s: Time to fetch vm and 2 rows = %f\n", __func__, (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw)); - DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (updated)\n", __func__, s->prefetch_bw1); -#endif - } - - // prefetch_bw2: VM + SW - if (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded > 0) { - s->prefetch_bw2 = (vm_bytes * p->HostVMInefficiencyFactor + *p->prefetch_sw_bytes) / - (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded); - s->Tsw_est2 = *p->prefetch_sw_bytes / s->prefetch_bw2; - } else - s->prefetch_bw2 = 0; - - DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f\n", __func__, s->prefetch_bw2); - if ((s->Tsw_est2 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime) > 0)) { - s->prefetch_bw2 = vm_bytes * p->HostVMInefficiencyFactor / (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime); - DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (updated)\n", __func__, s->prefetch_bw2); - } - - // prefetch_bw3: 2*R0 + SW - if (*p->Tpre_rounded - s->Tvm_trips_rounded > 0) { - s->prefetch_bw3 = (2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) + *p->prefetch_sw_bytes) / - (*p->Tpre_rounded - s->Tvm_trips_rounded); - s->Tsw_est3 = *p->prefetch_sw_bytes / s->prefetch_bw3; - } else - s->prefetch_bw3 = 0; - - DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f\n", __func__, s->prefetch_bw3); - if ((s->Tsw_est3 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded) > 0)) { - s->prefetch_bw3 = (2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded); - DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (updated)\n", __func__, s->prefetch_bw3); - } - - // prefetch_bw4: SW - if (*p->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded > 0) - s->prefetch_bw4 = *p->prefetch_sw_bytes / (*p->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded); - else - s->prefetch_bw4 = 0; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); - DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, s->dst_y_prefetch_equ * s->LineTime, *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); - DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); - DML_LOG_VERBOSE("DML::%s: Tr0_trips=%f Tr0_trips_rounded: %f, delta=%f\n", __func__, *p->Tr0_trips, s->Tr0_trips_rounded, (s->Tr0_trips_rounded - *p->Tr0_trips)); - DML_LOG_VERBOSE("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); - DML_LOG_VERBOSE("DML::%s: Tsw_est2: %f\n", __func__, s->Tsw_est2); - DML_LOG_VERBOSE("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); - DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (final)\n", __func__, s->prefetch_bw1); - DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (final)\n", __func__, s->prefetch_bw2); - DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (final)\n", __func__, s->prefetch_bw3); - DML_LOG_VERBOSE("DML::%s: prefetch_bw4: %f (final)\n", __func__, s->prefetch_bw4); -#endif - { - bool Case1OK = false; - bool Case2OK = false; - bool Case3OK = false; - - // get "equalized" bw among all stages (vm, r0, sw), so based is all 3 stages are just above the latency-based requirement - // so it is not too dis-portionally favor a particular stage, next is either r0 more agressive and next is vm more agressive, the worst is all are agressive - // vs the latency based number - - // prefetch_bw1: VM + 2*R0 + SW - // so prefetch_bw1 will have enough bw to transfer the necessary data within Tpre_rounded - Tno_bw (Tpre is the the worst-case latency based time to fetch the data) - // here is to make sure equ bw wont be more agressive than the latency-based requirement. - // check vm time >= vm_trips - // check r0 time >= r0_trips - - double total_row_bytes = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes); - - DML_LOG_VERBOSE("DML::%s: Tvm_trips_rounded = %f\n", __func__, s->Tvm_trips_rounded); - DML_LOG_VERBOSE("DML::%s: Tr0_trips_rounded = %f\n", __func__, s->Tr0_trips_rounded); - - if (s->prefetch_bw1 > 0) { - double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw1; - double row_transfer_time = total_row_bytes / s->prefetch_bw1; - DML_LOG_VERBOSE("DML::%s: Case1: vm_transfer_time = %f\n", __func__, vm_transfer_time); - DML_LOG_VERBOSE("DML::%s: Case1: row_transfer_time = %f\n", __func__, row_transfer_time); - if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { - Case1OK = true; - } - } - - // prefetch_bw2: VM + SW - // prefetch_bw2 will be enough bw to transfer VM and SW data within (Tpre_rounded - Tr0_trips_rounded - Tno_bw) - // check vm time >= vm_trips - // check r0 time < r0_trips - if (s->prefetch_bw2 > 0) { - double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw2; - double row_transfer_time = total_row_bytes / s->prefetch_bw2; - DML_LOG_VERBOSE("DML::%s: Case2: vm_transfer_time = %f\n", __func__, vm_transfer_time); - DML_LOG_VERBOSE("DML::%s: Case2: row_transfer_time = %f\n", __func__, row_transfer_time); - if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time < s->Tr0_trips_rounded) { - Case2OK = true; - } - } - - // prefetch_bw3: VM + 2*R0 - // check vm time < vm_trips - // check r0 time >= r0_trips - if (s->prefetch_bw3 > 0) { - double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw3; - double row_transfer_time = total_row_bytes / s->prefetch_bw3; - DML_LOG_VERBOSE("DML::%s: Case3: vm_transfer_time = %f\n", __func__, vm_transfer_time); - DML_LOG_VERBOSE("DML::%s: Case3: row_transfer_time = %f\n", __func__, row_transfer_time); - if (vm_transfer_time < s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { - Case3OK = true; - } - } - - if (Case1OK) { - s->prefetch_bw_equ = s->prefetch_bw1; - } else if (Case2OK) { - s->prefetch_bw_equ = s->prefetch_bw2; - } else if (Case3OK) { - s->prefetch_bw_equ = s->prefetch_bw3; - } else { - s->prefetch_bw_equ = s->prefetch_bw4; - } - - s->prefetch_bw_equ = math_max3(s->prefetch_bw_equ, - p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, - (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Case1OK: %u\n", __func__, Case1OK); - DML_LOG_VERBOSE("DML::%s: Case2OK: %u\n", __func__, Case2OK); - DML_LOG_VERBOSE("DML::%s: Case3OK: %u\n", __func__, Case3OK); - DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); -#endif - - if (s->prefetch_bw_equ > 0) { - if (p->display_cfg->gpuvm_enable == true) { - s->Tvm_equ = math_max3(*p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_equ, *p->Tvm_trips, s->LineTime / 4); - } else { - s->Tvm_equ = s->LineTime / 4; - } - - if (p->display_cfg->gpuvm_enable == true || dcc_mrq_enable || p->setup_for_tdlut) { - s->Tr0_equ = math_max3((p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_equ, // PixelPTEBytesPerRow is dpte_row_bytes - *p->Tr0_trips, - s->LineTime / 4); - } else { - s->Tr0_equ = s->LineTime / 4; - } - } else { - s->Tvm_equ = 0; - s->Tr0_equ = 0; - DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ equals 0!\n", __func__); - } - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Tvm_equ = %f\n", __func__, s->Tvm_equ); - DML_LOG_VERBOSE("DML::%s: Tr0_equ = %f\n", __func__, s->Tr0_equ); -#endif - // Use the more stressful prefetch schedule - if (s->dst_y_prefetch_oto < s->dst_y_prefetch_equ) { - *p->dst_y_prefetch = s->dst_y_prefetch_oto; - s->TimeForFetchingVM = s->Tvm_oto; - s->TimeForFetchingRowInVBlank = s->Tr0_oto; - - *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; - *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Using oto scheduling for prefetch\n", __func__); -#endif - - } else { - *p->dst_y_prefetch = s->dst_y_prefetch_equ; - - if (s->dst_y_prefetch_equ < s->dst_y_prefetch_equ_impacted) - *p->dst_y_prefetch = s->dst_y_prefetch_equ_impacted; - - s->TimeForFetchingVM = s->Tvm_equ; - s->TimeForFetchingRowInVBlank = s->Tr0_equ; - - *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; - *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; - - /* equ bw should be propagated so a ceiling of the equ bw is accounted for prior to mode programming. - * Overall bandwidth may be lower when going from mode support to mode programming but final pixel data - * bandwidth may end up higher than what was calculated in mode support. - */ - *p->RequiredPrefetchBWMax = math_max2(s->prefetch_bw_equ, *p->RequiredPrefetchBWMax); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Using equ bw scheduling for prefetch\n", __func__); -#endif - } - - // Lsw = dst_y_prefetch - (dst_y_per_vm_vblank + 2*dst_y_per_row_vblank) - s->LinesToRequestPrefetchPixelData = *p->dst_y_prefetch - *p->dst_y_per_vm_vblank - 2 * *p->dst_y_per_row_vblank; // Lsw - - s->cursor_prefetch_bytes = (unsigned int)math_max2(p->cursor_bytes_per_chunk, 4 * p->cursor_bytes_per_line); - *p->prefetch_cursor_bw = p->num_cursors * s->cursor_prefetch_bytes / (s->LinesToRequestPrefetchPixelData * s->LineTime); - *p->prefetch_swath_time_us = (s->LinesToRequestPrefetchPixelData * s->LineTime); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: TimeForFetchingVM = %f\n", __func__, s->TimeForFetchingVM); - DML_LOG_VERBOSE("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); - DML_LOG_VERBOSE("DML::%s: dst_y_prefetch = %f\n", __func__, *p->dst_y_prefetch); - DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); - DML_LOG_VERBOSE("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); - DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - DML_LOG_VERBOSE("DML::%s: prefetch_swath_time_us = %f\n", __func__, *p->prefetch_swath_time_us); - - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, p->cursor_bytes_per_chunk); - DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, p->cursor_bytes_per_line); - DML_LOG_VERBOSE("DML::%s: cursor_prefetch_bytes = %d\n", __func__, s->cursor_prefetch_bytes); - DML_LOG_VERBOSE("DML::%s: prefetch_cursor_bw = %f\n", __func__, *p->prefetch_cursor_bw); -#endif - DML_ASSERT(*p->dst_y_prefetch < 64); - - unsigned int min_lsw_required = (unsigned int)math_max2(2, p->tdlut_drain_time / s->LineTime); - if (s->LinesToRequestPrefetchPixelData >= min_lsw_required && s->prefetch_bw_equ > 0) { - *p->VRatioPrefetchY = (double)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData; - *p->VRatioPrefetchY = math_max2(*p->VRatioPrefetchY, 1.0); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - DML_LOG_VERBOSE("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); - DML_LOG_VERBOSE("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); -#endif - if ((p->SwathHeightY > 4) && (p->VInitPreFillY > 3)) { - if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillY - 3.0) / 2.0) { - *p->VRatioPrefetchY = math_max2(*p->VRatioPrefetchY, - (double)p->MaxNumSwathY * p->SwathHeightY / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillY - 3.0) / 2.0)); - } else { - s->NoTimeToPrefetch = true; - DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); - *p->VRatioPrefetchY = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); - DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); - DML_LOG_VERBOSE("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); -#endif - } - - *p->VRatioPrefetchC = (double)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData; - *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, 1.0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - DML_LOG_VERBOSE("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); - DML_LOG_VERBOSE("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); -#endif - if ((p->SwathHeightC > 4) && (p->VInitPreFillC > 3)) { - if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillC - 3.0) / 2.0) { - *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, (double)p->MaxNumSwathC * p->SwathHeightC / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillC - 3.0) / 2.0)); - } else { - s->NoTimeToPrefetch = true; - DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); - *p->VRatioPrefetchC = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); - DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); - DML_LOG_VERBOSE("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); -#endif - } - - *p->RequiredPrefetchPixelDataBWLuma = (double)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData * p->myPipe->BytePerPixelY * p->swath_width_luma_ub / s->LineTime; - *p->RequiredPrefetchPixelDataBWChroma = (double)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData * p->myPipe->BytePerPixelC * p->swath_width_chroma_ub / s->LineTime; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); - DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); - DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); - DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); -#endif - } else { - s->NoTimeToPrefetch = true; - DML_LOG_VERBOSE("DML::%s: No time to prefetch!, LinesToRequestPrefetchPixelData: %f, should be >= %d\n", __func__, s->LinesToRequestPrefetchPixelData, min_lsw_required); - DML_LOG_VERBOSE("DML::%s: No time to prefetch!, prefetch_bw_equ: %f, should be > 0\n", __func__, s->prefetch_bw_equ); - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixelDataBWLuma = 0; - *p->RequiredPrefetchPixelDataBWChroma = 0; - } - DML_LOG_VERBOSE("DML: Tpre: %fus - sum of time to request 2 x data pte, swaths\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingVM); - DML_LOG_VERBOSE("DML: Tvm: %fus - time to fetch vm\n", s->TimeForFetchingVM); - DML_LOG_VERBOSE("DML: Tr0: %fus - time to fetch first row of data pagetables\n", s->TimeForFetchingRowInVBlank); - DML_LOG_VERBOSE("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime); - DML_LOG_VERBOSE("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime); - DML_LOG_VERBOSE("DML: Tvstartup - TSetup - Tcalc - TWait - Tpre - To > 0\n"); - DML_LOG_VERBOSE("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingVM - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); - DML_LOG_VERBOSE("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); - - } else { - DML_LOG_VERBOSE("DML::%s: No time to prefetch! dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); - DML_LOG_VERBOSE("DML::%s: No time to prefetch! min_Lsw_equ_ok = %d, Tpre_rounded (%f) should be >= Tvm_trips_rounded (%f) + 2.0*Tr0_trips_rounded (%f) + min_Tsw_equ (%f)\n", - __func__, min_Lsw_equ_ok, *p->Tpre_rounded, s->Tvm_trips_rounded, 2.0*s->Tr0_trips_rounded, s->min_Lsw_equ*s->LineTime); - s->NoTimeToPrefetch = true; - s->TimeForFetchingVM = 0; - s->TimeForFetchingRowInVBlank = 0; - *p->dst_y_per_vm_vblank = 0; - *p->dst_y_per_row_vblank = 0; - s->LinesToRequestPrefetchPixelData = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixelDataBWLuma = 0; - *p->RequiredPrefetchPixelDataBWChroma = 0; - } - - { - double prefetch_vm_bw; - double prefetch_row_bw; - - if (vm_bytes == 0) { - prefetch_vm_bw = 0; - } else if (*p->dst_y_per_vm_vblank > 0) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); -#endif - prefetch_vm_bw = vm_bytes * p->HostVMInefficiencyFactor / (*p->dst_y_per_vm_vblank * s->LineTime); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); -#endif - } else { - prefetch_vm_bw = 0; - s->NoTimeToPrefetch = true; - DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_vm_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_vm_vblank); - } - - if (p->PixelPTEBytesPerRow == 0 && tdlut_row_bytes == 0) { - prefetch_row_bw = 0; - } else if (*p->dst_y_per_row_vblank > 0) { - prefetch_row_bw = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + tdlut_row_bytes) / (*p->dst_y_per_row_vblank * s->LineTime); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); - DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); - DML_LOG_VERBOSE("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); -#endif - } else { - prefetch_row_bw = 0; - s->NoTimeToPrefetch = true; - DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_row_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_row_vblank); - } - - *p->prefetch_vmrow_bw = math_max2(prefetch_vm_bw, prefetch_row_bw); - } - - if (s->NoTimeToPrefetch) { - s->TimeForFetchingVM = 0; - s->TimeForFetchingRowInVBlank = 0; - *p->dst_y_per_vm_vblank = 0; - *p->dst_y_per_row_vblank = 0; - *p->dst_y_prefetch = 0; - s->LinesToRequestPrefetchPixelData = 0; - *p->VRatioPrefetchY = 0; - *p->VRatioPrefetchC = 0; - *p->RequiredPrefetchPixelDataBWLuma = 0; - *p->RequiredPrefetchPixelDataBWChroma = 0; - *p->prefetch_vmrow_bw = 0; - } - - DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f (final)\n", __func__, *p->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f (final)\n", __func__, *p->dst_y_per_row_vblank); - DML_LOG_VERBOSE("DML::%s: prefetch_vmrow_bw = %f (final)\n", __func__, *p->prefetch_vmrow_bw); - DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); - DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); - DML_LOG_VERBOSE("DML::%s: NoTimeToPrefetch=%d\n", __func__, s->NoTimeToPrefetch); - - return s->NoTimeToPrefetch; -} - -static unsigned int get_num_lb_source_lines(unsigned int max_line_buffer_lines, - unsigned int line_buffer_size_bits, - unsigned int num_pipes, - unsigned int vp_width, - unsigned int vp_height, - double h_ratio, - enum dml2_rotation_angle rotation_angle) -{ - unsigned int num_lb_source_lines = 0; - double lb_bit_per_pixel = 57.0; - unsigned recin_width = vp_width/num_pipes; - - if (dml_is_vertical_rotation(rotation_angle)) - recin_width = vp_height/num_pipes; - - num_lb_source_lines = (unsigned int) math_min2((double) max_line_buffer_lines, - math_floor2(line_buffer_size_bits / lb_bit_per_pixel / (recin_width / math_max2(h_ratio, 1.0)), 1.0)); - - return num_lb_source_lines; -} - -static unsigned int find_max_impact_plane(unsigned int this_plane_idx, unsigned int num_planes, unsigned int Trpd_dcfclk_cycles[]) -{ - int max_value = -1; - int max_idx = -1; - for (unsigned int i = 0; i < num_planes; i++) { - if (i != this_plane_idx && (int) Trpd_dcfclk_cycles[i] > max_value) { - max_value = Trpd_dcfclk_cycles[i]; - max_idx = i; - } - } - if (max_idx <= 0) { - DML_ASSERT(max_idx >= 0); - max_idx = this_plane_idx; - } - - return max_idx; -} - -static double calculate_impacted_Tsw(unsigned int exclude_plane_idx, unsigned int num_planes, double *prefetch_swath_bytes, double bw_mbps) -{ - double sum = 0.; - for (unsigned int i = 0; i < num_planes; i++) { - if (i != exclude_plane_idx) { - sum += prefetch_swath_bytes[i]; - } - } - return sum / bw_mbps; -} - -// a global check against the aggregate effect of the per plane prefetch schedule -static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *p) -{ - struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_locals *s = &scratch->CheckGlobalPrefetchAdmissibility_locals; - unsigned int i, k; - - memset(s, 0, sizeof(struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_locals)); - - *p->recalc_prefetch_schedule = 0; - s->prefetch_global_check_passed = 1; - // worst case if the rob and cdb is fully hogged - s->max_Trpd_dcfclk_cycles = (unsigned int) math_ceil2((p->rob_buffer_size_kbytes*1024 + p->compressed_buffer_size_kbytes*DML_MAX_COMPRESSION_RATIO*1024)/64.0, 1.0); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: num_active_planes = %d\n", __func__, p->num_active_planes); - DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %d\n", __func__, p->rob_buffer_size_kbytes); - DML_LOG_VERBOSE("DML::%s: compressed_buffer_size_kbytes = %d\n", __func__, p->compressed_buffer_size_kbytes); - DML_LOG_VERBOSE("DML::%s: estimated_urg_bandwidth_required_mbps = %f\n", __func__, p->estimated_urg_bandwidth_required_mbps); - DML_LOG_VERBOSE("DML::%s: estimated_dcfclk_mhz = %f\n", __func__, p->estimated_dcfclk_mhz); - DML_LOG_VERBOSE("DML::%s: max_Trpd_dcfclk_cycles = %u\n", __func__, s->max_Trpd_dcfclk_cycles); -#endif - - // calculate the return impact from each plane, request is 256B per dcfclk - for (i = 0; i < p->num_active_planes; i++) { - s->src_detile_buf_size_bytes_l[i] = p->detile_buffer_size_bytes_l[i]; - s->src_detile_buf_size_bytes_c[i] = p->detile_buffer_size_bytes_c[i]; - s->src_swath_bytes_l[i] = p->full_swath_bytes_l[i]; - s->src_swath_bytes_c[i] = p->full_swath_bytes_c[i]; - - if (p->pixel_format[i] == dml2_420_10) { - s->src_detile_buf_size_bytes_l[i] = (unsigned int) (s->src_detile_buf_size_bytes_l[i] * 1.5); - s->src_detile_buf_size_bytes_c[i] = (unsigned int) (s->src_detile_buf_size_bytes_c[i] * 1.5); - s->src_swath_bytes_l[i] = (unsigned int) (s->src_swath_bytes_l[i] * 1.5); - s->src_swath_bytes_c[i] = (unsigned int) (s->src_swath_bytes_c[i] * 1.5); - } - - s->burst_bytes_to_fill_det = (unsigned int) (math_floor2(s->src_detile_buf_size_bytes_l[i] / p->chunk_bytes_l, 1) * p->chunk_bytes_l); - s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(p->lb_source_lines_l[i] / p->swath_height_l[i], 1) * s->src_swath_bytes_l[i]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: i=%u pixel_format = %d\n", __func__, i, p->pixel_format[i]); - DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_l = %d\n", __func__, i, p->chunk_bytes_l); - DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_l = %d\n", __func__, i, p->lb_source_lines_l[i]); - DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_l=%d\n", __func__, i, s->src_detile_buf_size_bytes_l[i]); - DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_l=%d\n", __func__, i, s->src_swath_bytes_l[i]); - DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d (luma)\n", __func__, i, s->burst_bytes_to_fill_det); -#endif - - if (s->src_swath_bytes_c[i] > 0) { // dual_plane - s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(s->src_detile_buf_size_bytes_c[i] / p->chunk_bytes_c, 1) * p->chunk_bytes_c); - - if (p->pixel_format[i] == dml2_422_planar_8 || p->pixel_format[i] == dml2_422_planar_10 || p->pixel_format[i] == dml2_422_planar_12) { - s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(p->lb_source_lines_c[i] / p->swath_height_c[i], 1) * s->src_swath_bytes_c[i]); - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_c = %d\n", __func__, i, p->chunk_bytes_c); - DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_c = %d\n", __func__, i, p->lb_source_lines_c[i]); - DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_c=%d\n", __func__, i, s->src_detile_buf_size_bytes_c[i]); - DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_c=%d\n", __func__, i, s->src_swath_bytes_c[i]); -#endif - } - - s->time_to_fill_det_us = (double) s->burst_bytes_to_fill_det / (256 * p->estimated_dcfclk_mhz); // fill time assume full burst at request rate - s->accumulated_return_path_dcfclk_cycles[i] = (unsigned int) math_ceil2(((DML_MAX_COMPRESSION_RATIO-1) * 64 * p->estimated_dcfclk_mhz) * s->time_to_fill_det_us / 64.0, 1.0); //for 64B per DCFClk - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d\n", __func__, i, s->burst_bytes_to_fill_det); - DML_LOG_VERBOSE("DML::%s: i=%u time_to_fill_det_us=%f\n", __func__, i, s->time_to_fill_det_us); - DML_LOG_VERBOSE("DML::%s: i=%u accumulated_return_path_dcfclk_cycles=%u\n", __func__, i, s->accumulated_return_path_dcfclk_cycles[i]); -#endif - // clamping to worst case delay which is one which occupy the full rob+cdb - if (s->accumulated_return_path_dcfclk_cycles[i] > s->max_Trpd_dcfclk_cycles) - s->accumulated_return_path_dcfclk_cycles[i] = s->max_Trpd_dcfclk_cycles; - } - - // Figure out the impacted prefetch time for each plane - // if impacted_Tre is > equ bw Tpre, we need to fail the prefetch schedule as we need a higher state to support the bw - for (i = 0; i < p->num_active_planes; i++) { - k = find_max_impact_plane(i, p->num_active_planes, s->accumulated_return_path_dcfclk_cycles); // plane k causes most impact to plane i - // the rest of planes (except for k) complete for bw - p->impacted_dst_y_pre[i] = s->accumulated_return_path_dcfclk_cycles[k]/p->estimated_dcfclk_mhz; - p->impacted_dst_y_pre[i] += calculate_impacted_Tsw(k, p->num_active_planes, p->prefetch_sw_bytes, p->estimated_urg_bandwidth_required_mbps); - p->impacted_dst_y_pre[i] = math_ceil2(p->impacted_dst_y_pre[i] / p->line_time[i], 0.25); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: i=%u impacted_Tpre=%f (k=%u)\n", __func__, i, p->impacted_dst_y_pre[i], k); -#endif - } - - if (p->Tpre_rounded != NULL && p->Tpre_oto != NULL) { - for (i = 0; i < p->num_active_planes; i++) { - if (p->impacted_dst_y_pre[i] > p->dst_y_prefetch[i]) { - s->prefetch_global_check_passed = 0; - *p->recalc_prefetch_schedule = 1; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: i=%u Tpre_rounded=%f\n", __func__, i, p->Tpre_rounded[i]); - DML_LOG_VERBOSE("DML::%s: i=%u Tpre_oto=%f\n", __func__, i, p->Tpre_oto[i]); -#endif - } - } else { - // likely a mode programming calls, assume support, and no recalc - not used anyways - s->prefetch_global_check_passed = 1; - *p->recalc_prefetch_schedule = 0; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: prefetch_global_check_passed=%u\n", __func__, s->prefetch_global_check_passed); - DML_LOG_VERBOSE("DML::%s: recalc_prefetch_schedule=%u\n", __func__, *p->recalc_prefetch_schedule); -#endif - - return s->prefetch_global_check_passed; -} - -static void calculate_peak_bandwidth_required( - struct dml2_core_internal_scratch *s, - struct dml2_core_calcs_calculate_peak_bandwidth_required_params *p) -{ - unsigned int n; - unsigned int m; - - struct dml2_core_shared_calculate_peak_bandwidth_required_locals *l = &s->calculate_peak_bandwidth_required_locals; - - memset(l, 0, sizeof(struct dml2_core_shared_calculate_peak_bandwidth_required_locals)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: inc_flip_bw = %d\n", __func__, p->inc_flip_bw); - DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %d\n", __func__, p->num_active_planes); -#endif - - for (unsigned int k = 0; k < p->num_active_planes; ++k) { - l->unity_array[k] = 1.0; - l->zero_array[k] = 0.0; - } - - for (m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (n = 0; n < dml2_core_internal_bw_max; n++) { - get_urgent_bandwidth_required( - &s->get_urgent_bandwidth_required_locals, - p->display_cfg, - m, - n, - 0, //inc_flip_bw, - 0, //use_qual_row_bw - p->num_active_planes, - p->num_of_dpp, - p->dcc_dram_bw_nom_overhead_factor_p0, - p->dcc_dram_bw_nom_overhead_factor_p1, - p->dcc_dram_bw_pref_overhead_factor_p0, - p->dcc_dram_bw_pref_overhead_factor_p1, - p->mall_prefetch_sdp_overhead_factor, - p->mall_prefetch_dram_overhead_factor, - p->surface_read_bandwidth_l, - p->surface_read_bandwidth_c, - l->zero_array, //PrefetchBandwidthLuma, - l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWMax - l->zero_array, - l->zero_array, - l->zero_array, - p->dpte_row_bw, - p->meta_row_bw, - l->zero_array, //prefetch_cursor_bw, - l->zero_array, //prefetch_vmrow_bw, - l->zero_array, //flip_bw, - l->zero_array, - l->zero_array, - l->zero_array, - l->zero_array, - l->zero_array, - l->zero_array, - p->surface_avg_vactive_required_bw[m][n], - p->surface_peak_required_bw[m][n]); - - p->urg_vactive_bandwidth_required[m][n] = get_urgent_bandwidth_required( - &s->get_urgent_bandwidth_required_locals, - p->display_cfg, - m, - n, - 0, //inc_flip_bw, - 0, //use_qual_row_bw - p->num_active_planes, - p->num_of_dpp, - p->dcc_dram_bw_nom_overhead_factor_p0, - p->dcc_dram_bw_nom_overhead_factor_p1, - p->dcc_dram_bw_pref_overhead_factor_p0, - p->dcc_dram_bw_pref_overhead_factor_p1, - p->mall_prefetch_sdp_overhead_factor, - p->mall_prefetch_dram_overhead_factor, - p->surface_read_bandwidth_l, - p->surface_read_bandwidth_c, - l->zero_array, //PrefetchBandwidthLuma, - l->zero_array, //PrefetchBandwidthChroma, - l->zero_array, //PrefetchBWMax - p->excess_vactive_fill_bw_l, - p->excess_vactive_fill_bw_c, - p->cursor_bw, - p->dpte_row_bw, - p->meta_row_bw, - l->zero_array, //prefetch_cursor_bw, - l->zero_array, //prefetch_vmrow_bw, - l->zero_array, //flip_bw, - p->urgent_burst_factor_l, - p->urgent_burst_factor_c, - p->urgent_burst_factor_cursor, - p->urgent_burst_factor_prefetch_l, - p->urgent_burst_factor_prefetch_c, - p->urgent_burst_factor_prefetch_cursor, - l->surface_dummy_bw, - p->surface_peak_required_bw[m][n]); - - p->urg_bandwidth_required[m][n] = get_urgent_bandwidth_required( - &s->get_urgent_bandwidth_required_locals, - p->display_cfg, - m, - n, - p->inc_flip_bw, - 0, //use_qual_row_bw - p->num_active_planes, - p->num_of_dpp, - p->dcc_dram_bw_nom_overhead_factor_p0, - p->dcc_dram_bw_nom_overhead_factor_p1, - p->dcc_dram_bw_pref_overhead_factor_p0, - p->dcc_dram_bw_pref_overhead_factor_p1, - p->mall_prefetch_sdp_overhead_factor, - p->mall_prefetch_dram_overhead_factor, - p->surface_read_bandwidth_l, - p->surface_read_bandwidth_c, - p->prefetch_bandwidth_l, - p->prefetch_bandwidth_c, - p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw - p->excess_vactive_fill_bw_l, - p->excess_vactive_fill_bw_c, - p->cursor_bw, - p->dpte_row_bw, - p->meta_row_bw, - p->prefetch_cursor_bw, - p->prefetch_vmrow_bw, - p->flip_bw, - p->urgent_burst_factor_l, - p->urgent_burst_factor_c, - p->urgent_burst_factor_cursor, - p->urgent_burst_factor_prefetch_l, - p->urgent_burst_factor_prefetch_c, - p->urgent_burst_factor_prefetch_cursor, - l->surface_dummy_bw, - p->surface_peak_required_bw[m][n]); - - p->urg_bandwidth_required_qual[m][n] = get_urgent_bandwidth_required( - &s->get_urgent_bandwidth_required_locals, - p->display_cfg, - m, - n, - 0, //inc_flip_bw - 1, //use_qual_row_bw - p->num_active_planes, - p->num_of_dpp, - p->dcc_dram_bw_nom_overhead_factor_p0, - p->dcc_dram_bw_nom_overhead_factor_p1, - p->dcc_dram_bw_pref_overhead_factor_p0, - p->dcc_dram_bw_pref_overhead_factor_p1, - p->mall_prefetch_sdp_overhead_factor, - p->mall_prefetch_dram_overhead_factor, - p->surface_read_bandwidth_l, - p->surface_read_bandwidth_c, - p->prefetch_bandwidth_l, - p->prefetch_bandwidth_c, - p->prefetch_bandwidth_max, // to prevent ms/mp mismatch where mp prefetch bw > ms prefetch bw - p->excess_vactive_fill_bw_l, - p->excess_vactive_fill_bw_c, - p->cursor_bw, - p->dpte_row_bw, - p->meta_row_bw, - p->prefetch_cursor_bw, - p->prefetch_vmrow_bw, - p->flip_bw, - p->urgent_burst_factor_l, - p->urgent_burst_factor_c, - p->urgent_burst_factor_cursor, - p->urgent_burst_factor_prefetch_l, - p->urgent_burst_factor_prefetch_c, - p->urgent_burst_factor_prefetch_cursor, - l->surface_dummy_bw, - p->surface_peak_required_bw[m][n]); - - p->non_urg_bandwidth_required[m][n] = get_urgent_bandwidth_required( - &s->get_urgent_bandwidth_required_locals, - p->display_cfg, - m, - n, - p->inc_flip_bw, - 0, //use_qual_row_bw - p->num_active_planes, - p->num_of_dpp, - p->dcc_dram_bw_nom_overhead_factor_p0, - p->dcc_dram_bw_nom_overhead_factor_p1, - p->dcc_dram_bw_pref_overhead_factor_p0, - p->dcc_dram_bw_pref_overhead_factor_p1, - p->mall_prefetch_sdp_overhead_factor, - p->mall_prefetch_dram_overhead_factor, - p->surface_read_bandwidth_l, - p->surface_read_bandwidth_c, - p->prefetch_bandwidth_l, - p->prefetch_bandwidth_c, - p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw - p->excess_vactive_fill_bw_l, - p->excess_vactive_fill_bw_c, - p->cursor_bw, - p->dpte_row_bw, - p->meta_row_bw, - p->prefetch_cursor_bw, - p->prefetch_vmrow_bw, - p->flip_bw, - l->unity_array, - l->unity_array, - l->unity_array, - l->unity_array, - l->unity_array, - l->unity_array, - l->surface_dummy_bw, - p->surface_peak_required_bw[m][n]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: urg_vactive_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_vactive_bandwidth_required[m][n]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_qual[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); - DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->non_urg_bandwidth_required[m][n]); -#endif - DML_ASSERT(p->urg_bandwidth_required[m][n] >= p->non_urg_bandwidth_required[m][n]); - } - } -} - -static void check_urgent_bandwidth_support( - double *frac_urg_bandwidth_nom, - double *frac_urg_bandwidth_mall, - bool *vactive_bandwidth_support_ok, // vactive ok - bool *bandwidth_support_ok,// max of vm, prefetch, vactive all ok - - unsigned int mall_allocated_for_dcn_mbytes, - double non_urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_vactive_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) -{ - double frac_urg_bandwidth_nom_sdp = non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] / urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - double frac_urg_bandwidth_nom_dram = non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] / urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; - double frac_urg_bandwidth_mall_sdp; - double frac_urg_bandwidth_mall_dram; - if (urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] > 0) - frac_urg_bandwidth_mall_sdp = non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] / urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; - else - frac_urg_bandwidth_mall_sdp = 0.0; - if (urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] > 0) - frac_urg_bandwidth_mall_dram = non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] / urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; - else - frac_urg_bandwidth_mall_dram = 0.0; - - *bandwidth_support_ok = 1; - *vactive_bandwidth_support_ok = 1; - - // Check urgent bandwidth required at sdp vs urgent bandwidth avail at sdp -> FractionOfUrgentBandwidth - // Check urgent bandwidth required at dram vs urgent bandwidth avail at dram - // Check urgent bandwidth required at sdp vs urgent bandwidth avail at sdp, svp_prefetch -> FractionOfUrgentBandwidthMALL - // Check urgent bandwidth required at dram vs urgent bandwidth avail at dram, svp_prefetch - - *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; - - if (mall_allocated_for_dcn_mbytes > 0) { - *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; - *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; - } - - *frac_urg_bandwidth_nom = math_max2(frac_urg_bandwidth_nom_sdp, frac_urg_bandwidth_nom_dram); - *frac_urg_bandwidth_mall = math_max2(frac_urg_bandwidth_mall_sdp, frac_urg_bandwidth_mall_dram); - - *bandwidth_support_ok &= (*frac_urg_bandwidth_nom <= 1.0); - - if (mall_allocated_for_dcn_mbytes > 0) - *bandwidth_support_ok &= (*frac_urg_bandwidth_mall <= 1.0); - - *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; - if (mall_allocated_for_dcn_mbytes > 0) { - *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; - *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_sdp = %f\n", __func__, frac_urg_bandwidth_nom_sdp); - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_dram = %f\n", __func__, frac_urg_bandwidth_nom_dram); - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom = %f\n", __func__, *frac_urg_bandwidth_nom); - - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_sdp = %f\n", __func__, frac_urg_bandwidth_mall_sdp); - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_dram = %f\n", __func__, frac_urg_bandwidth_mall_dram); - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall = %f\n", __func__, *frac_urg_bandwidth_mall); - DML_LOG_VERBOSE("DML::%s: bandwidth_support_ok = %d\n", __func__, *bandwidth_support_ok); - - for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { - DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", - __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), - urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required[m][n]) ? "<" : ">=", urg_bandwidth_required[m][n]); - } - } -#endif -} - -static double get_bandwidth_available_for_immediate_flip(enum dml2_core_internal_soc_state_type eval_state, - double urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], // no flip - double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) -{ - double flip_bw_available_mbps; - double flip_bw_available_sdp_mbps; - double flip_bw_available_dram_mbps; - - flip_bw_available_sdp_mbps = urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp] - urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]; - flip_bw_available_dram_mbps = urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram] - urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]; - flip_bw_available_mbps = flip_bw_available_sdp_mbps < flip_bw_available_dram_mbps ? flip_bw_available_sdp_mbps : flip_bw_available_dram_mbps; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_sdp_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_dram_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_sdp_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_dram_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]); - DML_LOG_VERBOSE("DML::%s: flip_bw_available_sdp_mbps = %f\n", __func__, flip_bw_available_sdp_mbps); - DML_LOG_VERBOSE("DML::%s: flip_bw_available_dram_mbps = %f\n", __func__, flip_bw_available_dram_mbps); - DML_LOG_VERBOSE("DML::%s: flip_bw_available_mbps = %f\n", __func__, flip_bw_available_mbps); -#endif - - return flip_bw_available_mbps; -} - -static void calculate_immediate_flip_bandwidth_support( - // Output - double *frac_urg_bandwidth_flip, - bool *flip_bandwidth_support_ok, - - // Input - enum dml2_core_internal_soc_state_type eval_state, - double urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], - double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) -{ - double frac_urg_bw_flip_sdp = non_urg_bandwidth_required_flip[eval_state][dml2_core_internal_bw_sdp] / urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]; - double frac_urg_bw_flip_dram = non_urg_bandwidth_required_flip[eval_state][dml2_core_internal_bw_dram] / urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]; - - *flip_bandwidth_support_ok = true; - for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { // check sdp and dram - *flip_bandwidth_support_ok &= urg_bandwidth_available[eval_state][n] >= urg_bandwidth_required_flip[eval_state][n]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: n = %s\n", __func__, dml2_core_internal_bw_type_str(n)); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available = %f\n", __func__, urg_bandwidth_available[eval_state][n]); - DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required_flip = %f\n", __func__, non_urg_bandwidth_required_flip[eval_state][n]); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_flip = %f\n", __func__, urg_bandwidth_required_flip[eval_state][n]); - DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); -#endif - DML_ASSERT(urg_bandwidth_required_flip[eval_state][n] >= non_urg_bandwidth_required_flip[eval_state][n]); - } - - *frac_urg_bandwidth_flip = (frac_urg_bw_flip_sdp > frac_urg_bw_flip_dram) ? frac_urg_bw_flip_sdp : frac_urg_bw_flip_dram; - *flip_bandwidth_support_ok &= (*frac_urg_bandwidth_flip <= 1.0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); - DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_sdp = %f\n", __func__, frac_urg_bw_flip_sdp); - DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_dram = %f\n", __func__, frac_urg_bw_flip_dram); - DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_flip = %f\n", __func__, *frac_urg_bandwidth_flip); - DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); - - for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { - DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s, urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", - __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), - urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required_flip[m][n]) ? "<" : ">=", urg_bandwidth_required_flip[m][n]); - } - } -#endif -} - -static void CalculateFlipSchedule( - struct dml2_core_internal_scratch *s, - bool iflip_enable, - bool use_lb_flip_bw, - double HostVMInefficiencyFactor, - double Tvm_trips_flip, - double Tr0_trips_flip, - double Tvm_trips_flip_rounded, - double Tr0_trips_flip_rounded, - bool GPUVMEnable, - double vm_bytes, // vm_bytes - double DPTEBytesPerRow, // dpte_row_bytes - double BandwidthAvailableForImmediateFlip, - unsigned int TotImmediateFlipBytes, - enum dml2_source_format_class SourcePixelFormat, - double LineTime, - double VRatio, - double VRatioChroma, - double Tno_bw_flip, - unsigned int dpte_row_height, - unsigned int dpte_row_height_chroma, - bool use_one_row_for_frame_flip, - unsigned int max_flip_time_us, - unsigned int max_flip_time_lines, - unsigned int per_pipe_flip_bytes, - unsigned int meta_row_bytes, - unsigned int meta_row_height, - unsigned int meta_row_height_chroma, - bool dcc_mrq_enable, - - // Output - double *dst_y_per_vm_flip, - double *dst_y_per_row_flip, - double *final_flip_bw, - bool *ImmediateFlipSupportedForPipe) -{ - struct dml2_core_shared_CalculateFlipSchedule_locals *l = &s->CalculateFlipSchedule_locals; - - l->dual_plane = dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha; - l->dpte_row_bytes = DPTEBytesPerRow; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); - DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_us = %d\n", __func__, max_flip_time_us); - DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_lines = %d\n", __func__, max_flip_time_lines); - DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); - DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); - DML_LOG_VERBOSE("DML::%s: use_lb_flip_bw = %u\n", __func__, use_lb_flip_bw); - DML_LOG_VERBOSE("DML::%s: iflip_enable = %u\n", __func__, iflip_enable); - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); - DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, Tno_bw_flip); - DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, Tvm_trips_flip); - DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, Tr0_trips_flip); - DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded = %f\n", __func__, Tvm_trips_flip_rounded); - DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip_rounded = %f\n", __func__, Tr0_trips_flip_rounded); - DML_LOG_VERBOSE("DML::%s: vm_bytes = %f\n", __func__, vm_bytes); - DML_LOG_VERBOSE("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); - DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, meta_row_bytes); - DML_LOG_VERBOSE("DML::%s: dpte_row_bytes = %f\n", __func__, l->dpte_row_bytes); - DML_LOG_VERBOSE("DML::%s: dpte_row_height = %d\n", __func__, dpte_row_height); - DML_LOG_VERBOSE("DML::%s: meta_row_height = %d\n", __func__, meta_row_height); - DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); -#endif - - if (TotImmediateFlipBytes > 0 && (GPUVMEnable || dcc_mrq_enable)) { - if (l->dual_plane) { - if (dcc_mrq_enable & GPUVMEnable) { - l->min_row_height = math_min2(dpte_row_height, meta_row_height); - l->min_row_height_chroma = math_min2(dpte_row_height_chroma, meta_row_height_chroma); - } else if (GPUVMEnable) { - l->min_row_height = dpte_row_height; - l->min_row_height_chroma = dpte_row_height_chroma; - } else { - l->min_row_height = meta_row_height; - l->min_row_height_chroma = meta_row_height_chroma; - } - l->min_row_time = math_min2(l->min_row_height * LineTime / VRatio, l->min_row_height_chroma * LineTime / VRatioChroma); - } else { - if (dcc_mrq_enable & GPUVMEnable) - l->min_row_height = math_min2(dpte_row_height, meta_row_height); - else if (GPUVMEnable) - l->min_row_height = dpte_row_height; - else - l->min_row_height = meta_row_height; - - l->min_row_time = l->min_row_height * LineTime / VRatio; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: min_row_time = %f\n", __func__, l->min_row_time); -#endif - DML_ASSERT(l->min_row_time > 0); - - if (use_lb_flip_bw) { - // For mode check, calculation the flip bw requirement with worst case flip time - l->max_flip_time = math_min2(math_min2(l->min_row_time, (double)max_flip_time_lines * LineTime / VRatio), - math_max2(Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded, (double)max_flip_time_us)); - - //The lower bound on flip bandwidth - // Note: The get_urgent_bandwidth_required already consider dpte_row_bw and meta_row_bw in bandwidth calculation, so leave final_flip_bw = 0 if iflip not required - l->lb_flip_bw = 0; - - if (iflip_enable) { - l->hvm_scaled_vm_bytes = vm_bytes * HostVMInefficiencyFactor; - l->num_rows = 2; - l->hvm_scaled_row_bytes = (l->num_rows * l->dpte_row_bytes * HostVMInefficiencyFactor + l->num_rows * meta_row_bytes); - l->hvm_scaled_vm_row_bytes = l->hvm_scaled_vm_bytes + l->hvm_scaled_row_bytes; - l->lb_flip_bw = math_max3( - l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip), - l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded), - l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: max_flip_time = %f\n", __func__, l->max_flip_time); - DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_bytes); - DML_LOG_VERBOSE("DML::%s: total row bytes (%f row, hvm ineff scaled) = %f\n", __func__, l->num_rows, l->hvm_scaled_row_bytes); - DML_LOG_VERBOSE("DML::%s: total vm+row bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_row_bytes); - DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm and row = %f\n", __func__, l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip)); - DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm = %f\n", __func__, l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded)); - DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row = %f\n", __func__, l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); - - if (l->lb_flip_bw > 0) { - DML_LOG_VERBOSE("DML::%s: mode_support est Tvm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw); - DML_LOG_VERBOSE("DML::%s: mode_support est Tr0_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / l->num_rows); - DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_vm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw / LineTime); - DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_row_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / LineTime / l->num_rows); - DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded + 2*Tr0_trips_flip_rounded = %f\n", __func__, (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded)); - } -#endif - l->lb_flip_bw = math_max3(l->lb_flip_bw, - l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip, - (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm reg limit = %f\n", __func__, l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip); - DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row reg limit = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); -#endif - } - - *final_flip_bw = l->lb_flip_bw; - - *dst_y_per_vm_flip = 1; // not used - *dst_y_per_row_flip = 1; // not used - *ImmediateFlipSupportedForPipe = l->min_row_time >= (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded); - } else { - if (iflip_enable) { - l->ImmediateFlipBW = (double)per_pipe_flip_bytes * BandwidthAvailableForImmediateFlip / (double)TotImmediateFlipBytes; // flip_bw(i) - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: per_pipe_flip_bytes = %d\n", __func__, per_pipe_flip_bytes); - DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); - DML_LOG_VERBOSE("DML::%s: ImmediateFlipBW = %f\n", __func__, l->ImmediateFlipBW); - DML_LOG_VERBOSE("DML::%s: portion of flip bw = %f\n", __func__, (double)per_pipe_flip_bytes / (double)TotImmediateFlipBytes); -#endif - if (l->ImmediateFlipBW == 0) { - l->Tvm_flip = 0; - l->Tr0_flip = 0; - } else { - l->Tvm_flip = math_max3(Tvm_trips_flip, - Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, - LineTime / 4.0); - - l->Tr0_flip = math_max3(Tr0_trips_flip, - (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, - LineTime / 4.0); - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, vm_bytes * HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: total row bytes (hvm ineff scaled, one row) = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes)); - - DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (bw-based), Tvm_trips_flip = %f (latency-based)\n", __func__, Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, Tvm_trips_flip); - DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (bw-based), Tr0_trips_flip = %f (latency-based)\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, Tr0_trips_flip); -#endif - *dst_y_per_vm_flip = math_ceil2(4.0 * (l->Tvm_flip / LineTime), 1.0) / 4.0; - *dst_y_per_row_flip = math_ceil2(4.0 * (l->Tr0_flip / LineTime), 1.0) / 4.0; - - *final_flip_bw = math_max2(vm_bytes * HostVMInefficiencyFactor / (*dst_y_per_vm_flip * LineTime), - (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (*dst_y_per_row_flip * LineTime)); - - if (*dst_y_per_vm_flip >= 32 || *dst_y_per_row_flip >= 16 || l->Tvm_flip + 2 * l->Tr0_flip > l->min_row_time) { - *ImmediateFlipSupportedForPipe = false; - } else { - *ImmediateFlipSupportedForPipe = iflip_enable; - } - } else { - l->Tvm_flip = 0; - l->Tr0_flip = 0; - *dst_y_per_vm_flip = 0; - *dst_y_per_row_flip = 0; - *final_flip_bw = 0; - *ImmediateFlipSupportedForPipe = iflip_enable; - } - } - } else { - l->Tvm_flip = 0; - l->Tr0_flip = 0; - *dst_y_per_vm_flip = 0; - *dst_y_per_row_flip = 0; - *final_flip_bw = 0; - *ImmediateFlipSupportedForPipe = iflip_enable; - } - -#ifdef __DML_VBA_DEBUG__ - if (!use_lb_flip_bw) { - DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_flip = %f (should be < 32)\n", __func__, *dst_y_per_vm_flip); - DML_LOG_VERBOSE("DML::%s: dst_y_per_row_flip = %f (should be < 16)\n", __func__, *dst_y_per_row_flip); - DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (final)\n", __func__, l->Tvm_flip); - DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (final)\n", __func__, l->Tr0_flip); - DML_LOG_VERBOSE("DML::%s: Tvm_flip + 2*Tr0_flip = %f (should be <= min_row_time=%f)\n", __func__, l->Tvm_flip + 2 * l->Tr0_flip, l->min_row_time); - } - DML_LOG_VERBOSE("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); - DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); -#endif -} - -static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( - struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *p) -{ - struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals *s = &scratch->CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals; - - enum dml2_uclk_pstate_change_strategy uclk_pstate_change_strategy; - double reserved_vblank_time_us; - bool FoundCriticalSurface = false; - - s->TotalActiveWriteback = 0; - p->Watermark->UrgentWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); -#endif - - p->Watermark->USRRetrainingWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency + p->mmSOCParameters.USRRetrainingLatency + p->mmSOCParameters.SMNLatency; - p->Watermark->DRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->Watermark->UrgentWatermark; - p->Watermark->FCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->Watermark->UrgentWatermark; - p->Watermark->StutterExitWatermark = p->mmSOCParameters.SRExitTime + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; - p->Watermark->StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitTime + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; - p->Watermark->Z8StutterExitWatermark = p->mmSOCParameters.SRExitZ8Time + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; - p->Watermark->Z8StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitZ8Time + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; - if (p->mmSOCParameters.qos_type == dml2_qos_param_type_dcn4x) { - p->Watermark->StutterExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; - p->Watermark->StutterEnterPlusExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; - p->Watermark->Z8StutterExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; - p->Watermark->Z8StutterEnterPlusExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; - } - p->Watermark->temp_read_or_ppt_watermark_us = p->mmSOCParameters.g6_temp_read_blackout_us + p->Watermark->UrgentWatermark; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); - DML_LOG_VERBOSE("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); - DML_LOG_VERBOSE("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); - DML_LOG_VERBOSE("DML::%s: SREnterPlusExitZ8Time = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitZ8Time); - DML_LOG_VERBOSE("DML::%s: SREnterPlusExitTime = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitTime); - DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); - DML_LOG_VERBOSE("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); - DML_LOG_VERBOSE("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); - DML_LOG_VERBOSE("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); - DML_LOG_VERBOSE("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); - DML_LOG_VERBOSE("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); - DML_LOG_VERBOSE("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); - DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); - DML_LOG_VERBOSE("DML::%s: temp_read_or_ppt_watermark_us = %f\n", __func__, p->Watermark->temp_read_or_ppt_watermark_us); -#endif - - s->TotalActiveWriteback = 0; - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - s->TotalActiveWriteback = s->TotalActiveWriteback + 1; - } - } - - if (s->TotalActiveWriteback <= 1) { - p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency; - } else { - p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; - } - if (p->USRRetrainingRequired) - p->Watermark->WritebackUrgentWatermark = p->Watermark->WritebackUrgentWatermark + p->mmSOCParameters.USRRetrainingLatency; - - if (s->TotalActiveWriteback <= 1) { - p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency; - p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency; - } else { - p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; - p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024 / 32 / p->SOCCLK; - } - - if (p->USRRetrainingRequired) - p->Watermark->WritebackDRAMClockChangeWatermark = p->Watermark->WritebackDRAMClockChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; - - if (p->USRRetrainingRequired) - p->Watermark->WritebackFCLKChangeWatermark = p->Watermark->WritebackFCLKChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); - DML_LOG_VERBOSE("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); - DML_LOG_VERBOSE("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); - DML_LOG_VERBOSE("DML::%s: USRRetrainingRequired = %u\n", __func__, p->USRRetrainingRequired); - DML_LOG_VERBOSE("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); -#endif - - s->TotalPixelBW = 0.0; - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - double h_total = (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; - double pixel_clock_mhz = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000.0; - double v_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - double v_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - s->TotalPixelBW = s->TotalPixelBW + p->DPPPerSurface[k] - * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * v_ratio + p->SwathWidthC[k] * p->BytePerPixelDETC[k] * v_ratio_c) / (h_total / pixel_clock_mhz); - } - - *p->global_fclk_change_supported = true; - *p->global_dram_clock_change_supported = true; - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - double h_total = (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; - double pixel_clock_mhz = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000.0; - double v_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - double v_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - double v_taps = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - double v_taps_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - double h_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio; - double h_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio; - double LBBitPerPixel = 57; - - s->LBLatencyHidingSourceLinesY[k] = (unsigned int)(math_min2((double)p->MaxLineBufferLines, math_floor2((double)p->LineBufferSize / LBBitPerPixel / ((double)p->SwathWidthY[k] / math_max2(h_ratio, 1.0)), 1)) - (v_taps - 1)); - s->LBLatencyHidingSourceLinesC[k] = (unsigned int)(math_min2((double)p->MaxLineBufferLines, math_floor2((double)p->LineBufferSize / LBBitPerPixel / ((double)p->SwathWidthC[k] / math_max2(h_ratio_c, 1.0)), 1)) - (v_taps_c - 1)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); - DML_LOG_VERBOSE("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); - DML_LOG_VERBOSE("DML::%s: k=%u, LBBitPerPixel = %f\n", __func__, k, LBBitPerPixel); - DML_LOG_VERBOSE("DML::%s: k=%u, HRatio = %f\n", __func__, k, h_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u, VTaps = %f\n", __func__, k, v_taps); -#endif - - s->EffectiveLBLatencyHidingY = s->LBLatencyHidingSourceLinesY[k] / v_ratio * (h_total / pixel_clock_mhz); - s->EffectiveLBLatencyHidingC = s->LBLatencyHidingSourceLinesC[k] / v_ratio_c * (h_total / pixel_clock_mhz); - - s->EffectiveDETBufferSizeY = p->DETBufferSizeY[k]; - if (p->UnboundedRequestEnabled) { - s->EffectiveDETBufferSizeY = s->EffectiveDETBufferSizeY + p->CompressedBufferSizeInkByte * 1024 * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * v_ratio) / (h_total / pixel_clock_mhz) / s->TotalPixelBW; - } - - s->LinesInDETY[k] = (double)s->EffectiveDETBufferSizeY / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; - s->LinesInDETYRoundedDownToSwath[k] = (unsigned int)(math_floor2(s->LinesInDETY[k], p->SwathHeightY[k])); - s->FullDETBufferingTimeY = s->LinesInDETYRoundedDownToSwath[k] * (h_total / pixel_clock_mhz) / v_ratio; - - s->ActiveClockChangeLatencyHidingY = s->EffectiveLBLatencyHidingY + s->FullDETBufferingTimeY - ((double)p->DSTXAfterScaler[k] / h_total + (double)p->DSTYAfterScaler[k]) * h_total / pixel_clock_mhz; - - if (p->NumberOfActiveSurfaces > 1) { - s->ActiveClockChangeLatencyHidingY = s->ActiveClockChangeLatencyHidingY - (1.0 - 1.0 / (double)p->NumberOfActiveSurfaces) * (double)p->SwathHeightY[k] * (double)h_total / pixel_clock_mhz / v_ratio; - } - - if (p->BytePerPixelDETC[k] > 0) { - s->LinesInDETC[k] = p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k]; - s->LinesInDETCRoundedDownToSwath[k] = (unsigned int)(math_floor2(s->LinesInDETC[k], p->SwathHeightC[k])); - s->FullDETBufferingTimeC = s->LinesInDETCRoundedDownToSwath[k] * (h_total / pixel_clock_mhz) / v_ratio_c; - s->ActiveClockChangeLatencyHidingC = s->EffectiveLBLatencyHidingC + s->FullDETBufferingTimeC - ((double)p->DSTXAfterScaler[k] / (double)h_total + (double)p->DSTYAfterScaler[k]) * (double)h_total / pixel_clock_mhz; - if (p->NumberOfActiveSurfaces > 1) { - s->ActiveClockChangeLatencyHidingC = s->ActiveClockChangeLatencyHidingC - (1.0 - 1.0 / (double)p->NumberOfActiveSurfaces) * (double)p->SwathHeightC[k] * (double)h_total / pixel_clock_mhz / v_ratio_c; - } - s->ActiveClockChangeLatencyHiding = math_min2(s->ActiveClockChangeLatencyHidingY, s->ActiveClockChangeLatencyHidingC); - } else { - s->ActiveClockChangeLatencyHiding = s->ActiveClockChangeLatencyHidingY; - } - - s->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->DRAMClockChangeWatermark; - s->ActiveFCLKChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->FCLKChangeWatermark; - s->USRRetrainingLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->USRRetrainingWatermark; - s->g6_temp_read_latency_margin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->temp_read_or_ppt_watermark_us; - - if (p->VActiveLatencyHidingMargin) - p->VActiveLatencyHidingMargin[k] = s->ActiveDRAMClockChangeLatencyMargin[k]; - - if (p->VActiveLatencyHidingUs) - p->VActiveLatencyHidingUs[k] = s->ActiveClockChangeLatencyHiding; - - if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - s->WritebackLatencyHiding = (double)p->WritebackInterfaceBufferSize * 1024.0 - / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height - * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width - / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height * (double)h_total / pixel_clock_mhz) * 4.0); - if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format == dml2_444_64) { - s->WritebackLatencyHiding = s->WritebackLatencyHiding / 2; - } - s->WritebackDRAMClockChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackDRAMClockChangeWatermark; - - s->WritebackFCLKChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackFCLKChangeWatermark; - - s->ActiveDRAMClockChangeLatencyMargin[k] = math_min2(s->ActiveDRAMClockChangeLatencyMargin[k], s->WritebackDRAMClockChangeLatencyMargin); - s->ActiveFCLKChangeLatencyMargin[k] = math_min2(s->ActiveFCLKChangeLatencyMargin[k], s->WritebackFCLKChangeLatencyMargin); - } - p->MaxActiveDRAMClockChangeLatencySupported[k] = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) ? 0 : (s->ActiveDRAMClockChangeLatencyMargin[k] + p->mmSOCParameters.DRAMClockChangeLatency); - - uclk_pstate_change_strategy = p->display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy; - reserved_vblank_time_us = (double)p->display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns / 1000; - - p->FCLKChangeSupport[k] = dml2_pstate_change_unsupported; - if (s->ActiveFCLKChangeLatencyMargin[k] > 0) - p->FCLKChangeSupport[k] = dml2_pstate_change_vactive; - else if (reserved_vblank_time_us >= p->mmSOCParameters.FCLKChangeLatency) - p->FCLKChangeSupport[k] = dml2_pstate_change_vblank; - - if (p->FCLKChangeSupport[k] == dml2_pstate_change_unsupported) - *p->global_fclk_change_supported = false; - - p->DRAMClockChangeSupport[k] = dml2_pstate_change_unsupported; - if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_auto) { - if (p->display_cfg->overrides.all_streams_blanked || - (s->ActiveDRAMClockChangeLatencyMargin[k] > 0 && reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency)) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank_and_vactive; - else if (s->ActiveDRAMClockChangeLatencyMargin[k] > 0) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_vactive; - else if (reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank; - } else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_vactive && s->ActiveDRAMClockChangeLatencyMargin[k] > 0) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_vactive; - else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_vblank && reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank; - else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_drr) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_drr; - else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_svp) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_mall_svp; - else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame) - p->DRAMClockChangeSupport[k] = dml2_pstate_change_mall_full_frame; - - if (p->DRAMClockChangeSupport[k] == dml2_pstate_change_unsupported) - *p->global_dram_clock_change_supported = false; - - s->dst_y_pstate = (unsigned int)(math_ceil2((p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.UrgentLatency) / (h_total / pixel_clock_mhz), 1)); - s->src_y_pstate_l = (unsigned int)(math_ceil2(s->dst_y_pstate * v_ratio, p->SwathHeightY[k])); - s->src_y_ahead_l = (unsigned int)(math_floor2(p->DETBufferSizeY[k] / p->BytePerPixelDETY[k] / p->SwathWidthY[k], p->SwathHeightY[k]) + s->LBLatencyHidingSourceLinesY[k]); - s->sub_vp_lines_l = s->src_y_pstate_l + s->src_y_ahead_l + p->meta_row_height_l[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); - DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); - DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); - DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_l = %u\n", __func__, k, p->meta_row_height_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); -#endif - p->SubViewportLinesNeededInMALL[k] = s->sub_vp_lines_l; - - if (p->BytePerPixelDETC[k] > 0) { - s->src_y_pstate_c = (unsigned int)(math_ceil2(s->dst_y_pstate * v_ratio_c, p->SwathHeightC[k])); - s->src_y_ahead_c = (unsigned int)(math_floor2(p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k], p->SwathHeightC[k]) + s->LBLatencyHidingSourceLinesC[k]); - s->sub_vp_lines_c = s->src_y_pstate_c + s->src_y_ahead_c + p->meta_row_height_c[k]; - - if (dml_is_420(p->display_cfg->plane_descriptors[k].pixel_format)) - p->SubViewportLinesNeededInMALL[k] = (unsigned int)(math_max2(s->sub_vp_lines_l, 2 * s->sub_vp_lines_c)); - else - p->SubViewportLinesNeededInMALL[k] = (unsigned int)(math_max2(s->sub_vp_lines_l, s->sub_vp_lines_c)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_c = %u\n", __func__, k, p->meta_row_height_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); - DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); - DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); -#endif - } - } - - *p->g6_temp_read_support = true; - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if ((!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) && - (s->g6_temp_read_latency_margin[k] < 0)) { - *p->g6_temp_read_support = false; - } - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if ((!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) && ((!FoundCriticalSurface) - || ((s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency) < *p->MaxActiveFCLKChangeLatencySupported))) { - FoundCriticalSurface = true; - *p->MaxActiveFCLKChangeLatencySupported = s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency; - } - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->global_dram_clock_change_supported); - DML_LOG_VERBOSE("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->global_fclk_change_supported); - DML_LOG_VERBOSE("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); - DML_LOG_VERBOSE("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); -#endif -} - -static void calculate_bytes_to_fetch_required_to_hide_latency( - struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *p) -{ - unsigned int dst_lines_to_hide; - unsigned int src_lines_to_hide_l; - unsigned int src_lines_to_hide_c; - unsigned int plane_index; - unsigned int stream_index; - - for (plane_index = 0; plane_index < p->num_active_planes; plane_index++) { - if (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[plane_index])) - continue; - - stream_index = p->display_cfg->plane_descriptors[plane_index].stream_index; - - dst_lines_to_hide = (unsigned int)math_ceil(p->latency_to_hide_us / - ((double)p->display_cfg->stream_descriptors[stream_index].timing.h_total / - (double)p->display_cfg->stream_descriptors[stream_index].timing.pixel_clock_khz * 1000.0)); - - src_lines_to_hide_l = (unsigned int)math_ceil2(p->display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane0.v_ratio * dst_lines_to_hide, - p->swath_height_l[plane_index]); - p->bytes_required_l[plane_index] = src_lines_to_hide_l * p->num_of_dpp[plane_index] * p->swath_width_l[plane_index] * p->byte_per_pix_l[plane_index]; - - src_lines_to_hide_c = (unsigned int)math_ceil2(p->display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane1.v_ratio * dst_lines_to_hide, - p->swath_height_c[plane_index]); - p->bytes_required_c[plane_index] = src_lines_to_hide_c * p->num_of_dpp[plane_index] * p->swath_width_c[plane_index] * p->byte_per_pix_c[plane_index]; - - if (p->display_cfg->plane_descriptors[plane_index].surface.dcc.enable && p->mrq_present) { - p->bytes_required_l[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_l / p->meta_row_height_l[plane_index]) * p->meta_row_bytes_per_row_ub_l[plane_index]; - if (p->meta_row_height_c[plane_index]) { - p->bytes_required_c[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_c / p->meta_row_height_c[plane_index]) * p->meta_row_bytes_per_row_ub_c[plane_index]; - } - } - - if (p->display_cfg->gpuvm_enable == true) { - p->bytes_required_l[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_l / p->dpte_row_height_l[plane_index]) * p->dpte_bytes_per_row_l[plane_index]; - if (p->dpte_row_height_c[plane_index]) { - p->bytes_required_c[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_c / p->dpte_row_height_c[plane_index]) * p->dpte_bytes_per_row_c[plane_index]; - } - } - } -} - -static noinline_for_stack void calculate_vactive_det_fill_latency( - const struct dml2_display_cfg *display_cfg, - unsigned int num_active_planes, - unsigned int bytes_required_l[], - unsigned int bytes_required_c[], - double dcc_dram_bw_nom_overhead_factor_p0[], - double dcc_dram_bw_nom_overhead_factor_p1[], - double surface_read_bw_l[], - double surface_read_bw_c[], - double (*surface_avg_vactive_required_bw)[dml2_core_internal_bw_max][DML2_MAX_PLANES], - double (*surface_peak_required_bw)[dml2_core_internal_bw_max][DML2_MAX_PLANES], - /* output */ - double vactive_det_fill_delay_us[]) -{ - double effective_excess_bandwidth; - double effective_excess_bandwidth_l; - double effective_excess_bandwidth_c; - double adj_factor; - unsigned int plane_index; - unsigned int soc_state; - unsigned int bw_type; - - for (plane_index = 0; plane_index < num_active_planes; plane_index++) { - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_index])) - continue; - - vactive_det_fill_delay_us[plane_index] = 0.0; - for (soc_state = 0; soc_state < dml2_core_internal_soc_state_max; soc_state++) { - for (bw_type = 0; bw_type < dml2_core_internal_bw_max; bw_type++) { - effective_excess_bandwidth = (surface_peak_required_bw[soc_state][bw_type][plane_index] - surface_avg_vactive_required_bw[soc_state][bw_type][plane_index]); - - /* luma */ - adj_factor = bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p0[plane_index] : 1.0; - - effective_excess_bandwidth_l = effective_excess_bandwidth * surface_read_bw_l[plane_index] / (surface_read_bw_l[plane_index] + surface_read_bw_c[plane_index]) / adj_factor; - if (effective_excess_bandwidth_l > 0.0) { - vactive_det_fill_delay_us[plane_index] = math_max2(vactive_det_fill_delay_us[plane_index], bytes_required_l[plane_index] / effective_excess_bandwidth_l); - } - - /* chroma */ - adj_factor = bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p1[plane_index] : 1.0; - - effective_excess_bandwidth_c = effective_excess_bandwidth * surface_read_bw_c[plane_index] / (surface_read_bw_l[plane_index] + surface_read_bw_c[plane_index]) / adj_factor; - if (effective_excess_bandwidth_c > 0.0) { - vactive_det_fill_delay_us[plane_index] = math_max2(vactive_det_fill_delay_us[plane_index], bytes_required_c[plane_index] / effective_excess_bandwidth_c); - } - } - } - } -} - -static void calculate_excess_vactive_bandwidth_required( - const struct dml2_display_cfg *display_cfg, - unsigned int num_active_planes, - unsigned int bytes_required_l[], - unsigned int bytes_required_c[], - /* outputs */ - double excess_vactive_fill_bw_l[], - double excess_vactive_fill_bw_c[]) -{ - unsigned int plane_index; - - for (plane_index = 0; plane_index < num_active_planes; plane_index++) { - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_index])) - continue; - - excess_vactive_fill_bw_l[plane_index] = 0.0; - excess_vactive_fill_bw_c[plane_index] = 0.0; - - if (display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us > 0) { - excess_vactive_fill_bw_l[plane_index] = (double)bytes_required_l[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; - excess_vactive_fill_bw_c[plane_index] = (double)bytes_required_c[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; - } - } -} - -static double uclk_khz_to_dram_bw_mbps(unsigned long uclk_khz, const struct dml2_dram_params *dram_config) -{ - double bw_mbps = 0; - bw_mbps = ((double)uclk_khz * dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock) / 1000.0; - - return bw_mbps; -} - -static double dram_bw_kbps_to_uclk_mhz(unsigned long long bw_kbps, const struct dml2_dram_params *dram_config) -{ - double uclk_mhz = 0; - - uclk_mhz = (double)bw_kbps / (dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock) / 1000.0; - - return uclk_mhz; -} - -static unsigned int get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params) -{ - unsigned int i; - unsigned int index = 0; - - for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); - - if (i == 0) - index = 0; - else - index = i - 1; - - if (uclk_freq_khz < per_uclk_dpm_params[i].minimum_uclk_khz || - per_uclk_dpm_params[i].minimum_uclk_khz == 0) { - break; - } - } - DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); - return index; -} - -static unsigned int get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) -{ - unsigned int i; - bool clk_entry_found = false; - - for (i = 0; i < clk_table->uclk.num_clk_values; i++) { - DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); - - if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { - clk_entry_found = true; - break; - } - } - - if (!clk_entry_found) - DML_ASSERT(clk_entry_found); -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); -#endif - return i; -} - -static unsigned int get_pipe_flip_bytes( - double hostvm_inefficiency_factor, - unsigned int vm_bytes, - unsigned int dpte_row_bytes, - unsigned int meta_row_bytes) -{ - unsigned int flip_bytes = 0; - - flip_bytes += (unsigned int) ((vm_bytes * hostvm_inefficiency_factor) + 2*meta_row_bytes); - flip_bytes += (unsigned int) (2*dpte_row_bytes * hostvm_inefficiency_factor); - - return flip_bytes; -} - -static void calculate_hostvm_inefficiency_factor( - double *HostVMInefficiencyFactor, - double *HostVMInefficiencyFactorPrefetch, - - bool gpuvm_enable, - bool hostvm_enable, - unsigned int remote_iommu_outstanding_translations, - unsigned int max_outstanding_reqs, - double urg_bandwidth_avail_active_pixel_and_vm, - double urg_bandwidth_avail_active_vm_only) -{ - *HostVMInefficiencyFactor = 1; - *HostVMInefficiencyFactorPrefetch = 1; - - if (gpuvm_enable && hostvm_enable) { - *HostVMInefficiencyFactor = urg_bandwidth_avail_active_pixel_and_vm / urg_bandwidth_avail_active_vm_only; - *HostVMInefficiencyFactorPrefetch = *HostVMInefficiencyFactor; - - if ((*HostVMInefficiencyFactorPrefetch < 4) && (remote_iommu_outstanding_translations < max_outstanding_reqs)) - *HostVMInefficiencyFactorPrefetch = 4; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_pixel_and_vm = %f\n", __func__, urg_bandwidth_avail_active_pixel_and_vm); - DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_vm_only = %f\n", __func__, urg_bandwidth_avail_active_vm_only); - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, *HostVMInefficiencyFactor); - DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactorPrefetch = %f\n", __func__, *HostVMInefficiencyFactorPrefetch); -#endif - } -} - -struct dml2_core_internal_g6_temp_read_blackouts_table { - struct { - unsigned int uclk_khz; - unsigned int blackout_us; - } entries[DML_MAX_CLK_TABLE_SIZE]; -}; - -struct dml2_core_internal_g6_temp_read_blackouts_table core_dcn4_g6_temp_read_blackout_table = { - .entries = { - { - .uclk_khz = 96000, - .blackout_us = 23, - }, - { - .uclk_khz = 435000, - .blackout_us = 10, - }, - { - .uclk_khz = 521000, - .blackout_us = 10, - }, - { - .uclk_khz = 731000, - .blackout_us = 8, - }, - { - .uclk_khz = 822000, - .blackout_us = 8, - }, - { - .uclk_khz = 962000, - .blackout_us = 5, - }, - { - .uclk_khz = 1069000, - .blackout_us = 5, - }, - { - .uclk_khz = 1187000, - .blackout_us = 5, - }, - }, -}; - -static double get_g6_temp_read_blackout_us( - struct dml2_soc_bb *soc, - unsigned int uclk_freq_khz, - unsigned int min_clk_index) -{ - unsigned int i; - unsigned int blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[0].blackout_us; - - if (soc->power_management_parameters.g6_temp_read_blackout_us[0] > 0.0) { - /* overrides are present in the SoC BB */ - return soc->power_management_parameters.g6_temp_read_blackout_us[min_clk_index]; - } - - /* use internal table */ - blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[0].blackout_us; - - for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - if (uclk_freq_khz < core_dcn4_g6_temp_read_blackout_table.entries[i].uclk_khz || - core_dcn4_g6_temp_read_blackout_table.entries[i].uclk_khz == 0) { - break; - } - - blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[i].blackout_us; - } - - return (double)blackout_us; -} - -static double get_max_urgent_latency_us( - struct dml2_dcn4x_soc_qos_params *dcn4x, - double uclk_freq_mhz, - double FabricClock, - unsigned int min_clk_index) -{ - double latency; - latency = dcn4x->per_uclk_dpm_params[min_clk_index].maximum_latency_when_urgent_uclk_cycles / uclk_freq_mhz - * (1 + dcn4x->umc_max_latency_margin / 100.0) - + dcn4x->mall_overhead_fclk_cycles / FabricClock - + dcn4x->max_round_trip_to_furthest_cs_fclk_cycles / FabricClock - * (1 + dcn4x->fabric_max_transport_latency_margin / 100.0); - return latency; -} - -static void calculate_pstate_keepout_dst_lines( - const struct dml2_display_cfg *display_cfg, - const struct dml2_core_internal_watermarks *watermarks, - unsigned int pstate_keepout_dst_lines[]) -{ - const struct dml2_stream_parameters *stream_descriptor; - unsigned int i; - - for (i = 0; i < display_cfg->num_planes; i++) { - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[i])) { - stream_descriptor = &display_cfg->stream_descriptors[display_cfg->plane_descriptors[i].stream_index]; - - pstate_keepout_dst_lines[i] = - (unsigned int)math_ceil(watermarks->DRAMClockChangeWatermark / ((double)stream_descriptor->timing.h_total * 1000.0 / (double)stream_descriptor->timing.pixel_clock_khz)); - - if (pstate_keepout_dst_lines[i] > stream_descriptor->timing.v_total - 1) { - pstate_keepout_dst_lines[i] = stream_descriptor->timing.v_total - 1; - } - } - } -} - -static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_internal_display_mode_lib *mode_lib, - const struct dml2_display_cfg *display_cfg) -{ - struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; - struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; - struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; - struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; -#ifdef DML_GLOBAL_PREFETCH_CHECK - struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; -#endif - struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; - - double min_return_bw_for_latency; - unsigned int k; - - mode_lib->ms.TimeCalc = 24 / mode_lib->ms.dcfclk_deepsleep; - - calculate_hostvm_inefficiency_factor( - &s->HostVMInefficiencyFactor, - &s->HostVMInefficiencyFactorPrefetch, - - display_cfg->gpuvm_enable, - display_cfg->hostvm_enable, - mode_lib->ip.remote_iommu_outstanding_translations, - mode_lib->soc.max_outstanding_reqs, - mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], - mode_lib->ms.support.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); - - mode_lib->ms.Total3dlutActive = 0; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) - mode_lib->ms.Total3dlutActive = mode_lib->ms.Total3dlutActive + 1; - - // Calculate tdlut schedule related terms - calculate_tdlut_setting_params->dispclk_mhz = mode_lib->ms.RequiredDISPCLK; - calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; - calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; - calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; - calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; - calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - calculate_tdlut_setting_params->tdlut_mpc_width_flag = display_cfg->plane_descriptors[k].tdlut.tdlut_mpc_width_flag; - calculate_tdlut_setting_params->is_gfx11 = dml_get_gfx_version(display_cfg->plane_descriptors[k].surface.tiling); - - // output - calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; - calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; - calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; - calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; - - calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); - } - - min_return_bw_for_latency = mode_lib->ms.support.urg_bandwidth_available_min_latency[dml2_core_internal_soc_state_sys_active]; - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) - s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); - - CalculateExtraLatency( - display_cfg, - mode_lib->ip.rob_buffer_size_kbytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, - s->ReorderingBytes, - mode_lib->ms.DCFCLK, - mode_lib->ms.FabricClock, - mode_lib->ip.pixel_chunk_size_kbytes, - min_return_bw_for_latency, - mode_lib->ms.num_active_planes, - mode_lib->ms.NoOfDPP, - mode_lib->ms.dpte_group_bytes, - s->tdlut_bytes_per_group, - s->HostVMInefficiencyFactor, - s->HostVMInefficiencyFactorPrefetch, - mode_lib->soc.hostvm_min_page_size_kbytes, - mode_lib->soc.qos_parameters.qos_type, - !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), - mode_lib->soc.max_outstanding_reqs, - mode_lib->ms.support.request_size_bytes_luma, - mode_lib->ms.support.request_size_bytes_chroma, - mode_lib->ip.meta_chunk_size_kbytes, - mode_lib->ip.dchub_arb_to_ret_delay, - mode_lib->ms.TripToMemory, - mode_lib->ip.hostvm_mode, - - // output - &mode_lib->ms.ExtraLatency, - &mode_lib->ms.ExtraLatency_sr, - &mode_lib->ms.ExtraLatencyPrefetch); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - s->impacted_dst_y_pre[k] = 0; - - s->recalc_prefetch_schedule = 0; - s->recalc_prefetch_done = 0; - do { - mode_lib->ms.support.PrefetchSupported = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; - - s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->ms.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane0.width, - display_cfg->plane_descriptors[k].composition.viewport.plane0.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->ms.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane1.width, - display_cfg->plane_descriptors[k].composition.viewport.plane1.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; - - mode_lib->ms.TWait[k] = CalculateTWait( - display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, - mode_lib->ms.UrgLatency, - mode_lib->ms.TripToMemory, - !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? - get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx) : 0.0); - - myPipe->Dppclk = mode_lib->ms.RequiredDPPCLK[k]; - myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK; - myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - myPipe->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; - myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[k]; - myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; - myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; - myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; - myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; - myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; - myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; - myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; - myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; - myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; - myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; - myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; - myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; - myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; - myPipe->ODMMode = mode_lib->ms.ODMMode[k]; - myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; - myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; - myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; - myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[k]); -#endif - CalculatePrefetchSchedule_params->display_cfg = display_cfg; - CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; - CalculatePrefetchSchedule_params->myPipe = myPipe; - CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelay[k]; - CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; - CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; - CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; - CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; - CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; - CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->ms.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; - CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; - CalculatePrefetchSchedule_params->VStartup = s->MaximumVStartup[k]; - CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; - CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; - CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; - CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; - CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; - CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; - CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->ms.ExtraLatencyPrefetch; - CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; - CalculatePrefetchSchedule_params->vm_bytes = mode_lib->ms.vm_bytes[k]; - CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[k]; - CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; - CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[k]; - CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; - CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC[k]; - CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub[k]; - CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub[k]; - CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightY[k]; - CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightC[k]; - CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait[k]; - CalculatePrefetchSchedule_params->Ttrip = mode_lib->ms.TripToMemory; - CalculatePrefetchSchedule_params->Turg = mode_lib->ms.UrgLatency; - CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; - CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; - CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); - CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; - CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; - CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; - CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; - CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->ms.meta_row_bytes[k]; - CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor[k]; - CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->ms.vactive_sw_bw_l[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->ms.vactive_sw_bw_c[k]; - - // output - CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; - CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; - CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->ms.dst_y_prefetch[k]; - CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->ms.LinesForVM[k]; - CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->ms.LinesForDPTERow[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c - CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &mode_lib->ms.RequiredPrefetchBWMax[k]; - CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; - CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; - CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; - CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; - CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; - CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; - CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; - CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; - CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; - CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; - CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; - CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->ms.prefetch_cursor_bw[k]; - CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; - CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; - CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; - CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->prefetch_swath_time_us[k]; - - mode_lib->ms.NoTimeForPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); - - mode_lib->ms.support.PrefetchSupported &= !mode_lib->ms.NoTimeForPrefetch[k]; - DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_vm_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_row_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_row_vblank); - } // for k num_planes - - CalculateDCFCLKDeepSleepTdlut( - display_cfg, - mode_lib->ms.num_active_planes, - mode_lib->ms.BytePerPixelY, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.SwathWidthY, - mode_lib->ms.SwathWidthC, - mode_lib->ms.NoOfDPP, - mode_lib->ms.PSCL_FACTOR, - mode_lib->ms.PSCL_FACTOR_CHROMA, - mode_lib->ms.RequiredDPPCLK, - mode_lib->ms.vactive_sw_bw_l, - mode_lib->ms.vactive_sw_bw_c, - mode_lib->soc.return_bus_width_bytes, - mode_lib->ms.RequiredDISPCLK, - s->tdlut_bytes_to_deliver, - s->prefetch_swath_time_us, - - /* Output */ - &mode_lib->ms.dcfclk_deepsleep); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.dst_y_prefetch[k] < 2.0 - || mode_lib->ms.LinesForVM[k] >= 32.0 - || mode_lib->ms.LinesForDPTERow[k] >= 16.0 - || mode_lib->ms.NoTimeForPrefetch[k] == true - || s->DSTYAfterScaler[k] > 8) { - mode_lib->ms.support.PrefetchSupported = false; - DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_prefetch=%f (should not be < 2)\n", __func__, k, mode_lib->ms.dst_y_prefetch[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, LinesForVM=%f (should not be >= 32)\n", __func__, k, mode_lib->ms.LinesForVM[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, LinesForDPTERow=%f (should not be >= 16)\n", __func__, k, mode_lib->ms.LinesForDPTERow[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, DSTYAfterScaler=%d (should be <= 8)\n", __func__, k, s->DSTYAfterScaler[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, NoTimeForPrefetch=%d\n", __func__, k, mode_lib->ms.NoTimeForPrefetch[k]); - } - } - - mode_lib->ms.support.DynamicMetadataSupported = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.NoTimeForDynamicMetadata[k] == true) { - mode_lib->ms.support.DynamicMetadataSupported = false; - } - } - - mode_lib->ms.support.VRatioInPrefetchSupported = true; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.VRatioPreY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || - mode_lib->ms.VRatioPreC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { - mode_lib->ms.support.VRatioInPrefetchSupported = false; - DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreY = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreY[k], __DML2_CALCS_MAX_VRATIO_PRE__); - DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreC = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreC[k], __DML2_CALCS_MAX_VRATIO_PRE__); - DML_LOG_VERBOSE("DML::%s: VRatioInPrefetchSupported = %u\n", __func__, mode_lib->ms.support.VRatioInPrefetchSupported); - } - } - - mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.VRatioInPrefetchSupported; - - // By default, do not recalc prefetch schedule - s->recalc_prefetch_schedule = 0; - - // Only do urg vs prefetch bandwidth check, flip schedule check, power saving feature support check IF the Prefetch Schedule Check is ok - if (mode_lib->ms.support.PrefetchSupported) { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - // Calculate Urgent burst factor for prefetch -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor (for prefetch)\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreY=%f\n", __func__, k, mode_lib->ms.VRatioPreY[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreC=%f\n", __func__, k, mode_lib->ms.VRatioPreC[k]); -#endif - CalculateUrgentBurstFactor( - &display_cfg->plane_descriptors[k], - mode_lib->ms.swath_width_luma_ub[k], - mode_lib->ms.swath_width_chroma_ub[k], - mode_lib->ms.SwathHeightY[k], - mode_lib->ms.SwathHeightC[k], - s->line_times[k], - mode_lib->ms.UrgLatency, - mode_lib->ms.VRatioPreY[k], - mode_lib->ms.VRatioPreC[k], - mode_lib->ms.BytePerPixelInDETY[k], - mode_lib->ms.BytePerPixelInDETC[k], - mode_lib->ms.DETBufferSizeY[k], - mode_lib->ms.DETBufferSizeC[k], - /* Output */ - &mode_lib->ms.UrgentBurstFactorLumaPre[k], - &mode_lib->ms.UrgentBurstFactorChromaPre[k], - &mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); - } - - // Calculate urgent bandwidth required, both urg and non urg peak bandwidth - // assume flip bw is 0 at this point - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - mode_lib->ms.final_flip_bw[k] = 0; - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->ms.support.urg_vactive_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->ms.support.urg_bandwidth_required_qual; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = mode_lib->ms.surface_avg_vactive_required_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 0; - calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - // Check urg peak bandwidth against available urg bw - // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) - check_urgent_bandwidth_support( - &s->dummy_single[0], // double* frac_urg_bandwidth - &s->dummy_single[1], // double* frac_urg_bandwidth_mall - &mode_lib->ms.support.UrgVactiveBandwidthSupport, - &mode_lib->ms.support.PrefetchBandwidthSupported, - - mode_lib->soc.mall_allocated_for_dcn_mbytes, - mode_lib->ms.support.non_urg_bandwidth_required, - mode_lib->ms.support.urg_vactive_bandwidth_required, - mode_lib->ms.support.urg_bandwidth_required, - mode_lib->ms.support.urg_bandwidth_available); - - mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.PrefetchBandwidthSupported; - DML_LOG_VERBOSE("DML::%s: PrefetchBandwidthSupported=%0d\n", __func__, mode_lib->ms.support.PrefetchBandwidthSupported); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]) { - mode_lib->ms.support.PrefetchSupported = false; - DML_LOG_VERBOSE("DML::%s: k=%d, NotEnoughUrgentLatencyHidingPre=%d\n", __func__, k, mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); - } - } - -#ifdef DML_GLOBAL_PREFETCH_CHECK - if (mode_lib->ms.support.PrefetchSupported && mode_lib->ms.num_active_planes > 1 && s->recalc_prefetch_done == 0) { - CheckGlobalPrefetchAdmissibility_params->num_active_planes = mode_lib->ms.num_active_planes; - CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; - CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->ms.SwathHeightY; - CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->ms.SwathHeightC; - CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->ms.CompressedBufferSizeInkByte; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->ms.DETBufferSizeY; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->ms.DETBufferSizeC; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; - CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; - CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = s->Tpre_rounded; - CheckGlobalPrefetchAdmissibility_params->Tpre_oto = s->Tpre_oto; - CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->ms.support.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; - CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->ms.dst_y_prefetch; - if (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps < 10 * 1024) - CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = 10 * 1024; - - CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps / (double) mode_lib->soc.return_bus_width_bytes) / - ((double)mode_lib->soc.qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0); - - // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible - CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->recalc_prefetch_schedule; - CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; - mode_lib->ms.support.PrefetchSupported = CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); - s->recalc_prefetch_done = 1; - s->recalc_prefetch_schedule = 1; - } -#endif - } // prefetch schedule ok, do urg bw and flip schedule - } while (s->recalc_prefetch_schedule); - - // Flip Schedule - // Both prefetch schedule and BW okay - if (mode_lib->ms.support.PrefetchSupported == true) { - mode_lib->ms.BandwidthAvailableForImmediateFlip = - get_bandwidth_available_for_immediate_flip( - dml2_core_internal_soc_state_sys_active, - mode_lib->ms.support.urg_bandwidth_required_qual, // no flip - mode_lib->ms.support.urg_bandwidth_available); - - mode_lib->ms.TotImmediateFlipBytes = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->plane_descriptors[k].immediate_flip) { - s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes( - s->HostVMInefficiencyFactor, - mode_lib->ms.vm_bytes[k], - mode_lib->ms.DPTEBytesPerRow[k], - mode_lib->ms.meta_row_bytes[k]); - } else { - s->per_pipe_flip_bytes[k] = 0; - } - mode_lib->ms.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->ms.NoOfDPP[k]; - - } - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - CalculateFlipSchedule( - &mode_lib->scratch, - display_cfg->plane_descriptors[k].immediate_flip, - 1, // use_lb_flip_bw - s->HostVMInefficiencyFactor, - s->Tvm_trips_flip[k], - s->Tr0_trips_flip[k], - s->Tvm_trips_flip_rounded[k], - s->Tr0_trips_flip_rounded[k], - display_cfg->gpuvm_enable, - mode_lib->ms.vm_bytes[k], - mode_lib->ms.DPTEBytesPerRow[k], - mode_lib->ms.BandwidthAvailableForImmediateFlip, - mode_lib->ms.TotImmediateFlipBytes, - display_cfg->plane_descriptors[k].pixel_format, - (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->ms.Tno_bw_flip[k], - mode_lib->ms.dpte_row_height[k], - mode_lib->ms.dpte_row_height_chroma[k], - mode_lib->ms.use_one_row_for_frame_flip[k], - mode_lib->ip.max_flip_time_us, - mode_lib->ip.max_flip_time_lines, - s->per_pipe_flip_bytes[k], - mode_lib->ms.meta_row_bytes[k], - s->meta_row_height_luma[k], - s->meta_row_height_chroma[k], - mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, - - /* Output */ - &mode_lib->ms.dst_y_per_vm_flip[k], - &mode_lib->ms.dst_y_per_row_flip[k], - &mode_lib->ms.final_flip_bw[k], - &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); - } - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 1; - calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - calculate_immediate_flip_bandwidth_support( - &s->dummy_single[0], // double* frac_urg_bandwidth_flip - &mode_lib->ms.support.ImmediateFlipSupport, - - dml2_core_internal_soc_state_sys_active, - mode_lib->ms.support.urg_bandwidth_required_flip, - mode_lib->ms.support.non_urg_bandwidth_required_flip, - mode_lib->ms.support.urg_bandwidth_available); - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].immediate_flip == true && mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false) - mode_lib->ms.support.ImmediateFlipSupport = false; - } - - } else { // if prefetch not support, assume iflip is not supported too - mode_lib->ms.support.ImmediateFlipSupport = false; - } - - s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; - s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; - s->mSOCParameters.ExtraLatency_sr = mode_lib->ms.ExtraLatency_sr; - s->mSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; - s->mSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; - s->mSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; - s->mSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; - s->mSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; - s->mSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; - s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; - s->mSOCParameters.USRRetrainingLatency = 0; - s->mSOCParameters.SMNLatency = 0; - s->mSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx); - s->mSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->ms.uclk_freq_mhz, mode_lib->ms.FabricClock, mode_lib->ms.state_idx); - s->mSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->ms.FabricClock; - s->mSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; - - CalculateWatermarks_params->display_cfg = display_cfg; - CalculateWatermarks_params->USRRetrainingRequired = false; - CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; - CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; - CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; - CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLK; - CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; - CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; - CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; - CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; - CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; - CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; - CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; - CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; - CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightY; - CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightC; - CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthY; - CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthC; - CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPP; - CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; - CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; - CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabled; - CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByte; - CalculateWatermarks_params->meta_row_height_l = s->meta_row_height_luma; - CalculateWatermarks_params->meta_row_height_c = s->meta_row_height_chroma; - - // Output - CalculateWatermarks_params->Watermark = &mode_lib->ms.support.watermarks; // Watermarks *Watermark - CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->ms.support.DRAMClockChangeSupport; - CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->ms.support.global_dram_clock_change_supported; - CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // double *MaxActiveDRAMClockChangeLatencySupported[] - CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->ms.SubViewportLinesNeededInMALL; // unsigned int SubViewportLinesNeededInMALL[] - CalculateWatermarks_params->FCLKChangeSupport = mode_lib->ms.support.FCLKChangeSupport; - CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->ms.support.global_fclk_change_supported; - CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // double *MaxActiveFCLKChangeLatencySupported - CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport; - CalculateWatermarks_params->g6_temp_read_support = &mode_lib->ms.support.g6_temp_read_support; - CalculateWatermarks_params->VActiveLatencyHidingMargin = mode_lib->ms.VActiveLatencyHidingMargin; - CalculateWatermarks_params->VActiveLatencyHidingUs = mode_lib->ms.VActiveLatencyHidingUs; - - CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); - - calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->ms.support.watermarks, s->dummy_integer_array[0]); - DML_LOG_VERBOSE("DML::%s: Done prefetch calculation\n", __func__); - -} - - -static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params) -{ - struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; - const struct dml2_display_cfg *display_cfg = in_out_params->in_display_cfg; - const struct dml2_mcg_min_clock_table *min_clk_table = in_out_params->min_clk_table; - - double outstanding_latency_us = 0; - - struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; - struct dml2_core_calcs_CalculateVMRowAndSwath_params *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; - struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; - struct dml2_core_calcs_calculate_mcache_setting_params *calculate_mcache_setting_params = &mode_lib->scratch.calculate_mcache_setting_params; - struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *calculate_bytes_to_fetch_required_to_hide_latency_params = &mode_lib->scratch.calculate_bytes_to_fetch_required_to_hide_latency_params; - unsigned int k, m, n; - - memset(&mode_lib->scratch, 0, sizeof(struct dml2_core_internal_scratch)); - memset(&mode_lib->ms, 0, sizeof(struct dml2_core_internal_mode_support)); - - mode_lib->ms.num_active_planes = display_cfg->num_planes; - get_stream_output_bpp(s->OutputBpp, display_cfg); - - mode_lib->ms.state_idx = in_out_params->min_clk_index; - mode_lib->ms.SOCCLK = ((double)mode_lib->soc.clk_table.socclk.clk_values_khz[0] / 1000); - mode_lib->ms.DCFCLK = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_dcfclk_khz / 1000); - mode_lib->ms.FabricClock = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz / 1000); - mode_lib->ms.MaxDCFCLK = (double)min_clk_table->max_clocks_khz.dcfclk / 1000; - mode_lib->ms.MaxFabricClock = (double)min_clk_table->max_clocks_khz.fclk / 1000; - mode_lib->ms.max_dispclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dispclk / 1000; - mode_lib->ms.max_dscclk_freq_mhz = (double)min_clk_table->max_clocks_khz.dscclk / 1000; - mode_lib->ms.max_dppclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dppclk / 1000; - mode_lib->ms.uclk_freq_mhz = dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config); - mode_lib->ms.dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps / 1000); - mode_lib->ms.max_dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[min_clk_table->dram_bw_table.num_entries - 1].pre_derate_dram_bw_kbps / 1000); - mode_lib->ms.qos_param_index = get_qos_param_index((unsigned int) (mode_lib->ms.uclk_freq_mhz * 1000.0), mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params); - mode_lib->ms.active_min_uclk_dpm_index = get_active_min_uclk_dpm_index((unsigned int) (mode_lib->ms.uclk_freq_mhz * 1000.0), &mode_lib->soc.clk_table); - -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); - DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); - DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); - DML_LOG_VERBOSE("DML::%s: qos_param_index = %0d\n", __func__, mode_lib->ms.qos_param_index); - DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); - DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->ms.dram_bw_mbps); - DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); - DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); - DML_LOG_VERBOSE("DML::%s: MaxDCFCLK = %f\n", __func__, mode_lib->ms.MaxDCFCLK); - DML_LOG_VERBOSE("DML::%s: max_dispclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dispclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: max_dscclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dscclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: max_dppclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dppclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: MaxFabricClock = %f\n", __func__, mode_lib->ms.MaxFabricClock); - DML_LOG_VERBOSE("DML::%s: ip.compressed_buffer_segment_size_in_kbytes = %u\n", __func__, mode_lib->ip.compressed_buffer_segment_size_in_kbytes); - DML_LOG_VERBOSE("DML::%s: ip.dcn_mrq_present = %u\n", __func__, mode_lib->ip.dcn_mrq_present); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); -#endif - - CalculateMaxDETAndMinCompressedBufferSize( - mode_lib->ip.config_return_buffer_size_in_kbytes, - mode_lib->ip.config_return_buffer_segment_size_in_kbytes, - mode_lib->ip.rob_buffer_size_kbytes, - mode_lib->ip.max_num_dpp, - display_cfg->overrides.hw.force_nom_det_size_kbytes.enable, - display_cfg->overrides.hw.force_nom_det_size_kbytes.value, - mode_lib->ip.dcn_mrq_present, - - /* Output */ - &mode_lib->ms.MaxTotalDETInKByte, - &mode_lib->ms.NomDETInKByte, - &mode_lib->ms.MinCompressedBufferSizeInKByte); - - PixelClockAdjustmentForProgressiveToInterlaceUnit(display_cfg, mode_lib->ip.ptoi_supported, s->PixelClockBackEnd); - - /*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/ - - /*Scale Ratio, taps Support Check*/ - mode_lib->ms.support.ScaleRatioAndTapsSupport = true; - // Many core tests are still setting scaling parameters "incorrectly" - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].composition.scaler_info.enabled == false - && (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio != 1.0 - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps != 1.0 - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio != 1.0 - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps != 1.0)) { - mode_lib->ms.support.ScaleRatioAndTapsSupport = false; - } else if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps < 1.0 || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps > 8.0 - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps < 1.0 || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps > 8.0 - || (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps > 1.0 && (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps % 2) == 1) - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio > mode_lib->ip.max_hscl_ratio - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio > mode_lib->ip.max_vscl_ratio - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps - || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps - || (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) - && (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps < 1 || display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps > 8 || - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps < 1 || display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps > 8 || - (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps > 1 && display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps % 2 == 1) || - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio > mode_lib->ip.max_hscl_ratio || - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio > mode_lib->ip.max_vscl_ratio || - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps || - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps))) { - mode_lib->ms.support.ScaleRatioAndTapsSupport = false; - } - } - - /*Source Format, Pixel Format and Scan Support Check*/ - mode_lib->ms.support.SourceFormatPixelAndScanSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear && dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { - mode_lib->ms.support.SourceFormatPixelAndScanSupport = false; - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateBytePerPixelAndBlockSizes( - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->plane_descriptors[k].surface.tiling, - display_cfg->plane_descriptors[k].surface.plane0.pitch, - display_cfg->plane_descriptors[k].surface.plane1.pitch, - - /* Output */ - &mode_lib->ms.BytePerPixelY[k], - &mode_lib->ms.BytePerPixelC[k], - &mode_lib->ms.BytePerPixelInDETY[k], - &mode_lib->ms.BytePerPixelInDETC[k], - &mode_lib->ms.Read256BlockHeightY[k], - &mode_lib->ms.Read256BlockHeightC[k], - &mode_lib->ms.Read256BlockWidthY[k], - &mode_lib->ms.Read256BlockWidthC[k], - &mode_lib->ms.MacroTileHeightY[k], - &mode_lib->ms.MacroTileHeightC[k], - &mode_lib->ms.MacroTileWidthY[k], - &mode_lib->ms.MacroTileWidthC[k], - &mode_lib->ms.surf_linear128_l[k], - &mode_lib->ms.surf_linear128_c[k]); - } - - /*Bandwidth Support Check*/ - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { - mode_lib->ms.SwathWidthYSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; - mode_lib->ms.SwathWidthCSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; - } else { - mode_lib->ms.SwathWidthYSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - mode_lib->ms.SwathWidthCSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - mode_lib->ms.vactive_sw_bw_l[k] = mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - mode_lib->ms.vactive_sw_bw_c[k] = mode_lib->ms.SwathWidthCSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - - mode_lib->ms.cursor_bw[k] = display_cfg->plane_descriptors[k].cursor.num_cursors * display_cfg->plane_descriptors[k].cursor.cursor_width * - display_cfg->plane_descriptors[k].cursor.cursor_bpp / 8.0 / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthLuma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthChroma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * math_ceil2(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio / 2.0); - DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_c[k]); -#endif - } - - // Writeback bandwidth - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format == dml2_444_64) { - mode_lib->ms.WriteBandwidth[k][0] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height - * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width - / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height - * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total - / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * 8.0; - } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - mode_lib->ms.WriteBandwidth[k][0] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height - * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width - / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height - * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total - / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * 4.0; - } else { - mode_lib->ms.WriteBandwidth[k][0] = 0.0; - } - } - - /*Writeback Latency support check*/ - mode_lib->ms.support.WritebackLatencySupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0 && - (mode_lib->ms.WriteBandwidth[k][0] > mode_lib->ip.writeback_interface_buffer_size_kbytes * 1024 / ((double)mode_lib->soc.qos_parameters.writeback.base_latency_us))) { - mode_lib->ms.support.WritebackLatencySupport = false; - } - } - - - /* Writeback Scale Ratio and Taps Support Check */ - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = true; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio > mode_lib->ip.writeback_max_hscl_ratio - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio > mode_lib->ip.writeback_max_vscl_ratio - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio < mode_lib->ip.writeback_min_hscl_ratio - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio < mode_lib->ip.writeback_min_vscl_ratio - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps > (unsigned int) mode_lib->ip.writeback_max_hscl_taps - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps > (unsigned int) mode_lib->ip.writeback_max_vscl_taps - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio > (unsigned int)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps - || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio > (unsigned int)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps - || (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps > 2.0 && ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps % 2) == 1))) { - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; - } - if (2.0 * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height * (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps - 1) * 57 > mode_lib->ip.writeback_line_buffer_buffer_size) { - mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; - } - } - } - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - CalculateSinglePipeDPPCLKAndSCLThroughput( - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->ip.max_dchub_pscl_bw_pix_per_clk, - mode_lib->ip.max_pscl_lb_bw_pix_per_clk, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps, - /* Output */ - &mode_lib->ms.PSCL_FACTOR[k], - &mode_lib->ms.PSCL_FACTOR_CHROMA[k], - &mode_lib->ms.MinDPPCLKUsingSingleDPP[k]); - } - - // Max Viewport Size support - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { - s->MaximumSwathWidthSupportLuma = 15360; - } else if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelC[k] > 0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { // horz video - s->MaximumSwathWidthSupportLuma = 7680 + 16; - } else if (dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelC[k] > 0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { // vert video - s->MaximumSwathWidthSupportLuma = 4320 + 16; - } else if (display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { // rgbe + alpha - s->MaximumSwathWidthSupportLuma = 5120 + 16; - } else if (dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelY[k] == 8 && display_cfg->plane_descriptors[k].surface.dcc.enable == true) { // vert 64bpp - s->MaximumSwathWidthSupportLuma = 3072 + 16; - } else { - s->MaximumSwathWidthSupportLuma = 6144 + 16; - } - - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { - s->MaximumSwathWidthSupportChroma = (unsigned int)(s->MaximumSwathWidthSupportLuma / 2.0); - } else { - s->MaximumSwathWidthSupportChroma = s->MaximumSwathWidthSupportLuma; - } - - unsigned lb_buffer_size_bits_luma = mode_lib->ip.line_buffer_size_bits; - unsigned lb_buffer_size_bits_chroma = mode_lib->ip.line_buffer_size_bits; - -/* -#if defined(DV_BUILD) - // Assume a memory config setting of 3 in 420 mode or get a new ip parameter that reflects the programming. - if (mode_lib->ms.BytePerPixelC[k] != 0.0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { - lb_buffer_size_bits_luma = 34620 * 57; - lb_buffer_size_bits_chroma = 13560 * 57; - } -#endif -*/ - mode_lib->ms.MaximumSwathWidthInLineBufferLuma = lb_buffer_size_bits_luma * math_max2(display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, 1.0) / 57 / - (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps + math_max2(math_ceil2(display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, 1.0) - 2, 0.0)); - if (mode_lib->ms.BytePerPixelC[k] == 0.0) { - mode_lib->ms.MaximumSwathWidthInLineBufferChroma = 0; - } else { - mode_lib->ms.MaximumSwathWidthInLineBufferChroma = lb_buffer_size_bits_chroma * math_max2(display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, 1.0) / 57 / - (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps + math_max2(math_ceil2(display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, 1.0) - 2, 0.0)); - } - - mode_lib->ms.MaximumSwathWidthLuma[k] = math_min2(s->MaximumSwathWidthSupportLuma, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); - mode_lib->ms.MaximumSwathWidthChroma[k] = math_min2(s->MaximumSwathWidthSupportChroma, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); - - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportLuma=%u\n", __func__, k, s->MaximumSwathWidthSupportLuma); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); - - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportChroma=%u\n", __func__, k, s->MaximumSwathWidthSupportChroma); - DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); - } - - /* Cursor Support Check */ - mode_lib->ms.support.CursorSupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { - if (display_cfg->plane_descriptors[k].cursor.cursor_bpp == 64 && mode_lib->ip.cursor_64bpp_support == false) - mode_lib->ms.support.CursorSupport = false; - } - } - - /* Valid Pitch Check */ - mode_lib->ms.support.PitchSupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - - // data pitch - unsigned int alignment_l = mode_lib->ms.MacroTileWidthY[k]; - - if (mode_lib->ms.surf_linear128_l[k]) - alignment_l = alignment_l / 2; - - mode_lib->ms.support.AlignedYPitch[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.plane0.pitch, display_cfg->plane_descriptors[k].surface.plane0.width), alignment_l); - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { - unsigned int alignment_c = mode_lib->ms.MacroTileWidthC[k]; - - if (mode_lib->ms.surf_linear128_c[k]) - alignment_c = alignment_c / 2; - mode_lib->ms.support.AlignedCPitch[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.plane1.pitch, display_cfg->plane_descriptors[k].surface.plane1.width), alignment_c); - } else { - mode_lib->ms.support.AlignedCPitch[k] = display_cfg->plane_descriptors[k].surface.plane1.pitch; - } - - if (mode_lib->ms.support.AlignedYPitch[k] > display_cfg->plane_descriptors[k].surface.plane0.pitch || - mode_lib->ms.support.AlignedCPitch[k] > display_cfg->plane_descriptors[k].surface.plane1.pitch) { - mode_lib->ms.support.PitchSupport = false; -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: k=%u AlignedYPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedYPitch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u PitchY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.pitch); - DML_LOG_VERBOSE("DML::%s: k=%u AlignedCPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedCPitch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u PitchC = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane1.pitch); - DML_LOG_VERBOSE("DML::%s: k=%u PitchSupport = %d\n", __func__, k, mode_lib->ms.support.PitchSupport); -#endif - } - - // meta pitch - if (mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable) { - mode_lib->ms.support.AlignedDCCMetaPitchY[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch, - display_cfg->plane_descriptors[k].surface.plane0.width), 64.0 * mode_lib->ms.Read256BlockWidthY[k]); - - if (mode_lib->ms.support.AlignedDCCMetaPitchY[k] > display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch) - mode_lib->ms.support.PitchSupport = false; - - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { - mode_lib->ms.support.AlignedDCCMetaPitchC[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch, - display_cfg->plane_descriptors[k].surface.plane1.width), 64.0 * mode_lib->ms.Read256BlockWidthC[k]); - - if (mode_lib->ms.support.AlignedDCCMetaPitchC[k] > display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch) - mode_lib->ms.support.PitchSupport = false; - } - } else { - mode_lib->ms.support.AlignedDCCMetaPitchY[k] = 0; - mode_lib->ms.support.AlignedDCCMetaPitchC[k] = 0; - } - } - - mode_lib->ms.support.ViewportExceedsSurface = false; - if (!display_cfg->overrides.hw.surface_viewport_size_check_disable) { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - if (display_cfg->plane_descriptors[k].composition.viewport.plane0.width > display_cfg->plane_descriptors[k].surface.plane0.width || - display_cfg->plane_descriptors[k].composition.viewport.plane0.height > display_cfg->plane_descriptors[k].surface.plane0.height) { - mode_lib->ms.support.ViewportExceedsSurface = true; -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); - DML_LOG_VERBOSE("DML::%s: k=%u SurfaceWidthY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.width); - DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); - DML_LOG_VERBOSE("DML::%s: k=%u SurfaceHeightY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.height); - DML_LOG_VERBOSE("DML::%s: k=%u ViewportExceedsSurface = %d\n", __func__, k, mode_lib->ms.support.ViewportExceedsSurface); -#endif - } - if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { - if (display_cfg->plane_descriptors[k].composition.viewport.plane1.width > display_cfg->plane_descriptors[k].surface.plane1.width || - display_cfg->plane_descriptors[k].composition.viewport.plane1.height > display_cfg->plane_descriptors[k].surface.plane1.height) { - mode_lib->ms.support.ViewportExceedsSurface = true; - } - } - } - } - - CalculateSwathAndDETConfiguration_params->display_cfg = display_cfg; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ip.config_return_buffer_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; - CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = 1; - CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ip.config_return_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByte = mode_lib->ip.compressed_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.vactive_sw_bw_l; - CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.vactive_sw_bw_c; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; - CalculateSwathAndDETConfiguration_params->surf_linear128_l = mode_lib->ms.surf_linear128_l; - CalculateSwathAndDETConfiguration_params->surf_linear128_c = mode_lib->ms.surf_linear128_c; - CalculateSwathAndDETConfiguration_params->ODMMode = s->dummy_odm_mode; - CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; - CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; - CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; - CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = s->dummy_integer_array[2]; - CalculateSwathAndDETConfiguration_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - // output - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = s->dummy_integer_array[0]; - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = s->dummy_integer_array[1]; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_integer_array[3]; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_integer_array[4]; - CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_integer_array[5]; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_integer_array[6]; - CalculateSwathAndDETConfiguration_params->SwathHeightY = s->dummy_integer_array[7]; - CalculateSwathAndDETConfiguration_params->SwathHeightC = s->dummy_integer_array[8]; - CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = s->dummy_integer_array[26]; - CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = s->dummy_integer_array[27]; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = s->dummy_integer_array[9]; - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = s->dummy_integer_array[10]; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = s->dummy_integer_array[11]; - CalculateSwathAndDETConfiguration_params->full_swath_bytes_l = s->full_swath_bytes_l; - CalculateSwathAndDETConfiguration_params->full_swath_bytes_c = s->full_swath_bytes_c; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &s->dummy_boolean[0]; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[1]; - CalculateSwathAndDETConfiguration_params->hw_debug5 = &s->dummy_boolean[2]; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &s->dummy_integer[0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = mode_lib->ms.SingleDPPViewportSizeSupportPerSurface; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[1]; - - // This calls is just to find out if there is enough DET space to support full vp in 1 pipe. - CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); - - mode_lib->ms.TotalNumberOfActiveDPP = 0; - mode_lib->ms.TotalNumberOfActiveOPP = 0; - mode_lib->ms.support.TotalAvailablePipesSupport = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - /*Number Of DSC Slices*/ - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable_if_necessary) { - - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices != 0) - mode_lib->ms.support.NumberOfDSCSlices[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices; - else { - if (s->PixelClockBackEnd[k] > 4800) { - mode_lib->ms.support.NumberOfDSCSlices[k] = (unsigned int)(math_ceil2(s->PixelClockBackEnd[k] / 600, 4)); - } else if (s->PixelClockBackEnd[k] > 2400) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 8; - } else if (s->PixelClockBackEnd[k] > 1200) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 4; - } else if (s->PixelClockBackEnd[k] > 340) { - mode_lib->ms.support.NumberOfDSCSlices[k] = 2; - } else { - mode_lib->ms.support.NumberOfDSCSlices[k] = 1; - } - } - } else { - mode_lib->ms.support.NumberOfDSCSlices[k] = 0; - } - - CalculateODMMode( - mode_lib->ip.maximum_pixels_per_line_per_dsc_unit, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode, - mode_lib->ms.max_dispclk_freq_mhz, - false, // DSCEnable - mode_lib->ms.TotalNumberOfActiveDPP, - mode_lib->ms.TotalNumberOfActiveOPP, - mode_lib->ip.max_num_dpp, - mode_lib->ip.max_num_opp, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - mode_lib->ms.support.NumberOfDSCSlices[k], - - /* Output */ - &s->TotalAvailablePipesSupportNoDSC, - &s->NumberOfDPPNoDSC, - &s->ODMModeNoDSC, - &s->RequiredDISPCLKPerSurfaceNoDSC); - - CalculateODMMode( - mode_lib->ip.maximum_pixels_per_line_per_dsc_unit, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode, - mode_lib->ms.max_dispclk_freq_mhz, - true, // DSCEnable - mode_lib->ms.TotalNumberOfActiveDPP, - mode_lib->ms.TotalNumberOfActiveOPP, - mode_lib->ip.max_num_dpp, - mode_lib->ip.max_num_opp, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - mode_lib->ms.support.NumberOfDSCSlices[k], - - /* Output */ - &s->TotalAvailablePipesSupportDSC, - &s->NumberOfDPPDSC, - &s->ODMModeDSC, - &s->RequiredDISPCLKPerSurfaceDSC); - - CalculateOutputLink( - &mode_lib->scratch, - ((double)mode_lib->soc.clk_table.phyclk.clk_values_khz[0] / 1000), - ((double)mode_lib->soc.clk_table.phyclk_d18.clk_values_khz[0] / 1000), - ((double)mode_lib->soc.clk_table.phyclk_d32.clk_values_khz[0] / 1000), - mode_lib->soc.phy_downspread_percent, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - s->PixelClockBackEnd[k], - s->OutputBpp[k], - mode_lib->ip.maximum_dsc_bits_per_component, - mode_lib->ms.support.NumberOfDSCSlices[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_rate, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_layout, - s->ODMModeNoDSC, - s->ODMModeDSC, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate, - - /* Output */ - &mode_lib->ms.RequiresDSC[k], - &mode_lib->ms.RequiresFEC[k], - &mode_lib->ms.OutputBpp[k], - &mode_lib->ms.OutputType[k], - &mode_lib->ms.OutputRate[k], - &mode_lib->ms.RequiredSlots[k]); - - if (s->OutputBpp[k] == 0.0) { - s->OutputBpp[k] = mode_lib->ms.OutputBpp[k]; - } - - if (mode_lib->ms.RequiresDSC[k] == false) { - mode_lib->ms.ODMMode[k] = s->ODMModeNoDSC; - mode_lib->ms.RequiredDISPCLKPerSurface[k] = s->RequiredDISPCLKPerSurfaceNoDSC; - if (!s->TotalAvailablePipesSupportNoDSC) - mode_lib->ms.support.TotalAvailablePipesSupport = false; - mode_lib->ms.TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP + s->NumberOfDPPNoDSC; - } else { - mode_lib->ms.ODMMode[k] = s->ODMModeDSC; - mode_lib->ms.RequiredDISPCLKPerSurface[k] = s->RequiredDISPCLKPerSurfaceDSC; - if (!s->TotalAvailablePipesSupportDSC) - mode_lib->ms.support.TotalAvailablePipesSupport = false; - mode_lib->ms.TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP + s->NumberOfDPPDSC; - } -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: k=%d RequiresDSC = %d\n", __func__, k, mode_lib->ms.RequiresDSC[k]); - DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); -#endif - - // ensure the number dsc slices is integer multiple based on ODM mode - mode_lib->ms.support.DSCSlicesODMModeSupported = true; - if (mode_lib->ms.RequiresDSC[k]) { - // fail a ms check if the override num_slices doesn't align with odm mode setting - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices != 0) { - if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) - mode_lib->ms.support.DSCSlicesODMModeSupported = ((mode_lib->ms.support.NumberOfDSCSlices[k] % 2) == 0); - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) - mode_lib->ms.support.DSCSlicesODMModeSupported = (mode_lib->ms.support.NumberOfDSCSlices[k] == 12); - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) - mode_lib->ms.support.DSCSlicesODMModeSupported = ((mode_lib->ms.support.NumberOfDSCSlices[k] % 4) == 0); -#if defined(__DML_VBA_DEBUG__) - if (!mode_lib->ms.support.DSCSlicesODMModeSupported) { - DML_LOG_VERBOSE("DML::%s: k=%d Invalid dsc num_slices and ODM mode setting\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: k=%d num_slices = %d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices); - DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); - } -#endif - } else { - // safe guard to ensure the dml derived dsc slices and odm setting are compatible - if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) - mode_lib->ms.support.NumberOfDSCSlices[k] = 2 * (unsigned int)math_ceil2(mode_lib->ms.support.NumberOfDSCSlices[k] / 2.0, 1.0); - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) - mode_lib->ms.support.NumberOfDSCSlices[k] = 12; - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) - mode_lib->ms.support.NumberOfDSCSlices[k] = 4 * (unsigned int)math_ceil2(mode_lib->ms.support.NumberOfDSCSlices[k] / 4.0, 1.0); - } - - } else { - mode_lib->ms.support.NumberOfDSCSlices[k] = 0; - } - } - - mode_lib->ms.support.incorrect_imall_usage = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) - mode_lib->ms.support.incorrect_imall_usage = 1; - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.MPCCombine[k] = false; - mode_lib->ms.NoOfDPP[k] = 1; - mode_lib->ms.NoOfOPP[k] = 1; - - if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) { - mode_lib->ms.MPCCombine[k] = false; - mode_lib->ms.NoOfDPP[k] = 4; - mode_lib->ms.NoOfOPP[k] = 4; - } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) { - mode_lib->ms.MPCCombine[k] = false; - mode_lib->ms.NoOfDPP[k] = 3; - mode_lib->ms.NoOfOPP[k] = 3; - } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) { - mode_lib->ms.MPCCombine[k] = false; - mode_lib->ms.NoOfDPP[k] = 2; - mode_lib->ms.NoOfOPP[k] = 2; - } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 2) { - mode_lib->ms.MPCCombine[k] = true; - mode_lib->ms.NoOfDPP[k] = 2; - } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 1) { - mode_lib->ms.MPCCombine[k] = false; - mode_lib->ms.NoOfDPP[k] = 1; - if (!mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { - DML_LOG_VERBOSE("WARNING: DML::%s: MPCC is override to disable but viewport is too large to be supported with single pipe!\n", __func__); - } - } else { - if ((mode_lib->ms.MinDPPCLKUsingSingleDPP[k] > mode_lib->ms.max_dppclk_freq_mhz) || !mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { - mode_lib->ms.MPCCombine[k] = true; - mode_lib->ms.NoOfDPP[k] = 2; - } - } -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: k=%d, NoOfDPP = %d\n", __func__, k, mode_lib->ms.NoOfDPP[k]); -#endif - } - - mode_lib->ms.TotalNumberOfActiveDPP = 0; - mode_lib->ms.TotalNumberOfActiveOPP = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.TotalNumberOfActiveDPP += mode_lib->ms.NoOfDPP[k]; - mode_lib->ms.TotalNumberOfActiveOPP += mode_lib->ms.NoOfOPP[k]; - } - if (mode_lib->ms.TotalNumberOfActiveDPP > (unsigned int)mode_lib->ip.max_num_dpp) - mode_lib->ms.support.TotalAvailablePipesSupport = false; - if (mode_lib->ms.TotalNumberOfActiveOPP > (unsigned int)mode_lib->ip.max_num_opp) - mode_lib->ms.support.TotalAvailablePipesSupport = false; - - - mode_lib->ms.TotalNumberOfSingleDPPSurfaces = 0; - for (k = 0; k < (unsigned int)mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.NoOfDPP[k] == 1) - mode_lib->ms.TotalNumberOfSingleDPPSurfaces = mode_lib->ms.TotalNumberOfSingleDPPSurfaces + 1; - } - - //DISPCLK/DPPCLK - mode_lib->ms.WritebackRequiredDISPCLK = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - mode_lib->ms.WritebackRequiredDISPCLK = math_max2(mode_lib->ms.WritebackRequiredDISPCLK, - CalculateWriteBackDISPCLK(display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_width, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, - mode_lib->ip.writeback_line_buffer_buffer_size)); - } - } - - mode_lib->ms.RequiredDISPCLK = mode_lib->ms.WritebackRequiredDISPCLK; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.RequiredDISPCLK = math_max2(mode_lib->ms.RequiredDISPCLK, mode_lib->ms.RequiredDISPCLKPerSurface[k]); - } - - mode_lib->ms.GlobalDPPCLK = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.RequiredDPPCLK[k] = mode_lib->ms.MinDPPCLKUsingSingleDPP[k] / mode_lib->ms.NoOfDPP[k]; - mode_lib->ms.GlobalDPPCLK = math_max2(mode_lib->ms.GlobalDPPCLK, mode_lib->ms.RequiredDPPCLK[k]); - } - - mode_lib->ms.support.DISPCLK_DPPCLK_Support = !((mode_lib->ms.RequiredDISPCLK > mode_lib->ms.max_dispclk_freq_mhz) || (mode_lib->ms.GlobalDPPCLK > mode_lib->ms.max_dppclk_freq_mhz)); - - /* Total Available OTG, Writeback, HDMIFRL, DP Support Check */ - s->TotalNumberOfActiveOTG = 0; - s->TotalNumberOfActiveHDMIFRL = 0; - s->TotalNumberOfActiveDP2p0 = 0; - s->TotalNumberOfActiveDP2p0Outputs = 0; - s->TotalNumberOfActiveWriteback = 0; - memset(s->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - if (!s->stream_visited[display_cfg->plane_descriptors[k].stream_index]) { - s->stream_visited[display_cfg->plane_descriptors[k].stream_index] = 1; - - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) - s->TotalNumberOfActiveWriteback = s->TotalNumberOfActiveWriteback + 1; - - s->TotalNumberOfActiveOTG = s->TotalNumberOfActiveOTG + 1; - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) - s->TotalNumberOfActiveHDMIFRL = s->TotalNumberOfActiveHDMIFRL + 1; - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0) { - s->TotalNumberOfActiveDP2p0 = s->TotalNumberOfActiveDP2p0 + 1; - // FIXME_STAGE2: SW not using backend related stuff, need mapping for mst setup - //if (display_cfg->output.OutputMultistreamId[k] == k || display_cfg->output.OutputMultistreamEn[k] == false) { - s->TotalNumberOfActiveDP2p0Outputs = s->TotalNumberOfActiveDP2p0Outputs + 1; - //} - } - } - } - } - - /* Writeback Mode Support Check */ - mode_lib->ms.support.EnoughWritebackUnits = 1; - if (s->TotalNumberOfActiveWriteback > (unsigned int)mode_lib->ip.max_num_wb) { - mode_lib->ms.support.EnoughWritebackUnits = false; - } - mode_lib->ms.support.NumberOfOTGSupport = (s->TotalNumberOfActiveOTG <= (unsigned int)mode_lib->ip.max_num_otg); - mode_lib->ms.support.NumberOfHDMIFRLSupport = (s->TotalNumberOfActiveHDMIFRL <= (unsigned int)mode_lib->ip.max_num_hdmi_frl_outputs); - mode_lib->ms.support.NumberOfDP2p0Support = (s->TotalNumberOfActiveDP2p0 <= (unsigned int)mode_lib->ip.max_num_dp2p0_streams && s->TotalNumberOfActiveDP2p0Outputs <= (unsigned int)mode_lib->ip.max_num_dp2p0_outputs); - - - mode_lib->ms.support.ExceededMultistreamSlots = false; - mode_lib->ms.support.LinkCapacitySupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_disabled == false && - (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) && mode_lib->ms.OutputBpp[k] == 0) { - mode_lib->ms.support.LinkCapacitySupport = false; - } - } - - mode_lib->ms.support.P2IWith420 = false; - mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = false; - mode_lib->ms.support.DSC422NativeNotSupported = false; - mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = false; - mode_lib->ms.support.LinkRateForMultistreamNotIndicated = false; - mode_lib->ms.support.BPPForMultistreamNotIndicated = false; - mode_lib->ms.support.MultistreamWithHDMIOreDP = false; - mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = false; - mode_lib->ms.support.NotEnoughLanesForMSO = false; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_420 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced == 1 && mode_lib->ip.ptoi_supported == true) - mode_lib->ms.support.P2IWith420 = true; - - if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable_if_necessary) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_n422 && !mode_lib->ip.dsc422_native_support) - mode_lib->ms.support.DSC422NativeNotSupported = true; - - if (((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr2 || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr3) && - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_edp) || - ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr10 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr13p5 || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr20) && - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp2p0)) - mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = true; - - // FIXME_STAGE2 - //if (display_cfg->output.OutputMultistreamEn[k] == 1) { - // if (display_cfg->output.OutputMultistreamId[k] == k && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_na) - // mode_lib->ms.support.LinkRateForMultistreamNotIndicated = true; - // if (display_cfg->output.OutputMultistreamId[k] == k && s->OutputBpp[k] == 0) - // mode_lib->ms.support.BPPForMultistreamNotIndicated = true; - // for (n = 0; n < mode_lib->ms.num_active_planes; ++n) { - // if (display_cfg->output.OutputMultistreamId[k] == n && s->OutputBpp[k] == 0) - // mode_lib->ms.support.BPPForMultistreamNotIndicated = true; - // } - //} - - if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl)) { - // FIXME_STAGE2 - //if (display_cfg->output.OutputMultistreamEn[k] == 1 && display_cfg->output.OutputMultistreamId[k] == k) - // mode_lib->ms.support.MultistreamWithHDMIOreDP = true; - //for (n = 0; n < mode_lib->ms.num_active_planes; ++n) { - // if (display_cfg->output.OutputMultistreamEn[k] == 1 && display_cfg->output.OutputMultistreamId[k] == n) - // mode_lib->ms.support.MultistreamWithHDMIOreDP = true; - //} - } - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp && (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_split_1to2 || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to2 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to4)) - mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = true; - - if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to2 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count < 2) || - (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to4 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count < 4)) - mode_lib->ms.support.NotEnoughLanesForMSO = true; - } - } - - mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl && - !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { - mode_lib->ms.RequiredDTBCLK[k] = RequiredDTBCLK( - mode_lib->ms.RequiresDSC[k], - s->PixelClockBackEnd[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - mode_lib->ms.OutputBpp[k], - mode_lib->ms.support.NumberOfDSCSlices[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_rate, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_layout); - - if (mode_lib->ms.RequiredDTBCLK[k] > ((double)min_clk_table->max_ss_clocks_khz.dtbclk / 1000)) { - mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = true; - } - } else { - /* Phantom DTBCLK can be calculated different from main because phantom has no DSC and thus - * will have a different output BPP. Ignore phantom DTBCLK requirement and only consider - * non-phantom DTBCLK requirements. In map_mode_to_soc_dpm we choose the highest DTBCLK - * required - by setting phantom dtbclk to 0 we ignore it. - */ - mode_lib->ms.RequiredDTBCLK[k] = 0; - } - } - - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = false; - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_420) { - s->DSCFormatFactor = 2; - } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_444) { - s->DSCFormatFactor = 1; - } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_n422 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { - s->DSCFormatFactor = 2; - } else { - s->DSCFormatFactor = 1; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); -#endif - if (mode_lib->ms.RequiresDSC[k] == true) { - s->PixelClockBackEndFactor = 3.0; - - if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) - s->PixelClockBackEndFactor = 12.0; - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) - s->PixelClockBackEndFactor = 9.0; - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) - s->PixelClockBackEndFactor = 6.0; - - mode_lib->ms.required_dscclk_freq_mhz[k] = s->PixelClockBackEnd[k] / s->PixelClockBackEndFactor / (double)s->DSCFormatFactor; - if (mode_lib->ms.required_dscclk_freq_mhz[k] > mode_lib->ms.max_dscclk_freq_mhz) { - mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, s->PixelClockBackEnd[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, required_dscclk_freq_mhz = %f\n", __func__, k, mode_lib->ms.required_dscclk_freq_mhz[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); - DML_LOG_VERBOSE("DML::%s: k=%u, DSCCLKRequiredMoreThanSupported = %u\n", __func__, k, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); -#endif - } - } - } - - /* Check DSC Unit and Slices Support */ - mode_lib->ms.support.NotEnoughDSCSlices = false; - s->TotalDSCUnitsRequired = 0; - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = true; - memset(s->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.RequiresDSC[k] == true && !s->stream_visited[display_cfg->plane_descriptors[k].stream_index]) { - s->NumDSCUnitRequired = 1; - - if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) - s->NumDSCUnitRequired = 4; - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) - s->NumDSCUnitRequired = 3; - else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) - s->NumDSCUnitRequired = 2; - - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active > s->NumDSCUnitRequired * (unsigned int)mode_lib->ip.maximum_pixels_per_line_per_dsc_unit) - mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; - s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + s->NumDSCUnitRequired; - - if (mode_lib->ms.support.NumberOfDSCSlices[k] > 4 * s->NumDSCUnitRequired) - mode_lib->ms.support.NotEnoughDSCSlices = true; - } - s->stream_visited[display_cfg->plane_descriptors[k].stream_index] = 1; - } - - mode_lib->ms.support.NotEnoughDSCUnits = false; - if (s->TotalDSCUnitsRequired > (unsigned int)mode_lib->ip.num_dsc) { - mode_lib->ms.support.NotEnoughDSCUnits = true; - } - - /*DSC Delay per state*/ - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.DSCDelay[k] = DSCDelayRequirement(mode_lib->ms.RequiresDSC[k], - mode_lib->ms.ODMMode[k], - mode_lib->ip.maximum_dsc_bits_per_component, - s->OutputBpp[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, - mode_lib->ms.support.NumberOfDSCSlices[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - s->PixelClockBackEnd[k]); - } - - // Figure out the swath and DET configuration after the num dpp per plane is figured out - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; - CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.ODMMode; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.NoOfDPP; - - // output - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = s->dummy_integer_array[0]; - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = s->dummy_integer_array[1]; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub; - CalculateSwathAndDETConfiguration_params->SwathWidth = mode_lib->ms.SwathWidthY; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = mode_lib->ms.SwathWidthC; - CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->ms.SwathHeightY; - CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->ms.SwathHeightC; - CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = mode_lib->ms.support.request_size_bytes_luma; - CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = mode_lib->ms.support.request_size_bytes_chroma; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->ms.DETBufferSizeInKByte; // FIXME: This is per pipe but the pipes in plane will use that - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->ms.UnboundedRequestEnabled; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = s->dummy_integer_array[3]; - CalculateSwathAndDETConfiguration_params->hw_debug5 = s->dummy_boolean_array[1]; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->ms.CompressedBufferSizeInkByte; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = s->dummy_boolean_array[0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &mode_lib->ms.support.ViewportSizeSupport; - - CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); - - if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0) { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) - mode_lib->ms.SurfaceSizeInMALL[k] = 0; - mode_lib->ms.support.ExceededMALLSize = 0; - } else { - CalculateSurfaceSizeInMall( - display_cfg, - mode_lib->ms.num_active_planes, - mode_lib->soc.mall_allocated_for_dcn_mbytes, - - mode_lib->ms.BytePerPixelY, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.Read256BlockWidthY, - mode_lib->ms.Read256BlockWidthC, - mode_lib->ms.Read256BlockHeightY, - mode_lib->ms.Read256BlockHeightC, - mode_lib->ms.MacroTileWidthY, - mode_lib->ms.MacroTileWidthC, - mode_lib->ms.MacroTileHeightY, - mode_lib->ms.MacroTileHeightC, - - /* Output */ - mode_lib->ms.SurfaceSizeInMALL, - &mode_lib->ms.support.ExceededMALLSize); - } - - mode_lib->ms.TotalNumberOfDCCActiveDPP = 0; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->plane_descriptors[k].surface.dcc.enable == true) { - mode_lib->ms.TotalNumberOfDCCActiveDPP = mode_lib->ms.TotalNumberOfDCCActiveDPP + mode_lib->ms.NoOfDPP[k]; - } - } - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->SurfParameters[k].PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - s->SurfParameters[k].DPPPerSurface = mode_lib->ms.NoOfDPP[k]; - s->SurfParameters[k].RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; - s->SurfParameters[k].ViewportHeight = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - s->SurfParameters[k].ViewportHeightC = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; - s->SurfParameters[k].BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; - s->SurfParameters[k].BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; - s->SurfParameters[k].BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; - s->SurfParameters[k].BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; - s->SurfParameters[k].BlockWidthY = mode_lib->ms.MacroTileWidthY[k]; - s->SurfParameters[k].BlockHeightY = mode_lib->ms.MacroTileHeightY[k]; - s->SurfParameters[k].BlockWidthC = mode_lib->ms.MacroTileWidthC[k]; - s->SurfParameters[k].BlockHeightC = mode_lib->ms.MacroTileHeightC[k]; - s->SurfParameters[k].InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; - s->SurfParameters[k].HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; - s->SurfParameters[k].DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; - s->SurfParameters[k].SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; - s->SurfParameters[k].SurfaceTiling = display_cfg->plane_descriptors[k].surface.tiling; - s->SurfParameters[k].BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; - s->SurfParameters[k].BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; - s->SurfParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - s->SurfParameters[k].VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - s->SurfParameters[k].VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - s->SurfParameters[k].VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - s->SurfParameters[k].VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - s->SurfParameters[k].PitchY = display_cfg->plane_descriptors[k].surface.plane0.pitch; - s->SurfParameters[k].PitchC = display_cfg->plane_descriptors[k].surface.plane1.pitch; - s->SurfParameters[k].ViewportStationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; - s->SurfParameters[k].ViewportXStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; - s->SurfParameters[k].ViewportYStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; - s->SurfParameters[k].ViewportXStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - s->SurfParameters[k].ViewportYStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - s->SurfParameters[k].FORCE_ONE_ROW_FOR_FRAME = display_cfg->plane_descriptors[k].overrides.hw.force_one_row_for_frame; - s->SurfParameters[k].SwathHeightY = mode_lib->ms.SwathHeightY[k]; - s->SurfParameters[k].SwathHeightC = mode_lib->ms.SwathHeightC[k]; - - s->SurfParameters[k].DCCMetaPitchY = display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch; - s->SurfParameters[k].DCCMetaPitchC = display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch; - } - - CalculateVMRowAndSwath_params->display_cfg = display_cfg; - CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; - CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; - CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma; - CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->soc.mall_allocated_for_dcn_mbytes; - CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthY; - CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthC; - CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ip.dcc_meta_buffer_size_bytes; - CalculateVMRowAndSwath_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - // output - CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceeded; - CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[12]; - CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[13]; - CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; - CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; - CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[14]; // VBA_DELTA - CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[15]; // VBA_DELTA - CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[16]; - CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; - CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[17]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[18]; - CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[19]; - CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[20]; - CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[21]; - CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[22]; - CalculateVMRowAndSwath_params->vmpg_width_y = s->vmpg_width_y; - CalculateVMRowAndSwath_params->vmpg_height_y = s->vmpg_height_y; - CalculateVMRowAndSwath_params->vmpg_width_c = s->vmpg_width_c; - CalculateVMRowAndSwath_params->vmpg_height_c = s->vmpg_height_c; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[23]; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[24]; - CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY; - CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC; - CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; - CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; - CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY; - CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC; - CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; - CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow; - CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_l = s->dpte_row_bytes_per_row_l; - CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_c = s->dpte_row_bytes_per_row_c; - CalculateVMRowAndSwath_params->vm_bytes = mode_lib->ms.vm_bytes; - CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame; - CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip; - CalculateVMRowAndSwath_params->is_using_mall_for_ss = s->dummy_boolean_array[0]; - CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; - CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[25]; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceeded; - CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bw; - CalculateVMRowAndSwath_params->meta_row_bytes = mode_lib->ms.meta_row_bytes; - CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; - CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; - CalculateVMRowAndSwath_params->meta_req_width_luma = s->dummy_integer_array[26]; - CalculateVMRowAndSwath_params->meta_req_height_luma = s->dummy_integer_array[27]; - CalculateVMRowAndSwath_params->meta_row_width_luma = s->dummy_integer_array[28]; - CalculateVMRowAndSwath_params->meta_row_height_luma = s->meta_row_height_luma; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[29]; - CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[30]; - CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[31]; - CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[32]; - CalculateVMRowAndSwath_params->meta_row_height_chroma = s->meta_row_height_chroma; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[33]; - - CalculateVMRowAndSwath(&mode_lib->scratch, CalculateVMRowAndSwath_params); - - mode_lib->ms.support.PTEBufferSizeNotExceeded = true; - mode_lib->ms.support.DCCMetaBufferSizeNotExceeded = true; - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.PTEBufferSizeNotExceeded[k] == false) - mode_lib->ms.support.PTEBufferSizeNotExceeded = false; - - if (mode_lib->ms.DCCMetaBufferSizeNotExceeded[k] == false) - mode_lib->ms.support.DCCMetaBufferSizeNotExceeded = false; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.PTEBufferSizeNotExceeded[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.DCCMetaBufferSizeNotExceeded[k]); -#endif - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: PTEBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.PTEBufferSizeNotExceeded); - DML_LOG_VERBOSE("DML::%s: DCCMetaBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.DCCMetaBufferSizeNotExceeded); -#endif - - /* VActive bytes to fetch for UCLK P-State */ - calculate_bytes_to_fetch_required_to_hide_latency_params->display_cfg = display_cfg; - calculate_bytes_to_fetch_required_to_hide_latency_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - calculate_bytes_to_fetch_required_to_hide_latency_params->num_active_planes = mode_lib->ms.num_active_planes; - calculate_bytes_to_fetch_required_to_hide_latency_params->num_of_dpp = mode_lib->ms.NoOfDPP; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_l = s->meta_row_height_luma; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_c = s->meta_row_height_chroma; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_l = mode_lib->ms.dpte_row_height; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_c = mode_lib->ms.dpte_row_height_chroma; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_l = s->dpte_row_bytes_per_row_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_c = s->dpte_row_bytes_per_row_c; - calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_l = mode_lib->ms.BytePerPixelY; - calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_c = mode_lib->ms.BytePerPixelC; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_l = mode_lib->ms.SwathWidthY; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->ms.SwathWidthC; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->ms.SwathHeightY; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->ms.SwathHeightC; - calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; - - /* outputs */ - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; - - calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); - - /* Excess VActive bandwidth required to fill DET */ - calculate_excess_vactive_bandwidth_required( - display_cfg, - mode_lib->ms.num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, - /* outputs */ - mode_lib->ms.excess_vactive_fill_bw_l, - mode_lib->ms.excess_vactive_fill_bw_c); - - mode_lib->ms.UrgLatency = CalculateUrgentLatency( - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_pixel_vm_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_vm_us, - mode_lib->soc.do_urgent_latency_adjustment, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_fclk_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_mhz, - mode_lib->ms.FabricClock, - mode_lib->ms.uclk_freq_mhz, - mode_lib->soc.qos_parameters.qos_type, - mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].urgent_ramp_uclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_urgent_ramp_latency_margin, - mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); - - mode_lib->ms.TripToMemory = CalculateTripToMemory( - mode_lib->ms.UrgLatency, - mode_lib->ms.FabricClock, - mode_lib->ms.uclk_freq_mhz, - mode_lib->soc.qos_parameters.qos_type, - mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].trip_to_memory_uclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, - mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); - - mode_lib->ms.TripToMemory = math_max2(mode_lib->ms.UrgLatency, mode_lib->ms.TripToMemory); - - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - double line_time_us = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - bool cursor_not_enough_urgent_latency_hiding = false; - - if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { - calculate_cursor_req_attributes( - display_cfg->plane_descriptors[k].cursor.cursor_width, - display_cfg->plane_descriptors[k].cursor.cursor_bpp, - - // output - &s->cursor_lines_per_chunk[k], - &s->cursor_bytes_per_line[k], - &s->cursor_bytes_per_chunk[k], - &s->cursor_bytes[k]); - - calculate_cursor_urgent_burst_factor( - mode_lib->ip.cursor_buffer_size, - display_cfg->plane_descriptors[k].cursor.cursor_width, - s->cursor_bytes_per_chunk[k], - s->cursor_lines_per_chunk[k], - line_time_us, - mode_lib->ms.UrgLatency, - - // output - &mode_lib->ms.UrgentBurstFactorCursor[k], - &cursor_not_enough_urgent_latency_hiding); - } - - mode_lib->ms.UrgentBurstFactorCursorPre[k] = mode_lib->ms.UrgentBurstFactorCursor[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: k=%d, VRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - DML_LOG_VERBOSE("DML::%s: k=%d, VRatioChroma=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); -#endif - - CalculateUrgentBurstFactor( - &display_cfg->plane_descriptors[k], - mode_lib->ms.swath_width_luma_ub[k], - mode_lib->ms.swath_width_chroma_ub[k], - mode_lib->ms.SwathHeightY[k], - mode_lib->ms.SwathHeightC[k], - line_time_us, - mode_lib->ms.UrgLatency, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->ms.BytePerPixelInDETY[k], - mode_lib->ms.BytePerPixelInDETC[k], - mode_lib->ms.DETBufferSizeY[k], - mode_lib->ms.DETBufferSizeC[k], - - // Output - &mode_lib->ms.UrgentBurstFactorLuma[k], - &mode_lib->ms.UrgentBurstFactorChroma[k], - &mode_lib->ms.NotEnoughUrgentLatencyHiding[k]); - - mode_lib->ms.NotEnoughUrgentLatencyHiding[k] = mode_lib->ms.NotEnoughUrgentLatencyHiding[k] || cursor_not_enough_urgent_latency_hiding; - } - - CalculateDCFCLKDeepSleep( - display_cfg, - mode_lib->ms.num_active_planes, - mode_lib->ms.BytePerPixelY, - mode_lib->ms.BytePerPixelC, - mode_lib->ms.SwathWidthY, - mode_lib->ms.SwathWidthC, - mode_lib->ms.NoOfDPP, - mode_lib->ms.PSCL_FACTOR, - mode_lib->ms.PSCL_FACTOR_CHROMA, - mode_lib->ms.RequiredDPPCLK, - mode_lib->ms.vactive_sw_bw_l, - mode_lib->ms.vactive_sw_bw_c, - mode_lib->soc.return_bus_width_bytes, - - /* Output */ - &mode_lib->ms.dcfclk_deepsleep); - - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - mode_lib->ms.WritebackDelayTime[k] = mode_lib->soc.qos_parameters.writeback.base_latency_us + CalculateWriteBackDelay( - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) / mode_lib->ms.RequiredDISPCLK; - } else { - mode_lib->ms.WritebackDelayTime[k] = 0.0; - } - } - - // MaximumVStartup is actually Tvstartup_min in DCN4 programming guide - for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { - bool isInterlaceTiming = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !mode_lib->ip.ptoi_supported); - s->MaximumVStartup[k] = CalculateMaxVStartup( - mode_lib->ip.ptoi_supported, - mode_lib->ip.vblank_nom_default_us, - &display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing, - mode_lib->ms.WritebackDelayTime[k]); - mode_lib->ms.MaxVStartupLines[k] = (isInterlaceTiming ? (2 * s->MaximumVStartup[k]) : s->MaximumVStartup[k]); - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, MaximumVStartup = %u\n", __func__, k, s->MaximumVStartup[k]); -#endif - - /* Immediate Flip and MALL parameters */ - s->ImmediateFlipRequired = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - s->ImmediateFlipRequired = s->ImmediateFlipRequired || display_cfg->plane_descriptors[k].immediate_flip; - } - - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = - mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe || - ((display_cfg->hostvm_enable == true || display_cfg->plane_descriptors[k].immediate_flip == true) && - (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame || dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]))); - } - - mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen || - ((display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable || display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_auto) && (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]))) || - ((display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_disable || display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_auto) && (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame)); - } - - s->FullFrameMALLPStateMethod = false; - s->SubViewportMALLPStateMethod = false; - s->PhantomPipeMALLPStateMethod = false; - s->SubViewportMALLRefreshGreaterThan120Hz = false; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame) - s->FullFrameMALLPStateMethod = true; - if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) { - s->SubViewportMALLPStateMethod = true; - if (!display_cfg->overrides.enable_subvp_implicit_pmo) { - // For dv, small frame tests will have very high refresh rate - unsigned long long refresh_rate = (unsigned long long) ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz * 1000 / - (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / - (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); - if (refresh_rate > 120) - s->SubViewportMALLRefreshGreaterThan120Hz = true; - } - } - if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) - s->PhantomPipeMALLPStateMethod = true; - } - mode_lib->ms.support.InvalidCombinationOfMALLUseForPState = (s->SubViewportMALLPStateMethod != s->PhantomPipeMALLPStateMethod) || - (s->SubViewportMALLPStateMethod && s->FullFrameMALLPStateMethod) || s->SubViewportMALLRefreshGreaterThan120Hz; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: SubViewportMALLPStateMethod = %u\n", __func__, s->SubViewportMALLPStateMethod); - DML_LOG_VERBOSE("DML::%s: PhantomPipeMALLPStateMethod = %u\n", __func__, s->PhantomPipeMALLPStateMethod); - DML_LOG_VERBOSE("DML::%s: FullFrameMALLPStateMethod = %u\n", __func__, s->FullFrameMALLPStateMethod); - DML_LOG_VERBOSE("DML::%s: SubViewportMALLRefreshGreaterThan120Hz = %u\n", __func__, s->SubViewportMALLRefreshGreaterThan120Hz); - DML_LOG_VERBOSE("DML::%s: InvalidCombinationOfMALLUseForPState = %u\n", __func__, mode_lib->ms.support.InvalidCombinationOfMALLUseForPState); - DML_LOG_VERBOSE("DML::%s: in_out_params->min_clk_index = %u\n", __func__, in_out_params->min_clk_index); - DML_LOG_VERBOSE("DML::%s: mode_lib->ms.DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); - DML_LOG_VERBOSE("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); - DML_LOG_VERBOSE("DML::%s: mode_lib->ms.uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); -#endif - - mode_lib->ms.support.OutstandingRequestsSupport = true; - mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = true; - - mode_lib->ms.support.avg_urgent_latency_us - = (mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].average_latency_when_urgent_uclk_cycles / mode_lib->ms.uclk_freq_mhz - * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_average_latency_margin / 100.0) - + mode_lib->soc.qos_parameters.qos_params.dcn4x.average_transport_distance_fclk_cycles / mode_lib->ms.FabricClock) - * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_average_transport_latency_margin / 100.0); - - mode_lib->ms.support.avg_non_urgent_latency_us - = (mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].average_latency_when_non_urgent_uclk_cycles / mode_lib->ms.uclk_freq_mhz - * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_average_latency_margin / 100.0) - + mode_lib->soc.qos_parameters.qos_params.dcn4x.average_transport_distance_fclk_cycles / mode_lib->ms.FabricClock) - * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_average_transport_latency_margin / 100.0); - - mode_lib->ms.support.max_non_urgent_latency_us - = mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].maximum_latency_when_non_urgent_uclk_cycles - / mode_lib->ms.uclk_freq_mhz * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin / 100.0) - + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles / mode_lib->ms.FabricClock - + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles / mode_lib->ms.FabricClock - * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin / 100.0); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { - outstanding_latency_us = (mode_lib->soc.max_outstanding_reqs * mode_lib->ms.support.request_size_bytes_luma[k] - / (mode_lib->ms.DCFCLK * mode_lib->soc.return_bus_width_bytes)); - - if (outstanding_latency_us < mode_lib->ms.support.avg_urgent_latency_us) { - mode_lib->ms.support.OutstandingRequestsSupport = false; - } - - if (outstanding_latency_us < mode_lib->ms.support.avg_non_urgent_latency_us) { - mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = false; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: avg_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_urgent_latency_us); - DML_LOG_VERBOSE("DML::%s: avg_non_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_non_urgent_latency_us); - DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_luma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_luma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (luma)\n", __func__, k, outstanding_latency_us); -#endif - } - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x && mode_lib->ms.BytePerPixelC[k] > 0) { - outstanding_latency_us = (mode_lib->soc.max_outstanding_reqs * mode_lib->ms.support.request_size_bytes_chroma[k] - / (mode_lib->ms.DCFCLK * mode_lib->soc.return_bus_width_bytes)); - - if (outstanding_latency_us < mode_lib->ms.support.avg_urgent_latency_us) { - mode_lib->ms.support.OutstandingRequestsSupport = false; - } - - if (outstanding_latency_us < mode_lib->ms.support.avg_non_urgent_latency_us) { - mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = false; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_chroma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_chroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (chroma)\n", __func__, k, outstanding_latency_us); -#endif - } - } - - memset(calculate_mcache_setting_params, 0, sizeof(struct dml2_core_calcs_calculate_mcache_setting_params)); - if (mode_lib->soc.mcache_size_bytes == 0 || mode_lib->ip.dcn_mrq_present) { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - mode_lib->ms.mall_prefetch_sdp_overhead_factor[k] = 1.0; - mode_lib->ms.mall_prefetch_dram_overhead_factor[k] = 1.0; - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0[k] = 1.0; - mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0[k] = 1.0; - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1[k] = 1.0; - mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1[k] = 1.0; - } - } else { - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - calculate_mcache_setting_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; - calculate_mcache_setting_params->num_chans = mode_lib->soc.clk_table.dram_config.channel_count; - calculate_mcache_setting_params->mem_word_bytes = mode_lib->soc.mem_word_bytes; - calculate_mcache_setting_params->mcache_size_bytes = mode_lib->soc.mcache_size_bytes; - calculate_mcache_setting_params->mcache_line_size_bytes = mode_lib->soc.mcache_line_size_bytes; - calculate_mcache_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; - calculate_mcache_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - - calculate_mcache_setting_params->source_format = display_cfg->plane_descriptors[k].pixel_format; - calculate_mcache_setting_params->surf_vert = dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle); - calculate_mcache_setting_params->vp_stationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; - calculate_mcache_setting_params->tiling_mode = display_cfg->plane_descriptors[k].surface.tiling; - calculate_mcache_setting_params->imall_enable = mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall; - - calculate_mcache_setting_params->vp_start_x_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; - calculate_mcache_setting_params->vp_start_y_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; - calculate_mcache_setting_params->full_vp_width_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; - calculate_mcache_setting_params->full_vp_height_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - calculate_mcache_setting_params->blk_width_l = mode_lib->ms.MacroTileWidthY[k]; - calculate_mcache_setting_params->blk_height_l = mode_lib->ms.MacroTileHeightY[k]; - calculate_mcache_setting_params->vmpg_width_l = s->vmpg_width_y[k]; - calculate_mcache_setting_params->vmpg_height_l = s->vmpg_height_y[k]; - calculate_mcache_setting_params->full_swath_bytes_l = s->full_swath_bytes_l[k]; - calculate_mcache_setting_params->bytes_per_pixel_l = mode_lib->ms.BytePerPixelY[k]; - - calculate_mcache_setting_params->vp_start_x_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.x_start; - calculate_mcache_setting_params->vp_start_y_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - calculate_mcache_setting_params->full_vp_width_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; - calculate_mcache_setting_params->full_vp_height_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; - calculate_mcache_setting_params->blk_width_c = mode_lib->ms.MacroTileWidthC[k]; - calculate_mcache_setting_params->blk_height_c = mode_lib->ms.MacroTileHeightC[k]; - calculate_mcache_setting_params->vmpg_width_c = s->vmpg_width_c[k]; - calculate_mcache_setting_params->vmpg_height_c = s->vmpg_height_c[k]; - calculate_mcache_setting_params->full_swath_bytes_c = s->full_swath_bytes_c[k]; - calculate_mcache_setting_params->bytes_per_pixel_c = mode_lib->ms.BytePerPixelC[k]; - - // output - calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_l = &mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0[k]; - calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_l = &mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0[k]; - calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_c = &mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1[k]; - calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_c = &mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1[k]; - - calculate_mcache_setting_params->num_mcaches_l = &mode_lib->ms.num_mcaches_l[k]; - calculate_mcache_setting_params->mcache_row_bytes_l = &mode_lib->ms.mcache_row_bytes_l[k]; - calculate_mcache_setting_params->mcache_row_bytes_per_channel_l = &mode_lib->ms.mcache_row_bytes_per_channel_l[k]; - calculate_mcache_setting_params->mcache_offsets_l = mode_lib->ms.mcache_offsets_l[k]; - calculate_mcache_setting_params->mcache_shift_granularity_l = &mode_lib->ms.mcache_shift_granularity_l[k]; - - calculate_mcache_setting_params->num_mcaches_c = &mode_lib->ms.num_mcaches_c[k]; - calculate_mcache_setting_params->mcache_row_bytes_c = &mode_lib->ms.mcache_row_bytes_c[k]; - calculate_mcache_setting_params->mcache_row_bytes_per_channel_c = &mode_lib->ms.mcache_row_bytes_per_channel_c[k]; - calculate_mcache_setting_params->mcache_offsets_c = mode_lib->ms.mcache_offsets_c[k]; - calculate_mcache_setting_params->mcache_shift_granularity_c = &mode_lib->ms.mcache_shift_granularity_c[k]; - - calculate_mcache_setting_params->mall_comb_mcache_l = &mode_lib->ms.mall_comb_mcache_l[k]; - calculate_mcache_setting_params->mall_comb_mcache_c = &mode_lib->ms.mall_comb_mcache_c[k]; - calculate_mcache_setting_params->lc_comb_mcache = &mode_lib->ms.lc_comb_mcache[k]; - - calculate_mcache_setting(&mode_lib->scratch, calculate_mcache_setting_params); - } - - calculate_mall_bw_overhead_factor( - mode_lib->ms.mall_prefetch_sdp_overhead_factor, - mode_lib->ms.mall_prefetch_dram_overhead_factor, - - // input - display_cfg, - mode_lib->ms.num_active_planes); - } - - // Calculate all the bandwidth available - // Need anothe bw for latency evaluation - calculate_bandwidth_available( - mode_lib->ms.support.avg_bandwidth_available_min, // not used - mode_lib->ms.support.avg_bandwidth_available, // not used - mode_lib->ms.support.urg_bandwidth_available_min_latency, - mode_lib->ms.support.urg_bandwidth_available, // not used - mode_lib->ms.support.urg_bandwidth_available_vm_only, // not used - mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm, // not used - - &mode_lib->soc, - display_cfg->hostvm_enable, - mode_lib->ms.DCFCLK, - mode_lib->ms.FabricClock, - mode_lib->ms.dram_bw_mbps); - - calculate_bandwidth_available( - mode_lib->ms.support.avg_bandwidth_available_min, - mode_lib->ms.support.avg_bandwidth_available, - mode_lib->ms.support.urg_bandwidth_available_min, - mode_lib->ms.support.urg_bandwidth_available, - mode_lib->ms.support.urg_bandwidth_available_vm_only, - mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm, - - &mode_lib->soc, - display_cfg->hostvm_enable, - mode_lib->ms.MaxDCFCLK, - mode_lib->ms.MaxFabricClock, -#ifdef DML_MODE_SUPPORT_USE_DPM_DRAM_BW - mode_lib->ms.dram_bw_mbps); -#else - mode_lib->ms.max_dram_bw_mbps); -#endif - - // Average BW support check - calculate_avg_bandwidth_required( - mode_lib->ms.support.avg_bandwidth_required, - // input - display_cfg, - mode_lib->ms.num_active_planes, - mode_lib->ms.vactive_sw_bw_l, - mode_lib->ms.vactive_sw_bw_c, - mode_lib->ms.cursor_bw, - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0, - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1, - mode_lib->ms.mall_prefetch_dram_overhead_factor, - mode_lib->ms.mall_prefetch_sdp_overhead_factor); - - for (m = 0; m < dml2_core_internal_bw_max; m++) { // check sdp and dram - mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_sys_idle][m] = 1; - mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_sys_active][m] = (mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][m] <= mode_lib->ms.support.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][m]); - mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_svp_prefetch][m] = (mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][m] <= mode_lib->ms.support.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][m]); - } - - mode_lib->ms.support.AvgBandwidthSupport = true; - mode_lib->ms.support.EnoughUrgentLatencyHidingSupport = true; - for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { - if (mode_lib->ms.NotEnoughUrgentLatencyHiding[k]) { - mode_lib->ms.support.EnoughUrgentLatencyHidingSupport = false; - DML_LOG_VERBOSE("DML::%s: k=%u NotEnoughUrgentLatencyHiding set\n", __func__, k); - - } - } - for (m = 0; m < dml2_core_internal_soc_state_max; m++) { - for (n = 0; n < dml2_core_internal_bw_max; n++) { // check sdp and dram - if (!mode_lib->ms.support.avg_bandwidth_support_ok[m][n] && (m == dml2_core_internal_soc_state_sys_active || mode_lib->soc.mall_allocated_for_dcn_mbytes > 0)) { - mode_lib->ms.support.AvgBandwidthSupport = false; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: avg_bandwidth_support_ok[%s][%s] not ok\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n)); -#endif - } - } - } - - dml_core_ms_prefetch_check(mode_lib, display_cfg); - - mode_lib->ms.support.max_urgent_latency_us = s->mSOCParameters.max_urgent_latency_us; - - //Re-ordering Buffer Support Check - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { - if (((mode_lib->ip.rob_buffer_size_kbytes - mode_lib->ip.pixel_chunk_size_kbytes) * 1024 - / mode_lib->ms.support.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]) >= s->mSOCParameters.max_urgent_latency_us) { - mode_lib->ms.support.ROBSupport = true; - } else { - mode_lib->ms.support.ROBSupport = false; - } - } else { - if (mode_lib->ip.rob_buffer_size_kbytes * 1024 >= mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles * mode_lib->soc.fabric_datapath_to_dcn_data_return_bytes) { - mode_lib->ms.support.ROBSupport = true; - } else { - mode_lib->ms.support.ROBSupport = false; - } - } - - /* VActive fill time calculations (informative) */ - calculate_vactive_det_fill_latency( - display_cfg, - mode_lib->ms.num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0, - mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1, - mode_lib->ms.vactive_sw_bw_l, - mode_lib->ms.vactive_sw_bw_c, - mode_lib->ms.surface_avg_vactive_required_bw, - mode_lib->ms.surface_peak_required_bw, - /* outputs */ - mode_lib->ms.dram_change_vactive_det_fill_delay_us); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: max_urgent_latency_us = %f\n", __func__, s->mSOCParameters.max_urgent_latency_us); - DML_LOG_VERBOSE("DML::%s: ROBSupport = %u\n", __func__, mode_lib->ms.support.ROBSupport); -#endif - - /*Mode Support, Voltage State and SOC Configuration*/ - { - if (mode_lib->ms.support.ScaleRatioAndTapsSupport - && mode_lib->ms.support.SourceFormatPixelAndScanSupport - && mode_lib->ms.support.ViewportSizeSupport - && !mode_lib->ms.support.LinkRateDoesNotMatchDPVersion - && !mode_lib->ms.support.LinkRateForMultistreamNotIndicated - && !mode_lib->ms.support.BPPForMultistreamNotIndicated - && !mode_lib->ms.support.MultistreamWithHDMIOreDP - && !mode_lib->ms.support.ExceededMultistreamSlots - && !mode_lib->ms.support.MSOOrODMSplitWithNonDPLink - && !mode_lib->ms.support.NotEnoughLanesForMSO - && !mode_lib->ms.support.P2IWith420 - && !mode_lib->ms.support.DSC422NativeNotSupported - && mode_lib->ms.support.DSCSlicesODMModeSupported - && !mode_lib->ms.support.NotEnoughDSCUnits - && !mode_lib->ms.support.NotEnoughDSCSlices - && !mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe - && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen - && !mode_lib->ms.support.DSCCLKRequiredMoreThanSupported - && mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport - && !mode_lib->ms.support.DTBCLKRequiredMoreThanSupported - && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPState - && mode_lib->ms.support.ROBSupport - && mode_lib->ms.support.OutstandingRequestsSupport - && mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance - && mode_lib->ms.support.DISPCLK_DPPCLK_Support - && mode_lib->ms.support.TotalAvailablePipesSupport - && mode_lib->ms.support.NumberOfOTGSupport - && mode_lib->ms.support.NumberOfHDMIFRLSupport - && mode_lib->ms.support.NumberOfDP2p0Support - && mode_lib->ms.support.EnoughWritebackUnits - && mode_lib->ms.support.WritebackLatencySupport - && mode_lib->ms.support.WritebackScaleRatioAndTapsSupport - && mode_lib->ms.support.CursorSupport - && mode_lib->ms.support.PitchSupport - && !mode_lib->ms.support.ViewportExceedsSurface - && mode_lib->ms.support.PrefetchSupported - && mode_lib->ms.support.EnoughUrgentLatencyHidingSupport - && mode_lib->ms.support.AvgBandwidthSupport - && mode_lib->ms.support.DynamicMetadataSupported - && mode_lib->ms.support.VRatioInPrefetchSupported - && mode_lib->ms.support.PTEBufferSizeNotExceeded - && mode_lib->ms.support.DCCMetaBufferSizeNotExceeded - && !mode_lib->ms.support.ExceededMALLSize - && mode_lib->ms.support.g6_temp_read_support - && ((!display_cfg->hostvm_enable && !s->ImmediateFlipRequired) || mode_lib->ms.support.ImmediateFlipSupport)) { - DML_LOG_VERBOSE("DML::%s: mode is supported\n", __func__); - mode_lib->ms.support.ModeSupport = true; - } else { - DML_LOG_VERBOSE("DML::%s: mode is NOT supported\n", __func__); - mode_lib->ms.support.ModeSupport = false; - } - } - - // Since now the mode_support work on 1 particular power state, so there is only 1 state idx (index 0). - DML_LOG_VERBOSE("DML::%s: ModeSupport = %u\n", __func__, mode_lib->ms.support.ModeSupport); - DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - mode_lib->ms.support.MPCCombineEnable[k] = mode_lib->ms.MPCCombine[k]; - mode_lib->ms.support.DPPPerSurface[k] = mode_lib->ms.NoOfDPP[k]; - } - - for (k = 0; k < mode_lib->ms.num_active_planes; k++) { - mode_lib->ms.support.ODMMode[k] = mode_lib->ms.ODMMode[k]; - mode_lib->ms.support.DSCEnabled[k] = mode_lib->ms.RequiresDSC[k]; - mode_lib->ms.support.FECEnabled[k] = mode_lib->ms.RequiresFEC[k]; - mode_lib->ms.support.OutputBpp[k] = mode_lib->ms.OutputBpp[k]; - mode_lib->ms.support.OutputType[k] = mode_lib->ms.OutputType[k]; - mode_lib->ms.support.OutputRate[k] = mode_lib->ms.OutputRate[k]; - -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: k=%d, ODMMode = %u\n", __func__, k, mode_lib->ms.support.ODMMode[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, DSCEnabled = %u\n", __func__, k, mode_lib->ms.support.DSCEnabled[k]); -#endif - } - -#if defined(__DML_VBA_DEBUG__) - if (!mode_lib->ms.support.ModeSupport) - dml2_print_mode_support_info(&mode_lib->ms.support, true); - - DML_LOG_VERBOSE("DML::%s: --- DONE --- \n", __func__); -#endif - - return mode_lib->ms.support.ModeSupport; -} - -unsigned int dml2_core_calcs_mode_support_ex(struct dml2_core_calcs_mode_support_ex *in_out_params) -{ - unsigned int result; - - DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); - result = dml_core_mode_support(in_out_params); - - if (result) - *in_out_params->out_evaluation_info = in_out_params->mode_lib->ms.support; - - DML_LOG_VERBOSE("DML::%s: is_mode_support = %u (min_clk_index=%d)\n", __func__, result, in_out_params->min_clk_index); - - for (unsigned int k = 0; k < in_out_params->in_display_cfg->num_planes; k++) - DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, in_out_params->in_display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); - - DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); - - return result; -} - -static void CalculatePixelDeliveryTimes( - const struct dml2_display_cfg *display_cfg, - const struct core_display_cfg_support_info *cfg_support_info, - unsigned int NumberOfActiveSurfaces, - double VRatioPrefetchY[], - double VRatioPrefetchC[], - unsigned int swath_width_luma_ub[], - unsigned int swath_width_chroma_ub[], - double PSCL_THROUGHPUT[], - double PSCL_THROUGHPUT_CHROMA[], - double Dppclk[], - unsigned int BytePerPixelC[], - unsigned int req_per_swath_ub_l[], - unsigned int req_per_swath_ub_c[], - - // Output - double DisplayPipeLineDeliveryTimeLuma[], - double DisplayPipeLineDeliveryTimeChroma[], - double DisplayPipeLineDeliveryTimeLumaPrefetch[], - double DisplayPipeLineDeliveryTimeChromaPrefetch[], - double DisplayPipeRequestDeliveryTimeLuma[], - double DisplayPipeRequestDeliveryTimeChroma[], - double DisplayPipeRequestDeliveryTimeLumaPrefetch[], - double DisplayPipeRequestDeliveryTimeChromaPrefetch[]) -{ - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u : HRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u : VRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchY = %f\n", __func__, k, VRatioPrefetchY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchC = %f\n", __func__, k, VRatioPrefetchC[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, cfg_support_info->plane_support_info[k].dpps_used); - DML_LOG_VERBOSE("DML::%s: k=%u : pixel_clock_mhz = %f\n", __func__, k, pixel_clock_mhz); - DML_LOG_VERBOSE("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); -#endif - if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio <= 1) { - DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_clock_mhz; - } else { - DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChroma[k] = 0; - } else { - if (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio <= 1) { - DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_clock_mhz; - } else { - DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } - - if (VRatioPrefetchY[k] <= 1) { - DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_clock_mhz; - } else { - DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; - } - - if (BytePerPixelC[k] == 0) { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0; - } else { - if (VRatioPrefetchC[k] <= 1) { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_clock_mhz; - } else { - DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; - } - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); -#endif - } - - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - - DisplayPipeRequestDeliveryTimeLuma[k] = DisplayPipeLineDeliveryTimeLuma[k] / req_per_swath_ub_l[k]; - DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = DisplayPipeLineDeliveryTimeLumaPrefetch[k] / req_per_swath_ub_l[k]; - if (BytePerPixelC[k] == 0) { - DisplayPipeRequestDeliveryTimeChroma[k] = 0; - DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = 0; - } else { - DisplayPipeRequestDeliveryTimeChroma[k] = DisplayPipeLineDeliveryTimeChroma[k] / req_per_swath_ub_c[k]; - DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub_c[k]; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_l = %d\n", __func__, k, req_per_swath_ub_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_c = %d\n", __func__, k, req_per_swath_ub_c[k]); -#endif - } -} - -static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTETimes_params *p) -{ - unsigned int meta_chunk_width; - unsigned int min_meta_chunk_width; - unsigned int meta_chunk_per_row_int; - unsigned int meta_row_remainder; - unsigned int meta_chunk_threshold; - unsigned int meta_chunks_per_row_ub; - unsigned int meta_chunk_width_chroma; - unsigned int min_meta_chunk_width_chroma; - unsigned int meta_chunk_per_row_int_chroma; - unsigned int meta_row_remainder_chroma; - unsigned int meta_chunk_threshold_chroma; - unsigned int meta_chunks_per_row_ub_chroma; - unsigned int dpte_group_width_luma; - unsigned int dpte_groups_per_row_luma_ub; - unsigned int dpte_group_width_chroma; - unsigned int dpte_groups_per_row_chroma_ub; - double pixel_clock_mhz; - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - p->DST_Y_PER_PTE_ROW_NOM_L[k] = p->dpte_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - if (p->BytePerPixelC[k] == 0) { - p->DST_Y_PER_PTE_ROW_NOM_C[k] = 0; - } else { - p->DST_Y_PER_PTE_ROW_NOM_C[k] = p->dpte_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - } - p->DST_Y_PER_META_ROW_NOM_L[k] = p->meta_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - if (p->BytePerPixelC[k] == 0) { - p->DST_Y_PER_META_ROW_NOM_C[k] = 0; - } else { - p->DST_Y_PER_META_ROW_NOM_C[k] = p->meta_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - } - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (p->display_cfg->plane_descriptors[k].surface.dcc.enable == true && p->mrq_present) { - meta_chunk_width = p->MetaChunkSize * 1024 * 256 / p->BytePerPixelY[k] / p->meta_row_height[k]; - min_meta_chunk_width = p->MinMetaChunkSizeBytes * 256 / p->BytePerPixelY[k] / p->meta_row_height[k]; - meta_chunk_per_row_int = p->meta_row_width[k] / meta_chunk_width; - meta_row_remainder = p->meta_row_width[k] % meta_chunk_width; - if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { - meta_chunk_threshold = 2 * min_meta_chunk_width - p->meta_req_width[k]; - } else { - meta_chunk_threshold = 2 * min_meta_chunk_width - p->meta_req_height[k]; - } - if (meta_row_remainder <= meta_chunk_threshold) { - meta_chunks_per_row_ub = meta_chunk_per_row_int + 1; - } else { - meta_chunks_per_row_ub = meta_chunk_per_row_int + 2; - } - p->TimePerMetaChunkNominal[k] = p->meta_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio * - p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / - (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; - p->TimePerMetaChunkVBlank[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / - (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; - p->TimePerMetaChunkFlip[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / - (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; - if (p->BytePerPixelC[k] == 0) { - p->TimePerChromaMetaChunkNominal[k] = 0; - p->TimePerChromaMetaChunkVBlank[k] = 0; - p->TimePerChromaMetaChunkFlip[k] = 0; - } else { - meta_chunk_width_chroma = p->MetaChunkSize * 1024 * 256 / p->BytePerPixelC[k] / p->meta_row_height_chroma[k]; - min_meta_chunk_width_chroma = p->MinMetaChunkSizeBytes * 256 / p->BytePerPixelC[k] / p->meta_row_height_chroma[k]; - meta_chunk_per_row_int_chroma = (unsigned int)((double)p->meta_row_width_chroma[k] / meta_chunk_width_chroma); - meta_row_remainder_chroma = p->meta_row_width_chroma[k] % meta_chunk_width_chroma; - if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { - meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - p->meta_req_width_chroma[k]; - } else { - meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - p->meta_req_height_chroma[k]; - } - if (meta_row_remainder_chroma <= meta_chunk_threshold_chroma) { - meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 1; - } else { - meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 2; - } - p->TimePerChromaMetaChunkNominal[k] = p->meta_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; - p->TimePerChromaMetaChunkVBlank[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; - p->TimePerChromaMetaChunkFlip[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; - } - } else { - p->TimePerMetaChunkNominal[k] = 0; - p->TimePerMetaChunkVBlank[k] = 0; - p->TimePerMetaChunkFlip[k] = 0; - p->TimePerChromaMetaChunkNominal[k] = 0; - p->TimePerChromaMetaChunkVBlank[k] = 0; - p->TimePerChromaMetaChunkFlip[k] = 0; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_L[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_C[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkNominal = %f\n", __func__, k, p->TimePerMetaChunkNominal[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkVBlank = %f\n", __func__, k, p->TimePerMetaChunkVBlank[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkFlip = %f\n", __func__, k, p->TimePerMetaChunkFlip[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkNominal = %f\n", __func__, k, p->TimePerChromaMetaChunkNominal[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, p->TimePerChromaMetaChunkVBlank[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkFlip = %f\n", __func__, k, p->TimePerChromaMetaChunkFlip[k]); -#endif - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - p->DST_Y_PER_PTE_ROW_NOM_L[k] = p->dpte_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - if (p->BytePerPixelC[k] == 0) { - p->DST_Y_PER_PTE_ROW_NOM_C[k] = 0; - } else { - p->DST_Y_PER_PTE_ROW_NOM_C[k] = p->dpte_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - } - } - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - pixel_clock_mhz = ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - - if (p->display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) - p->time_per_tdlut_group[k] = 2 * p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / p->tdlut_groups_per_2row_ub[k]; - else - p->time_per_tdlut_group[k] = 0; - - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_tdlut_group = %f\n", __func__, k, p->time_per_tdlut_group[k]); - - if (p->display_cfg->gpuvm_enable == true) { - if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { - dpte_group_width_luma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeY[k] * p->PixelPTEReqWidthY[k]); - } else { - dpte_group_width_luma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeY[k] * p->PixelPTEReqHeightY[k]); - } - if (p->use_one_row_for_frame[k]) { - dpte_groups_per_row_luma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_luma_ub[k] / (double)dpte_group_width_luma / 2.0, 1.0)); - } else { - dpte_groups_per_row_luma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_luma_ub[k] / (double)dpte_group_width_luma, 1.0)); - } - if (dpte_groups_per_row_luma_ub <= 2) { - dpte_groups_per_row_luma_ub = dpte_groups_per_row_luma_ub + 1; - } - DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, p->dpte_group_bytes[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, p->PTERequestSizeY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, p->PixelPTEReqWidthY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, p->PixelPTEReqHeightY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); - - p->time_per_pte_group_nom_luma[k] = p->DST_Y_PER_PTE_ROW_NOM_L[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; - p->time_per_pte_group_vblank_luma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; - p->time_per_pte_group_flip_luma[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; - if (p->BytePerPixelC[k] == 0) { - p->time_per_pte_group_nom_chroma[k] = 0; - p->time_per_pte_group_vblank_chroma[k] = 0; - p->time_per_pte_group_flip_chroma[k] = 0; - } else { - if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { - dpte_group_width_chroma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeC[k] * p->PixelPTEReqWidthC[k]); - } else { - dpte_group_width_chroma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeC[k] * p->PixelPTEReqHeightC[k]); - } - - if (p->use_one_row_for_frame[k]) { - dpte_groups_per_row_chroma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_chroma_ub[k] / (double)dpte_group_width_chroma / 2.0, 1.0)); - } else { - dpte_groups_per_row_chroma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_chroma_ub[k] / (double)dpte_group_width_chroma, 1.0)); - } - if (dpte_groups_per_row_chroma_ub <= 2) { - dpte_groups_per_row_chroma_ub = dpte_groups_per_row_chroma_ub + 1; - } - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); - DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); - - p->time_per_pte_group_nom_chroma[k] = p->DST_Y_PER_PTE_ROW_NOM_C[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; - p->time_per_pte_group_vblank_chroma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; - p->time_per_pte_group_flip_chroma[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; - } - } else { - p->time_per_pte_group_nom_luma[k] = 0; - p->time_per_pte_group_vblank_luma[k] = 0; - p->time_per_pte_group_flip_luma[k] = 0; - p->time_per_pte_group_nom_chroma[k] = 0; - p->time_per_pte_group_vblank_chroma[k] = 0; - p->time_per_pte_group_flip_chroma[k] = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_vblank = %f\n", __func__, k, p->dst_y_per_row_vblank[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_flip = %f\n", __func__, k, p->dst_y_per_row_flip[k]); - - DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_L[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_C[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, p->time_per_pte_group_nom_luma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, p->time_per_pte_group_vblank_luma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, p->time_per_pte_group_flip_luma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, p->time_per_pte_group_nom_chroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, p->time_per_pte_group_vblank_chroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, p->time_per_pte_group_flip_chroma[k]); -#endif - } -} // CalculateMetaAndPTETimes - -static void CalculateVMGroupAndRequestTimes( - const struct dml2_display_cfg *display_cfg, - unsigned int NumberOfActiveSurfaces, - unsigned int BytePerPixelC[], - double dst_y_per_vm_vblank[], - double dst_y_per_vm_flip[], - unsigned int dpte_row_width_luma_ub[], - unsigned int dpte_row_width_chroma_ub[], - unsigned int vm_group_bytes[], - unsigned int dpde0_bytes_per_frame_ub_l[], - unsigned int dpde0_bytes_per_frame_ub_c[], - unsigned int tdlut_pte_bytes_per_frame[], - unsigned int meta_pte_bytes_per_frame_ub_l[], - unsigned int meta_pte_bytes_per_frame_ub_c[], - bool mrq_present, - - // Output - double TimePerVMGroupVBlank[], - double TimePerVMGroupFlip[], - double TimePerVMRequestVBlank[], - double TimePerVMRequestFlip[]) -{ - unsigned int num_group_per_lower_vm_stage = 0; - unsigned int num_req_per_lower_vm_stage = 0; - unsigned int num_group_per_lower_vm_stage_flip; - unsigned int num_group_per_lower_vm_stage_pref; - unsigned int num_req_per_lower_vm_stage_flip; - unsigned int num_req_per_lower_vm_stage_pref; - double line_time; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); -#endif - for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { - double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - bool dcc_mrq_enable = display_cfg->plane_descriptors[k].surface.dcc.enable && mrq_present; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, dcc_mrq_enable = %u\n", __func__, k, dcc_mrq_enable); - DML_LOG_VERBOSE("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_l = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_c = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); -#endif - - if (display_cfg->gpuvm_enable) { - if (display_cfg->gpuvm_max_page_table_levels >= 2) { - num_group_per_lower_vm_stage += (unsigned int) math_ceil2((double) (dpde0_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1); - - if (BytePerPixelC[k] > 0) - num_group_per_lower_vm_stage += (unsigned int) math_ceil2((double) (dpde0_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1); - } - - if (dcc_mrq_enable) { - if (BytePerPixelC[k] > 0) { - num_group_per_lower_vm_stage += (unsigned int)(2.0 /*for each mpde0 group*/ + math_ceil2((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1) + - math_ceil2((double) (meta_pte_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1)); - } else { - num_group_per_lower_vm_stage += (unsigned int)(1.0 + math_ceil2((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1)); - } - } - - num_group_per_lower_vm_stage_flip = num_group_per_lower_vm_stage; - num_group_per_lower_vm_stage_pref = num_group_per_lower_vm_stage; - - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && display_cfg->gpuvm_enable) { - num_group_per_lower_vm_stage_pref += (unsigned int) math_ceil2(tdlut_pte_bytes_per_frame[k] / vm_group_bytes[k], 1); - if (display_cfg->gpuvm_max_page_table_levels >= 2) - num_group_per_lower_vm_stage_pref += 1; // tdpe0 group - } - - if (display_cfg->gpuvm_max_page_table_levels >= 2) { - num_req_per_lower_vm_stage += dpde0_bytes_per_frame_ub_l[k] / 64; - if (BytePerPixelC[k] > 0) - num_req_per_lower_vm_stage += dpde0_bytes_per_frame_ub_c[k]; - } - - if (dcc_mrq_enable) { - num_req_per_lower_vm_stage += meta_pte_bytes_per_frame_ub_l[k] / 64; - if (BytePerPixelC[k] > 0) - num_req_per_lower_vm_stage += meta_pte_bytes_per_frame_ub_c[k] / 64; - } - - num_req_per_lower_vm_stage_flip = num_req_per_lower_vm_stage; - num_req_per_lower_vm_stage_pref = num_req_per_lower_vm_stage; - - if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && display_cfg->gpuvm_enable) { - num_req_per_lower_vm_stage_pref += tdlut_pte_bytes_per_frame[k] / 64; - } - - line_time = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz; - - if (num_group_per_lower_vm_stage_pref > 0) - TimePerVMGroupVBlank[k] = dst_y_per_vm_vblank[k] * line_time / num_group_per_lower_vm_stage_pref; - else - TimePerVMGroupVBlank[k] = 0; - - if (num_group_per_lower_vm_stage_flip > 0) - TimePerVMGroupFlip[k] = dst_y_per_vm_flip[k] * line_time / num_group_per_lower_vm_stage_flip; - else - TimePerVMGroupFlip[k] = 0; - - if (num_req_per_lower_vm_stage_pref > 0) - TimePerVMRequestVBlank[k] = dst_y_per_vm_vblank[k] * line_time / num_req_per_lower_vm_stage_pref; - else - TimePerVMRequestVBlank[k] = 0.0; - if (num_req_per_lower_vm_stage_flip > 0) - TimePerVMRequestFlip[k] = dst_y_per_vm_flip[k] * line_time / num_req_per_lower_vm_stage_flip; - else - TimePerVMRequestFlip[k] = 0.0; - - DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_vblank = %f\n", __func__, k, dst_y_per_vm_vblank[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_flip = %f\n", __func__, k, dst_y_per_vm_flip[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, line_time = %f\n", __func__, k, line_time); - DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_pref = %d\n", __func__, k, num_group_per_lower_vm_stage_pref); - DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_flip = %d\n", __func__, k, num_group_per_lower_vm_stage_flip); - DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_pref = %d\n", __func__, k, num_req_per_lower_vm_stage_pref); - DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_flip = %d\n", __func__, k, num_req_per_lower_vm_stage_flip); - - if (display_cfg->gpuvm_max_page_table_levels > 2) { - TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; - TimePerVMGroupFlip[k] = TimePerVMGroupFlip[k] / 2; - TimePerVMRequestVBlank[k] = TimePerVMRequestVBlank[k] / 2; - TimePerVMRequestFlip[k] = TimePerVMRequestFlip[k] / 2; - } - - } else { - TimePerVMGroupVBlank[k] = 0; - TimePerVMGroupFlip[k] = 0; - TimePerVMRequestVBlank[k] = 0; - TimePerVMRequestFlip[k] = 0; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); -#endif - } -} - -static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratch, - struct dml2_core_calcs_CalculateStutterEfficiency_params *p) -{ - struct dml2_core_calcs_CalculateStutterEfficiency_locals *l = &scratch->CalculateStutterEfficiency_locals; - - unsigned int TotalNumberOfActiveOTG = 0; - double SinglePixelClock = 0; - unsigned int SingleHTotal = 0; - unsigned int SingleVTotal = 0; - bool SameTiming = true; - bool FoundCriticalSurface = false; - - memset(l, 0, sizeof(struct dml2_core_calcs_CalculateStutterEfficiency_locals)); - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { - if (p->display_cfg->plane_descriptors[k].surface.dcc.enable == true) { - if ((dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockWidth256BytesY[k] > p->SwathHeightY[k]) || (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockHeight256BytesY[k] > p->SwathHeightY[k]) || p->DCCYMaxUncompressedBlock[k] < 256) { - l->MaximumEffectiveCompressionLuma = 2; - } else { - l->MaximumEffectiveCompressionLuma = 4; - } - l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0, l->MaximumEffectiveCompressionLuma); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0); - DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, l->MaximumEffectiveCompressionLuma); -#endif - l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0; - l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0 / l->MaximumEffectiveCompressionLuma; - - if (p->ReadBandwidthSurfaceChroma[k] > 0) { - if ((dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockWidth256BytesC[k] > p->SwathHeightC[k]) || (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockHeight256BytesC[k] > p->SwathHeightC[k]) || p->DCCCMaxUncompressedBlock[k] < 256) { - l->MaximumEffectiveCompressionChroma = 2; - } else { - l->MaximumEffectiveCompressionChroma = 4; - } - l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1, l->MaximumEffectiveCompressionChroma); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1); - DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, l->MaximumEffectiveCompressionChroma); -#endif - l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1; - l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1 / l->MaximumEffectiveCompressionChroma; - } - } else { - l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] + p->ReadBandwidthSurfaceChroma[k]; - } - l->TotalRowReadBandwidth = l->TotalRowReadBandwidth + p->DPPPerSurface[k] * (p->meta_row_bw[k] + p->dpte_row_bw[k]); - } - } - - l->AverageDCCCompressionRate = p->TotalDataReadBandwidth / l->TotalCompressedReadBandwidth; - l->AverageDCCZeroSizeFraction = l->TotalZeroSizeRequestReadBandwidth / p->TotalDataReadBandwidth; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); - DML_LOG_VERBOSE("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, l->TotalCompressedReadBandwidth); - DML_LOG_VERBOSE("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, l->TotalZeroSizeRequestReadBandwidth); - DML_LOG_VERBOSE("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, l->TotalZeroSizeCompressedReadBandwidth); - DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, l->MaximumEffectiveCompressionLuma); - DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, l->MaximumEffectiveCompressionChroma); - DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); - DML_LOG_VERBOSE("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, l->AverageDCCZeroSizeFraction); - - DML_LOG_VERBOSE("DML::%s: CompbufReservedSpace64B = %u (%f kbytes)\n", __func__, p->CompbufReservedSpace64B, p->CompbufReservedSpace64B * 64 / 1024.0); - DML_LOG_VERBOSE("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); - DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u kbytes\n", __func__, p->CompressedBufferSizeInkByte); - DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u kbytes\n", __func__, p->ROBBufferSizeInKByte); -#endif - if (l->AverageDCCZeroSizeFraction == 1) { - l->AverageZeroSizeCompressionRate = l->TotalZeroSizeRequestReadBandwidth / l->TotalZeroSizeCompressedReadBandwidth; - l->EffectiveCompressedBufferSize = (double)p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageZeroSizeCompressionRate + ((double)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 * l->AverageZeroSizeCompressionRate; - - - } else if (l->AverageDCCZeroSizeFraction > 0) { - l->AverageZeroSizeCompressionRate = l->TotalZeroSizeRequestReadBandwidth / l->TotalZeroSizeCompressedReadBandwidth; - l->EffectiveCompressedBufferSize = math_min2((double)p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate, - (double)p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)) + - (p->rob_alloc_compressed ? math_min2(((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * l->AverageDCCCompressionRate, - ((double)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)) - : ((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); - - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); - DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)); - DML_LOG_VERBOSE("DML::%s: min 3 = %d\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); - DML_LOG_VERBOSE("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)); -#endif - } else { - l->EffectiveCompressedBufferSize = math_min2((double)p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate, - (double)p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate) + - ((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * (p->rob_alloc_compressed ? l->AverageDCCCompressionRate : 1.0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); - DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate); -#endif - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); - DML_LOG_VERBOSE("DML::%s: ZeroSizeBufferEntries = %u\n", __func__, p->ZeroSizeBufferEntries); - DML_LOG_VERBOSE("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, l->AverageZeroSizeCompressionRate); - DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); -#endif - - *p->StutterPeriod = 0; - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { - l->LinesInDETY = ((double)p->DETBufferSizeY[k] + (p->UnboundedRequestEnabled == true ? l->EffectiveCompressedBufferSize : 0) * p->ReadBandwidthSurfaceLuma[k] / p->TotalDataReadBandwidth) / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; - l->LinesInDETYRoundedDownToSwath = math_floor2(l->LinesInDETY, p->SwathHeightY[k]); - l->DETBufferingTimeY = l->LinesInDETYRoundedDownToSwath * ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u (%u kbytes)\n", __func__, k, p->DETBufferSizeY[k], p->DETBufferSizeY[k] / 1024); - DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); - DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, l->LinesInDETY); - DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, l->LinesInDETYRoundedDownToSwath); - DML_LOG_VERBOSE("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, l->DETBufferingTimeY); -#endif - - if (!FoundCriticalSurface || l->DETBufferingTimeY < *p->StutterPeriod) { - bool isInterlaceTiming = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !p->ProgressiveToInterlaceUnitInOPP; - - FoundCriticalSurface = true; - *p->StutterPeriod = l->DETBufferingTimeY; - l->FrameTimeCriticalSurface = (isInterlaceTiming ? math_floor2((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total / 2.0, 1.0) : p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total) * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - l->VActiveTimeCriticalSurface = (isInterlaceTiming ? math_floor2((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_active / 2.0, 1.0) : p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_active) * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - l->BytePerPixelYCriticalSurface = p->BytePerPixelY[k]; - l->SwathWidthYCriticalSurface = p->SwathWidthY[k]; - l->SwathHeightYCriticalSurface = p->SwathHeightY[k]; - l->BlockWidth256BytesYCriticalSurface = p->BlockWidth256BytesY[k]; - l->DETBufferSizeYCriticalSurface = p->DETBufferSizeY[k]; - l->MinTTUVBlankCriticalSurface = p->MinTTUVBlank[k]; - l->SinglePlaneCriticalSurface = (p->ReadBandwidthSurfaceChroma[k] == 0); - l->SinglePipeCriticalSurface = (p->DPPPerSurface[k] == 1); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); - DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, l->MinTTUVBlankCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, FrameTimeCriticalSurface= %f\n", __func__, k, l->FrameTimeCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, l->VActiveTimeCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, l->BytePerPixelYCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, l->SwathWidthYCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, l->SwathHeightYCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, l->BlockWidth256BytesYCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, l->SinglePlaneCriticalSurface); - DML_LOG_VERBOSE("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, l->SinglePipeCriticalSurface); -#endif - } - } - } - - // for bounded req, the stutter period is calculated only based on DET size, but during burst there can be some return inside ROB/compressed buffer - // stutter period is calculated only on the det sizing - // if (cdb + rob >= det) the stutter burst will be absorbed by the cdb + rob which is before decompress - // else - // the cdb + rob part will be in compressed rate with urg bw (idea bw) - // the det part will be return at uncompressed rate with 64B/dcfclk - // - // for unbounded req, the stutter period should be calculated as total of CDB+ROB+DET, so the term "PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer" - // should be == EffectiveCompressedBufferSize which will returned a compressed rate, the rest of stutter period is from the DET will be returned at uncompressed rate with 64B/dcfclk - - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = math_min2(*p->StutterPeriod * p->TotalDataReadBandwidth, l->EffectiveCompressedBufferSize); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); - DML_LOG_VERBOSE("DML::%s: StutterPeriod*TotalDataReadBandwidth = %f (%f kbytes)\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth, (*p->StutterPeriod * p->TotalDataReadBandwidth) / 1024.0); - DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); - DML_LOG_VERBOSE("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f (%f kbytes)\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / 1024); - DML_LOG_VERBOSE("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); - DML_LOG_VERBOSE("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); - DML_LOG_VERBOSE("DML::%s: TotalRowReadBandwidth = %f\n", __func__, l->TotalRowReadBandwidth); - DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); -#endif - - l->StutterBurstTime = l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer - / (p->ReturnBW * (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)) + - (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) - / math_min2(p->DCFCLK * 64, p->ReturnBW * (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)) + - *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Part 1 = %f\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / p->ReturnBW / (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)); - DML_LOG_VERBOSE("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); - DML_LOG_VERBOSE("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW); - DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); -#endif - l->TotalActiveWriteback = 0; - memset(l->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); - - for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { - if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { - if (!l->stream_visited[p->display_cfg->plane_descriptors[k].stream_index]) { - - if (p->display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) - l->TotalActiveWriteback = l->TotalActiveWriteback + 1; - - if (TotalNumberOfActiveOTG == 0) { // first otg - SinglePixelClock = ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - SingleHTotal = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; - SingleVTotal = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total; - } else if (SinglePixelClock != ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) || - SingleHTotal != p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total || - SingleVTotal != p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total) { - SameTiming = false; - } - TotalNumberOfActiveOTG = TotalNumberOfActiveOTG + 1; - l->stream_visited[p->display_cfg->plane_descriptors[k].stream_index] = 1; - } - } - } - - if (l->TotalActiveWriteback == 0) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); - DML_LOG_VERBOSE("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); - DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); -#endif - *p->StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitTime + l->StutterBurstTime) / *p->StutterPeriod) * 100; - *p->Z8StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitZ8Time + l->StutterBurstTime) / *p->StutterPeriod) * 100; - *p->NumberOfStutterBurstsPerFrame = (*p->StutterEfficiencyNotIncludingVBlank > 0 ? (unsigned int)(math_ceil2(l->VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); - *p->Z8NumberOfStutterBurstsPerFrame = (*p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? (unsigned int)(math_ceil2(l->VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); - } else { - *p->StutterEfficiencyNotIncludingVBlank = 0.; - *p->Z8StutterEfficiencyNotIncludingVBlank = 0.; - *p->NumberOfStutterBurstsPerFrame = 0; - *p->Z8NumberOfStutterBurstsPerFrame = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, l->VActiveTimeCriticalSurface); - DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); - DML_LOG_VERBOSE("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); - DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); -#endif - - if (*p->StutterEfficiencyNotIncludingVBlank > 0) { - if (!((p->SynchronizeTimings || TotalNumberOfActiveOTG == 1) && SameTiming)) { - *p->StutterEfficiency = *p->StutterEfficiencyNotIncludingVBlank; - } else { - *p->StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitTime + l->StutterBurstTime * l->VActiveTimeCriticalSurface / *p->StutterPeriod) / l->FrameTimeCriticalSurface) * 100; - } - } else { - *p->StutterEfficiency = 0; - *p->NumberOfStutterBurstsPerFrame = 0; - } - - if (*p->Z8StutterEfficiencyNotIncludingVBlank > 0) { - //LastZ8StutterPeriod = l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; - if (!((p->SynchronizeTimings || TotalNumberOfActiveOTG == 1) && SameTiming)) { - *p->Z8StutterEfficiency = *p->Z8StutterEfficiencyNotIncludingVBlank; - } else { - *p->Z8StutterEfficiency = (1 - (*p->Z8NumberOfStutterBurstsPerFrame * p->SRExitZ8Time + l->StutterBurstTime * l->VActiveTimeCriticalSurface / *p->StutterPeriod) / l->FrameTimeCriticalSurface) * 100; - } - } else { - *p->Z8StutterEfficiency = 0.; - *p->Z8NumberOfStutterBurstsPerFrame = 0; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: TotalNumberOfActiveOTG = %u\n", __func__, TotalNumberOfActiveOTG); - DML_LOG_VERBOSE("DML::%s: SameTiming = %u\n", __func__, SameTiming); - DML_LOG_VERBOSE("DML::%s: SynchronizeTimings = %u\n", __func__, p->SynchronizeTimings); - DML_LOG_VERBOSE("DML::%s: LastZ8StutterPeriod = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod : 0); - DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); - DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); - DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); - DML_LOG_VERBOSE("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); - DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); - DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); - DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); -#endif - - *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = !(!p->UnboundedRequestEnabled && (p->NumberOfActiveSurfaces == 1) && l->SinglePlaneCriticalSurface && l->SinglePipeCriticalSurface); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, l->DETBufferSizeYCriticalSurface); - DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); - DML_LOG_VERBOSE("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); -#endif -} - -static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex *in_out_params) -{ - const struct dml2_display_cfg *display_cfg = in_out_params->in_display_cfg; - const struct dml2_mcg_min_clock_table *min_clk_table = in_out_params->min_clk_table; - const struct core_display_cfg_support_info *cfg_support_info = in_out_params->cfg_support_info; - struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; - struct dml2_display_cfg_programming *programming = in_out_params->programming; - - struct dml2_core_calcs_mode_programming_locals *s = &mode_lib->scratch.dml_core_mode_programming_locals; - struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; - struct dml2_core_calcs_CalculateVMRowAndSwath_params *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; - struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; - struct dml2_core_calcs_CalculateStutterEfficiency_params *CalculateStutterEfficiency_params = &mode_lib->scratch.CalculateStutterEfficiency_params; - struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; - struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; - struct dml2_core_calcs_calculate_mcache_setting_params *calculate_mcache_setting_params = &mode_lib->scratch.calculate_mcache_setting_params; - struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; - struct dml2_core_shared_CalculateMetaAndPTETimes_params *CalculateMetaAndPTETimes_params = &mode_lib->scratch.CalculateMetaAndPTETimes_params; - struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; - struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *calculate_bytes_to_fetch_required_to_hide_latency_params = &mode_lib->scratch.calculate_bytes_to_fetch_required_to_hide_latency_params; - - unsigned int k; - bool must_support_iflip; - const long min_return_uclk_cycles = 83; - const long min_return_fclk_cycles = 75; - const double max_fclk_mhz = min_clk_table->max_clocks_khz.fclk / 1000.0; - double hard_minimum_dcfclk_mhz = (double)min_clk_table->dram_bw_table.entries[0].min_dcfclk_khz / 1000.0; - double max_uclk_mhz = 0; - double min_return_latency_in_DCFCLK_cycles = 0; - - DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); - - memset(&mode_lib->scratch, 0, sizeof(struct dml2_core_internal_scratch)); - memset(&mode_lib->mp, 0, sizeof(struct dml2_core_internal_mode_program)); - - s->num_active_planes = display_cfg->num_planes; - get_stream_output_bpp(s->OutputBpp, display_cfg); - - mode_lib->mp.num_active_pipes = dml_get_num_active_pipes(display_cfg->num_planes, cfg_support_info); - dml_calc_pipe_plane_mapping(cfg_support_info, mode_lib->mp.pipe_plane); - - mode_lib->mp.Dcfclk = programming->min_clocks.dcn4x.active.dcfclk_khz / 1000.0; - mode_lib->mp.FabricClock = programming->min_clocks.dcn4x.active.fclk_khz / 1000.0; - mode_lib->mp.dram_bw_mbps = uclk_khz_to_dram_bw_mbps(programming->min_clocks.dcn4x.active.uclk_khz, &mode_lib->soc.clk_table.dram_config); - mode_lib->mp.uclk_freq_mhz = programming->min_clocks.dcn4x.active.uclk_khz / 1000.0; - mode_lib->mp.GlobalDPPCLK = programming->min_clocks.dcn4x.dpprefclk_khz / 1000.0; - s->SOCCLK = (double)programming->min_clocks.dcn4x.socclk_khz / 1000; - mode_lib->mp.qos_param_index = get_qos_param_index(programming->min_clocks.dcn4x.active.uclk_khz, mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params); - mode_lib->mp.active_min_uclk_dpm_index = get_active_min_uclk_dpm_index(programming->min_clocks.dcn4x.active.uclk_khz, &mode_lib->soc.clk_table); - - for (k = 0; k < s->num_active_planes; ++k) { - unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; - DML_ASSERT(cfg_support_info->stream_support_info[stream_index].odms_used <= 4); - DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4 || - cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 2 || - cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); - - if (cfg_support_info->stream_support_info[stream_index].odms_used > 1) - DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); - - switch (cfg_support_info->stream_support_info[stream_index].odms_used) { - case (4): - mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_4to1; - break; - case (3): - mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_3to1; - break; - case (2): - mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_2to1; - break; - default: - if (cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4) - mode_lib->mp.ODMMode[k] = dml2_odm_mode_mso_1to4; - else if (cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 2) - mode_lib->mp.ODMMode[k] = dml2_odm_mode_mso_1to2; - else - mode_lib->mp.ODMMode[k] = dml2_odm_mode_bypass; - break; - } - } - - for (k = 0; k < s->num_active_planes; ++k) { - mode_lib->mp.NoOfDPP[k] = cfg_support_info->plane_support_info[k].dpps_used; - mode_lib->mp.Dppclk[k] = programming->plane_programming[k].min_clocks.dcn4x.dppclk_khz / 1000.0; - DML_ASSERT(mode_lib->mp.Dppclk[k] > 0); - } - - for (k = 0; k < s->num_active_planes; ++k) { - unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; - mode_lib->mp.DSCCLK[k] = programming->stream_programming[stream_index].min_clocks.dcn4x.dscclk_khz / 1000.0; - DML_LOG_VERBOSE("DML::%s: k=%d stream_index=%d, mode_lib->mp.DSCCLK = %f\n", __func__, k, stream_index, mode_lib->mp.DSCCLK[k]); - } - - mode_lib->mp.Dispclk = programming->min_clocks.dcn4x.dispclk_khz / 1000.0; - mode_lib->mp.DCFCLKDeepSleep = programming->min_clocks.dcn4x.deepsleep_dcfclk_khz / 1000.0; - - DML_ASSERT(mode_lib->mp.Dcfclk > 0); - DML_ASSERT(mode_lib->mp.FabricClock > 0); - DML_ASSERT(mode_lib->mp.dram_bw_mbps > 0); - DML_ASSERT(mode_lib->mp.uclk_freq_mhz > 0); - DML_ASSERT(mode_lib->mp.GlobalDPPCLK > 0); - DML_ASSERT(mode_lib->mp.Dispclk > 0); - DML_ASSERT(mode_lib->mp.DCFCLKDeepSleep > 0); - DML_ASSERT(s->SOCCLK > 0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, s->num_active_planes); - DML_LOG_VERBOSE("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); - DML_LOG_VERBOSE("DML::%s: Dcfclk = %f\n", __func__, mode_lib->mp.Dcfclk); - DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->mp.FabricClock); - DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->mp.dram_bw_mbps); - DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->mp.uclk_freq_mhz); - DML_LOG_VERBOSE("DML::%s: Dispclk = %f\n", __func__, mode_lib->mp.Dispclk); - for (k = 0; k < s->num_active_planes; ++k) { - DML_LOG_VERBOSE("DML::%s: Dppclk[%0d] = %f\n", __func__, k, mode_lib->mp.Dppclk[k]); - } - DML_LOG_VERBOSE("DML::%s: GlobalDPPCLK = %f\n", __func__, mode_lib->mp.GlobalDPPCLK); - DML_LOG_VERBOSE("DML::%s: DCFCLKDeepSleep = %f\n", __func__, mode_lib->mp.DCFCLKDeepSleep); - DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, s->SOCCLK); - DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); - DML_LOG_VERBOSE("DML::%s: min_clk_table min_fclk_khz = %ld\n", __func__, min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz); - DML_LOG_VERBOSE("DML::%s: min_clk_table uclk_mhz = %f\n", __func__, dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config)); - for (k = 0; k < mode_lib->mp.num_active_pipes; ++k) { - DML_LOG_VERBOSE("DML::%s: pipe=%d is in plane=%d\n", __func__, k, mode_lib->mp.pipe_plane[k]); - DML_LOG_VERBOSE("DML::%s: Per-plane DPPPerSurface[%0d] = %d\n", __func__, k, mode_lib->mp.NoOfDPP[k]); - } - - for (k = 0; k < s->num_active_planes; k++) - DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); -#endif - - CalculateMaxDETAndMinCompressedBufferSize( - mode_lib->ip.config_return_buffer_size_in_kbytes, - mode_lib->ip.config_return_buffer_segment_size_in_kbytes, - mode_lib->ip.rob_buffer_size_kbytes, - mode_lib->ip.max_num_dpp, - display_cfg->overrides.hw.force_nom_det_size_kbytes.enable, - display_cfg->overrides.hw.force_nom_det_size_kbytes.value, - mode_lib->ip.dcn_mrq_present, - - /* Output */ - &s->MaxTotalDETInKByte, - &s->NomDETInKByte, - &s->MinCompressedBufferSizeInKByte); - - - PixelClockAdjustmentForProgressiveToInterlaceUnit(display_cfg, mode_lib->ip.ptoi_supported, s->PixelClockBackEnd); - - for (k = 0; k < s->num_active_planes; ++k) { - CalculateSinglePipeDPPCLKAndSCLThroughput( - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->ip.max_dchub_pscl_bw_pix_per_clk, - mode_lib->ip.max_pscl_lb_bw_pix_per_clk, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps, - - /* Output */ - &mode_lib->mp.PSCL_THROUGHPUT[k], - &mode_lib->mp.PSCL_THROUGHPUT_CHROMA[k], - &mode_lib->mp.DPPCLKUsingSingleDPP[k]); - } - - for (k = 0; k < s->num_active_planes; ++k) { - CalculateBytePerPixelAndBlockSizes( - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->plane_descriptors[k].surface.tiling, - display_cfg->plane_descriptors[k].surface.plane0.pitch, - display_cfg->plane_descriptors[k].surface.plane1.pitch, - - // Output - &mode_lib->mp.BytePerPixelY[k], - &mode_lib->mp.BytePerPixelC[k], - &mode_lib->mp.BytePerPixelInDETY[k], - &mode_lib->mp.BytePerPixelInDETC[k], - &mode_lib->mp.Read256BlockHeightY[k], - &mode_lib->mp.Read256BlockHeightC[k], - &mode_lib->mp.Read256BlockWidthY[k], - &mode_lib->mp.Read256BlockWidthC[k], - &mode_lib->mp.MacroTileHeightY[k], - &mode_lib->mp.MacroTileHeightC[k], - &mode_lib->mp.MacroTileWidthY[k], - &mode_lib->mp.MacroTileWidthC[k], - &mode_lib->mp.surf_linear128_l[k], - &mode_lib->mp.surf_linear128_c[k]); - } - - CalculateSwathWidth( - display_cfg, - false, // ForceSingleDPP - s->num_active_planes, - mode_lib->mp.ODMMode, - mode_lib->mp.BytePerPixelY, - mode_lib->mp.BytePerPixelC, - mode_lib->mp.Read256BlockHeightY, - mode_lib->mp.Read256BlockHeightC, - mode_lib->mp.Read256BlockWidthY, - mode_lib->mp.Read256BlockWidthC, - mode_lib->mp.surf_linear128_l, - mode_lib->mp.surf_linear128_c, - mode_lib->mp.NoOfDPP, - - /* Output */ - mode_lib->mp.req_per_swath_ub_l, - mode_lib->mp.req_per_swath_ub_c, - mode_lib->mp.SwathWidthSingleDPPY, - mode_lib->mp.SwathWidthSingleDPPC, - mode_lib->mp.SwathWidthY, - mode_lib->mp.SwathWidthC, - s->dummy_integer_array[0], // unsigned int MaximumSwathHeightY[] - s->dummy_integer_array[1], // unsigned int MaximumSwathHeightC[] - mode_lib->mp.swath_width_luma_ub, - mode_lib->mp.swath_width_chroma_ub); - - for (k = 0; k < s->num_active_planes; ++k) { - mode_lib->mp.cursor_bw[k] = display_cfg->plane_descriptors[k].cursor.num_cursors * display_cfg->plane_descriptors[k].cursor.cursor_width * display_cfg->plane_descriptors[k].cursor.cursor_bpp / 8.0 / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); - mode_lib->mp.vactive_sw_bw_l[k] = mode_lib->mp.SwathWidthSingleDPPY[k] * mode_lib->mp.BytePerPixelY[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - mode_lib->mp.vactive_sw_bw_c[k] = mode_lib->mp.SwathWidthSingleDPPC[k] * mode_lib->mp.BytePerPixelC[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); - } - - CalculateSwathAndDETConfiguration_params->display_cfg = display_cfg; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ip.config_return_buffer_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = s->MaxTotalDETInKByte; - CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = s->MinCompressedBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; - CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; - CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = s->num_active_planes; - CalculateSwathAndDETConfiguration_params->nomDETInKByte = s->NomDETInKByte; - CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ip.config_return_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByte = mode_lib->ip.compressed_buffer_segment_size_in_kbytes; - CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->mp.vactive_sw_bw_l; - CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->mp.vactive_sw_bw_c; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = s->dummy_single_array[0]; - CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = s->dummy_single_array[1]; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->mp.Read256BlockHeightY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->mp.Read256BlockHeightC; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->mp.Read256BlockWidthY; - CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->mp.Read256BlockWidthC; - CalculateSwathAndDETConfiguration_params->surf_linear128_l = mode_lib->mp.surf_linear128_l; - CalculateSwathAndDETConfiguration_params->surf_linear128_c = mode_lib->mp.surf_linear128_c; - CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->mp.ODMMode; - CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->mp.NoOfDPP; - CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->mp.BytePerPixelY; - CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->mp.BytePerPixelC; - CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->mp.BytePerPixelInDETY; - CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->mp.BytePerPixelInDETC; - CalculateSwathAndDETConfiguration_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - // output - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = mode_lib->mp.req_per_swath_ub_l; - CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = mode_lib->mp.req_per_swath_ub_c; - CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_long_array[0]; - CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_long_array[1]; - CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_long_array[2]; - CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_long_array[3]; - CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->mp.SwathHeightY; - CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->mp.SwathHeightC; - CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = mode_lib->mp.request_size_bytes_luma; - CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = mode_lib->mp.request_size_bytes_chroma; - CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->mp.DETBufferSizeInKByte; - CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; - CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->mp.DETBufferSizeC; - CalculateSwathAndDETConfiguration_params->full_swath_bytes_l = s->full_swath_bytes_l; - CalculateSwathAndDETConfiguration_params->full_swath_bytes_c = s->full_swath_bytes_c; - CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->mp.UnboundedRequestEnabled; - CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &mode_lib->mp.compbuf_reserved_space_64b; - CalculateSwathAndDETConfiguration_params->hw_debug5 = &mode_lib->mp.hw_debug5; - CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->mp.CompressedBufferSizeInkByte; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = &s->dummy_boolean_array[0][0]; - CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[0]; - - // Calculate DET size, swath height here. - CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); - - // DSC Delay - for (k = 0; k < s->num_active_planes; ++k) { - mode_lib->mp.DSCDelay[k] = DSCDelayRequirement(cfg_support_info->stream_support_info[display_cfg->plane_descriptors[k].stream_index].dsc_enable, - mode_lib->mp.ODMMode[k], - mode_lib->ip.maximum_dsc_bits_per_component, - s->OutputBpp[k], - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, - cfg_support_info->stream_support_info[display_cfg->plane_descriptors[k].stream_index].num_dsc_slices, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - s->PixelClockBackEnd[k]); - } - - // Prefetch - if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0) { - for (k = 0; k < s->num_active_planes; ++k) - mode_lib->mp.SurfaceSizeInTheMALL[k] = 0; - } else { - CalculateSurfaceSizeInMall( - display_cfg, - s->num_active_planes, - mode_lib->soc.mall_allocated_for_dcn_mbytes, - mode_lib->mp.BytePerPixelY, - mode_lib->mp.BytePerPixelC, - mode_lib->mp.Read256BlockWidthY, - mode_lib->mp.Read256BlockWidthC, - mode_lib->mp.Read256BlockHeightY, - mode_lib->mp.Read256BlockHeightC, - mode_lib->mp.MacroTileWidthY, - mode_lib->mp.MacroTileWidthC, - mode_lib->mp.MacroTileHeightY, - mode_lib->mp.MacroTileHeightC, - - /* Output */ - mode_lib->mp.SurfaceSizeInTheMALL, - &s->dummy_boolean[0]); /* bool *ExceededMALLSize */ - } - - for (k = 0; k < s->num_active_planes; ++k) { - s->SurfaceParameters[k].PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - s->SurfaceParameters[k].DPPPerSurface = mode_lib->mp.NoOfDPP[k]; - s->SurfaceParameters[k].RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; - s->SurfaceParameters[k].ViewportHeight = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - s->SurfaceParameters[k].ViewportHeightC = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; - s->SurfaceParameters[k].BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY[k]; - s->SurfaceParameters[k].BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY[k]; - s->SurfaceParameters[k].BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC[k]; - s->SurfaceParameters[k].BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC[k]; - s->SurfaceParameters[k].BlockWidthY = mode_lib->mp.MacroTileWidthY[k]; - s->SurfaceParameters[k].BlockHeightY = mode_lib->mp.MacroTileHeightY[k]; - s->SurfaceParameters[k].BlockWidthC = mode_lib->mp.MacroTileWidthC[k]; - s->SurfaceParameters[k].BlockHeightC = mode_lib->mp.MacroTileHeightC[k]; - s->SurfaceParameters[k].InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; - s->SurfaceParameters[k].HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; - s->SurfaceParameters[k].DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; - s->SurfaceParameters[k].SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; - s->SurfaceParameters[k].SurfaceTiling = display_cfg->plane_descriptors[k].surface.tiling; - s->SurfaceParameters[k].BytePerPixelY = mode_lib->mp.BytePerPixelY[k]; - s->SurfaceParameters[k].BytePerPixelC = mode_lib->mp.BytePerPixelC[k]; - s->SurfaceParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - s->SurfaceParameters[k].VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - s->SurfaceParameters[k].VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - s->SurfaceParameters[k].VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - s->SurfaceParameters[k].VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - s->SurfaceParameters[k].PitchY = display_cfg->plane_descriptors[k].surface.plane0.pitch; - s->SurfaceParameters[k].PitchC = display_cfg->plane_descriptors[k].surface.plane1.pitch; - s->SurfaceParameters[k].ViewportStationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; - s->SurfaceParameters[k].ViewportXStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; - s->SurfaceParameters[k].ViewportYStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; - s->SurfaceParameters[k].ViewportXStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - s->SurfaceParameters[k].ViewportYStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - s->SurfaceParameters[k].FORCE_ONE_ROW_FOR_FRAME = display_cfg->plane_descriptors[k].overrides.hw.force_one_row_for_frame; - s->SurfaceParameters[k].SwathHeightY = mode_lib->mp.SwathHeightY[k]; - s->SurfaceParameters[k].SwathHeightC = mode_lib->mp.SwathHeightC[k]; - s->SurfaceParameters[k].DCCMetaPitchY = display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch; - s->SurfaceParameters[k].DCCMetaPitchC = display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch; - } - - CalculateVMRowAndSwath_params->display_cfg = display_cfg; - CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = s->num_active_planes; - CalculateVMRowAndSwath_params->myPipe = s->SurfaceParameters; - CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->mp.SurfaceSizeInTheMALL; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; - CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma; - CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->soc.mall_allocated_for_dcn_mbytes; - CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->mp.SwathWidthY; - CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->mp.SwathWidthC; - CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ip.dcc_meta_buffer_size_bytes; - CalculateVMRowAndSwath_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - // output - CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = s->dummy_boolean_array[0]; - CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = mode_lib->mp.dpte_row_width_luma_ub; - CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = mode_lib->mp.dpte_row_width_chroma_ub; - CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->mp.dpte_row_height; - CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->mp.dpte_row_height_chroma; - CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = mode_lib->mp.dpte_row_height_linear; - CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = mode_lib->mp.dpte_row_height_linear_chroma; - CalculateVMRowAndSwath_params->vm_group_bytes = mode_lib->mp.vm_group_bytes; - CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; - CalculateVMRowAndSwath_params->PixelPTEReqWidthY = mode_lib->mp.PixelPTEReqWidthY; - CalculateVMRowAndSwath_params->PixelPTEReqHeightY = mode_lib->mp.PixelPTEReqHeightY; - CalculateVMRowAndSwath_params->PTERequestSizeY = mode_lib->mp.PTERequestSizeY; - CalculateVMRowAndSwath_params->PixelPTEReqWidthC = mode_lib->mp.PixelPTEReqWidthC; - CalculateVMRowAndSwath_params->PixelPTEReqHeightC = mode_lib->mp.PixelPTEReqHeightC; - CalculateVMRowAndSwath_params->PTERequestSizeC = mode_lib->mp.PTERequestSizeC; - CalculateVMRowAndSwath_params->vmpg_width_y = s->vmpg_width_y; - CalculateVMRowAndSwath_params->vmpg_height_y = s->vmpg_height_y; - CalculateVMRowAndSwath_params->vmpg_width_c = s->vmpg_width_c; - CalculateVMRowAndSwath_params->vmpg_height_c = s->vmpg_height_c; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = mode_lib->mp.dpde0_bytes_per_frame_ub_l; - CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = mode_lib->mp.dpde0_bytes_per_frame_ub_c; - CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->mp.PrefetchSourceLinesY; - CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->mp.PrefetchSourceLinesC; - CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->mp.VInitPreFillY; - CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->mp.VInitPreFillC; - CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->mp.MaxNumSwathY; - CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->mp.MaxNumSwathC; - CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; - CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->mp.PixelPTEBytesPerRow; - CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_l = s->dpte_row_bytes_per_row_l; - CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_c = s->dpte_row_bytes_per_row_c; - CalculateVMRowAndSwath_params->vm_bytes = mode_lib->mp.vm_bytes; - CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->mp.use_one_row_for_frame; - CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->mp.use_one_row_for_frame_flip; - CalculateVMRowAndSwath_params->is_using_mall_for_ss = mode_lib->mp.is_using_mall_for_ss; - CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = mode_lib->mp.PTE_BUFFER_MODE; - CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = mode_lib->mp.BIGK_FRAGMENT_SIZE; - CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = s->dummy_boolean_array[1]; - CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->mp.meta_row_bw; - CalculateVMRowAndSwath_params->meta_row_bytes = mode_lib->mp.meta_row_bytes; - CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; - CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; - CalculateVMRowAndSwath_params->meta_req_width_luma = mode_lib->mp.meta_req_width; - CalculateVMRowAndSwath_params->meta_req_height_luma = mode_lib->mp.meta_req_height; - CalculateVMRowAndSwath_params->meta_row_width_luma = mode_lib->mp.meta_row_width; - CalculateVMRowAndSwath_params->meta_row_height_luma = mode_lib->mp.meta_row_height; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = mode_lib->mp.meta_pte_bytes_per_frame_ub_l; - CalculateVMRowAndSwath_params->meta_req_width_chroma = mode_lib->mp.meta_req_width_chroma; - CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->mp.meta_row_height_chroma; - CalculateVMRowAndSwath_params->meta_row_width_chroma = mode_lib->mp.meta_row_width_chroma; - CalculateVMRowAndSwath_params->meta_req_height_chroma = mode_lib->mp.meta_req_height_chroma; - CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = mode_lib->mp.meta_pte_bytes_per_frame_ub_c; - - CalculateVMRowAndSwath(&mode_lib->scratch, CalculateVMRowAndSwath_params); - - memset(calculate_mcache_setting_params, 0, sizeof(struct dml2_core_calcs_calculate_mcache_setting_params)); - if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0 || mode_lib->ip.dcn_mrq_present) { - for (k = 0; k < s->num_active_planes; k++) { - mode_lib->mp.mall_prefetch_sdp_overhead_factor[k] = 1.0; - mode_lib->mp.mall_prefetch_dram_overhead_factor[k] = 1.0; - mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0[k] = 1.0; - mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0[k] = 1.0; - mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1[k] = 1.0; - mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1[k] = 1.0; - } - } else { - for (k = 0; k < s->num_active_planes; k++) { - calculate_mcache_setting_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; - calculate_mcache_setting_params->num_chans = mode_lib->soc.clk_table.dram_config.channel_count; - calculate_mcache_setting_params->mem_word_bytes = mode_lib->soc.mem_word_bytes; - calculate_mcache_setting_params->mcache_size_bytes = mode_lib->soc.mcache_size_bytes; - calculate_mcache_setting_params->mcache_line_size_bytes = mode_lib->soc.mcache_line_size_bytes; - calculate_mcache_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; - calculate_mcache_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - - calculate_mcache_setting_params->source_format = display_cfg->plane_descriptors[k].pixel_format; - calculate_mcache_setting_params->surf_vert = dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle); - calculate_mcache_setting_params->vp_stationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; - calculate_mcache_setting_params->tiling_mode = display_cfg->plane_descriptors[k].surface.tiling; - calculate_mcache_setting_params->imall_enable = mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall; - - calculate_mcache_setting_params->vp_start_x_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; - calculate_mcache_setting_params->vp_start_y_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; - calculate_mcache_setting_params->full_vp_width_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; - calculate_mcache_setting_params->full_vp_height_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; - calculate_mcache_setting_params->blk_width_l = mode_lib->mp.MacroTileWidthY[k]; - calculate_mcache_setting_params->blk_height_l = mode_lib->mp.MacroTileHeightY[k]; - calculate_mcache_setting_params->vmpg_width_l = s->vmpg_width_y[k]; - calculate_mcache_setting_params->vmpg_height_l = s->vmpg_height_y[k]; - calculate_mcache_setting_params->full_swath_bytes_l = s->full_swath_bytes_l[k]; - calculate_mcache_setting_params->bytes_per_pixel_l = mode_lib->mp.BytePerPixelY[k]; - - calculate_mcache_setting_params->vp_start_x_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - calculate_mcache_setting_params->vp_start_y_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; - calculate_mcache_setting_params->full_vp_width_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; - calculate_mcache_setting_params->full_vp_height_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; - calculate_mcache_setting_params->blk_width_c = mode_lib->mp.MacroTileWidthC[k]; - calculate_mcache_setting_params->blk_height_c = mode_lib->mp.MacroTileHeightC[k]; - calculate_mcache_setting_params->vmpg_width_c = s->vmpg_width_c[k]; - calculate_mcache_setting_params->vmpg_height_c = s->vmpg_height_c[k]; - calculate_mcache_setting_params->full_swath_bytes_c = s->full_swath_bytes_c[k]; - calculate_mcache_setting_params->bytes_per_pixel_c = mode_lib->mp.BytePerPixelC[k]; - - // output - calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_l = &mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0[k]; - calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_l = &mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0[k]; - calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_c = &mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1[k]; - calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_c = &mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1[k]; - - calculate_mcache_setting_params->num_mcaches_l = &mode_lib->mp.num_mcaches_l[k]; - calculate_mcache_setting_params->mcache_row_bytes_l = &mode_lib->mp.mcache_row_bytes_l[k]; - calculate_mcache_setting_params->mcache_row_bytes_per_channel_l = &mode_lib->mp.mcache_row_bytes_per_channel_l[k]; - calculate_mcache_setting_params->mcache_offsets_l = mode_lib->mp.mcache_offsets_l[k]; - calculate_mcache_setting_params->mcache_shift_granularity_l = &mode_lib->mp.mcache_shift_granularity_l[k]; - - calculate_mcache_setting_params->num_mcaches_c = &mode_lib->mp.num_mcaches_c[k]; - calculate_mcache_setting_params->mcache_row_bytes_c = &mode_lib->mp.mcache_row_bytes_c[k]; - calculate_mcache_setting_params->mcache_row_bytes_per_channel_c = &mode_lib->mp.mcache_row_bytes_per_channel_c[k]; - calculate_mcache_setting_params->mcache_offsets_c = mode_lib->mp.mcache_offsets_c[k]; - calculate_mcache_setting_params->mcache_shift_granularity_c = &mode_lib->mp.mcache_shift_granularity_c[k]; - - calculate_mcache_setting_params->mall_comb_mcache_l = &mode_lib->mp.mall_comb_mcache_l[k]; - calculate_mcache_setting_params->mall_comb_mcache_c = &mode_lib->mp.mall_comb_mcache_c[k]; - calculate_mcache_setting_params->lc_comb_mcache = &mode_lib->mp.lc_comb_mcache[k]; - calculate_mcache_setting(&mode_lib->scratch, calculate_mcache_setting_params); - } - - calculate_mall_bw_overhead_factor( - mode_lib->mp.mall_prefetch_sdp_overhead_factor, - mode_lib->mp.mall_prefetch_dram_overhead_factor, - - // input - display_cfg, - s->num_active_planes); - } - - // Calculate all the bandwidth availabe - calculate_bandwidth_available( - mode_lib->mp.avg_bandwidth_available_min, - mode_lib->mp.avg_bandwidth_available, - mode_lib->mp.urg_bandwidth_available_min, - mode_lib->mp.urg_bandwidth_available, - mode_lib->mp.urg_bandwidth_available_vm_only, - mode_lib->mp.urg_bandwidth_available_pixel_and_vm, - - &mode_lib->soc, - display_cfg->hostvm_enable, - mode_lib->mp.Dcfclk, - mode_lib->mp.FabricClock, - mode_lib->mp.dram_bw_mbps); - - - calculate_hostvm_inefficiency_factor( - &s->HostVMInefficiencyFactor, - &s->HostVMInefficiencyFactorPrefetch, - - display_cfg->gpuvm_enable, - display_cfg->hostvm_enable, - mode_lib->ip.remote_iommu_outstanding_translations, - mode_lib->soc.max_outstanding_reqs, - mode_lib->mp.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], - mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); - - s->TotalDCCActiveDPP = 0; - s->TotalActiveDPP = 0; - for (k = 0; k < s->num_active_planes; ++k) { - s->TotalActiveDPP = s->TotalActiveDPP + mode_lib->mp.NoOfDPP[k]; - if (display_cfg->plane_descriptors[k].surface.dcc.enable) - s->TotalDCCActiveDPP = s->TotalDCCActiveDPP + mode_lib->mp.NoOfDPP[k]; - } - // Calculate tdlut schedule related terms - for (k = 0; k <= s->num_active_planes - 1; k++) { - calculate_tdlut_setting_params->dispclk_mhz = mode_lib->mp.Dispclk; - calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; - calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; - calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; - calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; - calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; - - // output - calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; - calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; - calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; - calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; - calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; - calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; - calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); - } - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) - s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); - - CalculateExtraLatency( - display_cfg, - mode_lib->ip.rob_buffer_size_kbytes, - mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, - s->ReorderingBytes, - mode_lib->mp.Dcfclk, - mode_lib->mp.FabricClock, - mode_lib->ip.pixel_chunk_size_kbytes, - mode_lib->mp.urg_bandwidth_available_min[dml2_core_internal_soc_state_sys_active], - s->num_active_planes, - mode_lib->mp.NoOfDPP, - mode_lib->mp.dpte_group_bytes, - s->tdlut_bytes_per_group, - s->HostVMInefficiencyFactor, - s->HostVMInefficiencyFactorPrefetch, - mode_lib->soc.hostvm_min_page_size_kbytes, - mode_lib->soc.qos_parameters.qos_type, - !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), - mode_lib->soc.max_outstanding_reqs, - mode_lib->mp.request_size_bytes_luma, - mode_lib->mp.request_size_bytes_chroma, - mode_lib->ip.meta_chunk_size_kbytes, - mode_lib->ip.dchub_arb_to_ret_delay, - mode_lib->mp.TripToMemory, - mode_lib->ip.hostvm_mode, - - // output - &mode_lib->mp.ExtraLatency, - &mode_lib->mp.ExtraLatency_sr, - &mode_lib->mp.ExtraLatencyPrefetch); - - mode_lib->mp.TCalc = 24.0 / mode_lib->mp.DCFCLKDeepSleep; - - for (k = 0; k < s->num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - mode_lib->mp.WritebackDelay[k] = - mode_lib->soc.qos_parameters.writeback.base_latency_us - + CalculateWriteBackDelay( - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) / mode_lib->mp.Dispclk; - } else - mode_lib->mp.WritebackDelay[k] = 0; - } - - /* VActive bytes to fetch for UCLK P-State */ - calculate_bytes_to_fetch_required_to_hide_latency_params->display_cfg = display_cfg; - calculate_bytes_to_fetch_required_to_hide_latency_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - calculate_bytes_to_fetch_required_to_hide_latency_params->num_active_planes = s->num_active_planes; - calculate_bytes_to_fetch_required_to_hide_latency_params->num_of_dpp = mode_lib->mp.NoOfDPP; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_l = mode_lib->mp.meta_row_height; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_c = mode_lib->mp.meta_row_height_chroma; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_l = mode_lib->mp.dpte_row_height; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_c = mode_lib->mp.dpte_row_height_chroma; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_l = s->dpte_row_bytes_per_row_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_c = s->dpte_row_bytes_per_row_c; - calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_l = mode_lib->mp.BytePerPixelY; - calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_c = mode_lib->mp.BytePerPixelC; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_l = mode_lib->mp.SwathWidthY; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->mp.SwathWidthC; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->mp.SwathHeightY; - calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->mp.SwathHeightC; - calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; - - /* outputs */ - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; - - calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); - - /* Excess VActive bandwidth required to fill DET */ - calculate_excess_vactive_bandwidth_required( - display_cfg, - s->num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, - /* outputs */ - mode_lib->mp.excess_vactive_fill_bw_l, - mode_lib->mp.excess_vactive_fill_bw_c); - - mode_lib->mp.UrgentLatency = CalculateUrgentLatency( - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_pixel_vm_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_vm_us, - mode_lib->soc.do_urgent_latency_adjustment, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_fclk_us, - mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_mhz, - mode_lib->mp.FabricClock, - mode_lib->mp.uclk_freq_mhz, - mode_lib->soc.qos_parameters.qos_type, - mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].urgent_ramp_uclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_urgent_ramp_latency_margin, - mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); - - mode_lib->mp.TripToMemory = CalculateTripToMemory( - mode_lib->mp.UrgentLatency, - mode_lib->mp.FabricClock, - mode_lib->mp.uclk_freq_mhz, - mode_lib->soc.qos_parameters.qos_type, - mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].trip_to_memory_uclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, - mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); - - mode_lib->mp.TripToMemory = math_max2(mode_lib->mp.UrgentLatency, mode_lib->mp.TripToMemory); - - mode_lib->mp.MetaTripToMemory = CalculateMetaTripToMemory( - mode_lib->mp.UrgentLatency, - mode_lib->mp.FabricClock, - mode_lib->mp.uclk_freq_mhz, - mode_lib->soc.qos_parameters.qos_type, - mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].meta_trip_to_memory_uclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.meta_trip_adder_fclk_cycles, - mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, - mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); - - for (k = 0; k < s->num_active_planes; ++k) { - bool cursor_not_enough_urgent_latency_hiding = false; - s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - - s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; - - s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->mp.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane0.width, - display_cfg->plane_descriptors[k].composition.viewport.plane0.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, - mode_lib->mp.NoOfDPP[k], - display_cfg->plane_descriptors[k].composition.viewport.plane1.width, - display_cfg->plane_descriptors[k].composition.viewport.plane1.height, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, - display_cfg->plane_descriptors[k].composition.rotation_angle); - - if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { - calculate_cursor_req_attributes( - display_cfg->plane_descriptors[k].cursor.cursor_width, - display_cfg->plane_descriptors[k].cursor.cursor_bpp, - - // output - &s->cursor_lines_per_chunk[k], - &s->cursor_bytes_per_line[k], - &s->cursor_bytes_per_chunk[k], - &s->cursor_bytes[k]); - - calculate_cursor_urgent_burst_factor( - mode_lib->ip.cursor_buffer_size, - display_cfg->plane_descriptors[k].cursor.cursor_width, - s->cursor_bytes_per_chunk[k], - s->cursor_lines_per_chunk[k], - s->line_times[k], - mode_lib->mp.UrgentLatency, - - // output - &mode_lib->mp.UrgentBurstFactorCursor[k], - &cursor_not_enough_urgent_latency_hiding); - } - mode_lib->mp.UrgentBurstFactorCursorPre[k] = mode_lib->mp.UrgentBurstFactorCursor[k]; - - CalculateUrgentBurstFactor( - &display_cfg->plane_descriptors[k], - mode_lib->mp.swath_width_luma_ub[k], - mode_lib->mp.swath_width_chroma_ub[k], - mode_lib->mp.SwathHeightY[k], - mode_lib->mp.SwathHeightC[k], - s->line_times[k], - mode_lib->mp.UrgentLatency, - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->mp.BytePerPixelInDETY[k], - mode_lib->mp.BytePerPixelInDETC[k], - mode_lib->mp.DETBufferSizeY[k], - mode_lib->mp.DETBufferSizeC[k], - - /* output */ - &mode_lib->mp.UrgentBurstFactorLuma[k], - &mode_lib->mp.UrgentBurstFactorChroma[k], - &mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); - - mode_lib->mp.NotEnoughUrgentLatencyHiding[k] = mode_lib->mp.NotEnoughUrgentLatencyHiding[k] || cursor_not_enough_urgent_latency_hiding; - } - - for (k = 0; k < s->num_active_planes; ++k) { - s->MaxVStartupLines[k] = CalculateMaxVStartup( - mode_lib->ip.ptoi_supported, - mode_lib->ip.vblank_nom_default_us, - &display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing, - mode_lib->mp.WritebackDelay[k]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - DML_LOG_VERBOSE("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, mode_lib->mp.WritebackDelay[k]); -#endif - } - - s->immediate_flip_required = false; - for (k = 0; k < s->num_active_planes; ++k) { - s->immediate_flip_required = s->immediate_flip_required || display_cfg->plane_descriptors[k].immediate_flip; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: immediate_flip_required = %u\n", __func__, s->immediate_flip_required); -#endif - - if (s->num_active_planes > 1) { - CheckGlobalPrefetchAdmissibility_params->num_active_planes = s->num_active_planes; - CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; - CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; - CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->mp.SwathHeightY; - CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->mp.SwathHeightC; - CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; - CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->mp.CompressedBufferSizeInkByte; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->mp.DETBufferSizeY; - CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->mp.DETBufferSizeC; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; - CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; - CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; - CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = 0; // don't care - CheckGlobalPrefetchAdmissibility_params->Tpre_oto = 0; // don't care - CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; - CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = mode_lib->mp.Dcfclk; - CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; - CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->mp.dst_y_prefetch; - - // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible - CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->dummy_boolean[0]; - CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; - CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); // dont care about the check output for mode programming - } - - { - s->DestinationLineTimesForPrefetchLessThan2 = false; - s->VRatioPrefetchMoreThanMax = false; - - DML_LOG_VERBOSE("DML::%s: Start one iteration of prefetch schedule evaluation\n", __func__); - - for (k = 0; k < s->num_active_planes; ++k) { - struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; - - DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - mode_lib->mp.TWait[k] = CalculateTWait( - display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, - mode_lib->mp.UrgentLatency, - mode_lib->mp.TripToMemory, - !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? - get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->mp.uclk_freq_mhz * 1000), in_out_params->min_clk_index) : 0.0); - - myPipe->Dppclk = mode_lib->mp.Dppclk[k]; - myPipe->Dispclk = mode_lib->mp.Dispclk; - myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - myPipe->DCFClkDeepSleep = mode_lib->mp.DCFCLKDeepSleep; - myPipe->DPPPerSurface = mode_lib->mp.NoOfDPP[k]; - myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; - myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; - myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; - myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; - myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; - myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; - myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; - myPipe->BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY[k]; - myPipe->BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY[k]; - myPipe->BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC[k]; - myPipe->BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC[k]; - myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; - myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; - myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; - myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; - myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; - myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; - myPipe->ODMMode = mode_lib->mp.ODMMode[k]; - myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; - myPipe->BytePerPixelY = mode_lib->mp.BytePerPixelY[k]; - myPipe->BytePerPixelC = mode_lib->mp.BytePerPixelC[k]; - myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); -#endif - CalculatePrefetchSchedule_params->display_cfg = display_cfg; - CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; - CalculatePrefetchSchedule_params->myPipe = myPipe; - CalculatePrefetchSchedule_params->DSCDelay = mode_lib->mp.DSCDelay[k]; - CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; - CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; - CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; - CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; - CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; - CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->mp.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); - CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; - CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; - CalculatePrefetchSchedule_params->VStartup = s->MaxVStartupLines[k]; - CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; - CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; - CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; - CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; - CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; - CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->mp.UrgentLatency; - CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->mp.ExtraLatencyPrefetch; - CalculatePrefetchSchedule_params->TCalc = mode_lib->mp.TCalc; - CalculatePrefetchSchedule_params->vm_bytes = mode_lib->mp.vm_bytes[k]; - CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->mp.PixelPTEBytesPerRow[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->mp.PrefetchSourceLinesY[k]; - CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->mp.VInitPreFillY[k]; - CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->mp.MaxNumSwathY[k]; - CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->mp.PrefetchSourceLinesC[k]; - CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->mp.VInitPreFillC[k]; - CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->mp.MaxNumSwathC[k]; - CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->mp.swath_width_luma_ub[k]; - CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->mp.swath_width_chroma_ub[k]; - CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->mp.SwathHeightY[k]; - CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->mp.SwathHeightC[k]; - CalculatePrefetchSchedule_params->TWait = mode_lib->mp.TWait[k]; - CalculatePrefetchSchedule_params->Ttrip = mode_lib->mp.TripToMemory; - CalculatePrefetchSchedule_params->Turg = mode_lib->mp.UrgentLatency; - CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; - CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; - CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; - CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; - CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); - CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; - CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; - CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; - CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; - CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->mp.meta_row_bytes[k]; - CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor[k]; - CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->mp.vactive_sw_bw_l[k]; - CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->mp.vactive_sw_bw_c[k]; - - // output - CalculatePrefetchSchedule_params->DSTXAfterScaler = &mode_lib->mp.DSTXAfterScaler[k]; - CalculatePrefetchSchedule_params->DSTYAfterScaler = &mode_lib->mp.DSTYAfterScaler[k]; - CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->mp.dst_y_prefetch[k]; - CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->mp.dst_y_per_vm_vblank[k]; - CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->mp.dst_y_per_row_vblank[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->mp.VRatioPrefetchY[k]; - CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->mp.VRatioPrefetchC[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]; - CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &s->dummy_single_array[0][k]; - CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]; - CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->mp.Tno_bw[k]; - CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->mp.Tno_bw_flip[k]; - CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->mp.prefetch_vmrow_bw[k]; - CalculatePrefetchSchedule_params->Tdmdl_vm = &mode_lib->mp.Tdmdl_vm[k]; - CalculatePrefetchSchedule_params->Tdmdl = &mode_lib->mp.Tdmdl[k]; - CalculatePrefetchSchedule_params->TSetup = &mode_lib->mp.TSetup[k]; - CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; - CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; - CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; - CalculatePrefetchSchedule_params->VUpdateOffsetPix = &mode_lib->mp.VUpdateOffsetPix[k]; - CalculatePrefetchSchedule_params->VUpdateWidthPix = &mode_lib->mp.VUpdateWidthPix[k]; - CalculatePrefetchSchedule_params->VReadyOffsetPix = &mode_lib->mp.VReadyOffsetPix[k]; - CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->mp.prefetch_cursor_bw[k]; - CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; - CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; - CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; - CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->dummy_single[0]; - - mode_lib->mp.NoTimeToPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); - - if (s->impacted_dst_y_pre[k] > 0) - mode_lib->mp.impacted_prefetch_margin_us[k] = (mode_lib->mp.dst_y_prefetch[k] - s->impacted_dst_y_pre[k]) * s->line_times[k]; - else - mode_lib->mp.impacted_prefetch_margin_us[k] = 0; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); -#endif - mode_lib->mp.VStartupMin[k] = s->MaxVStartupLines[k]; - } // for k - - mode_lib->mp.PrefetchModeSupported = true; - for (k = 0; k < s->num_active_planes; ++k) { - if (mode_lib->mp.NoTimeToPrefetch[k] == true || - mode_lib->mp.NotEnoughTimeForDynamicMetadata[k] || - mode_lib->mp.DSTYAfterScaler[k] > 8) { - DML_LOG_VERBOSE("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, DSTYAfterScaler=%u (should be <= 0)\n", __func__, k, mode_lib->mp.DSTYAfterScaler[k]); - mode_lib->mp.PrefetchModeSupported = false; - } - if (mode_lib->mp.dst_y_prefetch[k] < 2) - s->DestinationLineTimesForPrefetchLessThan2 = true; - - if (mode_lib->mp.VRatioPrefetchY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || - mode_lib->mp.VRatioPrefetchC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { - s->VRatioPrefetchMoreThanMax = true; - DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchY=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k], __DML2_CALCS_MAX_VRATIO_PRE__); - DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchC=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchC[k], __DML2_CALCS_MAX_VRATIO_PRE__); - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); - } - - if (mode_lib->mp.NotEnoughUrgentLatencyHiding[k]) { - DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHiding = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); - mode_lib->mp.PrefetchModeSupported = false; - } - } - - if (s->VRatioPrefetchMoreThanMax == true || s->DestinationLineTimesForPrefetchLessThan2 == true) { - DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); - DML_LOG_VERBOSE("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); - mode_lib->mp.PrefetchModeSupported = false; - } - - DML_LOG_VERBOSE("DML::%s: Prefetch schedule is %sOK at vstartup = %u\n", __func__, - mode_lib->mp.PrefetchModeSupported ? "" : "NOT ", CalculatePrefetchSchedule_params->VStartup); - - // Prefetch schedule OK, now check prefetch bw - if (mode_lib->mp.PrefetchModeSupported == true) { - for (k = 0; k < s->num_active_planes; ++k) { - double line_time_us = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - CalculateUrgentBurstFactor( - &display_cfg->plane_descriptors[k], - mode_lib->mp.swath_width_luma_ub[k], - mode_lib->mp.swath_width_chroma_ub[k], - mode_lib->mp.SwathHeightY[k], - mode_lib->mp.SwathHeightC[k], - line_time_us, - mode_lib->mp.UrgentLatency, - mode_lib->mp.VRatioPrefetchY[k], - mode_lib->mp.VRatioPrefetchC[k], - mode_lib->mp.BytePerPixelInDETY[k], - mode_lib->mp.BytePerPixelInDETC[k], - mode_lib->mp.DETBufferSizeY[k], - mode_lib->mp.DETBufferSizeC[k], - /* Output */ - &mode_lib->mp.UrgentBurstFactorLumaPre[k], - &mode_lib->mp.UrgentBurstFactorChromaPre[k], - &mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->mp.NoOfDPP[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLuma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChroma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLumaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLumaPre[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChromaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChromaPre[k]); - - DML_LOG_VERBOSE("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u VRatioY=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); - - DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, mode_lib->mp.prefetch_vmrow_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_l=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_c=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, mode_lib->mp.cursor_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, mode_lib->mp.dpte_row_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, mode_lib->mp.meta_row_bw[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWLuma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWChroma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]); - DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_cursor_bw=%f\n", __func__, k, mode_lib->mp.prefetch_cursor_bw[k]); -#endif - } - - for (k = 0; k <= s->num_active_planes - 1; k++) - mode_lib->mp.final_flip_bw[k] = 0; - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->mp.urg_vactive_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->mp.urg_bandwidth_required; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->mp.urg_bandwidth_required_qual; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->mp.non_urg_bandwidth_required; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = s->surface_dummy_bw0; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 0; - calculate_peak_bandwidth_params->num_active_planes = s->num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->mp.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->mp.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->mp.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->mp.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->mp.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->mp.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->mp.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - // Check urg peak bandwidth against available urg bw - // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) - check_urgent_bandwidth_support( - &mode_lib->mp.FractionOfUrgentBandwidth, // double* frac_urg_bandwidth - &mode_lib->mp.FractionOfUrgentBandwidthMALL, // double* frac_urg_bandwidth_mall - &s->dummy_boolean[1], // vactive bw ok - &mode_lib->mp.PrefetchModeSupported, // prefetch bw ok - - mode_lib->soc.mall_allocated_for_dcn_mbytes, - mode_lib->mp.non_urg_bandwidth_required, - mode_lib->mp.urg_vactive_bandwidth_required, - mode_lib->mp.urg_bandwidth_required, - mode_lib->mp.urg_bandwidth_available); - - if (!mode_lib->mp.PrefetchModeSupported) - DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for prefetch!\n", __func__); - - for (k = 0; k < s->num_active_planes; ++k) { - if (mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]) { - DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHidingPre = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); - mode_lib->mp.PrefetchModeSupported = false; - } - } - } // prefetch schedule ok - - // Prefetch schedule and prefetch bw ok, now check flip bw - if (mode_lib->mp.PrefetchModeSupported == true) { // prefetch schedule and prefetch bw ok, now check flip bw - - mode_lib->mp.BandwidthAvailableForImmediateFlip = - get_bandwidth_available_for_immediate_flip( - dml2_core_internal_soc_state_sys_active, - mode_lib->mp.urg_bandwidth_required_qual, // no flip - mode_lib->mp.urg_bandwidth_available); - mode_lib->mp.TotImmediateFlipBytes = 0; - for (k = 0; k < s->num_active_planes; ++k) { - if (display_cfg->plane_descriptors[k].immediate_flip) { - s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes(s->HostVMInefficiencyFactor, - mode_lib->mp.vm_bytes[k], - mode_lib->mp.PixelPTEBytesPerRow[k], - mode_lib->mp.meta_row_bytes[k]); - } else { - s->per_pipe_flip_bytes[k] = 0; - } - mode_lib->mp.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->mp.NoOfDPP[k]; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k = %u\n", __func__, k); - DML_LOG_VERBOSE("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->mp.NoOfDPP[k]); - DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, mode_lib->mp.vm_bytes[k]); - DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, mode_lib->mp.PixelPTEBytesPerRow[k]); - DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %u\n", __func__, mode_lib->mp.meta_row_bytes[k]); - DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, mode_lib->mp.TotImmediateFlipBytes); -#endif - } - for (k = 0; k < s->num_active_planes; ++k) { - CalculateFlipSchedule( - &mode_lib->scratch, - display_cfg->plane_descriptors[k].immediate_flip, - 0, // use_lb_flip_bw - s->HostVMInefficiencyFactor, - s->Tvm_trips_flip[k], - s->Tr0_trips_flip[k], - s->Tvm_trips_flip_rounded[k], - s->Tr0_trips_flip_rounded[k], - display_cfg->gpuvm_enable, - mode_lib->mp.vm_bytes[k], - mode_lib->mp.PixelPTEBytesPerRow[k], - mode_lib->mp.BandwidthAvailableForImmediateFlip, - mode_lib->mp.TotImmediateFlipBytes, - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), - display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, - display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, - mode_lib->mp.Tno_bw[k], - mode_lib->mp.dpte_row_height[k], - mode_lib->mp.dpte_row_height_chroma[k], - mode_lib->mp.use_one_row_for_frame_flip[k], - mode_lib->ip.max_flip_time_us, - mode_lib->ip.max_flip_time_lines, - s->per_pipe_flip_bytes[k], - mode_lib->mp.meta_row_bytes[k], - mode_lib->mp.meta_row_height[k], - mode_lib->mp.meta_row_height_chroma[k], - mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, - - // Output - &mode_lib->mp.dst_y_per_vm_flip[k], - &mode_lib->mp.dst_y_per_row_flip[k], - &mode_lib->mp.final_flip_bw[k], - &mode_lib->mp.ImmediateFlipSupportedForPipe[k]); - } - - calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; - calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->mp.urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; - calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->mp.non_urg_bandwidth_required_flip; - calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; - calculate_peak_bandwidth_params->surface_peak_required_bw = s->surface_dummy_bw0; - - calculate_peak_bandwidth_params->display_cfg = display_cfg; - calculate_peak_bandwidth_params->inc_flip_bw = 1; - calculate_peak_bandwidth_params->num_active_planes = s->num_active_planes; - calculate_peak_bandwidth_params->num_of_dpp = mode_lib->mp.NoOfDPP; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0; - calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1; - calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor; - calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->mp.mall_prefetch_dram_overhead_factor; - - calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->mp.vactive_sw_bw_l; - calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; - calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; - calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; - calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; - calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; - calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; - calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; - calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; - calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; - calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; - calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; - calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; - calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; - calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->mp.UrgentBurstFactorCursor; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->mp.UrgentBurstFactorLumaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->mp.UrgentBurstFactorChromaPre; - calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->mp.UrgentBurstFactorCursorPre; - - calculate_peak_bandwidth_required( - &mode_lib->scratch, - calculate_peak_bandwidth_params); - - calculate_immediate_flip_bandwidth_support( - &mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip, // double* frac_urg_bandwidth_flip - &mode_lib->mp.ImmediateFlipSupported, // bool* flip_bandwidth_support_ok - - dml2_core_internal_soc_state_sys_active, - mode_lib->mp.urg_bandwidth_required_flip, - mode_lib->mp.non_urg_bandwidth_required_flip, - mode_lib->mp.urg_bandwidth_available); - - if (!mode_lib->mp.ImmediateFlipSupported) - DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for flip!", __func__); - - for (k = 0; k < s->num_active_planes; ++k) { - if (display_cfg->plane_descriptors[k].immediate_flip && mode_lib->mp.ImmediateFlipSupportedForPipe[k] == false) { - mode_lib->mp.ImmediateFlipSupported = false; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Pipe %0d not supporting iflip!\n", __func__, k); -#endif - } - } - } else { // flip or prefetch not support - mode_lib->mp.ImmediateFlipSupported = false; - } - - // consider flip support is okay if the flip bw is ok or (when user does't require a iflip and there is no host vm) - must_support_iflip = display_cfg->hostvm_enable || s->immediate_flip_required; - mode_lib->mp.PrefetchAndImmediateFlipSupported = (mode_lib->mp.PrefetchModeSupported == true && (!must_support_iflip || mode_lib->mp.ImmediateFlipSupported)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: PrefetchModeSupported = %u\n", __func__, mode_lib->mp.PrefetchModeSupported); - for (k = 0; k < s->num_active_planes; ++k) - DML_LOG_VERBOSE("DML::%s: immediate_flip_required[%u] = %u\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); - DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, display_cfg->hostvm_enable); - DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupported = %u\n", __func__, mode_lib->mp.ImmediateFlipSupported); - DML_LOG_VERBOSE("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); -#endif - DML_LOG_VERBOSE("DML::%s: Done one iteration: k=%d, MaxVStartupLines=%u\n", __func__, k, s->MaxVStartupLines[k]); - } - - for (k = 0; k < s->num_active_planes; ++k) - DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - - if (!mode_lib->mp.PrefetchAndImmediateFlipSupported) { - DML_LOG_VERBOSE("DML::%s: Bad, Prefetch and flip scheduling solution NOT found!\n", __func__); - } else { - DML_LOG_VERBOSE("DML::%s: Good, Prefetch and flip scheduling solution found\n", __func__); - - // DCC Configuration - for (k = 0; k < s->num_active_planes; ++k) { -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); -#endif - CalculateDCCConfiguration( - display_cfg->plane_descriptors[k].surface.dcc.enable, - display_cfg->overrides.dcc_programming_assumes_scan_direction_unknown, - display_cfg->plane_descriptors[k].pixel_format, - display_cfg->plane_descriptors[k].surface.plane0.width, - display_cfg->plane_descriptors[k].surface.plane1.width, - display_cfg->plane_descriptors[k].surface.plane0.height, - display_cfg->plane_descriptors[k].surface.plane1.height, - s->NomDETInKByte, - mode_lib->mp.Read256BlockHeightY[k], - mode_lib->mp.Read256BlockHeightC[k], - display_cfg->plane_descriptors[k].surface.tiling, - mode_lib->mp.BytePerPixelY[k], - mode_lib->mp.BytePerPixelC[k], - mode_lib->mp.BytePerPixelInDETY[k], - mode_lib->mp.BytePerPixelInDETC[k], - display_cfg->plane_descriptors[k].composition.rotation_angle, - - /* Output */ - &mode_lib->mp.RequestLuma[k], - &mode_lib->mp.RequestChroma[k], - &mode_lib->mp.DCCYMaxUncompressedBlock[k], - &mode_lib->mp.DCCCMaxUncompressedBlock[k], - &mode_lib->mp.DCCYMaxCompressedBlock[k], - &mode_lib->mp.DCCCMaxCompressedBlock[k], - &mode_lib->mp.DCCYIndependentBlock[k], - &mode_lib->mp.DCCCIndependentBlock[k]); - } - - //Watermarks and NB P-State/DRAM Clock Change Support - s->mmSOCParameters.UrgentLatency = mode_lib->mp.UrgentLatency; - s->mmSOCParameters.ExtraLatency = mode_lib->mp.ExtraLatency; - s->mmSOCParameters.ExtraLatency_sr = mode_lib->mp.ExtraLatency_sr; - s->mmSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; - s->mmSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; - s->mmSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; - s->mmSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; - s->mmSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; - s->mmSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; - s->mmSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; - s->mmSOCParameters.USRRetrainingLatency = 0; - s->mmSOCParameters.SMNLatency = 0; - s->mmSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->mp.uclk_freq_mhz * 1000), in_out_params->min_clk_index); - s->mmSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->mp.uclk_freq_mhz, mode_lib->mp.FabricClock, in_out_params->min_clk_index); - s->mmSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->mp.FabricClock; - s->mmSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; - - CalculateWatermarks_params->display_cfg = display_cfg; - CalculateWatermarks_params->USRRetrainingRequired = false; - CalculateWatermarks_params->NumberOfActiveSurfaces = s->num_active_planes; - CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; - CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; - CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; - CalculateWatermarks_params->DCFCLK = mode_lib->mp.Dcfclk; - CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; - CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; - CalculateWatermarks_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; - CalculateWatermarks_params->mmSOCParameters = s->mmSOCParameters; - CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; - CalculateWatermarks_params->SOCCLK = s->SOCCLK; - CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->mp.DCFCLKDeepSleep; - CalculateWatermarks_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; - CalculateWatermarks_params->DETBufferSizeC = mode_lib->mp.DETBufferSizeC; - CalculateWatermarks_params->SwathHeightY = mode_lib->mp.SwathHeightY; - CalculateWatermarks_params->SwathHeightC = mode_lib->mp.SwathHeightC; - CalculateWatermarks_params->SwathWidthY = mode_lib->mp.SwathWidthY; - CalculateWatermarks_params->SwathWidthC = mode_lib->mp.SwathWidthC; - CalculateWatermarks_params->BytePerPixelDETY = mode_lib->mp.BytePerPixelInDETY; - CalculateWatermarks_params->BytePerPixelDETC = mode_lib->mp.BytePerPixelInDETC; - CalculateWatermarks_params->DSTXAfterScaler = mode_lib->mp.DSTXAfterScaler; - CalculateWatermarks_params->DSTYAfterScaler = mode_lib->mp.DSTYAfterScaler; - CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->mp.UnboundedRequestEnabled; - CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->mp.CompressedBufferSizeInkByte; - CalculateWatermarks_params->meta_row_height_l = mode_lib->mp.meta_row_height; - CalculateWatermarks_params->meta_row_height_c = mode_lib->mp.meta_row_height_chroma; - CalculateWatermarks_params->DPPPerSurface = mode_lib->mp.NoOfDPP; - - // Output - CalculateWatermarks_params->Watermark = &mode_lib->mp.Watermark; - CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->mp.DRAMClockChangeSupport; - CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->mp.global_dram_clock_change_supported; - CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported; - CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->mp.SubViewportLinesNeededInMALL; - CalculateWatermarks_params->FCLKChangeSupport = mode_lib->mp.FCLKChangeSupport; - CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->mp.global_fclk_change_supported; - CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &mode_lib->mp.MaxActiveFCLKChangeLatencySupported; - CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->mp.USRRetrainingSupport; - CalculateWatermarks_params->g6_temp_read_support = &mode_lib->mp.g6_temp_read_support; - CalculateWatermarks_params->VActiveLatencyHidingMargin = 0; - CalculateWatermarks_params->VActiveLatencyHidingUs = 0; - - CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); - - for (k = 0; k < s->num_active_planes; ++k) { - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { - mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k] = math_max2(0, mode_lib->mp.VStartupMin[k] * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) - mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); - mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k] = math_max2(0, mode_lib->mp.VStartupMin[k] * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) - mode_lib->mp.Watermark.WritebackFCLKChangeWatermark); - } else { - mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k] = 0; - mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k] = 0; - } - } - - calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->mp.Watermark, mode_lib->mp.pstate_keepout_dst_lines); - - DML_LOG_VERBOSE("DML::%s: DEBUG stream_index = %0d\n", __func__, display_cfg->plane_descriptors[0].stream_index); - DML_LOG_VERBOSE("DML::%s: DEBUG PixelClock = %ld kHz\n", __func__, (display_cfg->stream_descriptors[display_cfg->plane_descriptors[0].stream_index].timing.pixel_clock_khz)); - - //Display Pipeline Delivery Time in Prefetch, Groups - CalculatePixelDeliveryTimes( - display_cfg, - cfg_support_info, - s->num_active_planes, - mode_lib->mp.VRatioPrefetchY, - mode_lib->mp.VRatioPrefetchC, - mode_lib->mp.swath_width_luma_ub, - mode_lib->mp.swath_width_chroma_ub, - mode_lib->mp.PSCL_THROUGHPUT, - mode_lib->mp.PSCL_THROUGHPUT_CHROMA, - mode_lib->mp.Dppclk, - mode_lib->mp.BytePerPixelC, - mode_lib->mp.req_per_swath_ub_l, - mode_lib->mp.req_per_swath_ub_c, - - /* Output */ - mode_lib->mp.DisplayPipeLineDeliveryTimeLuma, - mode_lib->mp.DisplayPipeLineDeliveryTimeChroma, - mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch, - mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch, - mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma, - mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma, - mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch, - mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch); - - CalculateMetaAndPTETimes_params->scratch = &mode_lib->scratch; - CalculateMetaAndPTETimes_params->display_cfg = display_cfg; - CalculateMetaAndPTETimes_params->NumberOfActiveSurfaces = s->num_active_planes; - CalculateMetaAndPTETimes_params->use_one_row_for_frame = mode_lib->mp.use_one_row_for_frame; - CalculateMetaAndPTETimes_params->dst_y_per_row_vblank = mode_lib->mp.dst_y_per_row_vblank; - CalculateMetaAndPTETimes_params->dst_y_per_row_flip = mode_lib->mp.dst_y_per_row_flip; - CalculateMetaAndPTETimes_params->BytePerPixelY = mode_lib->mp.BytePerPixelY; - CalculateMetaAndPTETimes_params->BytePerPixelC = mode_lib->mp.BytePerPixelC; - CalculateMetaAndPTETimes_params->dpte_row_height = mode_lib->mp.dpte_row_height; - CalculateMetaAndPTETimes_params->dpte_row_height_chroma = mode_lib->mp.dpte_row_height_chroma; - CalculateMetaAndPTETimes_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; - CalculateMetaAndPTETimes_params->PTERequestSizeY = mode_lib->mp.PTERequestSizeY; - CalculateMetaAndPTETimes_params->PTERequestSizeC = mode_lib->mp.PTERequestSizeC; - CalculateMetaAndPTETimes_params->PixelPTEReqWidthY = mode_lib->mp.PixelPTEReqWidthY; - CalculateMetaAndPTETimes_params->PixelPTEReqHeightY = mode_lib->mp.PixelPTEReqHeightY; - CalculateMetaAndPTETimes_params->PixelPTEReqWidthC = mode_lib->mp.PixelPTEReqWidthC; - CalculateMetaAndPTETimes_params->PixelPTEReqHeightC = mode_lib->mp.PixelPTEReqHeightC; - CalculateMetaAndPTETimes_params->dpte_row_width_luma_ub = mode_lib->mp.dpte_row_width_luma_ub; - CalculateMetaAndPTETimes_params->dpte_row_width_chroma_ub = mode_lib->mp.dpte_row_width_chroma_ub; - CalculateMetaAndPTETimes_params->tdlut_groups_per_2row_ub = s->tdlut_groups_per_2row_ub; - CalculateMetaAndPTETimes_params->mrq_present = mode_lib->ip.dcn_mrq_present; - - CalculateMetaAndPTETimes_params->MetaChunkSize = mode_lib->ip.meta_chunk_size_kbytes; - CalculateMetaAndPTETimes_params->MinMetaChunkSizeBytes = mode_lib->ip.min_meta_chunk_size_bytes; - CalculateMetaAndPTETimes_params->meta_row_width = mode_lib->mp.meta_row_width; - CalculateMetaAndPTETimes_params->meta_row_width_chroma = mode_lib->mp.meta_row_width_chroma; - CalculateMetaAndPTETimes_params->meta_row_height = mode_lib->mp.meta_row_height; - CalculateMetaAndPTETimes_params->meta_row_height_chroma = mode_lib->mp.meta_row_height_chroma; - CalculateMetaAndPTETimes_params->meta_req_width = mode_lib->mp.meta_req_width; - CalculateMetaAndPTETimes_params->meta_req_width_chroma = mode_lib->mp.meta_req_width_chroma; - CalculateMetaAndPTETimes_params->meta_req_height = mode_lib->mp.meta_req_height; - CalculateMetaAndPTETimes_params->meta_req_height_chroma = mode_lib->mp.meta_req_height_chroma; - - CalculateMetaAndPTETimes_params->time_per_tdlut_group = mode_lib->mp.time_per_tdlut_group; - CalculateMetaAndPTETimes_params->DST_Y_PER_PTE_ROW_NOM_L = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L; - CalculateMetaAndPTETimes_params->DST_Y_PER_PTE_ROW_NOM_C = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C; - CalculateMetaAndPTETimes_params->time_per_pte_group_nom_luma = mode_lib->mp.time_per_pte_group_nom_luma; - CalculateMetaAndPTETimes_params->time_per_pte_group_vblank_luma = mode_lib->mp.time_per_pte_group_vblank_luma; - CalculateMetaAndPTETimes_params->time_per_pte_group_flip_luma = mode_lib->mp.time_per_pte_group_flip_luma; - CalculateMetaAndPTETimes_params->time_per_pte_group_nom_chroma = mode_lib->mp.time_per_pte_group_nom_chroma; - CalculateMetaAndPTETimes_params->time_per_pte_group_vblank_chroma = mode_lib->mp.time_per_pte_group_vblank_chroma; - CalculateMetaAndPTETimes_params->time_per_pte_group_flip_chroma = mode_lib->mp.time_per_pte_group_flip_chroma; - CalculateMetaAndPTETimes_params->DST_Y_PER_META_ROW_NOM_L = mode_lib->mp.DST_Y_PER_META_ROW_NOM_L; - CalculateMetaAndPTETimes_params->DST_Y_PER_META_ROW_NOM_C = mode_lib->mp.DST_Y_PER_META_ROW_NOM_C; - CalculateMetaAndPTETimes_params->TimePerMetaChunkNominal = mode_lib->mp.TimePerMetaChunkNominal; - CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkNominal = mode_lib->mp.TimePerChromaMetaChunkNominal; - CalculateMetaAndPTETimes_params->TimePerMetaChunkVBlank = mode_lib->mp.TimePerMetaChunkVBlank; - CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkVBlank = mode_lib->mp.TimePerChromaMetaChunkVBlank; - CalculateMetaAndPTETimes_params->TimePerMetaChunkFlip = mode_lib->mp.TimePerMetaChunkFlip; - CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkFlip = mode_lib->mp.TimePerChromaMetaChunkFlip; - - CalculateMetaAndPTETimes(CalculateMetaAndPTETimes_params); - - CalculateVMGroupAndRequestTimes( - display_cfg, - s->num_active_planes, - mode_lib->mp.BytePerPixelC, - mode_lib->mp.dst_y_per_vm_vblank, - mode_lib->mp.dst_y_per_vm_flip, - mode_lib->mp.dpte_row_width_luma_ub, - mode_lib->mp.dpte_row_width_chroma_ub, - mode_lib->mp.vm_group_bytes, - mode_lib->mp.dpde0_bytes_per_frame_ub_l, - mode_lib->mp.dpde0_bytes_per_frame_ub_c, - s->tdlut_pte_bytes_per_frame, - mode_lib->mp.meta_pte_bytes_per_frame_ub_l, - mode_lib->mp.meta_pte_bytes_per_frame_ub_c, - mode_lib->ip.dcn_mrq_present, - - /* Output */ - mode_lib->mp.TimePerVMGroupVBlank, - mode_lib->mp.TimePerVMGroupFlip, - mode_lib->mp.TimePerVMRequestVBlank, - mode_lib->mp.TimePerVMRequestFlip); - - // VStartup Adjustment - for (k = 0; k < s->num_active_planes; ++k) { - bool isInterlaceTiming; - - mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.TWait[k] + mode_lib->mp.ExtraLatency; - if (!display_cfg->plane_descriptors[k].dynamic_meta_data.enable) - mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.TCalc + mode_lib->mp.MinTTUVBlank[k]; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); -#endif - s->Tvstartup_margin = (s->MaxVStartupLines[k] - mode_lib->mp.VStartupMin[k]) * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); - mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.MinTTUVBlank[k] + s->Tvstartup_margin; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); - DML_LOG_VERBOSE("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); -#endif - - mode_lib->mp.Tdmdl[k] = mode_lib->mp.Tdmdl[k] + s->Tvstartup_margin; - if (display_cfg->plane_descriptors[k].dynamic_meta_data.enable && mode_lib->ip.dynamic_metadata_vm_enabled) { - mode_lib->mp.Tdmdl_vm[k] = mode_lib->mp.Tdmdl_vm[k] + s->Tvstartup_margin; - } - - isInterlaceTiming = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !mode_lib->ip.ptoi_supported); - - // The actual positioning of the vstartup - mode_lib->mp.VStartup[k] = (isInterlaceTiming ? (2 * s->MaxVStartupLines[k]) : s->MaxVStartupLines[k]); - - s->dlg_vblank_start = ((isInterlaceTiming ? math_floor2((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch) / 2.0, 1.0) : - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total) - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); - s->LSetup = math_floor2(4.0 * mode_lib->mp.TSetup[k] / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), 1.0) / 4.0; - s->blank_lines_remaining = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active) - mode_lib->mp.VStartup[k]; - - if (s->blank_lines_remaining < 0) { - DML_LOG_VERBOSE("ERROR: Vstartup is larger than vblank!?\n"); - s->blank_lines_remaining = 0; - DML_ASSERT(0); - } - mode_lib->mp.MIN_DST_Y_NEXT_START[k] = s->dlg_vblank_start + s->blank_lines_remaining + s->LSetup; - - // debug only - if (((mode_lib->mp.VUpdateOffsetPix[k] + mode_lib->mp.VUpdateWidthPix[k] + mode_lib->mp.VReadyOffsetPix[k]) / (double) display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) <= - (isInterlaceTiming ? - math_floor2((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch - mode_lib->mp.VStartup[k]) / 2.0, 1.0) : - (int)(display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch - mode_lib->mp.VStartup[k]))) { - mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k] = true; - } else { - mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k] = false; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, mode_lib->mp.VStartup[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, mode_lib->mp.VStartupMin[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, mode_lib->mp.VUpdateOffsetPix[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, mode_lib->mp.VUpdateWidthPix[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, mode_lib->mp.VReadyOffsetPix[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, HTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total); - DML_LOG_VERBOSE("DML::%s: k=%u, VTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); - DML_LOG_VERBOSE("DML::%s: k=%u, VActive = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active); - DML_LOG_VERBOSE("DML::%s: k=%u, VFrontPorch = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); - DML_LOG_VERBOSE("DML::%s: k=%u, TSetup = %f\n", __func__, k, mode_lib->mp.TSetup[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, mode_lib->mp.MIN_DST_Y_NEXT_START[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]); -#endif - } - - //Maximum Bandwidth Used - mode_lib->mp.TotalWRBandwidth = 0; - for (k = 0; k < display_cfg->num_streams; ++k) { - s->WRBandwidth = 0; - if (display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) { - s->WRBandwidth = display_cfg->stream_descriptors[k].writeback.writeback_stream[0].output_height - * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].output_width / - (display_cfg->stream_descriptors[k].timing.h_total * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].input_height - / ((double)display_cfg->stream_descriptors[k].timing.pixel_clock_khz / 1000)) - * (display_cfg->stream_descriptors[k].writeback.writeback_stream[0].pixel_format == dml2_444_32 ? 4.0 : 8.0); - mode_lib->mp.TotalWRBandwidth = mode_lib->mp.TotalWRBandwidth + s->WRBandwidth; - } - } - - mode_lib->mp.TotalDataReadBandwidth = 0; - for (k = 0; k < s->num_active_planes; ++k) { - mode_lib->mp.TotalDataReadBandwidth = mode_lib->mp.TotalDataReadBandwidth + mode_lib->mp.vactive_sw_bw_l[k] + mode_lib->mp.vactive_sw_bw_c[k]; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, mode_lib->mp.TotalDataReadBandwidth); - DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); - DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); -#endif - } - - CalculateStutterEfficiency_params->display_cfg = display_cfg; - CalculateStutterEfficiency_params->CompressedBufferSizeInkByte = mode_lib->mp.CompressedBufferSizeInkByte; - CalculateStutterEfficiency_params->UnboundedRequestEnabled = mode_lib->mp.UnboundedRequestEnabled; - CalculateStutterEfficiency_params->MetaFIFOSizeInKEntries = mode_lib->ip.meta_fifo_size_in_kentries; - CalculateStutterEfficiency_params->ZeroSizeBufferEntries = mode_lib->ip.zero_size_buffer_entries; - CalculateStutterEfficiency_params->PixelChunkSizeInKByte = mode_lib->ip.pixel_chunk_size_kbytes; - CalculateStutterEfficiency_params->NumberOfActiveSurfaces = s->num_active_planes; - CalculateStutterEfficiency_params->ROBBufferSizeInKByte = mode_lib->ip.rob_buffer_size_kbytes; - CalculateStutterEfficiency_params->TotalDataReadBandwidth = mode_lib->mp.TotalDataReadBandwidth; - CalculateStutterEfficiency_params->DCFCLK = mode_lib->mp.Dcfclk; - CalculateStutterEfficiency_params->ReturnBW = mode_lib->mp.urg_bandwidth_available_min[dml2_core_internal_soc_state_sys_active]; - CalculateStutterEfficiency_params->CompbufReservedSpace64B = mode_lib->mp.compbuf_reserved_space_64b; - CalculateStutterEfficiency_params->CompbufReservedSpaceZs = mode_lib->ip.compbuf_reserved_space_zs; - CalculateStutterEfficiency_params->SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; - CalculateStutterEfficiency_params->SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; - CalculateStutterEfficiency_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; - CalculateStutterEfficiency_params->StutterEnterPlusExitWatermark = mode_lib->mp.Watermark.StutterEnterPlusExitWatermark; - CalculateStutterEfficiency_params->Z8StutterEnterPlusExitWatermark = mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark; - CalculateStutterEfficiency_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; - CalculateStutterEfficiency_params->MinTTUVBlank = mode_lib->mp.MinTTUVBlank; - CalculateStutterEfficiency_params->DPPPerSurface = mode_lib->mp.NoOfDPP; - CalculateStutterEfficiency_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; - CalculateStutterEfficiency_params->BytePerPixelY = mode_lib->mp.BytePerPixelY; - CalculateStutterEfficiency_params->BytePerPixelDETY = mode_lib->mp.BytePerPixelInDETY; - CalculateStutterEfficiency_params->SwathWidthY = mode_lib->mp.SwathWidthY; - CalculateStutterEfficiency_params->SwathHeightY = mode_lib->mp.SwathHeightY; - CalculateStutterEfficiency_params->SwathHeightC = mode_lib->mp.SwathHeightC; - CalculateStutterEfficiency_params->BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY; - CalculateStutterEfficiency_params->BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY; - CalculateStutterEfficiency_params->BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC; - CalculateStutterEfficiency_params->BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC; - CalculateStutterEfficiency_params->DCCYMaxUncompressedBlock = mode_lib->mp.DCCYMaxUncompressedBlock; - CalculateStutterEfficiency_params->DCCCMaxUncompressedBlock = mode_lib->mp.DCCCMaxUncompressedBlock; - CalculateStutterEfficiency_params->ReadBandwidthSurfaceLuma = mode_lib->mp.vactive_sw_bw_l; - CalculateStutterEfficiency_params->ReadBandwidthSurfaceChroma = mode_lib->mp.vactive_sw_bw_c; - CalculateStutterEfficiency_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; - CalculateStutterEfficiency_params->meta_row_bw = mode_lib->mp.meta_row_bw; - CalculateStutterEfficiency_params->rob_alloc_compressed = mode_lib->ip.dcn_mrq_present; - - // output - CalculateStutterEfficiency_params->StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.StutterEfficiencyNotIncludingVBlank; - CalculateStutterEfficiency_params->StutterEfficiency = &mode_lib->mp.StutterEfficiency; - CalculateStutterEfficiency_params->NumberOfStutterBurstsPerFrame = &mode_lib->mp.NumberOfStutterBurstsPerFrame; - CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlank; - CalculateStutterEfficiency_params->Z8StutterEfficiency = &mode_lib->mp.Z8StutterEfficiency; - CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &mode_lib->mp.Z8NumberOfStutterBurstsPerFrame; - CalculateStutterEfficiency_params->StutterPeriod = &mode_lib->mp.StutterPeriod; - CalculateStutterEfficiency_params->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = &mode_lib->mp.DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE; - - // Stutter Efficiency - CalculateStutterEfficiency(&mode_lib->scratch, CalculateStutterEfficiency_params); - -#ifdef __DML_VBA_ALLOW_DELTA__ - // Calculate z8 stutter eff assuming 0 reserved space - CalculateStutterEfficiency_params->CompbufReservedSpace64B = 0; - CalculateStutterEfficiency_params->CompbufReservedSpaceZs = 0; - - CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlankBestCase; - CalculateStutterEfficiency_params->Z8StutterEfficiency = &mode_lib->mp.Z8StutterEfficiencyBestCase; - CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase; - CalculateStutterEfficiency_params->StutterPeriod = &mode_lib->mp.StutterPeriodBestCase; - - // Stutter Efficiency - CalculateStutterEfficiency(&mode_lib->scratch, CalculateStutterEfficiency_params); -#else - mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlankBestCase = mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlank; - mode_lib->mp.Z8StutterEfficiencyBestCase = mode_lib->mp.Z8StutterEfficiency; - mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase = mode_lib->mp.Z8NumberOfStutterBurstsPerFrame; - mode_lib->mp.StutterPeriodBestCase = mode_lib->mp.StutterPeriod; -#endif - } // PrefetchAndImmediateFlipSupported - - max_uclk_mhz = mode_lib->soc.clk_table.uclk.clk_values_khz[mode_lib->soc.clk_table.uclk.num_clk_values - 1] / 1000.0; - min_return_latency_in_DCFCLK_cycles = (min_return_uclk_cycles / max_uclk_mhz + min_return_fclk_cycles / max_fclk_mhz) * hard_minimum_dcfclk_mhz; - mode_lib->mp.min_return_latency_in_dcfclk = (unsigned int)min_return_latency_in_DCFCLK_cycles; - mode_lib->mp.dcfclk_deep_sleep_hysteresis = (unsigned int)math_max2(32, (double)mode_lib->ip.pixel_chunk_size_kbytes * 1024 * 3 / 4 / 64 - min_return_latency_in_DCFCLK_cycles); - DML_ASSERT(mode_lib->mp.dcfclk_deep_sleep_hysteresis < 256); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: max_fclk_mhz = %f\n", __func__, max_fclk_mhz); - DML_LOG_VERBOSE("DML::%s: max_uclk_mhz = %f\n", __func__, max_uclk_mhz); - DML_LOG_VERBOSE("DML::%s: hard_minimum_dcfclk_mhz = %f\n", __func__, hard_minimum_dcfclk_mhz); - DML_LOG_VERBOSE("DML::%s: min_return_uclk_cycles = %ld\n", __func__, min_return_uclk_cycles); - DML_LOG_VERBOSE("DML::%s: min_return_fclk_cycles = %ld\n", __func__, min_return_fclk_cycles); - DML_LOG_VERBOSE("DML::%s: min_return_latency_in_DCFCLK_cycles = %f\n", __func__, min_return_latency_in_DCFCLK_cycles); - DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d \n", __func__, mode_lib->mp.dcfclk_deep_sleep_hysteresis); - DML_LOG_VERBOSE("DML::%s: --- END --- \n", __func__); -#endif - return (in_out_params->mode_lib->mp.PrefetchAndImmediateFlipSupported); -} - -bool dml2_core_calcs_mode_programming_ex(struct dml2_core_calcs_mode_programming_ex *in_out_params) -{ - DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); - bool result = dml_core_mode_programming(in_out_params); - - DML_LOG_VERBOSE("DML::%s: result = %0d\n", __func__, result); - DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); - return result; -} - -void dml2_core_calcs_get_dpte_row_height( - unsigned int *dpte_row_height, - struct dml2_core_internal_display_mode_lib *mode_lib, - bool is_plane1, - enum dml2_source_format_class SourcePixelFormat, - enum dml2_swizzle_mode SurfaceTiling, - enum dml2_rotation_angle ScanDirection, - unsigned int pitch, - unsigned int GPUVMMinPageSizeKBytes) -{ - unsigned int BytePerPixelY; - unsigned int BytePerPixelC; - double BytePerPixelInDETY; - double BytePerPixelInDETC; - unsigned int BlockHeight256BytesY; - unsigned int BlockHeight256BytesC; - unsigned int BlockWidth256BytesY; - unsigned int BlockWidth256BytesC; - unsigned int MacroTileWidthY; - unsigned int MacroTileWidthC; - unsigned int MacroTileHeightY; - unsigned int MacroTileHeightC; - bool surf_linear_128_l = false; - bool surf_linear_128_c = false; - - CalculateBytePerPixelAndBlockSizes( - SourcePixelFormat, - SurfaceTiling, - pitch, - pitch, - - /* Output */ - &BytePerPixelY, - &BytePerPixelC, - &BytePerPixelInDETY, - &BytePerPixelInDETC, - &BlockHeight256BytesY, - &BlockHeight256BytesC, - &BlockWidth256BytesY, - &BlockWidth256BytesC, - &MacroTileHeightY, - &MacroTileHeightC, - &MacroTileWidthY, - &MacroTileWidthC, - &surf_linear_128_l, - &surf_linear_128_c); - - unsigned int BytePerPixel = is_plane1 ? BytePerPixelC : BytePerPixelY; - unsigned int BlockHeight256Bytes = is_plane1 ? BlockHeight256BytesC : BlockHeight256BytesY; - unsigned int BlockWidth256Bytes = is_plane1 ? BlockWidth256BytesC : BlockWidth256BytesY; - unsigned int MacroTileWidth = is_plane1 ? MacroTileWidthC : MacroTileWidthY; - unsigned int MacroTileHeight = is_plane1 ? MacroTileHeightC : MacroTileHeightY; - unsigned int PTEBufferSizeInRequests = is_plane1 ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML: %s: is_plane1 = %u\n", __func__, is_plane1); - DML_LOG_VERBOSE("DML: %s: BytePerPixel = %u\n", __func__, BytePerPixel); - DML_LOG_VERBOSE("DML: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); - DML_LOG_VERBOSE("DML: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); - DML_LOG_VERBOSE("DML: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); - DML_LOG_VERBOSE("DML: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); - DML_LOG_VERBOSE("DML: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); - DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); - DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); - DML_LOG_VERBOSE("DML: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); -#endif - unsigned int dummy_integer[21]; - - mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportStationary = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.DCCEnable = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.NumberOfDPPs = 1; - mode_lib->scratch.calculate_vm_and_row_bytes_params.BlockHeight256Bytes = BlockHeight256Bytes; - mode_lib->scratch.calculate_vm_and_row_bytes_params.BlockWidth256Bytes = BlockWidth256Bytes; - mode_lib->scratch.calculate_vm_and_row_bytes_params.SourcePixelFormat = SourcePixelFormat; - mode_lib->scratch.calculate_vm_and_row_bytes_params.SurfaceTiling = SurfaceTiling; - mode_lib->scratch.calculate_vm_and_row_bytes_params.BytePerPixel = BytePerPixel; - mode_lib->scratch.calculate_vm_and_row_bytes_params.RotationAngle = ScanDirection; - mode_lib->scratch.calculate_vm_and_row_bytes_params.SwathWidth = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportHeight = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportXStart = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportYStart = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMEnable = 1; - mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = 4; - mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = GPUVMMinPageSizeKBytes; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = PTEBufferSizeInRequests; - mode_lib->scratch.calculate_vm_and_row_bytes_params.Pitch = pitch; - mode_lib->scratch.calculate_vm_and_row_bytes_params.MacroTileWidth = MacroTileWidth; - mode_lib->scratch.calculate_vm_and_row_bytes_params.MacroTileHeight = MacroTileHeight; - mode_lib->scratch.calculate_vm_and_row_bytes_params.is_phantom = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.DCCMetaPitch = 0; - mode_lib->scratch.calculate_vm_and_row_bytes_params.mrq_present = 0; - - mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &dummy_integer[1]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &dummy_integer[2]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_width_ub = &dummy_integer[3]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height = dpte_row_height; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height_linear = &dummy_integer[4]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &dummy_integer[5]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &dummy_integer[6]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &dummy_integer[7]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.vmpg_width = &dummy_integer[8]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.vmpg_height = &dummy_integer[9]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &dummy_integer[11]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &dummy_integer[12]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.PTERequestSize = &dummy_integer[13]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &dummy_integer[14]; - - mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_bytes = &dummy_integer[15]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.MetaRequestWidth = &dummy_integer[16]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.MetaRequestHeight = &dummy_integer[17]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_width = &dummy_integer[18]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_height = &dummy_integer[19]; - mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &dummy_integer[20]; - - // just supply with enough parameters to calculate dpte - CalculateVMAndRowBytes(&mode_lib->scratch.calculate_vm_and_row_bytes_params); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); -#endif -} - -static bool is_dual_plane(enum dml2_source_format_class source_format) -{ - bool ret_val = false; - - if ((source_format == dml2_420_12) || (source_format == dml2_420_8) || (source_format == dml2_420_10) || (source_format == dml2_rgbe_alpha)) - ret_val = true; - - return ret_val; -} - -static unsigned int dml_get_plane_idx(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) -{ - unsigned int plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; - return plane_idx; -} - -static void rq_dlg_get_wm_regs(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *wm_regs) -{ - double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; - - wm_regs->fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); - wm_regs->sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - wm_regs->sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); - wm_regs->sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - wm_regs->sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); - wm_regs->temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); - wm_regs->uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); - wm_regs->urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - wm_regs->usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); - wm_regs->refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.UrgentLatency * refclk_freq_in_mhz); - wm_regs->refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.MetaTripToMemory * refclk_freq_in_mhz); - wm_regs->frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); - wm_regs->frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); - wm_regs->frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); -} - -static unsigned int log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend) -{ - if (a == 0) - return 0; - - return (math_log2_approx(a) - subtrahend); -} - -void dml2_core_calcs_cursor_dlg_reg(struct dml2_cursor_dlg_regs *cursor_dlg_regs, const struct dml2_get_cursor_dlg_reg *p) -{ - int dst_x_offset = (int) ((p->cursor_x_position + (p->cursor_stereo_en == 0 ? 0 : math_max2(p->cursor_primary_offset, p->cursor_secondary_offset)) - - (p->cursor_hotspot_x * (p->cursor_2x_magnify == 0 ? 1 : 2))) * p->dlg_refclk_mhz / p->pixel_rate_mhz / p->hratio); - cursor_dlg_regs->dst_x_offset = (unsigned int) ((dst_x_offset > 0) ? dst_x_offset : 0); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG::%s: cursor_x_position=%d\n", __func__, p->cursor_x_position); - DML_LOG_VERBOSE("DML_DLG::%s: dlg_refclk_mhz=%f\n", __func__, p->dlg_refclk_mhz); - DML_LOG_VERBOSE("DML_DLG::%s: pixel_rate_mhz=%f\n", __func__, p->pixel_rate_mhz); - DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d\n", __func__, dst_x_offset); - DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d (reg)\n", __func__, cursor_dlg_regs->dst_x_offset); -#endif - - cursor_dlg_regs->chunk_hdl_adjust = 3; - cursor_dlg_regs->dst_y_offset = 0; - - cursor_dlg_regs->qos_level_fixed = 8; - cursor_dlg_regs->qos_ramp_disable = 0; -} - -static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, - const struct dml2_display_cfg *display_cfg, - const struct dml2_core_internal_display_mode_lib *mode_lib, - unsigned int pipe_idx) -{ - unsigned int plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); - enum dml2_source_format_class source_format = display_cfg->plane_descriptors[plane_idx].pixel_format; - enum dml2_swizzle_mode sw_mode = display_cfg->plane_descriptors[plane_idx].surface.tiling; - bool dual_plane = is_dual_plane((enum dml2_source_format_class)(source_format)); - - unsigned int pixel_chunk_bytes = 0; - unsigned int min_pixel_chunk_bytes = 0; - unsigned int dpte_group_bytes = 0; - unsigned int mpte_group_bytes = 0; - - unsigned int p1_pixel_chunk_bytes = 0; - unsigned int p1_min_pixel_chunk_bytes = 0; - unsigned int p1_dpte_group_bytes = 0; - unsigned int p1_mpte_group_bytes = 0; - - unsigned int detile_buf_plane1_addr = 0; - unsigned int detile_buf_size_in_bytes; - double stored_swath_l_bytes; - double stored_swath_c_bytes; - bool is_phantom_pipe; - - DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); - - pixel_chunk_bytes = (unsigned int)(mode_lib->ip.pixel_chunk_size_kbytes * 1024); - min_pixel_chunk_bytes = (unsigned int)(mode_lib->ip.min_pixel_chunk_size_bytes); - - if (pixel_chunk_bytes == 64 * 1024) - min_pixel_chunk_bytes = 0; - - dpte_group_bytes = (unsigned int)(dml_get_dpte_group_size_in_bytes(mode_lib, pipe_idx)); - mpte_group_bytes = (unsigned int)(dml_get_vm_group_size_in_bytes(mode_lib, pipe_idx)); - - p1_pixel_chunk_bytes = pixel_chunk_bytes; - p1_min_pixel_chunk_bytes = min_pixel_chunk_bytes; - p1_dpte_group_bytes = dpte_group_bytes; - p1_mpte_group_bytes = mpte_group_bytes; - - if (source_format == dml2_rgbe_alpha) - p1_pixel_chunk_bytes = (unsigned int)(mode_lib->ip.alpha_pixel_chunk_size_kbytes * 1024); - - rq_regs->unbounded_request_enabled = dml_get_unbounded_request_enabled(mode_lib); - rq_regs->rq_regs_l.chunk_size = log_and_substract_if_non_zero(pixel_chunk_bytes, 10); - rq_regs->rq_regs_c.chunk_size = log_and_substract_if_non_zero(p1_pixel_chunk_bytes, 10); - - if (min_pixel_chunk_bytes == 0) - rq_regs->rq_regs_l.min_chunk_size = 0; - else - rq_regs->rq_regs_l.min_chunk_size = log_and_substract_if_non_zero(min_pixel_chunk_bytes, 8 - 1); - - if (p1_min_pixel_chunk_bytes == 0) - rq_regs->rq_regs_c.min_chunk_size = 0; - else - rq_regs->rq_regs_c.min_chunk_size = log_and_substract_if_non_zero(p1_min_pixel_chunk_bytes, 8 - 1); - - rq_regs->rq_regs_l.dpte_group_size = log_and_substract_if_non_zero(dpte_group_bytes, 6); - rq_regs->rq_regs_l.mpte_group_size = log_and_substract_if_non_zero(mpte_group_bytes, 6); - rq_regs->rq_regs_c.dpte_group_size = log_and_substract_if_non_zero(p1_dpte_group_bytes, 6); - rq_regs->rq_regs_c.mpte_group_size = log_and_substract_if_non_zero(p1_mpte_group_bytes, 6); - - detile_buf_size_in_bytes = (unsigned int)(dml_get_det_buffer_size_kbytes(mode_lib, pipe_idx) * 1024); - - if (sw_mode == dml2_sw_linear && display_cfg->gpuvm_enable) { - unsigned int p0_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_l(mode_lib, pipe_idx)); -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG: %s: p0_pte_row_height_linear = %u\n", __func__, p0_pte_row_height_linear); -#endif - DML_ASSERT(p0_pte_row_height_linear >= 8); - - rq_regs->rq_regs_l.pte_row_height_linear = math_log2_approx(p0_pte_row_height_linear) - 3; - if (dual_plane) { - unsigned int p1_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_c(mode_lib, pipe_idx)); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG: %s: p1_pte_row_height_linear = %u\n", __func__, p1_pte_row_height_linear); -#endif - if (sw_mode == dml2_sw_linear) { - DML_ASSERT(p1_pte_row_height_linear >= 8); - } - rq_regs->rq_regs_c.pte_row_height_linear = math_log2_approx(p1_pte_row_height_linear) - 3; - } - } else { - rq_regs->rq_regs_l.pte_row_height_linear = 0; - rq_regs->rq_regs_c.pte_row_height_linear = 0; - } - - rq_regs->rq_regs_l.swath_height = log_and_substract_if_non_zero(dml_get_swath_height_l(mode_lib, pipe_idx), 0); - rq_regs->rq_regs_c.swath_height = log_and_substract_if_non_zero(dml_get_swath_height_c(mode_lib, pipe_idx), 0); - - // FIXME_DCN4, programming guide has dGPU condition - if (pixel_chunk_bytes >= 32 * 1024 || (dual_plane && p1_pixel_chunk_bytes >= 32 * 1024)) { //32kb - rq_regs->drq_expansion_mode = 0; - } else { - rq_regs->drq_expansion_mode = 2; - } - rq_regs->prq_expansion_mode = 1; - rq_regs->crq_expansion_mode = 1; - rq_regs->mrq_expansion_mode = 1; - - stored_swath_l_bytes = dml_get_det_stored_buffer_size_l_bytes(mode_lib, pipe_idx); - stored_swath_c_bytes = dml_get_det_stored_buffer_size_c_bytes(mode_lib, pipe_idx); - is_phantom_pipe = dml_get_is_phantom_pipe(display_cfg, mode_lib, pipe_idx); - - // Note: detile_buf_plane1_addr is in unit of 1KB - if (dual_plane) { - if (is_phantom_pipe) { - detile_buf_plane1_addr = (unsigned int)((1024.0 * 1024.0) / 2.0 / 1024.0); // half to chroma - } else { - if (stored_swath_l_bytes / stored_swath_c_bytes <= 1.5) { - detile_buf_plane1_addr = (unsigned int)(detile_buf_size_in_bytes / 2.0 / 1024.0); // half to chroma -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); -#endif - } else { - detile_buf_plane1_addr = (unsigned int)(dml_round_to_multiple((unsigned int)((2.0 * detile_buf_size_in_bytes) / 3.0), 1024, 0) / 1024.0); // 2/3 to luma -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); -#endif - } - } - } - rq_regs->plane1_base_address = detile_buf_plane1_addr; - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); - DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); - DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); - DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); - DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); - DML_LOG_VERBOSE("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); -#endif - //DML_LOG_VERBOSE_rq_regs_st(rq_regs); - DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); -} - -static void rq_dlg_get_dlg_reg( - struct dml2_core_internal_scratch *s, - struct dml2_display_dlg_regs *disp_dlg_regs, - struct dml2_display_ttu_regs *disp_ttu_regs, - const struct dml2_display_cfg *display_cfg, - const struct dml2_core_internal_display_mode_lib *mode_lib, - const unsigned int pipe_idx) -{ - struct dml2_core_shared_rq_dlg_get_dlg_reg_locals *l = &s->rq_dlg_get_dlg_reg_locals; - - memset(l, 0, sizeof(struct dml2_core_shared_rq_dlg_get_dlg_reg_locals)); - - DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); - - l->plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); - DML_ASSERT(l->plane_idx < DML2_MAX_PLANES); - - l->source_format = dml2_444_8; - l->odm_mode = dml2_odm_mode_bypass; - l->dual_plane = false; - l->htotal = 0; - l->hactive = 0; - l->hblank_end = 0; - l->vblank_end = 0; - l->interlaced = false; - l->pclk_freq_in_mhz = 0.0; - l->refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; - l->ref_freq_to_pix_freq = 0.0; - - if (l->plane_idx < DML2_MAX_PLANES) { - - l->timing = &display_cfg->stream_descriptors[display_cfg->plane_descriptors[l->plane_idx].stream_index].timing; - l->source_format = display_cfg->plane_descriptors[l->plane_idx].pixel_format; - l->odm_mode = mode_lib->mp.ODMMode[l->plane_idx]; - - l->dual_plane = is_dual_plane(l->source_format); - - l->htotal = l->timing->h_total; - l->hactive = l->timing->h_active; - l->hblank_end = l->timing->h_blank_end; - l->vblank_end = l->timing->v_blank_end; - l->interlaced = l->timing->interlaced; - l->pclk_freq_in_mhz = (double)l->timing->pixel_clock_khz / 1000; - l->ref_freq_to_pix_freq = l->refclk_freq_in_mhz / l->pclk_freq_in_mhz; - - DML_LOG_VERBOSE("DML_DLG::%s: plane_idx = %d\n", __func__, l->plane_idx); - DML_LOG_VERBOSE("DML_DLG: %s: htotal = %d\n", __func__, l->htotal); - DML_LOG_VERBOSE("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, l->refclk_freq_in_mhz); - DML_LOG_VERBOSE("DML_DLG: %s: dlg_ref_clk_mhz = %3.2f\n", __func__, display_cfg->overrides.hw.dlg_ref_clk_mhz); - DML_LOG_VERBOSE("DML_DLG: %s: soc.refclk_mhz = %d\n", __func__, mode_lib->soc.dchub_refclk_mhz); - DML_LOG_VERBOSE("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, l->pclk_freq_in_mhz); - DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); - DML_LOG_VERBOSE("DML_DLG: %s: interlaced = %d\n", __func__, l->interlaced); - - DML_ASSERT(l->refclk_freq_in_mhz != 0); - DML_ASSERT(l->pclk_freq_in_mhz != 0); - DML_ASSERT(l->ref_freq_to_pix_freq < 4.0); - - // Need to figure out which side of odm combine we're in - // Assume the pipe instance under the same plane is in order - - if (l->odm_mode == dml2_odm_mode_bypass) { - disp_dlg_regs->refcyc_h_blank_end = (unsigned int)((double)l->hblank_end * l->ref_freq_to_pix_freq); - } else if (l->odm_mode == dml2_odm_mode_combine_2to1 || l->odm_mode == dml2_odm_mode_combine_3to1 || l->odm_mode == dml2_odm_mode_combine_4to1) { - // find out how many pipe are in this plane - l->num_active_pipes = mode_lib->mp.num_active_pipes; - l->first_pipe_idx_in_plane = DML2_MAX_PLANES; - l->pipe_idx_in_combine = 0; // pipe index within the plane - l->odm_combine_factor = 2; - - if (l->odm_mode == dml2_odm_mode_combine_3to1) - l->odm_combine_factor = 3; - else if (l->odm_mode == dml2_odm_mode_combine_4to1) - l->odm_combine_factor = 4; - - for (unsigned int i = 0; i < l->num_active_pipes; i++) { - if (dml_get_plane_idx(mode_lib, i) == l->plane_idx) { - if (i < l->first_pipe_idx_in_plane) { - l->first_pipe_idx_in_plane = i; - } - } - } - l->pipe_idx_in_combine = pipe_idx - l->first_pipe_idx_in_plane; // DML assumes the pipes in the same plane will have continuous indexing (i.e. plane 0 use pipe 0, 1, and plane 1 uses pipe 2, 3, etc.) - - disp_dlg_regs->refcyc_h_blank_end = (unsigned int)(((double)l->hblank_end + (double)l->pipe_idx_in_combine * (double)l->hactive / (double)l->odm_combine_factor) * l->ref_freq_to_pix_freq); - DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); - DML_LOG_VERBOSE("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, l->first_pipe_idx_in_plane); - DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, l->pipe_idx_in_combine); - DML_LOG_VERBOSE("DML_DLG: %s: odm_combine_factor = %d\n", __func__, l->odm_combine_factor); - } - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); - - DML_ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int)math_pow(2, 13)); - - disp_dlg_regs->ref_freq_to_pix_freq = (unsigned int)(l->ref_freq_to_pix_freq * math_pow(2, 19)); - disp_dlg_regs->refcyc_per_htotal = (unsigned int)(l->ref_freq_to_pix_freq * (double)l->htotal * math_pow(2, 8)); - disp_dlg_regs->dlg_vblank_end = l->interlaced ? (l->vblank_end / 2) : l->vblank_end; // 15 bits - - l->min_ttu_vblank = mode_lib->mp.MinTTUVBlank[mode_lib->mp.pipe_plane[pipe_idx]]; - l->min_dst_y_next_start = (unsigned int)(mode_lib->mp.MIN_DST_Y_NEXT_START[mode_lib->mp.pipe_plane[pipe_idx]]); - - DML_LOG_VERBOSE("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, l->min_ttu_vblank); - DML_LOG_VERBOSE("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, l->min_dst_y_next_start); - DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); - - l->vready_after_vcount0 = (unsigned int)(mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[mode_lib->mp.pipe_plane[pipe_idx]]); - disp_dlg_regs->vready_after_vcount0 = l->vready_after_vcount0; - - DML_LOG_VERBOSE("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); - - l->dst_x_after_scaler = (unsigned int)(mode_lib->mp.DSTXAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); - l->dst_y_after_scaler = (unsigned int)(mode_lib->mp.DSTYAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); - - DML_LOG_VERBOSE("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, l->dst_x_after_scaler); - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, l->dst_y_after_scaler); - - l->dst_y_prefetch = mode_lib->mp.dst_y_prefetch[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_vm_vblank = mode_lib->mp.dst_y_per_vm_vblank[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_row_vblank = mode_lib->mp.dst_y_per_row_vblank[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_vm_flip = mode_lib->mp.dst_y_per_vm_flip[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_row_flip = mode_lib->mp.dst_y_per_row_flip[mode_lib->mp.pipe_plane[pipe_idx]]; - - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, l->dst_y_prefetch); - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, l->dst_y_per_vm_flip); - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, l->dst_y_per_row_flip); - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, l->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, l->dst_y_per_row_vblank); - - if (l->dst_y_prefetch > 0 && l->dst_y_per_vm_vblank > 0 && l->dst_y_per_row_vblank > 0) { - DML_ASSERT(l->dst_y_prefetch > (l->dst_y_per_vm_vblank + l->dst_y_per_row_vblank)); - } - - l->vratio_pre_l = mode_lib->mp.VRatioPrefetchY[mode_lib->mp.pipe_plane[pipe_idx]]; - l->vratio_pre_c = mode_lib->mp.VRatioPrefetchC[mode_lib->mp.pipe_plane[pipe_idx]]; - - DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, l->vratio_pre_l); - DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, l->vratio_pre_c); - - // Active - l->refcyc_per_line_delivery_pre_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_line_delivery_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_l); - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_l); - - l->refcyc_per_line_delivery_pre_c = 0.0; - l->refcyc_per_line_delivery_c = 0.0; - - if (l->dual_plane) { - l->refcyc_per_line_delivery_pre_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_line_delivery_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_c); - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_c); - } - - disp_dlg_regs->refcyc_per_vm_dmdata = (unsigned int)(mode_lib->mp.Tdmdl_vm[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); - disp_dlg_regs->dmdata_dl_delta = (unsigned int)(mode_lib->mp.Tdmdl[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); - - l->refcyc_per_req_delivery_pre_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_req_delivery_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_l); - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_l); - - l->refcyc_per_req_delivery_pre_c = 0.0; - l->refcyc_per_req_delivery_c = 0.0; - if (l->dual_plane) { - l->refcyc_per_req_delivery_pre_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_req_delivery_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_c); - DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_c); - } - - // TTU - Cursor - DML_ASSERT(display_cfg->plane_descriptors[l->plane_idx].cursor.num_cursors <= 1); - - // Assign to register structures - disp_dlg_regs->min_dst_y_next_start = (unsigned int)((double)l->min_dst_y_next_start * math_pow(2, 2)); - DML_ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int)math_pow(2, 18)); - - disp_dlg_regs->dst_y_after_scaler = l->dst_y_after_scaler; // in terms of line - disp_dlg_regs->refcyc_x_after_scaler = (unsigned int)((double)l->dst_x_after_scaler * l->ref_freq_to_pix_freq); // in terms of refclk - disp_dlg_regs->dst_y_prefetch = (unsigned int)(l->dst_y_prefetch * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_vm_vblank = (unsigned int)(l->dst_y_per_vm_vblank * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_row_vblank = (unsigned int)(l->dst_y_per_row_vblank * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_vm_flip = (unsigned int)(l->dst_y_per_vm_flip * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_row_flip = (unsigned int)(l->dst_y_per_row_flip * math_pow(2, 2)); - - disp_dlg_regs->vratio_prefetch = (unsigned int)(l->vratio_pre_l * math_pow(2, 19)); - disp_dlg_regs->vratio_prefetch_c = (unsigned int)(l->vratio_pre_c * math_pow(2, 19)); - - DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); - DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); - DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); - DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); - - disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(mode_lib->mp.TimePerVMGroupVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); - disp_dlg_regs->refcyc_per_vm_group_flip = (unsigned int)(mode_lib->mp.TimePerVMGroupFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); - disp_dlg_regs->refcyc_per_vm_req_vblank = (unsigned int)(mode_lib->mp.TimePerVMRequestVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz * math_pow(2, 10)); - disp_dlg_regs->refcyc_per_vm_req_flip = (unsigned int)(mode_lib->mp.TimePerVMRequestFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz * math_pow(2, 10)); - - l->dst_y_per_pte_row_nom_l = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_pte_row_nom_c = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C[mode_lib->mp.pipe_plane[pipe_idx]]; - l->refcyc_per_pte_group_nom_l = mode_lib->mp.time_per_pte_group_nom_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_pte_group_nom_c = mode_lib->mp.time_per_pte_group_nom_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_pte_group_vblank_l = mode_lib->mp.time_per_pte_group_vblank_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_pte_group_vblank_c = mode_lib->mp.time_per_pte_group_vblank_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_pte_group_flip_l = mode_lib->mp.time_per_pte_group_flip_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_pte_group_flip_c = mode_lib->mp.time_per_pte_group_flip_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_tdlut_group = mode_lib->mp.time_per_tdlut_group[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - disp_dlg_regs->dst_y_per_pte_row_nom_l = (unsigned int)(l->dst_y_per_pte_row_nom_l * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_pte_row_nom_c = (unsigned int)(l->dst_y_per_pte_row_nom_c * math_pow(2, 2)); - - disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int)(l->refcyc_per_pte_group_nom_l); - disp_dlg_regs->refcyc_per_pte_group_nom_c = (unsigned int)(l->refcyc_per_pte_group_nom_c); - disp_dlg_regs->refcyc_per_pte_group_vblank_l = (unsigned int)(l->refcyc_per_pte_group_vblank_l); - disp_dlg_regs->refcyc_per_pte_group_vblank_c = (unsigned int)(l->refcyc_per_pte_group_vblank_c); - disp_dlg_regs->refcyc_per_pte_group_flip_l = (unsigned int)(l->refcyc_per_pte_group_flip_l); - disp_dlg_regs->refcyc_per_pte_group_flip_c = (unsigned int)(l->refcyc_per_pte_group_flip_c); - disp_dlg_regs->refcyc_per_line_delivery_pre_l = (unsigned int)math_floor2(l->refcyc_per_line_delivery_pre_l, 1); - disp_dlg_regs->refcyc_per_line_delivery_l = (unsigned int)math_floor2(l->refcyc_per_line_delivery_l, 1); - disp_dlg_regs->refcyc_per_line_delivery_pre_c = (unsigned int)math_floor2(l->refcyc_per_line_delivery_pre_c, 1); - disp_dlg_regs->refcyc_per_line_delivery_c = (unsigned int)math_floor2(l->refcyc_per_line_delivery_c, 1); - - l->dst_y_per_meta_row_nom_l = mode_lib->mp.DST_Y_PER_META_ROW_NOM_L[mode_lib->mp.pipe_plane[pipe_idx]]; - l->dst_y_per_meta_row_nom_c = mode_lib->mp.DST_Y_PER_META_ROW_NOM_C[mode_lib->mp.pipe_plane[pipe_idx]]; - l->refcyc_per_meta_chunk_nom_l = mode_lib->mp.TimePerMetaChunkNominal[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_meta_chunk_nom_c = mode_lib->mp.TimePerChromaMetaChunkNominal[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_meta_chunk_vblank_l = mode_lib->mp.TimePerMetaChunkVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_meta_chunk_vblank_c = mode_lib->mp.TimePerChromaMetaChunkVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_meta_chunk_flip_l = mode_lib->mp.TimePerMetaChunkFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - l->refcyc_per_meta_chunk_flip_c = mode_lib->mp.TimePerChromaMetaChunkFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; - - disp_dlg_regs->dst_y_per_meta_row_nom_l = (unsigned int)(l->dst_y_per_meta_row_nom_l * math_pow(2, 2)); - disp_dlg_regs->dst_y_per_meta_row_nom_c = (unsigned int)(l->dst_y_per_meta_row_nom_c * math_pow(2, 2)); - disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (unsigned int)(l->refcyc_per_meta_chunk_nom_l); - disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (unsigned int)(l->refcyc_per_meta_chunk_nom_c); - disp_dlg_regs->refcyc_per_meta_chunk_vblank_l = (unsigned int)(l->refcyc_per_meta_chunk_vblank_l); - disp_dlg_regs->refcyc_per_meta_chunk_vblank_c = (unsigned int)(l->refcyc_per_meta_chunk_vblank_c); - disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (unsigned int)(l->refcyc_per_meta_chunk_flip_l); - disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (unsigned int)(l->refcyc_per_meta_chunk_flip_c); - - disp_dlg_regs->refcyc_per_tdlut_group = (unsigned int)(l->refcyc_per_tdlut_group); - disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off - - disp_ttu_regs->refcyc_per_req_delivery_pre_l = (unsigned int)(l->refcyc_per_req_delivery_pre_l * math_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_l = (unsigned int)(l->refcyc_per_req_delivery_l * math_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_pre_c = (unsigned int)(l->refcyc_per_req_delivery_pre_c * math_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_c = (unsigned int)(l->refcyc_per_req_delivery_c * math_pow(2, 10)); - disp_ttu_regs->qos_level_low_wm = 0; - - disp_ttu_regs->qos_level_high_wm = (unsigned int)(4.0 * (double)l->htotal * l->ref_freq_to_pix_freq); - - disp_ttu_regs->qos_level_flip = 14; - disp_ttu_regs->qos_level_fixed_l = 8; - disp_ttu_regs->qos_level_fixed_c = 8; - disp_ttu_regs->qos_ramp_disable_l = 0; - disp_ttu_regs->qos_ramp_disable_c = 0; - disp_ttu_regs->min_ttu_vblank = (unsigned int)(l->min_ttu_vblank * l->refclk_freq_in_mhz); - - // CHECK for HW registers' range, DML_ASSERT or clamp - DML_ASSERT(l->refcyc_per_req_delivery_pre_l < math_pow(2, 13)); - DML_ASSERT(l->refcyc_per_req_delivery_l < math_pow(2, 13)); - DML_ASSERT(l->refcyc_per_req_delivery_pre_c < math_pow(2, 13)); - DML_ASSERT(l->refcyc_per_req_delivery_c < math_pow(2, 13)); - if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(math_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_group_flip >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_group_flip = (unsigned int)(math_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_req_vblank >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_req_vblank = (unsigned int)(math_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_req_flip >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_req_flip = (unsigned int)(math_pow(2, 23) - 1); - - - DML_ASSERT(disp_dlg_regs->dst_y_after_scaler < (unsigned int)8); - DML_ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int)math_pow(2, 13)); - - if (disp_dlg_regs->dst_y_per_pte_row_nom_l >= (unsigned int)math_pow(2, 17)) { - DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_L %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_l, (unsigned int)math_pow(2, 17) - 1); - l->dst_y_per_pte_row_nom_l = (unsigned int)math_pow(2, 17) - 1; - } - if (l->dual_plane) { - if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int)math_pow(2, 17)) { - DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_C %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (unsigned int)math_pow(2, 17) - 1); - l->dst_y_per_pte_row_nom_c = (unsigned int)math_pow(2, 17) - 1; - } - } - - if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int)(math_pow(2, 23) - 1); - if (l->dual_plane) { - if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int)math_pow(2, 23)) - disp_dlg_regs->refcyc_per_pte_group_nom_c = (unsigned int)(math_pow(2, 23) - 1); - } - DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int)math_pow(2, 13)); - if (l->dual_plane) { - DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (unsigned int)math_pow(2, 13)); - } - - DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int)math_pow(2, 13)); - DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int)math_pow(2, 13)); - DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int)math_pow(2, 13)); - DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int)math_pow(2, 13)); - DML_ASSERT(disp_ttu_regs->qos_level_low_wm < (unsigned int)math_pow(2, 14)); - DML_ASSERT(disp_ttu_regs->qos_level_high_wm < (unsigned int)math_pow(2, 14)); - DML_ASSERT(disp_ttu_regs->min_ttu_vblank < (unsigned int)math_pow(2, 24)); - - DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); - - } -} - -static void rq_dlg_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *arb_param) -{ - double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; - - arb_param->max_req_outstanding = mode_lib->soc.max_outstanding_reqs; - arb_param->min_req_outstanding = mode_lib->soc.max_outstanding_reqs; // turn off the sat level feature if this set to max - arb_param->sdpif_request_rate_limit = (3 * mode_lib->ip.words_per_channel * mode_lib->soc.clk_table.dram_config.channel_count) / 4; - arb_param->sdpif_request_rate_limit = arb_param->sdpif_request_rate_limit < 96 ? 96 : arb_param->sdpif_request_rate_limit; - arb_param->sat_level_us = 60; - arb_param->hvm_max_qos_commit_threshold = 0xf; - arb_param->hvm_min_req_outstand_commit_threshold = 0xa; - arb_param->compbuf_reserved_space_kbytes = dml_get_compbuf_reserved_space_64b(mode_lib) * 64 / 1024; - arb_param->compbuf_size = mode_lib->mp.CompressedBufferSizeInkByte / mode_lib->ip.compressed_buffer_segment_size_in_kbytes; - arb_param->allow_sdpif_rate_limit_when_cstate_req = dml_get_hw_debug5(mode_lib); - arb_param->dcfclk_deep_sleep_hysteresis = dml_get_dcfclk_deep_sleep_hysteresis(mode_lib); - arb_param->pstate_stall_threshold = (unsigned int)(mode_lib->ip_caps.fams2.max_allow_delay_us * refclk_freq_in_mhz); - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: max_req_outstanding = %d\n", __func__, arb_param->max_req_outstanding); - DML_LOG_VERBOSE("DML::%s: sdpif_request_rate_limit = %d\n", __func__, arb_param->sdpif_request_rate_limit); - DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_kbytes = %d\n", __func__, arb_param->compbuf_reserved_space_kbytes); - DML_LOG_VERBOSE("DML::%s: allow_sdpif_rate_limit_when_cstate_req = %d\n", __func__, arb_param->allow_sdpif_rate_limit_when_cstate_req); - DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d\n", __func__, arb_param->dcfclk_deep_sleep_hysteresis); -#endif - -} - -void dml2_core_calcs_get_watermarks(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *out) -{ - rq_dlg_get_wm_regs(display_cfg, mode_lib, out); -} - -void dml2_core_calcs_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *out) -{ - rq_dlg_get_arb_params(display_cfg, mode_lib, out); -} - -void dml2_core_calcs_get_pipe_regs(const struct dml2_display_cfg *display_cfg, - struct dml2_core_internal_display_mode_lib *mode_lib, - struct dml2_dchub_per_pipe_register_set *out, int pipe_index) -{ - rq_dlg_get_rq_reg(&out->rq_regs, display_cfg, mode_lib, pipe_index); - rq_dlg_get_dlg_reg(&mode_lib->scratch, &out->dlg_regs, &out->ttu_regs, display_cfg, mode_lib, pipe_index); - out->det_size = dml_get_det_buffer_size_kbytes(mode_lib, pipe_index) / mode_lib->ip.config_return_buffer_segment_size_in_kbytes; -} - -void dml2_core_calcs_get_global_sync_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, union dml2_global_sync_programming *out, int pipe_index) -{ - out->dcn4x.vready_offset_pixels = dml_get_vready_offset(mode_lib, pipe_index); - out->dcn4x.vstartup_lines = dml_get_vstartup_calculated(mode_lib, pipe_index); - out->dcn4x.vupdate_offset_pixels = dml_get_vupdate_offset(mode_lib, pipe_index); - out->dcn4x.vupdate_vupdate_width_pixels = dml_get_vupdate_width(mode_lib, pipe_index); - out->dcn4x.pstate_keepout_start_lines = dml_get_pstate_keepout_dst_lines(mode_lib, pipe_index); -} - -void dml2_core_calcs_get_stream_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_per_stream_programming *out, int pipe_index) -{ - dml2_core_calcs_get_global_sync_programming(mode_lib, &out->global_sync, pipe_index); -} - -void dml2_core_calcs_get_global_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, - const struct display_configuation_with_meta *display_cfg, - struct dmub_cmd_fams2_global_config *fams2_global_config) -{ - fams2_global_config->features.bits.enable = display_cfg->stage3.fams2_required; - - if (fams2_global_config->features.bits.enable) { - fams2_global_config->features.bits.enable_stall_recovery = true; - fams2_global_config->features.bits.allow_delay_check_mode = FAMS2_ALLOW_DELAY_CHECK_FROM_START; - - fams2_global_config->max_allow_delay_us = mode_lib->ip_caps.fams2.max_allow_delay_us; - fams2_global_config->lock_wait_time_us = mode_lib->ip_caps.fams2.lock_timeout_us; - fams2_global_config->recovery_timeout_us = mode_lib->ip_caps.fams2.recovery_timeout_us; - fams2_global_config->hwfq_flip_programming_delay_us = mode_lib->ip_caps.fams2.flip_programming_delay_us; - - fams2_global_config->num_streams = display_cfg->display_config.num_streams; - } -} - -void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, - const struct display_configuation_with_meta *display_cfg, - union dmub_cmd_fams2_config *fams2_base_programming, - union dmub_cmd_fams2_config *fams2_sub_programming, - enum dml2_pstate_method pstate_method, - int plane_index) -{ - const struct dml2_plane_parameters *plane_descriptor = &display_cfg->display_config.plane_descriptors[plane_index]; - const struct dml2_stream_parameters *stream_descriptor = &display_cfg->display_config.stream_descriptors[plane_descriptor->stream_index]; - const struct dml2_pstate_meta *stream_pstate_meta = &display_cfg->stage3.stream_pstate_meta[plane_descriptor->stream_index]; - - struct dmub_fams2_cmd_stream_static_base_state *base_programming = &fams2_base_programming->stream_v1.base; - union dmub_fams2_cmd_stream_static_sub_state *sub_programming = &fams2_sub_programming->stream_v1.sub_state; - - unsigned int i; - - if (display_cfg->display_config.overrides.all_streams_blanked) { - /* stream is blanked, so do nothing */ - return; - } - - /* from display configuration */ - base_programming->htotal = (uint16_t)stream_descriptor->timing.h_total; - base_programming->vtotal = (uint16_t)stream_descriptor->timing.v_total; - base_programming->vblank_start = (uint16_t)(stream_pstate_meta->nom_vtotal - - stream_descriptor->timing.v_front_porch); - base_programming->vblank_end = (uint16_t)(stream_pstate_meta->nom_vtotal - - stream_descriptor->timing.v_front_porch - - stream_descriptor->timing.v_active); - base_programming->config.bits.is_drr = stream_descriptor->timing.drr_config.enabled; - - /* from meta */ - base_programming->otg_vline_time_ns = - (unsigned int)(stream_pstate_meta->otg_vline_time_us * 1000.0); - base_programming->scheduling_delay_otg_vlines = (uint8_t)stream_pstate_meta->scheduling_delay_otg_vlines; - base_programming->contention_delay_otg_vlines = (uint8_t)stream_pstate_meta->contention_delay_otg_vlines; - base_programming->vline_int_ack_delay_otg_vlines = (uint8_t)stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines; - base_programming->drr_keepout_otg_vline = (uint16_t)(stream_pstate_meta->nom_vtotal - - stream_descriptor->timing.v_front_porch - - stream_pstate_meta->method_drr.programming_delay_otg_vlines); - base_programming->allow_to_target_delay_otg_vlines = (uint8_t)stream_pstate_meta->allow_to_target_delay_otg_vlines; - base_programming->max_vtotal = (uint16_t)stream_pstate_meta->max_vtotal; - - /* from core */ - base_programming->config.bits.min_ttu_vblank_usable = true; - for (i = 0; i < display_cfg->display_config.num_planes; i++) { - /* check if all planes support p-state in blank */ - if (display_cfg->display_config.plane_descriptors[i].stream_index == plane_descriptor->stream_index && - mode_lib->mp.MinTTUVBlank[i] <= mode_lib->mp.Watermark.DRAMClockChangeWatermark) { - base_programming->config.bits.min_ttu_vblank_usable = false; - break; - } - } - - switch (pstate_method) { - case dml2_pstate_method_vactive: - case dml2_pstate_method_fw_vactive_drr: - /* legacy vactive */ - base_programming->type = FAMS2_STREAM_TYPE_VACTIVE; - sub_programming->legacy.vactive_det_fill_delay_otg_vlines = - (uint8_t)stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; - base_programming->allow_start_otg_vline = - (uint16_t)stream_pstate_meta->method_vactive.common.allow_start_otg_vline; - base_programming->allow_end_otg_vline = - (uint16_t)stream_pstate_meta->method_vactive.common.allow_end_otg_vline; - base_programming->config.bits.clamp_vtotal_min = true; - break; - case dml2_pstate_method_vblank: - case dml2_pstate_method_fw_vblank_drr: - /* legacy vblank */ - base_programming->type = FAMS2_STREAM_TYPE_VBLANK; - base_programming->allow_start_otg_vline = - (uint16_t)stream_pstate_meta->method_vblank.common.allow_start_otg_vline; - base_programming->allow_end_otg_vline = - (uint16_t)stream_pstate_meta->method_vblank.common.allow_end_otg_vline; - base_programming->config.bits.clamp_vtotal_min = true; - break; - case dml2_pstate_method_fw_drr: - /* drr */ - base_programming->type = FAMS2_STREAM_TYPE_DRR; - sub_programming->drr.programming_delay_otg_vlines = - (uint8_t)stream_pstate_meta->method_drr.programming_delay_otg_vlines; - sub_programming->drr.nom_stretched_vtotal = - (uint16_t)stream_pstate_meta->method_drr.stretched_vtotal; - base_programming->allow_start_otg_vline = - (uint16_t)stream_pstate_meta->method_drr.common.allow_start_otg_vline; - base_programming->allow_end_otg_vline = - (uint16_t)stream_pstate_meta->method_drr.common.allow_end_otg_vline; - /* drr only clamps to vtotal min for single display */ - base_programming->config.bits.clamp_vtotal_min = display_cfg->display_config.num_streams == 1; - sub_programming->drr.only_stretch_if_required = true; - break; - case dml2_pstate_method_fw_svp: - case dml2_pstate_method_fw_svp_drr: - /* subvp */ - base_programming->type = FAMS2_STREAM_TYPE_SUBVP; - sub_programming->subvp.vratio_numerator = - (uint16_t)(plane_descriptor->composition.scaler_info.plane0.v_ratio * 1000.0); - sub_programming->subvp.vratio_denominator = 1000; - sub_programming->subvp.programming_delay_otg_vlines = - (uint8_t)stream_pstate_meta->method_subvp.programming_delay_otg_vlines; - sub_programming->subvp.prefetch_to_mall_otg_vlines = - (uint8_t)stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines; - sub_programming->subvp.phantom_vtotal = - (uint16_t)stream_pstate_meta->method_subvp.phantom_vtotal; - sub_programming->subvp.phantom_vactive = - (uint16_t)stream_pstate_meta->method_subvp.phantom_vactive; - sub_programming->subvp.config.bits.is_multi_planar = - plane_descriptor->surface.plane1.height > 0; - sub_programming->subvp.config.bits.is_yuv420 = - plane_descriptor->pixel_format == dml2_420_8 || - plane_descriptor->pixel_format == dml2_420_10 || - plane_descriptor->pixel_format == dml2_420_12; - - base_programming->allow_start_otg_vline = - (uint16_t)stream_pstate_meta->method_subvp.common.allow_start_otg_vline; - base_programming->allow_end_otg_vline = - (uint16_t)stream_pstate_meta->method_subvp.common.allow_end_otg_vline; - base_programming->config.bits.clamp_vtotal_min = true; - break; - case dml2_pstate_method_reserved_hw: - case dml2_pstate_method_reserved_fw: - case dml2_pstate_method_reserved_fw_drr_clamped: - case dml2_pstate_method_reserved_fw_drr_var: - case dml2_pstate_method_na: - case dml2_pstate_method_count: - default: - /* this should never happen */ - break; - } -} - -void dml2_core_calcs_get_mcache_allocation(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_mcache_surface_allocation *out, int plane_idx) -{ - unsigned int n; - - out->num_mcaches_plane0 = dml_get_plane_num_mcaches_plane0(mode_lib, plane_idx); - out->num_mcaches_plane1 = dml_get_plane_num_mcaches_plane1(mode_lib, plane_idx); - out->shift_granularity.p0 = dml_get_plane_mcache_shift_granularity_plane0(mode_lib, plane_idx); - out->shift_granularity.p1 = dml_get_plane_mcache_shift_granularity_plane1(mode_lib, plane_idx); - - for (n = 0; n < out->num_mcaches_plane0; n++) - out->mcache_x_offsets_plane0[n] = dml_get_plane_array_mcache_offsets_plane0(mode_lib, plane_idx, n); - - for (n = 0; n < out->num_mcaches_plane1; n++) - out->mcache_x_offsets_plane1[n] = dml_get_plane_array_mcache_offsets_plane1(mode_lib, plane_idx, n); - - out->last_slice_sharing.mall_comb_mcache_p0 = dml_get_plane_mall_comb_mcache_l(mode_lib, plane_idx); - out->last_slice_sharing.mall_comb_mcache_p1 = dml_get_plane_mall_comb_mcache_c(mode_lib, plane_idx); - out->last_slice_sharing.plane0_plane1 = dml_get_plane_lc_comb_mcache(mode_lib, plane_idx); - out->informative.meta_row_bytes_plane0 = dml_get_plane_mcache_row_bytes_plane0(mode_lib, plane_idx); - out->informative.meta_row_bytes_plane1 = dml_get_plane_mcache_row_bytes_plane1(mode_lib, plane_idx); - - out->valid = true; -} - -void dml2_core_calcs_get_mall_allocation(struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int *out, int pipe_index) -{ - *out = dml_get_surface_size_in_mall_bytes(mode_lib, pipe_index); -} - -void dml2_core_calcs_get_plane_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_plane_support_info *out, int plane_idx) -{ - out->mall_svp_size_requirement_ways = 0; - - out->nominal_vblank_pstate_latency_hiding_us = - (int)(display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_idx].stream_index].timing.h_total / - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_idx].stream_index].timing.pixel_clock_khz / 1000) * mode_lib->ms.TWait[plane_idx]); - - out->dram_change_latency_hiding_margin_in_active = (int)mode_lib->ms.VActiveLatencyHidingMargin[plane_idx]; - - out->active_latency_hiding_us = (int)mode_lib->ms.VActiveLatencyHidingUs[plane_idx]; - - out->dram_change_vactive_det_fill_delay_us = (unsigned int)math_ceil(mode_lib->ms.dram_change_vactive_det_fill_delay_us[plane_idx]); -} - -void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_stream_support_info *out, int plane_index) -{ - double phantom_processing_delay_pix; - unsigned int phantom_processing_delay_lines; - unsigned int phantom_min_v_active_lines; - unsigned int phantom_v_active_lines; - unsigned int phantom_v_startup_lines; - unsigned int phantom_v_blank_lines; - unsigned int main_v_blank_lines; - unsigned int rem; - - phantom_processing_delay_pix = (double)((mode_lib->ip.subvp_fw_processing_delay_us + mode_lib->ip.subvp_pstate_allow_width_us) * - ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.pixel_clock_khz / 1000)); - phantom_processing_delay_lines = (unsigned int)(phantom_processing_delay_pix / (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.h_total); - dml2_core_div_rem(phantom_processing_delay_pix, - display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.h_total, - &rem); - if (rem) - phantom_processing_delay_lines++; - - phantom_v_startup_lines = dml_get_plane_max_vstartup_lines(mode_lib, plane_index); - phantom_min_v_active_lines = (unsigned int)math_ceil((double)dml_get_plane_subviewport_lines_needed_in_mall(mode_lib, plane_index) / - display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane0.v_ratio); - phantom_v_active_lines = phantom_processing_delay_lines + phantom_min_v_active_lines + mode_lib->ip.subvp_swath_height_margin_lines; - - // phantom_vblank = max(vbp(vstartup) + vactive + vfp(always 1) + vsync(can be 1), main_vblank) - phantom_v_blank_lines = phantom_v_startup_lines + 1 + 1; - main_v_blank_lines = display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.v_active; - if (phantom_v_blank_lines > main_v_blank_lines) - phantom_v_blank_lines = main_v_blank_lines; - - out->phantom_v_active = phantom_v_active_lines; - // phantom_vtotal = vactive + vblank - out->phantom_v_total = phantom_v_active_lines + phantom_v_blank_lines; - - out->phantom_min_v_active = phantom_min_v_active_lines; - out->phantom_v_startup = phantom_v_startup_lines; - - out->vblank_reserved_time_us = display_cfg->plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000; -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: subvp_fw_processing_delay_us = %d\n", __func__, mode_lib->ip.subvp_fw_processing_delay_us); - DML_LOG_VERBOSE("DML::%s: subvp_pstate_allow_width_us = %d\n", __func__, mode_lib->ip.subvp_pstate_allow_width_us); - DML_LOG_VERBOSE("DML::%s: subvp_swath_height_margin_lines = %d\n", __func__, mode_lib->ip.subvp_swath_height_margin_lines); - DML_LOG_VERBOSE("DML::%s: vblank_reserved_time_us = %u\n", __func__, out->vblank_reserved_time_us); -#endif -} - -void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_cfg_programming *out) -{ - unsigned int k, n; - - out->informative.mode_support_info.ModeIsSupported = mode_lib->ms.support.ModeSupport; - out->informative.mode_support_info.ImmediateFlipSupport = mode_lib->ms.support.ImmediateFlipSupport; - out->informative.mode_support_info.WritebackLatencySupport = mode_lib->ms.support.WritebackLatencySupport; - out->informative.mode_support_info.ScaleRatioAndTapsSupport = mode_lib->ms.support.ScaleRatioAndTapsSupport; - out->informative.mode_support_info.SourceFormatPixelAndScanSupport = mode_lib->ms.support.SourceFormatPixelAndScanSupport; - out->informative.mode_support_info.P2IWith420 = mode_lib->ms.support.P2IWith420; - out->informative.mode_support_info.DSCOnlyIfNecessaryWithBPP = false; - out->informative.mode_support_info.DSC422NativeNotSupported = mode_lib->ms.support.DSC422NativeNotSupported; - out->informative.mode_support_info.LinkRateDoesNotMatchDPVersion = mode_lib->ms.support.LinkRateDoesNotMatchDPVersion; - out->informative.mode_support_info.LinkRateForMultistreamNotIndicated = mode_lib->ms.support.LinkRateForMultistreamNotIndicated; - out->informative.mode_support_info.BPPForMultistreamNotIndicated = mode_lib->ms.support.BPPForMultistreamNotIndicated; - out->informative.mode_support_info.MultistreamWithHDMIOreDP = mode_lib->ms.support.MultistreamWithHDMIOreDP; - out->informative.mode_support_info.MSOOrODMSplitWithNonDPLink = mode_lib->ms.support.MSOOrODMSplitWithNonDPLink; - out->informative.mode_support_info.NotEnoughLanesForMSO = mode_lib->ms.support.NotEnoughLanesForMSO; - out->informative.mode_support_info.NumberOfOTGSupport = mode_lib->ms.support.NumberOfOTGSupport; - out->informative.mode_support_info.NumberOfHDMIFRLSupport = mode_lib->ms.support.NumberOfHDMIFRLSupport; - out->informative.mode_support_info.NumberOfDP2p0Support = mode_lib->ms.support.NumberOfDP2p0Support; - out->informative.mode_support_info.WritebackScaleRatioAndTapsSupport = mode_lib->ms.support.WritebackScaleRatioAndTapsSupport; - out->informative.mode_support_info.CursorSupport = mode_lib->ms.support.CursorSupport; - out->informative.mode_support_info.PitchSupport = mode_lib->ms.support.PitchSupport; - out->informative.mode_support_info.ViewportExceedsSurface = mode_lib->ms.support.ViewportExceedsSurface; - out->informative.mode_support_info.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = false; - out->informative.mode_support_info.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe; - out->informative.mode_support_info.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen; - out->informative.mode_support_info.InvalidCombinationOfMALLUseForPState = mode_lib->ms.support.InvalidCombinationOfMALLUseForPState; - out->informative.mode_support_info.ExceededMALLSize = mode_lib->ms.support.ExceededMALLSize; - out->informative.mode_support_info.EnoughWritebackUnits = mode_lib->ms.support.EnoughWritebackUnits; - out->informative.mode_support_info.temp_read_or_ppt_support = mode_lib->ms.support.temp_read_or_ppt_support; - out->informative.mode_support_info.g6_temp_read_support = mode_lib->ms.support.g6_temp_read_support; - - out->informative.mode_support_info.ExceededMultistreamSlots = mode_lib->ms.support.ExceededMultistreamSlots; - out->informative.mode_support_info.NotEnoughDSCUnits = mode_lib->ms.support.NotEnoughDSCUnits; - out->informative.mode_support_info.NotEnoughDSCSlices = mode_lib->ms.support.NotEnoughDSCSlices; - out->informative.mode_support_info.PixelsPerLinePerDSCUnitSupport = mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport; - out->informative.mode_support_info.DSCCLKRequiredMoreThanSupported = mode_lib->ms.support.DSCCLKRequiredMoreThanSupported; - out->informative.mode_support_info.DTBCLKRequiredMoreThanSupported = mode_lib->ms.support.DTBCLKRequiredMoreThanSupported; - out->informative.mode_support_info.LinkCapacitySupport = mode_lib->ms.support.LinkCapacitySupport; - - out->informative.mode_support_info.ROBSupport = mode_lib->ms.support.ROBSupport; - out->informative.mode_support_info.OutstandingRequestsSupport = mode_lib->ms.support.OutstandingRequestsSupport; - out->informative.mode_support_info.OutstandingRequestsUrgencyAvoidance = mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance; - out->informative.mode_support_info.PTEBufferSizeNotExceeded = mode_lib->ms.support.PTEBufferSizeNotExceeded; - out->informative.mode_support_info.DCCMetaBufferSizeNotExceeded = mode_lib->ms.support.DCCMetaBufferSizeNotExceeded; - - out->informative.mode_support_info.TotalVerticalActiveBandwidthSupport = mode_lib->ms.support.AvgBandwidthSupport; - out->informative.mode_support_info.VActiveBandwidthSupport = mode_lib->ms.support.UrgVactiveBandwidthSupport; - out->informative.mode_support_info.USRRetrainingSupport = mode_lib->ms.support.USRRetrainingSupport; - - out->informative.mode_support_info.PrefetchSupported = mode_lib->ms.support.PrefetchSupported; - out->informative.mode_support_info.DynamicMetadataSupported = mode_lib->ms.support.DynamicMetadataSupported; - out->informative.mode_support_info.VRatioInPrefetchSupported = mode_lib->ms.support.VRatioInPrefetchSupported; - out->informative.mode_support_info.DISPCLK_DPPCLK_Support = mode_lib->ms.support.DISPCLK_DPPCLK_Support; - out->informative.mode_support_info.TotalAvailablePipesSupport = mode_lib->ms.support.TotalAvailablePipesSupport; - out->informative.mode_support_info.NumberOfTDLUT33cubeSupport = mode_lib->ms.support.NumberOfTDLUT33cubeSupport; - out->informative.mode_support_info.ViewportSizeSupport = mode_lib->ms.support.ViewportSizeSupport; - out->informative.mode_support_info.qos_bandwidth_support = mode_lib->ms.support.qos_bandwidth_support; - out->informative.mode_support_info.dcfclk_support = mode_lib->ms.support.dcfclk_support; - - for (k = 0; k < out->display_config.num_planes; k++) { - - out->informative.mode_support_info.FCLKChangeSupport[k] = mode_lib->ms.support.FCLKChangeSupport[k]; - out->informative.mode_support_info.MPCCombineEnable[k] = mode_lib->ms.support.MPCCombineEnable[k]; - out->informative.mode_support_info.ODMMode[k] = mode_lib->ms.support.ODMMode[k]; - out->informative.mode_support_info.DPPPerSurface[k] = mode_lib->ms.support.DPPPerSurface[k]; - out->informative.mode_support_info.DSCEnabled[k] = mode_lib->ms.support.DSCEnabled[k]; - out->informative.mode_support_info.FECEnabled[k] = mode_lib->ms.support.FECEnabled[k]; - out->informative.mode_support_info.NumberOfDSCSlices[k] = mode_lib->ms.support.NumberOfDSCSlices[k]; - out->informative.mode_support_info.OutputBpp[k] = mode_lib->ms.support.OutputBpp[k]; - - if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_unknown) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_unknown; - else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_dp) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_dp; - else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_edp) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_edp; - else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_dp2p0) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_dp2p0; - else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_hdmi) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_hdmi; - else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_hdmifrl) - out->informative.mode_support_info.OutputType[k] = dml2_output_type_hdmifrl; - - if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_unknown) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_unknown; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr2) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr2; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr3) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr3; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr10) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr10; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr13p5) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr13p5; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr20) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr20; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_3x3) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_3x3; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_6x3) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_6x3; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_6x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_6x4; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_8x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_8x4; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_10x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_10x4; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_12x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_12x4; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_16x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_16x4; - else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_20x4) - out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_20x4; - - out->informative.mode_support_info.AlignedYPitch[k] = mode_lib->ms.support.AlignedYPitch[k]; - out->informative.mode_support_info.AlignedCPitch[k] = mode_lib->ms.support.AlignedCPitch[k]; - } - - out->informative.watermarks.urgent_us = dml_get_wm_urgent(mode_lib); - out->informative.watermarks.writeback_urgent_us = dml_get_wm_writeback_urgent(mode_lib); - out->informative.watermarks.writeback_pstate_us = dml_get_wm_writeback_dram_clock_change(mode_lib); - out->informative.watermarks.writeback_fclk_pstate_us = dml_get_wm_writeback_fclk_change(mode_lib); - - out->informative.watermarks.cstate_exit_us = dml_get_wm_stutter_exit(mode_lib); - out->informative.watermarks.cstate_enter_plus_exit_us = dml_get_wm_stutter_enter_exit(mode_lib); - out->informative.watermarks.z8_cstate_exit_us = dml_get_wm_z8_stutter_exit(mode_lib); - out->informative.watermarks.z8_cstate_enter_plus_exit_us = dml_get_wm_z8_stutter_enter_exit(mode_lib); - out->informative.watermarks.pstate_change_us = dml_get_wm_dram_clock_change(mode_lib); - out->informative.watermarks.fclk_pstate_change_us = dml_get_wm_fclk_change(mode_lib); - out->informative.watermarks.usr_retraining_us = dml_get_wm_usr_retraining(mode_lib); - out->informative.watermarks.temp_read_or_ppt_watermark_us = dml_get_wm_temp_read_or_ppt(mode_lib); - - out->informative.mall.total_surface_size_in_mall_bytes = 0; - out->informative.dpp.total_num_dpps_required = 0; - for (k = 0; k < out->display_config.num_planes; ++k) { - out->informative.mall.total_surface_size_in_mall_bytes += mode_lib->mp.SurfaceSizeInTheMALL[k]; - out->informative.dpp.total_num_dpps_required += mode_lib->mp.NoOfDPP[k]; - } - - out->informative.qos.min_return_latency_in_dcfclk = mode_lib->mp.min_return_latency_in_dcfclk; - out->informative.qos.urgent_latency_us = dml_get_urgent_latency(mode_lib); - - out->informative.qos.max_urgent_latency_us = dml_get_max_urgent_latency_us(mode_lib); - out->informative.qos.avg_non_urgent_latency_us = dml_get_avg_non_urgent_latency_us(mode_lib); - out->informative.qos.avg_urgent_latency_us = dml_get_avg_urgent_latency_us(mode_lib); - - out->informative.qos.wm_memory_trip_us = dml_get_wm_memory_trip(mode_lib); - out->informative.qos.meta_trip_memory_us = dml_get_meta_trip_memory_us(mode_lib); - out->informative.qos.fraction_of_urgent_bandwidth = dml_get_fraction_of_urgent_bandwidth(mode_lib); - out->informative.qos.fraction_of_urgent_bandwidth_immediate_flip = dml_get_fraction_of_urgent_bandwidth_imm_flip(mode_lib); - out->informative.qos.fraction_of_urgent_bandwidth_mall = dml_get_fraction_of_urgent_bandwidth_mall(mode_lib); - - out->informative.qos.avg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_avg_bw_required_sdp(mode_lib); - out->informative.qos.avg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_avg_bw_required_dram(mode_lib); - out->informative.qos.avg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_avg_bw_required_sdp(mode_lib); - out->informative.qos.avg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_avg_bw_required_dram(mode_lib); - - out->informative.qos.avg_bw_available.sys_active.sdp_bw_mbps = dml_get_sys_active_avg_bw_available_sdp(mode_lib); - out->informative.qos.avg_bw_available.sys_active.dram_bw_mbps = dml_get_sys_active_avg_bw_available_dram(mode_lib); - out->informative.qos.avg_bw_available.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_avg_bw_available_sdp(mode_lib); - out->informative.qos.avg_bw_available.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_avg_bw_available_dram(mode_lib); - - out->informative.qos.urg_bw_available.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_available_sdp(mode_lib); - out->informative.qos.urg_bw_available.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_available_dram(mode_lib); - out->informative.qos.urg_bw_available.sys_active.dram_vm_only_bw_mbps = dml_get_sys_active_urg_bw_available_dram_vm_only(mode_lib); - - out->informative.qos.urg_bw_available.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_available_sdp(mode_lib); - out->informative.qos.urg_bw_available.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_available_dram(mode_lib); - out->informative.qos.urg_bw_available.svp_prefetch.dram_vm_only_bw_mbps = dml_get_svp_prefetch_urg_bw_available_dram_vm_only(mode_lib); - - out->informative.qos.urg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_required_sdp(mode_lib); - out->informative.qos.urg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_required_dram(mode_lib); - out->informative.qos.urg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_required_sdp(mode_lib); - out->informative.qos.urg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_required_dram(mode_lib); - - out->informative.qos.non_urg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_non_urg_required_sdp(mode_lib); - out->informative.qos.non_urg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_non_urg_required_dram(mode_lib); - out->informative.qos.non_urg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_sdp(mode_lib); - out->informative.qos.non_urg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_dram(mode_lib); - - out->informative.qos.urg_bw_required_with_flip.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_required_sdp_flip(mode_lib); - out->informative.qos.urg_bw_required_with_flip.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_required_dram_flip(mode_lib); - out->informative.qos.urg_bw_required_with_flip.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_required_sdp_flip(mode_lib); - out->informative.qos.urg_bw_required_with_flip.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_required_dram_flip(mode_lib); - - out->informative.qos.non_urg_bw_required_with_flip.sys_active.sdp_bw_mbps = dml_get_sys_active_non_urg_required_sdp_flip(mode_lib); - out->informative.qos.non_urg_bw_required_with_flip.sys_active.dram_bw_mbps = dml_get_sys_active_non_urg_required_dram_flip(mode_lib); - out->informative.qos.non_urg_bw_required_with_flip.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_sdp_flip(mode_lib); - out->informative.qos.non_urg_bw_required_with_flip.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_dram_flip(mode_lib); - - out->informative.crb.comp_buffer_size_kbytes = dml_get_comp_buffer_size_kbytes(mode_lib); - out->informative.crb.UnboundedRequestEnabled = dml_get_unbounded_request_enabled(mode_lib); - - out->informative.crb.compbuf_reserved_space_64b = dml_get_compbuf_reserved_space_64b(mode_lib); - out->informative.misc.hw_debug5 = dml_get_hw_debug5(mode_lib); - out->informative.misc.dcfclk_deep_sleep_hysteresis = dml_get_dcfclk_deep_sleep_hysteresis(mode_lib); - - out->informative.power_management.stutter_efficiency = dml_get_stutter_efficiency_no_vblank(mode_lib); - out->informative.power_management.stutter_efficiency_with_vblank = dml_get_stutter_efficiency(mode_lib); - out->informative.power_management.stutter_num_bursts = dml_get_stutter_num_bursts(mode_lib); - - out->informative.power_management.z8.stutter_efficiency = dml_get_stutter_efficiency_z8(mode_lib); - out->informative.power_management.z8.stutter_efficiency_with_vblank = dml_get_stutter_efficiency(mode_lib); - out->informative.power_management.z8.stutter_num_bursts = dml_get_stutter_num_bursts_z8(mode_lib); - out->informative.power_management.z8.stutter_period = dml_get_stutter_period(mode_lib); - - out->informative.power_management.z8.bestcase.stutter_efficiency = dml_get_stutter_efficiency_z8_bestcase(mode_lib); - out->informative.power_management.z8.bestcase.stutter_num_bursts = dml_get_stutter_num_bursts_z8_bestcase(mode_lib); - out->informative.power_management.z8.bestcase.stutter_period = dml_get_stutter_period_bestcase(mode_lib); - - out->informative.misc.cstate_max_cap_mode = dml_get_cstate_max_cap_mode(mode_lib); - - out->min_clocks.dcn4x.dpprefclk_khz = (int unsigned)dml_get_global_dppclk_khz(mode_lib); - - out->informative.qos.max_active_fclk_change_latency_supported = dml_get_fclk_change_latency(mode_lib); - - out->informative.misc.LowestPrefetchMargin = 10 * 1000 * 1000; - - for (k = 0; k < out->display_config.num_planes; k++) { - - if ((out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us) - && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.fclk_change_blackout_us) - && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us)) - out->informative.misc.PrefetchMode[k] = 0; - else if ((out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.fclk_change_blackout_us) - && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us)) - out->informative.misc.PrefetchMode[k] = 1; - else if (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us) - out->informative.misc.PrefetchMode[k] = 2; - else - out->informative.misc.PrefetchMode[k] = 3; - - out->informative.misc.min_ttu_vblank_us[k] = mode_lib->mp.MinTTUVBlank[k]; - out->informative.mall.subviewport_lines_needed_in_mall[k] = mode_lib->mp.SubViewportLinesNeededInMALL[k]; - out->informative.crb.det_size_in_kbytes[k] = mode_lib->mp.DETBufferSizeInKByte[k]; - out->informative.crb.DETBufferSizeY[k] = mode_lib->mp.DETBufferSizeY[k]; - out->informative.misc.ImmediateFlipSupportedForPipe[k] = mode_lib->mp.ImmediateFlipSupportedForPipe[k]; - out->informative.misc.UsesMALLForStaticScreen[k] = mode_lib->mp.is_using_mall_for_ss[k]; - out->informative.plane_info[k].dpte_row_height_plane0 = mode_lib->mp.dpte_row_height[k]; - out->informative.plane_info[k].dpte_row_height_plane1 = mode_lib->mp.dpte_row_height_chroma[k]; - out->informative.plane_info[k].meta_row_height_plane0 = mode_lib->mp.meta_row_height[k]; - out->informative.plane_info[k].meta_row_height_plane1 = mode_lib->mp.meta_row_height_chroma[k]; - out->informative.dcc_control[k].max_uncompressed_block_plane0 = mode_lib->mp.DCCYMaxUncompressedBlock[k]; - out->informative.dcc_control[k].max_compressed_block_plane0 = mode_lib->mp.DCCYMaxCompressedBlock[k]; - out->informative.dcc_control[k].independent_block_plane0 = mode_lib->mp.DCCYIndependentBlock[k]; - out->informative.dcc_control[k].max_uncompressed_block_plane1 = mode_lib->mp.DCCCMaxUncompressedBlock[k]; - out->informative.dcc_control[k].max_compressed_block_plane1 = mode_lib->mp.DCCCMaxCompressedBlock[k]; - out->informative.dcc_control[k].independent_block_plane1 = mode_lib->mp.DCCCIndependentBlock[k]; - out->informative.misc.dst_x_after_scaler[k] = mode_lib->mp.DSTXAfterScaler[k]; - out->informative.misc.dst_y_after_scaler[k] = mode_lib->mp.DSTYAfterScaler[k]; - out->informative.misc.prefetch_source_lines_plane0[k] = mode_lib->mp.PrefetchSourceLinesY[k]; - out->informative.misc.prefetch_source_lines_plane1[k] = mode_lib->mp.PrefetchSourceLinesC[k]; - out->informative.misc.vready_at_or_after_vsync[k] = mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]; - out->informative.misc.min_dst_y_next_start[k] = mode_lib->mp.MIN_DST_Y_NEXT_START[k]; - out->informative.plane_info[k].swath_width_plane0 = mode_lib->mp.SwathWidthY[k]; - out->informative.plane_info[k].swath_height_plane0 = mode_lib->mp.SwathHeightY[k]; - out->informative.plane_info[k].swath_height_plane1 = mode_lib->mp.SwathHeightC[k]; - out->informative.misc.CursorDstXOffset[k] = mode_lib->mp.CursorDstXOffset[k]; - out->informative.misc.CursorDstYOffset[k] = mode_lib->mp.CursorDstYOffset[k]; - out->informative.misc.CursorChunkHDLAdjust[k] = mode_lib->mp.CursorChunkHDLAdjust[k]; - out->informative.misc.dpte_group_bytes[k] = mode_lib->mp.dpte_group_bytes[k]; - out->informative.misc.vm_group_bytes[k] = mode_lib->mp.vm_group_bytes[k]; - out->informative.misc.DisplayPipeRequestDeliveryTimeLuma[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma[k]; - out->informative.misc.DisplayPipeRequestDeliveryTimeChroma[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma[k]; - out->informative.misc.DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch[k]; - out->informative.misc.DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch[k]; - out->informative.misc.TimePerVMGroupVBlank[k] = mode_lib->mp.TimePerVMGroupVBlank[k]; - out->informative.misc.TimePerVMGroupFlip[k] = mode_lib->mp.TimePerVMGroupFlip[k]; - out->informative.misc.TimePerVMRequestVBlank[k] = mode_lib->mp.TimePerVMRequestVBlank[k]; - out->informative.misc.TimePerVMRequestFlip[k] = mode_lib->mp.TimePerVMRequestFlip[k]; - out->informative.misc.Tdmdl_vm[k] = mode_lib->mp.Tdmdl_vm[k]; - out->informative.misc.Tdmdl[k] = mode_lib->mp.Tdmdl[k]; - out->informative.misc.VStartup[k] = mode_lib->mp.VStartup[k]; - out->informative.misc.VUpdateOffsetPix[k] = mode_lib->mp.VUpdateOffsetPix[k]; - out->informative.misc.VUpdateWidthPix[k] = mode_lib->mp.VUpdateWidthPix[k]; - out->informative.misc.VReadyOffsetPix[k] = mode_lib->mp.VReadyOffsetPix[k]; - - out->informative.misc.DST_Y_PER_PTE_ROW_NOM_L[k] = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L[k]; - out->informative.misc.DST_Y_PER_PTE_ROW_NOM_C[k] = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C[k]; - out->informative.misc.time_per_pte_group_nom_luma[k] = mode_lib->mp.time_per_pte_group_nom_luma[k]; - out->informative.misc.time_per_pte_group_nom_chroma[k] = mode_lib->mp.time_per_pte_group_nom_chroma[k]; - out->informative.misc.time_per_pte_group_vblank_luma[k] = mode_lib->mp.time_per_pte_group_vblank_luma[k]; - out->informative.misc.time_per_pte_group_vblank_chroma[k] = mode_lib->mp.time_per_pte_group_vblank_chroma[k]; - out->informative.misc.time_per_pte_group_flip_luma[k] = mode_lib->mp.time_per_pte_group_flip_luma[k]; - out->informative.misc.time_per_pte_group_flip_chroma[k] = mode_lib->mp.time_per_pte_group_flip_chroma[k]; - out->informative.misc.VRatioPrefetchY[k] = mode_lib->mp.VRatioPrefetchY[k]; - out->informative.misc.VRatioPrefetchC[k] = mode_lib->mp.VRatioPrefetchC[k]; - out->informative.misc.DestinationLinesForPrefetch[k] = mode_lib->mp.dst_y_prefetch[k]; - out->informative.misc.DestinationLinesToRequestVMInVBlank[k] = mode_lib->mp.dst_y_per_vm_vblank[k]; - out->informative.misc.DestinationLinesToRequestRowInVBlank[k] = mode_lib->mp.dst_y_per_row_vblank[k]; - out->informative.misc.DestinationLinesToRequestVMInImmediateFlip[k] = mode_lib->mp.dst_y_per_vm_flip[k]; - out->informative.misc.DestinationLinesToRequestRowInImmediateFlip[k] = mode_lib->mp.dst_y_per_row_flip[k]; - out->informative.misc.DisplayPipeLineDeliveryTimeLuma[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLuma[k]; - out->informative.misc.DisplayPipeLineDeliveryTimeChroma[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChroma[k]; - out->informative.misc.DisplayPipeLineDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[k]; - out->informative.misc.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[k]; - - out->informative.misc.WritebackRequiredBandwidth = mode_lib->mp.TotalWRBandwidth / 1000.0; - out->informative.misc.WritebackAllowDRAMClockChangeEndPosition[k] = mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k]; - out->informative.misc.WritebackAllowFCLKChangeEndPosition[k] = mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k]; - out->informative.misc.DSCCLK_calculated[k] = mode_lib->mp.DSCCLK[k]; - out->informative.misc.BIGK_FRAGMENT_SIZE[k] = mode_lib->mp.BIGK_FRAGMENT_SIZE[k]; - out->informative.misc.PTE_BUFFER_MODE[k] = mode_lib->mp.PTE_BUFFER_MODE[k]; - out->informative.misc.DSCDelay[k] = mode_lib->mp.DSCDelay[k]; - out->informative.misc.MaxActiveDRAMClockChangeLatencySupported[k] = mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported[k]; - - if (mode_lib->mp.impacted_prefetch_margin_us[k] < out->informative.misc.LowestPrefetchMargin) - out->informative.misc.LowestPrefetchMargin = mode_lib->mp.impacted_prefetch_margin_us[k]; - } - - // For this DV informative layer, all pipes in the same planes will just use the same id - // will have the optimization and helper layer later on - // only work when we can have high "mcache" that fit everything without thrashing the cache - for (k = 0; k < out->display_config.num_planes; k++) { - out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane0 = dml_get_plane_num_mcaches_plane0(mode_lib, k); - out->informative.non_optimized_mcache_allocation[k].informative.meta_row_bytes_plane0 = dml_get_plane_mcache_row_bytes_plane0(mode_lib, k); - - for (n = 0; n < out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane0; n++) { - out->informative.non_optimized_mcache_allocation[k].mcache_x_offsets_plane0[n] = dml_get_plane_array_mcache_offsets_plane0(mode_lib, k, n); - out->informative.non_optimized_mcache_allocation[k].global_mcache_ids_plane0[n] = k; - } - - out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane1 = dml_get_plane_num_mcaches_plane1(mode_lib, k); - out->informative.non_optimized_mcache_allocation[k].informative.meta_row_bytes_plane1 = dml_get_plane_mcache_row_bytes_plane1(mode_lib, k); - - for (n = 0; n < out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane1; n++) { - out->informative.non_optimized_mcache_allocation[k].mcache_x_offsets_plane1[n] = dml_get_plane_array_mcache_offsets_plane1(mode_lib, k, n); - out->informative.non_optimized_mcache_allocation[k].global_mcache_ids_plane1[n] = k; - } - } - out->informative.qos.max_non_urgent_latency_us = dml_get_max_non_urgent_latency_us(mode_lib); - - if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { - if (((mode_lib->ip.rob_buffer_size_kbytes - mode_lib->ip.pixel_chunk_size_kbytes) * 1024 - / mode_lib->ms.support.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]) >= out->informative.qos.max_non_urgent_latency_us) { - out->informative.misc.ROBUrgencyAvoidance = true; - } else { - out->informative.misc.ROBUrgencyAvoidance = false; - } - } else { - out->informative.misc.ROBUrgencyAvoidance = true; - } -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h deleted file mode 100644 index 27ef0e096b25..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h +++ /dev/null @@ -1,39 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_CORE_DCN4_CALCS_H__ -#define __DML2_CORE_DCN4_CALCS_H__ - -#include "dml2_core_shared_types.h" - -struct dml2_dchub_watermark_regs; -struct dml2_display_arb_regs; -struct dml2_per_stream_programming; -struct dml2_dchub_per_pipe_register_set; -struct core_plane_support_info; -struct core_stream_support_info; -struct dml2_cursor_dlg_regs; -struct display_configuation_with_meta; - -unsigned int dml2_core_calcs_mode_support_ex(struct dml2_core_calcs_mode_support_ex *in_out_params); -bool dml2_core_calcs_mode_programming_ex(struct dml2_core_calcs_mode_programming_ex *in_out_params); -void dml2_core_calcs_get_watermarks(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *out); -void dml2_core_calcs_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *out); -void dml2_core_calcs_get_pipe_regs(const struct dml2_display_cfg *dml2_display_cfg, struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_per_pipe_register_set *out, int pipe_index); -void dml2_core_calcs_get_stream_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_per_stream_programming *out, int pipe_index); -void dml2_core_calcs_get_global_sync_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, union dml2_global_sync_programming *out, int pipe_index); -void dml2_core_calcs_get_mcache_allocation(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_mcache_surface_allocation *out, int plane_index); -void dml2_core_calcs_get_plane_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_plane_support_info *out, int plane_index); -void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_cfg_programming *out); -void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_stream_support_info *out, int plane_index); -void dml2_core_calcs_get_mall_allocation(struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int *out, int pipe_index); -void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, const struct display_configuation_with_meta *display_cfg, union dmub_cmd_fams2_config *fams2_base_programming, union dmub_cmd_fams2_config *fams2_sub_programming, enum dml2_pstate_method pstate_method, int plane_index); -void dml2_core_calcs_get_global_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, const struct display_configuation_with_meta *display_cfg, struct dmub_cmd_fams2_global_config *fams2_global_config); - -void dml2_core_calcs_get_dpte_row_height(unsigned int *dpte_row_height, struct dml2_core_internal_display_mode_lib *mode_lib, bool is_plane1, enum dml2_source_format_class SourcePixelFormat, enum dml2_swizzle_mode SurfaceTiling, enum dml2_rotation_angle ScanDirection, unsigned int pitch, unsigned int GPUVMMinPageSizeKBytes); -void dml2_core_calcs_cursor_dlg_reg(struct dml2_cursor_dlg_regs *cursor_dlg_regs, const struct dml2_get_cursor_dlg_reg *p); -const char *dml2_core_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type); -const char *dml2_core_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c deleted file mode 100644 index cc4f0663c6d6..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c +++ /dev/null @@ -1,39 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_core_factory.h" -#include "dml2_core_dcn4.h" -#include "dml2_external_lib_deps.h" - -bool dml2_core_create(enum dml2_project_id project_id, struct dml2_core_instance *out) -{ - bool result = false; - - if (!out) - return false; - - memset(out, 0, sizeof(struct dml2_core_instance)); - - out->project_id = project_id; - - switch (project_id) { - case dml2_project_dcn4x_stage1: - result = false; - break; - case dml2_project_dcn4x_stage2: - case dml2_project_dcn4x_stage2_auto_drr_svp: - out->initialize = &core_dcn4_initialize; - out->mode_support = &core_dcn4_mode_support; - out->mode_programming = &core_dcn4_mode_programming; - out->populate_informative = &core_dcn4_populate_informative; - out->calculate_mcache_allocation = &core_dcn4_calculate_mcache_allocation; - result = true; - break; - case dml2_project_invalid: - default: - break; - } - - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h deleted file mode 100644 index 411c514fe65c..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_CORE_FACTORY_H__ -#define __DML2_CORE_FACTORY_H__ - -#include "dml2_internal_shared_types.h" -#include "dml_top_types.h" - -bool dml2_core_create(enum dml2_project_id project_id, struct dml2_core_instance *out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h deleted file mode 100644 index 051c31ec2f0e..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ /dev/null @@ -1,2335 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_CORE_SHARED_TYPES_H__ -#define __DML2_CORE_SHARED_TYPES_H__ - -#include "dml2_external_lib_deps.h" -#include "dml_top_display_cfg_types.h" -#include "dml_top_types.h" - -#define __DML_VBA_DEBUG__ -#define __DML2_CALCS_MAX_VRATIO_PRE_OTO__ 4.0 // 0); - return dividend / divisor; - -} - -const char *dml2_core_utils_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type) -{ - switch (bw_type) { - case (dml2_core_internal_bw_sdp): - return("dml2_core_internal_bw_sdp"); - case (dml2_core_internal_bw_dram): - return("dml2_core_internal_bw_dram"); - case (dml2_core_internal_bw_max): - return("dml2_core_internal_bw_max"); - default: - return("dml2_core_internal_bw_unknown"); - } -} - -bool dml2_core_utils_is_420(enum dml2_source_format_class source_format) -{ - bool val = false; - - switch (source_format) { - case dml2_444_8: - val = 0; - break; - case dml2_444_16: - val = 0; - break; - case dml2_444_32: - val = 0; - break; - case dml2_444_64: - val = 0; - break; - case dml2_420_8: - val = 1; - break; - case dml2_420_10: - val = 1; - break; - case dml2_420_12: - val = 1; - break; - case dml2_rgbe_alpha: - val = 0; - break; - case dml2_rgbe: - val = 0; - break; - case dml2_mono_8: - val = 0; - break; - case dml2_mono_16: - val = 0; - break; - case dml2_422_planar_8: - val = 0; - break; - case dml2_422_planar_10: - val = 0; - break; - case dml2_422_planar_12: - val = 0; - break; - case dml2_422_packed_8: - val = 0; - break; - case dml2_422_packed_10: - val = 0; - break; - case dml2_422_packed_12: - val = 0; - break; - default: - DML_ASSERT(0); - break; - } - return val; -} - -bool dml2_core_utils_is_422_planar(enum dml2_source_format_class source_format) -{ - bool val = false; - - switch (source_format) { - case dml2_444_8: - val = 0; - break; - case dml2_444_16: - val = 0; - break; - case dml2_444_32: - val = 0; - break; - case dml2_444_64: - val = 0; - break; - case dml2_420_8: - val = 0; - break; - case dml2_420_10: - val = 0; - break; - case dml2_420_12: - val = 0; - break; - case dml2_rgbe_alpha: - val = 0; - break; - case dml2_rgbe: - val = 0; - break; - case dml2_mono_8: - val = 0; - break; - case dml2_mono_16: - val = 0; - break; - case dml2_422_planar_8: - val = 1; - break; - case dml2_422_planar_10: - val = 1; - break; - case dml2_422_planar_12: - val = 1; - break; - case dml2_422_packed_8: - val = 0; - break; - case dml2_422_packed_10: - val = 0; - break; - case dml2_422_packed_12: - val = 0; - break; - default: - DML_ASSERT(0); - break; - } - return val; -} - -bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format) -{ - bool val = false; - - switch (source_format) { - case dml2_444_8: - val = 0; - break; - case dml2_444_16: - val = 0; - break; - case dml2_444_32: - val = 0; - break; - case dml2_444_64: - val = 0; - break; - case dml2_420_8: - val = 0; - break; - case dml2_420_10: - val = 0; - break; - case dml2_420_12: - val = 0; - break; - case dml2_rgbe_alpha: - val = 0; - break; - case dml2_rgbe: - val = 0; - break; - case dml2_mono_8: - val = 0; - break; - case dml2_mono_16: - val = 0; - break; - case dml2_422_planar_8: - val = 0; - break; - case dml2_422_planar_10: - val = 0; - break; - case dml2_422_planar_12: - val = 0; - break; - case dml2_422_packed_8: - val = 1; - break; - case dml2_422_packed_10: - val = 1; - break; - case dml2_422_packed_12: - val = 1; - break; - default: - DML_ASSERT(0); - break; - } - return val; -} - -void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) -{ - DML_LOG_VERBOSE("DML: ===================================== \n"); - DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); - if (!fail_only || support->ScaleRatioAndTapsSupport == 0) - DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); - if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) - DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); - if (!fail_only || support->ViewportSizeSupport == 0) - DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); - if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) - DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); - if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) - DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); - if (!fail_only || support->BPPForMultistreamNotIndicated == 1) - DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); - if (!fail_only || support->MultistreamWithHDMIOreDP == 1) - DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); - if (!fail_only || support->ExceededMultistreamSlots == 1) - DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); - if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) - DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); - if (!fail_only || support->NotEnoughLanesForMSO == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); - if (!fail_only || support->P2IWith420 == 1) - DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); - if (!fail_only || support->DSC422NativeNotSupported == 1) - DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); - if (!fail_only || support->DSCSlicesODMModeSupported == 0) - DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); - if (!fail_only || support->NotEnoughDSCUnits == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); - if (!fail_only || support->NotEnoughDSCSlices == 1) - DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); - if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) - DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); - if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) - DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); - if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) - DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); - if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) - DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); - if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) - DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); - if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) - DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); - if (!fail_only || support->ROBSupport == 0) - DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); - if (!fail_only || support->OutstandingRequestsSupport == 0) - DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); - if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) - DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); - if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) - DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); - if (!fail_only || support->TotalAvailablePipesSupport == 0) - DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); - if (!fail_only || support->NumberOfOTGSupport == 0) - DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); - if (!fail_only || support->NumberOfHDMIFRLSupport == 0) - DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); - if (!fail_only || support->NumberOfDP2p0Support == 0) - DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); - if (!fail_only || support->EnoughWritebackUnits == 0) - DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); - if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) - DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); - if (!fail_only || support->WritebackLatencySupport == 0) - DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); - if (!fail_only || support->CursorSupport == 0) - DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); - if (!fail_only || support->PitchSupport == 0) - DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); - if (!fail_only || support->ViewportExceedsSurface == 1) - DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); - if (!fail_only || support->PrefetchSupported == 0) - DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); - if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) - DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); - if (!fail_only || support->AvgBandwidthSupport == 0) - DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); - if (!fail_only || support->DynamicMetadataSupported == 0) - DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); - if (!fail_only || support->VRatioInPrefetchSupported == 0) - DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); - if (!fail_only || support->PTEBufferSizeNotExceeded == 0) - DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); - if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) - DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); - if (!fail_only || support->ExceededMALLSize == 1) - DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); - if (!fail_only || support->g6_temp_read_support == 0) - DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); - if (!fail_only || support->ImmediateFlipSupport == 0) - DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); - if (!fail_only || support->LinkCapacitySupport == 0) - DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); - - if (!fail_only || support->ModeSupport == 0) - DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); - DML_LOG_VERBOSE("DML: ===================================== \n"); -} - -const char *dml2_core_utils_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type) -{ - switch (dml2_core_internal_soc_state_type) { - case (dml2_core_internal_soc_state_sys_idle): - return("dml2_core_internal_soc_state_sys_idle"); - case (dml2_core_internal_soc_state_sys_active): - return("dml2_core_internal_soc_state_sys_active"); - case (dml2_core_internal_soc_state_svp_prefetch): - return("dml2_core_internal_soc_state_svp_prefetch"); - case dml2_core_internal_soc_state_max: - default: - return("dml2_core_internal_soc_state_unknown"); - } -} - - -void dml2_core_utils_get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg) -{ - for (unsigned int k = 0; k < display_cfg->num_planes; k++) { - double bpc = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.bpc; - if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_disable) { - switch (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format) { - case dml2_444: - out_bpp[k] = bpc * 3; - break; - case dml2_s422: - out_bpp[k] = bpc * 2; - break; - case dml2_n422: - out_bpp[k] = bpc * 2; - break; - case dml2_420: - default: - out_bpp[k] = bpc * 1.5; - break; - } - } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable) { - out_bpp[k] = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.dsc_compressed_bpp_x16 / 16; - } else { - out_bpp[k] = 0; - } -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); - DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); - DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); -#endif - } -} - -unsigned int dml2_core_utils_round_to_multiple(unsigned int num, unsigned int multiple, bool up) -{ - unsigned int remainder; - - if (multiple == 0) - return num; - - remainder = num % multiple; - if (remainder == 0) - return num; - - if (up) - return (num + multiple - remainder); - else - return (num - remainder); -} - -unsigned int dml2_core_util_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info) -{ - unsigned int num_active_pipes = 0; - - for (unsigned int k = 0; k < num_planes; k++) { - num_active_pipes = num_active_pipes + (unsigned int)cfg_support_info->plane_support_info[k].dpps_used; - } - -#ifdef __DML_VBA_DEBUG__ - DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); -#endif - return num_active_pipes; -} - -void dml2_core_utils_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane) -{ - unsigned int pipe_idx = 0; - - for (unsigned int k = 0; k < DML2_MAX_PLANES; ++k) { - pipe_plane[k] = __DML2_CALCS_PIPE_NO_PLANE__; - } - - for (unsigned int plane_idx = 0; plane_idx < DML2_MAX_PLANES; plane_idx++) { - for (int i = 0; i < cfg_support_info->plane_support_info[plane_idx].dpps_used; i++) { - pipe_plane[pipe_idx] = plane_idx; - pipe_idx++; - } - } -} - -bool dml2_core_utils_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg) -{ - bool is_phantom = false; - - if (plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe || - plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) { - is_phantom = true; - } - - return is_phantom; -} - -unsigned int dml2_core_utils_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel) -{ - - if (sw_mode == dml2_sw_linear) - return 256; - else if (sw_mode == dml2_sw_256b_2d) - return 256; - else if (sw_mode == dml2_sw_4kb_2d) - return 4096; - else if (sw_mode == dml2_sw_64kb_2d) - return 65536; - else if (sw_mode == dml2_sw_256kb_2d) - return 262144; - else if (sw_mode == dml2_gfx11_sw_linear) - return 256; - else if (sw_mode == dml2_gfx11_sw_64kb_d) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_d_t) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_d_x) - return 65536; - else if (sw_mode == dml2_gfx11_sw_64kb_r_x) - return 65536; - else if (sw_mode == dml2_gfx11_sw_256kb_d_x) - return 262144; - else if (sw_mode == dml2_gfx11_sw_256kb_r_x) - return 262144; - else { - DML_ASSERT(0); - return 256; - }; -} - -bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel) -{ - return (byte_per_pixel != 2); -} - -bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode) -{ - return sw_mode == dml2_sw_linear; -}; - - -bool dml2_core_utils_is_vertical_rotation(enum dml2_rotation_angle Scan) -{ - bool is_vert = false; - if (Scan == dml2_rotation_90 || Scan == dml2_rotation_270) { - is_vert = true; - } else { - is_vert = false; - } - return is_vert; -} - -int unsigned dml2_core_utils_get_gfx_version(enum dml2_swizzle_mode sw_mode) -{ - int unsigned version = 0; - - if (sw_mode == dml2_sw_linear || - sw_mode == dml2_sw_256b_2d || - sw_mode == dml2_sw_4kb_2d || - sw_mode == dml2_sw_64kb_2d || - sw_mode == dml2_sw_256kb_2d) - version = 12; - else if (sw_mode == dml2_gfx11_sw_linear || - sw_mode == dml2_gfx11_sw_64kb_d || - sw_mode == dml2_gfx11_sw_64kb_d_t || - sw_mode == dml2_gfx11_sw_64kb_d_x || - sw_mode == dml2_gfx11_sw_64kb_r_x || - sw_mode == dml2_gfx11_sw_256kb_d_x || - sw_mode == dml2_gfx11_sw_256kb_r_x) - version = 11; - else { - DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); - DML_ASSERT(0); - } - - return version; -} - -unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params) -{ - unsigned int i; - unsigned int index = 0; - - for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { - DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); - - if (i == 0) - index = 0; - else - index = i - 1; - - if (uclk_freq_khz < per_uclk_dpm_params[i].minimum_uclk_khz || - per_uclk_dpm_params[i].minimum_uclk_khz == 0) { - break; - } - } -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); -#endif - return index; -} - -unsigned int dml2_core_utils_get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) -{ - unsigned int i; - bool clk_entry_found = false; - - for (i = 0; i < clk_table->uclk.num_clk_values; i++) { - DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); - - if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { - clk_entry_found = true; - break; - } - } - - if (!clk_entry_found) - DML_ASSERT(clk_entry_found); -#if defined(__DML_VBA_DEBUG__) - DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); - DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); -#endif - return i; -} - -bool dml2_core_utils_is_dual_plane(enum dml2_source_format_class source_format) -{ - bool ret_val = false; - - if (dml2_core_utils_is_420(source_format) || dml2_core_utils_is_422_planar(source_format) || (source_format == dml2_rgbe_alpha)) - ret_val = true; - - return ret_val; -} - -unsigned int dml2_core_utils_log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend) -{ - if (a == 0) - return 0; - - return (math_log2_approx(a) - subtrahend); -} - -static void create_phantom_stream_from_main_stream(struct dml2_stream_parameters *phantom, const struct dml2_stream_parameters *main, - const struct dml2_implicit_svp_meta *meta) -{ - memcpy(phantom, main, sizeof(struct dml2_stream_parameters)); - - phantom->timing.v_total = meta->v_total; - phantom->timing.v_active = meta->v_active; - phantom->timing.v_front_porch = meta->v_front_porch; - phantom->timing.v_blank_end = phantom->timing.v_total - phantom->timing.v_front_porch - phantom->timing.v_active; - phantom->timing.vblank_nom = phantom->timing.v_total - phantom->timing.v_active; - phantom->timing.drr_config.enabled = false; -} - -static void create_phantom_plane_from_main_plane(struct dml2_plane_parameters *phantom, const struct dml2_plane_parameters *main, - const struct dml2_stream_parameters *phantom_stream, int phantom_stream_index, const struct dml2_stream_parameters *main_stream) -{ - memcpy(phantom, main, sizeof(struct dml2_plane_parameters)); - - phantom->stream_index = phantom_stream_index; - phantom->overrides.refresh_from_mall = dml2_refresh_from_mall_mode_override_force_disable; - phantom->overrides.legacy_svp_config = dml2_svp_mode_override_phantom_pipe_no_data_return; - phantom->composition.viewport.plane0.height = (long int unsigned) math_min2(math_ceil2( - (double)main->composition.scaler_info.plane0.v_ratio * (double)phantom_stream->timing.v_active, 16.0), - (double)main->composition.viewport.plane0.height); - phantom->composition.viewport.plane1.height = (long int unsigned) math_min2(math_ceil2( - (double)main->composition.scaler_info.plane1.v_ratio * (double)phantom_stream->timing.v_active, 16.0), - (double)main->composition.viewport.plane1.height); - phantom->immediate_flip = false; - phantom->dynamic_meta_data.enable = false; - phantom->cursor.num_cursors = 0; - phantom->cursor.cursor_width = 0; - phantom->tdlut.setup_for_tdlut = false; -} - -void dml2_core_utils_expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, - struct dml2_core_scratch *scratch) -{ - unsigned int stream_index, plane_index; - const struct dml2_plane_parameters *main_plane; - const struct dml2_stream_parameters *main_stream; - const struct dml2_stream_parameters *phantom_stream; - - memcpy(svp_expanded_display_cfg, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); - memset(scratch->main_stream_index_from_svp_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); - memset(scratch->svp_stream_index_from_main_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); - memset(scratch->main_plane_index_to_phantom_plane_index, 0, sizeof(int) * DML2_MAX_PLANES); - - if (!display_cfg->display_config.overrides.enable_subvp_implicit_pmo) - return; - - /* disable unbounded requesting for all planes until stage 3 has been performed */ - if (!display_cfg->stage3.performed) { - svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.enable = true; - svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.value = false; - } - // Create the phantom streams - for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { - main_stream = &display_cfg->display_config.stream_descriptors[stream_index]; - scratch->main_stream_index_from_svp_stream_index[stream_index] = stream_index; - scratch->svp_stream_index_from_main_stream_index[stream_index] = stream_index; - - if (display_cfg->stage3.stream_svp_meta[stream_index].valid) { - // Create the phantom stream - create_phantom_stream_from_main_stream(&svp_expanded_display_cfg->stream_descriptors[svp_expanded_display_cfg->num_streams], - main_stream, &display_cfg->stage3.stream_svp_meta[stream_index]); - - // Associate this phantom stream to the main stream - scratch->main_stream_index_from_svp_stream_index[svp_expanded_display_cfg->num_streams] = stream_index; - scratch->svp_stream_index_from_main_stream_index[stream_index] = svp_expanded_display_cfg->num_streams; - - // Increment num streams - svp_expanded_display_cfg->num_streams++; - } - } - - // Create the phantom planes - for (plane_index = 0; plane_index < display_cfg->display_config.num_planes; plane_index++) { - main_plane = &display_cfg->display_config.plane_descriptors[plane_index]; - - if (display_cfg->stage3.stream_svp_meta[main_plane->stream_index].valid) { - main_stream = &display_cfg->display_config.stream_descriptors[main_plane->stream_index]; - phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index]]; - create_phantom_plane_from_main_plane(&svp_expanded_display_cfg->plane_descriptors[svp_expanded_display_cfg->num_planes], - main_plane, phantom_stream, scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index], main_stream); - - // Associate this phantom plane to the main plane - scratch->phantom_plane_index_to_main_plane_index[svp_expanded_display_cfg->num_planes] = plane_index; - scratch->main_plane_index_to_phantom_plane_index[plane_index] = svp_expanded_display_cfg->num_planes; - - // Increment num planes - svp_expanded_display_cfg->num_planes++; - - // Adjust the main plane settings - svp_expanded_display_cfg->plane_descriptors[plane_index].overrides.legacy_svp_config = dml2_svp_mode_override_main_pipe; - } - } -} - -bool dml2_core_utils_is_stream_encoder_required(const struct dml2_stream_parameters *stream_descriptor) -{ - switch (stream_descriptor->output.output_encoder) { - case dml2_dp: - case dml2_dp2p0: - case dml2_edp: - case dml2_hdmi: - case dml2_hdmifrl: - return true; - case dml2_none: - default: - return false; - } -} -bool dml2_core_utils_is_encoder_dsc_capable(const struct dml2_stream_parameters *stream_descriptor) -{ - switch (stream_descriptor->output.output_encoder) { - case dml2_dp: - case dml2_dp2p0: - case dml2_edp: - case dml2_hdmifrl: - return true; - case dml2_hdmi: - case dml2_none: - default: - return false; - } -} - - -bool dml2_core_utils_is_dio_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) -{ - switch (stream_descriptor->output.output_encoder) { - case dml2_dp: - case dml2_edp: - return true; - case dml2_dp2p0: - case dml2_hdmi: - case dml2_hdmifrl: - case dml2_none: - default: - return false; - } -} - -bool dml2_core_utils_is_hpo_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) -{ - switch (stream_descriptor->output.output_encoder) { - case dml2_dp2p0: - return true; - case dml2_dp: - case dml2_edp: - case dml2_hdmi: - case dml2_hdmifrl: - case dml2_none: - default: - return false; - } -} - -bool dml2_core_utils_is_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) -{ - return dml2_core_utils_is_dio_dp_encoder(stream_descriptor) - || dml2_core_utils_is_hpo_dp_encoder(stream_descriptor); -} - - -bool dml2_core_utils_is_dp_8b_10b_link_rate(enum dml2_output_link_dp_rate rate) -{ - switch (rate) { - case dml2_dp_rate_hbr: - case dml2_dp_rate_hbr2: - case dml2_dp_rate_hbr3: - return true; - case dml2_dp_rate_na: - case dml2_dp_rate_uhbr10: - case dml2_dp_rate_uhbr13p5: - case dml2_dp_rate_uhbr20: - default: - return false; - } -} - -bool dml2_core_utils_is_dp_128b_132b_link_rate(enum dml2_output_link_dp_rate rate) -{ - switch (rate) { - case dml2_dp_rate_uhbr10: - case dml2_dp_rate_uhbr13p5: - case dml2_dp_rate_uhbr20: - return true; - case dml2_dp_rate_hbr: - case dml2_dp_rate_hbr2: - case dml2_dp_rate_hbr3: - case dml2_dp_rate_na: - default: - return false; - } -} - -bool dml2_core_utils_is_odm_split(enum dml2_odm_mode odm_mode) -{ - switch (odm_mode) { - case dml2_odm_mode_split_1to2: - case dml2_odm_mode_mso_1to2: - case dml2_odm_mode_mso_1to4: - return true; - case dml2_odm_mode_auto: - case dml2_odm_mode_bypass: - case dml2_odm_mode_combine_2to1: - case dml2_odm_mode_combine_3to1: - case dml2_odm_mode_combine_4to1: - default: - return false; - } -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h deleted file mode 100644 index 95f0d017add4..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h +++ /dev/null @@ -1,43 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_CORE_UTILS_H__ -#define __DML2_CORE_UTILS_H__ -#include "dml2_internal_shared_types.h" -#include "dml2_debug.h" -#include "lib_float_math.h" - -double dml2_core_utils_div_rem(double dividend, unsigned int divisor, unsigned int *remainder); -const char *dml2_core_utils_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type); -bool dml2_core_utils_is_420(enum dml2_source_format_class source_format); -bool dml2_core_utils_is_422_planar(enum dml2_source_format_class source_format); -bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format); -void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only); -const char *dml2_core_utils_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type); -void dml2_core_utils_get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg); -unsigned int dml2_core_utils_round_to_multiple(unsigned int num, unsigned int multiple, bool up); -unsigned int dml2_core_util_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info); -void dml2_core_utils_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane); -bool dml2_core_utils_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg); -unsigned int dml2_core_utils_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel); -bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel); -bool dml2_core_utils_is_vertical_rotation(enum dml2_rotation_angle Scan); -bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode); -int unsigned dml2_core_utils_get_gfx_version(enum dml2_swizzle_mode sw_mode); -unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params); -unsigned int dml2_core_utils_get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table); -bool dml2_core_utils_is_dual_plane(enum dml2_source_format_class source_format); -unsigned int dml2_core_utils_log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend); -void dml2_core_utils_expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, - struct dml2_core_scratch *scratch); -bool dml2_core_utils_is_stream_encoder_required(const struct dml2_stream_parameters *stream_descriptor); -bool dml2_core_utils_is_encoder_dsc_capable(const struct dml2_stream_parameters *stream_descriptor); -bool dml2_core_utils_is_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); -bool dml2_core_utils_is_dio_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); -bool dml2_core_utils_is_hpo_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); -bool dml2_core_utils_is_dp_8b_10b_link_rate(enum dml2_output_link_dp_rate rate); -bool dml2_core_utils_is_dp_128b_132b_link_rate(enum dml2_output_link_dp_rate rate); -bool dml2_core_utils_is_odm_split(enum dml2_odm_mode odm_mode); - -#endif /* __DML2_CORE_UTILS_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c deleted file mode 100644 index 22969a533a7b..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c +++ /dev/null @@ -1,785 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_dpmm_dcn4.h" -#include "dml2_internal_shared_types.h" -#include "dml_top_types.h" -#include "lib_float_math.h" - -static double dram_bw_kbps_to_uclk_khz(unsigned long long bandwidth_kbps, const struct dml2_dram_params *dram_config) -{ - double uclk_khz = 0; - unsigned long uclk_mbytes_per_tick = 0; - - uclk_mbytes_per_tick = dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock; - - uclk_khz = (double)bandwidth_kbps / uclk_mbytes_per_tick; - - return uclk_khz; -} - -static void get_minimum_clocks_for_latency(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, - double *uclk, - double *fclk, - double *dcfclk) -{ - int min_clock_index_for_latency; - - if (in_out->display_cfg->stage3.success) - min_clock_index_for_latency = in_out->display_cfg->stage3.min_clk_index_for_latency; - else - min_clock_index_for_latency = in_out->display_cfg->stage1.min_clk_index_for_latency; - - *dcfclk = in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].min_dcfclk_khz; - *fclk = in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].min_fclk_khz; - *uclk = dram_bw_kbps_to_uclk_khz(in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].pre_derate_dram_bw_kbps, - &in_out->soc_bb->clk_table.dram_config); -} - -static unsigned long dml_round_up(double a) -{ - if (a - (unsigned long)a > 0) { - return ((unsigned long)a) + 1; - } - return (unsigned long)a; -} - -static void calculate_system_active_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - double min_uclk_avg, min_uclk_urgent, min_uclk_bw; - double min_fclk_avg, min_fclk_urgent, min_fclk_bw; - double min_dcfclk_avg, min_dcfclk_urgent, min_dcfclk_bw; - double min_uclk_latency, min_fclk_latency, min_dcfclk_latency; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100); - - min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - if (in_out->display_cfg->display_config.hostvm_enable) - min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel_and_vm / 100); - else - min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100); - - min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; - - min_fclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100); - - min_fclk_urgent = (double)mode_support_result->global.active.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100); - - min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; - - min_dcfclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100); - - min_dcfclk_urgent = (double)mode_support_result->global.active.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100); - - min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; - - get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); - - in_out->programming->min_clocks.dcn4x.active.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); - in_out->programming->min_clocks.dcn4x.active.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); - in_out->programming->min_clocks.dcn4x.active.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); -} - -static void calculate_svp_prefetch_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - double min_uclk_avg, min_uclk_urgent, min_uclk_bw; - double min_fclk_avg, min_fclk_urgent, min_fclk_bw; - double min_dcfclk_avg, min_dcfclk_urgent, min_dcfclk_bw; - double min_fclk_latency, min_dcfclk_latency; - double min_uclk_latency; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - /* assumes DF throttling is enabled */ - min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.dram_derate_percent_pixel / 100); - - min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100); - - min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; - - min_fclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.fclk_derate_percent / 100); - - min_fclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.fclk_derate_percent / 100); - - min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; - - min_dcfclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.dcfclk_derate_percent / 100); - - min_dcfclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dcfclk_derate_percent / 100); - - min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; - - get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); - - in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); - in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); - in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); - - /* assumes DF throttling is disabled */ - min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100); - - min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100); - - min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; - - min_fclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100); - - min_fclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100); - - min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; - - min_dcfclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100); - - min_dcfclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100); - - min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; - - get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); - - in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); - in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); - in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); -} - -static void calculate_idle_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - double min_uclk_avg; - double min_fclk_avg; - double min_dcfclk_avg; - double min_uclk_latency, min_fclk_latency, min_dcfclk_latency; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); - min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.dram_derate_percent_pixel / 100); - - min_fclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; - min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.fclk_derate_percent / 100); - - min_dcfclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; - min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.dcfclk_derate_percent / 100); - - get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); - - in_out->programming->min_clocks.dcn4x.idle.uclk_khz = dml_round_up(min_uclk_avg > min_uclk_latency ? min_uclk_avg : min_uclk_latency); - in_out->programming->min_clocks.dcn4x.idle.fclk_khz = dml_round_up(min_fclk_avg > min_fclk_latency ? min_fclk_avg : min_fclk_latency); - in_out->programming->min_clocks.dcn4x.idle.dcfclk_khz = dml_round_up(min_dcfclk_avg > min_dcfclk_latency ? min_dcfclk_avg : min_dcfclk_latency); -} - -static bool add_margin_and_round_to_dfs_grainularity(double clock_khz, double margin, unsigned long vco_freq_khz, unsigned long *rounded_khz, uint32_t *divider_id) -{ - enum dentist_divider_range { - DFS_DIVIDER_RANGE_1_START = 8, /* 2.00 */ - DFS_DIVIDER_RANGE_1_STEP = 1, /* 0.25 */ - DFS_DIVIDER_RANGE_2_START = 64, /* 16.00 */ - DFS_DIVIDER_RANGE_2_STEP = 2, /* 0.50 */ - DFS_DIVIDER_RANGE_3_START = 128, /* 32.00 */ - DFS_DIVIDER_RANGE_3_STEP = 4, /* 1.00 */ - DFS_DIVIDER_RANGE_4_START = 248, /* 62.00 */ - DFS_DIVIDER_RANGE_4_STEP = 264, /* 66.00 */ - DFS_DIVIDER_RANGE_SCALE_FACTOR = 4 - }; - - enum DFS_base_divider_id { - DFS_BASE_DID_1 = 0x08, - DFS_BASE_DID_2 = 0x40, - DFS_BASE_DID_3 = 0x60, - DFS_BASE_DID_4 = 0x7e, - DFS_MAX_DID = 0x7f - }; - - unsigned int divider; - - if (clock_khz < 1 || vco_freq_khz < 1 || clock_khz > vco_freq_khz) - return false; - - clock_khz *= 1.0 + margin; - - divider = (unsigned int)((int)DFS_DIVIDER_RANGE_SCALE_FACTOR * (vco_freq_khz / clock_khz)); - - /* we want to floor here to get higher clock than required rather than lower */ - if (divider < DFS_DIVIDER_RANGE_2_START) { - if (divider < DFS_DIVIDER_RANGE_1_START) - *divider_id = DFS_BASE_DID_1; - else - *divider_id = DFS_BASE_DID_1 + ((divider - DFS_DIVIDER_RANGE_1_START) / DFS_DIVIDER_RANGE_1_STEP); - } else if (divider < DFS_DIVIDER_RANGE_3_START) { - *divider_id = DFS_BASE_DID_2 + ((divider - DFS_DIVIDER_RANGE_2_START) / DFS_DIVIDER_RANGE_2_STEP); - } else if (divider < DFS_DIVIDER_RANGE_4_START) { - *divider_id = DFS_BASE_DID_3 + ((divider - DFS_DIVIDER_RANGE_3_START) / DFS_DIVIDER_RANGE_3_STEP); - } else { - *divider_id = DFS_BASE_DID_4 + ((divider - DFS_DIVIDER_RANGE_4_START) / DFS_DIVIDER_RANGE_4_STEP); - if (*divider_id > DFS_MAX_DID) - *divider_id = DFS_MAX_DID; - } - - *rounded_khz = vco_freq_khz * DFS_DIVIDER_RANGE_SCALE_FACTOR / divider; - - return true; -} - -static bool round_to_non_dfs_granularity(unsigned long dispclk_khz, unsigned long dpprefclk_khz, unsigned long dtbrefclk_khz, - unsigned long *rounded_dispclk_khz, unsigned long *rounded_dpprefclk_khz, unsigned long *rounded_dtbrefclk_khz) -{ - unsigned long pll_frequency_khz; - - pll_frequency_khz = (unsigned long) math_max2(600000, math_ceil2(math_max3(dispclk_khz, dpprefclk_khz, dtbrefclk_khz), 1000)); - - *rounded_dispclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dispclk_khz, 32); - - *rounded_dpprefclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dpprefclk_khz, 32); - - if (dtbrefclk_khz > 0) { - *rounded_dtbrefclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dtbrefclk_khz, 32); - } else { - *rounded_dtbrefclk_khz = 0; - } - - return true; -} - -static bool round_up_and_copy_to_next_dpm(unsigned long min_value, unsigned long *rounded_value, const struct dml2_clk_table *clock_table) -{ - bool result = false; - int index = 0; - - if (clock_table->num_clk_values > 2) { - while (index < clock_table->num_clk_values && clock_table->clk_values_khz[index] < min_value) - index++; - - if (index < clock_table->num_clk_values) { - *rounded_value = clock_table->clk_values_khz[index]; - result = true; - } - } else if (clock_table->clk_values_khz[clock_table->num_clk_values - 1] >= min_value) { - *rounded_value = min_value; - result = true; - } - return result; -} - -static bool round_up_to_next_dpm(unsigned long *clock_value, const struct dml2_clk_table *clock_table) -{ - return round_up_and_copy_to_next_dpm(*clock_value, clock_value, clock_table); -} - -static bool map_soc_min_clocks_to_dpm_fine_grained(struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) -{ - bool result; - - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.dcfclk_khz, &state_table->dcfclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.fclk_khz, &state_table->fclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.uclk_khz, &state_table->uclk); - - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.dcfclk_khz, &state_table->dcfclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.fclk_khz, &state_table->fclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.uclk_khz, &state_table->uclk); - - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.dcfclk_khz, &state_table->dcfclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.fclk_khz, &state_table->fclk); - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.uclk_khz, &state_table->uclk); - - /* these clocks are optional, so they can fail to map, in which case map all to 0 */ - if (result) { - if (!round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz, &state_table->dcfclk) || - !round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz, &state_table->fclk) || - !round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz, &state_table->uclk)) { - display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz = 0; - display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz = 0; - display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz = 0; - } - } - - return result; -} - -static bool map_soc_min_clocks_to_dpm_coarse_grained(struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) -{ - bool result; - int index; - - result = false; - for (index = 0; index < state_table->uclk.num_clk_values; index++) { - if (display_cfg->min_clocks.dcn4x.active.dcfclk_khz <= state_table->dcfclk.clk_values_khz[index] && - display_cfg->min_clocks.dcn4x.active.fclk_khz <= state_table->fclk.clk_values_khz[index] && - display_cfg->min_clocks.dcn4x.active.uclk_khz <= state_table->uclk.clk_values_khz[index]) { - display_cfg->min_clocks.dcn4x.active.dcfclk_khz = state_table->dcfclk.clk_values_khz[index]; - display_cfg->min_clocks.dcn4x.active.fclk_khz = state_table->fclk.clk_values_khz[index]; - display_cfg->min_clocks.dcn4x.active.uclk_khz = state_table->uclk.clk_values_khz[index]; - result = true; - break; - } - } - - if (result) { - result = false; - for (index = 0; index < state_table->uclk.num_clk_values; index++) { - if (display_cfg->min_clocks.dcn4x.idle.dcfclk_khz <= state_table->dcfclk.clk_values_khz[index] && - display_cfg->min_clocks.dcn4x.idle.fclk_khz <= state_table->fclk.clk_values_khz[index] && - display_cfg->min_clocks.dcn4x.idle.uclk_khz <= state_table->uclk.clk_values_khz[index]) { - display_cfg->min_clocks.dcn4x.idle.dcfclk_khz = state_table->dcfclk.clk_values_khz[index]; - display_cfg->min_clocks.dcn4x.idle.fclk_khz = state_table->fclk.clk_values_khz[index]; - display_cfg->min_clocks.dcn4x.idle.uclk_khz = state_table->uclk.clk_values_khz[index]; - result = true; - break; - } - } - } - - // SVP is not supported on any coarse grained SoCs - display_cfg->min_clocks.dcn4x.svp_prefetch.dcfclk_khz = 0; - display_cfg->min_clocks.dcn4x.svp_prefetch.fclk_khz = 0; - display_cfg->min_clocks.dcn4x.svp_prefetch.uclk_khz = 0; - - return result; -} - -static bool map_min_clocks_to_dpm(const struct dml2_core_mode_support_result *mode_support_result, struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) -{ - bool result = false; - bool dcfclk_fine_grained = false, fclk_fine_grained = false, clock_state_count_identical = false; - unsigned int i; - - if (!state_table || !display_cfg) - return false; - - if (state_table->dcfclk.num_clk_values == 2) { - dcfclk_fine_grained = true; - } - - if (state_table->fclk.num_clk_values == 2) { - fclk_fine_grained = true; - } - - if (state_table->fclk.num_clk_values == state_table->dcfclk.num_clk_values && - state_table->fclk.num_clk_values == state_table->uclk.num_clk_values) { - clock_state_count_identical = true; - } - - if (dcfclk_fine_grained || fclk_fine_grained || !clock_state_count_identical) - result = map_soc_min_clocks_to_dpm_fine_grained(display_cfg, state_table); - else - result = map_soc_min_clocks_to_dpm_coarse_grained(display_cfg, state_table); - - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dispclk_khz, &state_table->dispclk); - - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { - if (result) - result = round_up_to_next_dpm(&display_cfg->plane_programming[i].min_clocks.dcn4x.dppclk_khz, &state_table->dppclk); - } - - for (i = 0; i < display_cfg->display_config.num_streams; i++) { - if (result) - result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].dscclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.dscclk_khz, &state_table->dscclk); - if (result) - result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].dtbclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.dtbclk_khz, &state_table->dtbclk); - if (result) - result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].phyclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.phyclk_khz, &state_table->phyclk); - } - - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dpprefclk_khz, &state_table->dppclk); - - if (result) - result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dtbrefclk_khz, &state_table->dtbclk); - - return result; -} - -static bool are_timings_trivially_synchronizable(struct dml2_display_cfg *display_config, int mask) -{ - unsigned int i; - bool identical = true; - bool contains_drr = false; - unsigned int remap_array[DML2_MAX_PLANES]; - unsigned int remap_array_size = 0; - - // Create a remap array to enable simple iteration through only masked stream indicies - for (i = 0; i < display_config->num_streams; i++) { - if (mask & (0x1 << i)) { - remap_array[remap_array_size++] = i; - } - } - - // 0 or 1 display is always trivially synchronizable - if (remap_array_size <= 1) - return true; - - // Check that all displays timings are the same - for (i = 1; i < remap_array_size; i++) { - if (memcmp(&display_config->stream_descriptors[remap_array[i - 1]].timing, &display_config->stream_descriptors[remap_array[i]].timing, sizeof(struct dml2_timing_cfg))) { - identical = false; - break; - } - } - - // Check if any displays are drr - for (i = 0; i < remap_array_size; i++) { - if (display_config->stream_descriptors[remap_array[i]].timing.drr_config.enabled) { - contains_drr = true; - break; - } - } - - // Trivial sync is possible if all displays are identical and none are DRR - return !contains_drr && identical; -} - -static int find_smallest_idle_time_in_vblank_us(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int mask) -{ - unsigned int i; - int min_idle_us = 0; - unsigned int remap_array[DML2_MAX_PLANES]; - unsigned int remap_array_size = 0; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - // Create a remap array to enable simple iteration through only masked stream indicies - for (i = 0; i < in_out->programming->display_config.num_streams; i++) { - if (mask & (0x1 << i)) { - remap_array[remap_array_size++] = i; - } - } - - if (remap_array_size == 0) - return 0; - - min_idle_us = mode_support_result->cfg_support_info.stream_support_info[remap_array[0]].vblank_reserved_time_us; - - for (i = 1; i < remap_array_size; i++) { - if (min_idle_us > mode_support_result->cfg_support_info.stream_support_info[remap_array[i]].vblank_reserved_time_us) - min_idle_us = mode_support_result->cfg_support_info.stream_support_info[remap_array[i]].vblank_reserved_time_us; - } - - return min_idle_us; -} - -static bool determine_power_management_features_with_vblank_only(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - int min_idle_us; - - if (are_timings_trivially_synchronizable(&in_out->programming->display_config, 0xF)) { - min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, 0xF); - - if (min_idle_us >= in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us) - in_out->programming->uclk_pstate_supported = true; - - if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) - in_out->programming->fclk_pstate_supported = true; - } - - return true; -} - -static int get_displays_without_vactive_margin_mask(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int latency_hiding_requirement_us) -{ - unsigned int i; - int displays_without_vactive_margin_mask = 0x0; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - for (i = 0; i < in_out->programming->display_config.num_planes; i++) { - if (mode_support_result->cfg_support_info.plane_support_info[i].active_latency_hiding_us - < latency_hiding_requirement_us) - displays_without_vactive_margin_mask |= (0x1 << i); - } - - return displays_without_vactive_margin_mask; -} - -static int get_displays_with_fams_mask(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int latency_hiding_requirement_us) -{ - unsigned int i; - int displays_with_fams_mask = 0x0; - - for (i = 0; i < in_out->programming->display_config.num_planes; i++) { - if (in_out->programming->display_config.plane_descriptors->overrides.legacy_svp_config != dml2_svp_mode_override_auto) - displays_with_fams_mask |= (0x1 << i); - } - - return displays_with_fams_mask; -} - -static bool determine_power_management_features_with_vactive_and_vblank(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - int displays_without_vactive_margin_mask = 0x0; - int min_idle_us = 0; - - if (in_out->programming->uclk_pstate_supported == false) { - displays_without_vactive_margin_mask = - get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); - - if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { - min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); - - if (min_idle_us >= in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us) - in_out->programming->uclk_pstate_supported = true; - } - } - - if (in_out->programming->fclk_pstate_supported == false) { - displays_without_vactive_margin_mask = - get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.fclk_change_blackout_us)); - - if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { - min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); - - if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) - in_out->programming->fclk_pstate_supported = true; - } - } - - return true; -} - -static bool determine_power_management_features_with_fams(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - int displays_without_vactive_margin_mask = 0x0; - int displays_without_fams_mask = 0x0; - - displays_without_vactive_margin_mask = - get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); - - displays_without_fams_mask = - get_displays_with_fams_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); - - if ((displays_without_vactive_margin_mask & ~displays_without_fams_mask) == 0) - in_out->programming->uclk_pstate_supported = true; - - return true; -} - -static void clamp_uclk_to_max(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - in_out->programming->min_clocks.dcn4x.active.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; - in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; - in_out->programming->min_clocks.dcn4x.idle.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; -} - -static void clamp_fclk_to_max(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - in_out->programming->min_clocks.dcn4x.active.fclk_khz = in_out->soc_bb->clk_table.fclk.clk_values_khz[in_out->soc_bb->clk_table.fclk.num_clk_values - 1]; - in_out->programming->min_clocks.dcn4x.idle.fclk_khz = in_out->soc_bb->clk_table.fclk.clk_values_khz[in_out->soc_bb->clk_table.fclk.num_clk_values - 1]; -} - -static bool map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - int i; - bool result; - double dispclk_khz; - const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; - - calculate_system_active_minimums(in_out); - calculate_svp_prefetch_minimums(in_out); - calculate_idle_minimums(in_out); - - // In NV4, there's no support for FCLK or DCFCLK DPM change before SVP prefetch starts, therefore - // active minimums must be boosted to prefetch minimums - if (in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz > in_out->programming->min_clocks.dcn4x.active.uclk_khz) - in_out->programming->min_clocks.dcn4x.active.uclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz; - - if (in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz > in_out->programming->min_clocks.dcn4x.active.fclk_khz) - in_out->programming->min_clocks.dcn4x.active.fclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz; - - if (in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz > in_out->programming->min_clocks.dcn4x.active.dcfclk_khz) - in_out->programming->min_clocks.dcn4x.active.dcfclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz; - - // need some massaging for the dispclk ramping cases: - dispclk_khz = mode_support_result->global.dispclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0) * (1.0 + in_out->ip->dispclk_ramp_margin_percent / 100.0); - // ramping margin should not make dispclk exceed the maximum dispclk speed: - dispclk_khz = math_min2(dispclk_khz, in_out->min_clk_table->max_clocks_khz.dispclk); - // but still the required dispclk can be more than the maximum dispclk speed: - dispclk_khz = math_max2(dispclk_khz, mode_support_result->global.dispclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); - - // DPP Ref is always set to max of all DPP clocks - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { - if (in_out->programming->min_clocks.dcn4x.dpprefclk_khz < mode_support_result->per_plane[i].dppclk_khz) - in_out->programming->min_clocks.dcn4x.dpprefclk_khz = mode_support_result->per_plane[i].dppclk_khz; - } - in_out->programming->min_clocks.dcn4x.dpprefclk_khz = (unsigned long) (in_out->programming->min_clocks.dcn4x.dpprefclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); - - // DTB Ref is always set to max of all DTB clocks - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { - if (in_out->programming->min_clocks.dcn4x.dtbrefclk_khz < mode_support_result->per_stream[i].dtbclk_khz) - in_out->programming->min_clocks.dcn4x.dtbrefclk_khz = mode_support_result->per_stream[i].dtbclk_khz; - } - in_out->programming->min_clocks.dcn4x.dtbrefclk_khz = (unsigned long)(in_out->programming->min_clocks.dcn4x.dtbrefclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); - - if (in_out->soc_bb->no_dfs) { - round_to_non_dfs_granularity((unsigned long)dispclk_khz, in_out->programming->min_clocks.dcn4x.dpprefclk_khz, in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, - &in_out->programming->min_clocks.dcn4x.dispclk_khz, &in_out->programming->min_clocks.dcn4x.dpprefclk_khz, &in_out->programming->min_clocks.dcn4x.dtbrefclk_khz); - } else { - add_margin_and_round_to_dfs_grainularity(dispclk_khz, 0.0, - (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dispclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dispclk_did); - - add_margin_and_round_to_dfs_grainularity(in_out->programming->min_clocks.dcn4x.dpprefclk_khz, 0.0, - (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dpprefclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dpprefclk_did); - - add_margin_and_round_to_dfs_grainularity(in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, 0.0, - (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dtbrefclk_did); - } - - - for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { - in_out->programming->plane_programming[i].min_clocks.dcn4x.dppclk_khz = (unsigned long)(in_out->programming->min_clocks.dcn4x.dpprefclk_khz / 255.0 - * math_ceil2(in_out->display_cfg->mode_support_result.per_plane[i].dppclk_khz * (1.0 + in_out->soc_bb->dcn_downspread_percent / 100.0) * 255.0 / in_out->programming->min_clocks.dcn4x.dpprefclk_khz, 1.0)); - } - - in_out->programming->min_clocks.dcn4x.deepsleep_dcfclk_khz = mode_support_result->global.dcfclk_deepsleep_khz; - in_out->programming->min_clocks.dcn4x.socclk_khz = mode_support_result->global.socclk_khz; - - result = map_min_clocks_to_dpm(mode_support_result, in_out->programming, &in_out->soc_bb->clk_table); - - // By default, all power management features are not enabled - in_out->programming->fclk_pstate_supported = false; - in_out->programming->uclk_pstate_supported = false; - - return result; -} - -bool dpmm_dcn3_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - bool result; - - result = map_mode_to_soc_dpm(in_out); - - // Check if any can be enabled by nominal vblank idle time - determine_power_management_features_with_vblank_only(in_out); - - // Check if any can be enabled in vactive/vblank - determine_power_management_features_with_vactive_and_vblank(in_out); - - // Check if any can be enabled via fams - determine_power_management_features_with_fams(in_out); - - if (in_out->programming->uclk_pstate_supported == false) - clamp_uclk_to_max(in_out); - - if (in_out->programming->fclk_pstate_supported == false) - clamp_fclk_to_max(in_out); - - return result; -} - -bool dpmm_dcn4_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - bool result; - int displays_without_vactive_margin_mask = 0x0; - int min_idle_us = 0; - - result = map_mode_to_soc_dpm(in_out); - - if (in_out->display_cfg->stage3.success) - in_out->programming->uclk_pstate_supported = true; - - displays_without_vactive_margin_mask = - get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.fclk_change_blackout_us)); - - if (displays_without_vactive_margin_mask == 0) { - in_out->programming->fclk_pstate_supported = true; - } else { - if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { - min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); - - if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) - in_out->programming->fclk_pstate_supported = true; - } - } - - if (in_out->programming->uclk_pstate_supported == false) - clamp_uclk_to_max(in_out); - - if (in_out->programming->fclk_pstate_supported == false) - clamp_fclk_to_max(in_out); - - min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, 0xFF); - if (in_out->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && - min_idle_us >= in_out->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us) - in_out->programming->stutter.supported_in_blank = true; - else - in_out->programming->stutter.supported_in_blank = false; - - // TODO: Fix me Sam - if (in_out->soc_bb->power_management_parameters.z8_min_idle_time > 0 && - in_out->programming->informative.power_management.z8.stutter_period >= in_out->soc_bb->power_management_parameters.z8_min_idle_time) - in_out->programming->z8_stutter.meets_eco = true; - else - in_out->programming->z8_stutter.meets_eco = false; - - if (in_out->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && - min_idle_us >= in_out->soc_bb->power_management_parameters.z8_stutter_exit_latency_us) - in_out->programming->z8_stutter.supported_in_blank = true; - else - in_out->programming->z8_stutter.supported_in_blank = false; - - return result; -} - -bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out) -{ - const struct dml2_display_cfg *display_cfg = &in_out->display_cfg->display_config; - const struct dml2_core_internal_display_mode_lib *mode_lib = &in_out->core->clean_me_up.mode_lib; - struct dml2_dchub_global_register_set *dchubbub_regs = &in_out->programming->global_regs; - - double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; - - /* set A */ - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); - - /* set B */ - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); - dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); - - dchubbub_regs->num_watermark_sets = 2; - - return true; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h deleted file mode 100644 index e7b58f2efda4..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h +++ /dev/null @@ -1,14 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_DPMM_DCN4_H__ -#define __DML2_DPMM_DCN4_H__ - -#include "dml2_internal_shared_types.h" - -bool dpmm_dcn3_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); -bool dpmm_dcn4_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); -bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c deleted file mode 100644 index dfd01440737d..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c +++ /dev/null @@ -1,50 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_dpmm_factory.h" -#include "dml2_dpmm_dcn4.h" -#include "dml2_external_lib_deps.h" - -static bool dummy_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) -{ - return true; -} - -static bool dummy_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out) -{ - return true; -} - -bool dml2_dpmm_create(enum dml2_project_id project_id, struct dml2_dpmm_instance *out) -{ - bool result = false; - - if (!out) - return false; - - memset(out, 0, sizeof(struct dml2_dpmm_instance)); - - switch (project_id) { - case dml2_project_dcn4x_stage1: - out->map_mode_to_soc_dpm = &dummy_map_mode_to_soc_dpm; - out->map_watermarks = &dummy_map_watermarks; - result = true; - break; - case dml2_project_dcn4x_stage2: - out->map_mode_to_soc_dpm = &dpmm_dcn3_map_mode_to_soc_dpm; - out->map_watermarks = &dummy_map_watermarks; - result = true; - break; - case dml2_project_dcn4x_stage2_auto_drr_svp: - out->map_mode_to_soc_dpm = &dpmm_dcn4_map_mode_to_soc_dpm; - out->map_watermarks = &dpmm_dcn4_map_watermarks; - result = true; - break; - case dml2_project_invalid: - default: - break; - } - - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h deleted file mode 100644 index 20ba2e446f1d..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_DPMM_FACTORY_H__ -#define __DML2_DPMM_FACTORY_H__ - -#include "dml2_internal_shared_types.h" -#include "dml_top_types.h" - -bool dml2_dpmm_create(enum dml2_project_id project_id, struct dml2_dpmm_instance *out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c deleted file mode 100644 index a265f254152c..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c +++ /dev/null @@ -1,198 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_mcg_dcn4.h" -#include "dml_top_soc_parameter_types.h" - -static bool build_min_clock_table(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table); - -bool mcg_dcn4_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out) -{ - return build_min_clock_table(in_out->soc_bb, in_out->min_clk_table); -} - -static unsigned long long uclk_to_dram_bw_kbps(unsigned long uclk_khz, const struct dml2_dram_params *dram_config) -{ - unsigned long long bw_kbps = 0; - - bw_kbps = (unsigned long long) uclk_khz * dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock; - - return bw_kbps; -} - -static unsigned long round_up_to_quantized_values(unsigned long value, const unsigned long *quantized_values, int num_quantized_values) -{ - int i; - - if (!quantized_values) - return 0; - - for (i = 0; i < num_quantized_values; i++) { - if (quantized_values[i] > value) - return quantized_values[i]; - } - - return 0; -} - -static bool build_min_clk_table_fine_grained(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) -{ - bool dcfclk_fine_grained = false, fclk_fine_grained = false; - - int i; - unsigned int j; - - unsigned long min_dcfclk_khz = 0; - unsigned long min_fclk_khz = 0; - unsigned long prev_100, cur_50; - - if (soc_bb->clk_table.dcfclk.num_clk_values == 2) { - dcfclk_fine_grained = true; - } - - if (soc_bb->clk_table.fclk.num_clk_values == 2) { - fclk_fine_grained = true; - } - - min_dcfclk_khz = soc_bb->clk_table.dcfclk.clk_values_khz[0]; - min_fclk_khz = soc_bb->clk_table.fclk.clk_values_khz[0]; - - // First calculate the table for "balanced" bandwidths across UCLK/FCLK - for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { - min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps = uclk_to_dram_bw_kbps(soc_bb->clk_table.uclk.clk_values_khz[i], &soc_bb->clk_table.dram_config); - - min_table->dram_bw_table.entries[i].min_fclk_khz = (unsigned long)((((double)min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps * soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100) / ((double)soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100)) / soc_bb->fabric_datapath_to_dcn_data_return_bytes); - } - min_table->dram_bw_table.num_entries = soc_bb->clk_table.uclk.num_clk_values; - - // To create the minium table, effectively shift "up" all the dcfclk/fclk entries by 1, and then replace the lowest entry with min fclk/dcfclk - for (i = min_table->dram_bw_table.num_entries - 1; i > 0; i--) { - prev_100 = min_table->dram_bw_table.entries[i - 1].min_fclk_khz; - cur_50 = min_table->dram_bw_table.entries[i].min_fclk_khz / 2; - min_table->dram_bw_table.entries[i].min_fclk_khz = prev_100 > cur_50 ? prev_100 : cur_50; - - if (!fclk_fine_grained) { - min_table->dram_bw_table.entries[i].min_fclk_khz = round_up_to_quantized_values(min_table->dram_bw_table.entries[i].min_fclk_khz, soc_bb->clk_table.fclk.clk_values_khz, soc_bb->clk_table.fclk.num_clk_values); - } - } - min_table->dram_bw_table.entries[0].min_fclk_khz /= 2; - - // Clamp to minimums and maximums - for (i = 0; i < (int)min_table->dram_bw_table.num_entries; i++) { - if (min_table->dram_bw_table.entries[i].min_dcfclk_khz < min_dcfclk_khz) - min_table->dram_bw_table.entries[i].min_dcfclk_khz = min_dcfclk_khz; - - if (min_table->dram_bw_table.entries[i].min_fclk_khz < min_fclk_khz) - min_table->dram_bw_table.entries[i].min_fclk_khz = min_fclk_khz; - - if (soc_bb->max_fclk_for_uclk_dpm_khz > 0 && - min_table->dram_bw_table.entries[i].min_fclk_khz > soc_bb->max_fclk_for_uclk_dpm_khz) - min_table->dram_bw_table.entries[i].min_fclk_khz = soc_bb->max_fclk_for_uclk_dpm_khz; - - min_table->dram_bw_table.entries[i].min_dcfclk_khz = - min_table->dram_bw_table.entries[i].min_fclk_khz * - soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent; - - min_table->dram_bw_table.entries[i].min_dcfclk_khz = - min_table->dram_bw_table.entries[i].min_dcfclk_khz * soc_bb->fabric_datapath_to_dcn_data_return_bytes / soc_bb->return_bus_width_bytes; - - if (!dcfclk_fine_grained) { - min_table->dram_bw_table.entries[i].min_dcfclk_khz = round_up_to_quantized_values(min_table->dram_bw_table.entries[i].min_dcfclk_khz, soc_bb->clk_table.dcfclk.clk_values_khz, soc_bb->clk_table.dcfclk.num_clk_values); - } - } - - // Prune states which are invalid (some clocks exceed maximum) - for (i = 0; i < (int)min_table->dram_bw_table.num_entries; i++) { - if (min_table->dram_bw_table.entries[i].min_dcfclk_khz > min_table->max_clocks_khz.dcfclk || - min_table->dram_bw_table.entries[i].min_fclk_khz > min_table->max_clocks_khz.fclk) { - min_table->dram_bw_table.num_entries = i; - break; - } - } - - // Prune duplicate states - for (i = 0; i < (int)min_table->dram_bw_table.num_entries - 1; i++) { - if (min_table->dram_bw_table.entries[i].min_dcfclk_khz == min_table->dram_bw_table.entries[i + 1].min_dcfclk_khz && - min_table->dram_bw_table.entries[i].min_fclk_khz == min_table->dram_bw_table.entries[i + 1].min_fclk_khz && - min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps == min_table->dram_bw_table.entries[i + 1].pre_derate_dram_bw_kbps) { - - // i + 1 is the same state as i, so shift everything - for (j = i + 1; j < min_table->dram_bw_table.num_entries; j++) { - min_table->dram_bw_table.entries[j].min_dcfclk_khz = min_table->dram_bw_table.entries[j + 1].min_dcfclk_khz; - min_table->dram_bw_table.entries[j].min_fclk_khz = min_table->dram_bw_table.entries[j + 1].min_fclk_khz; - min_table->dram_bw_table.entries[j].pre_derate_dram_bw_kbps = min_table->dram_bw_table.entries[j + 1].pre_derate_dram_bw_kbps; - } - min_table->dram_bw_table.num_entries--; - } - } - - return true; -} - -static bool build_min_clk_table_coarse_grained(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) -{ - int i; - - for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { - min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps = uclk_to_dram_bw_kbps(soc_bb->clk_table.uclk.clk_values_khz[i], &soc_bb->clk_table.dram_config); - min_table->dram_bw_table.entries[i].min_dcfclk_khz = soc_bb->clk_table.dcfclk.clk_values_khz[i]; - min_table->dram_bw_table.entries[i].min_fclk_khz = soc_bb->clk_table.fclk.clk_values_khz[i]; - } - min_table->dram_bw_table.num_entries = soc_bb->clk_table.uclk.num_clk_values; - - return true; -} - -static bool build_min_clock_table(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) -{ - bool result; - bool dcfclk_fine_grained = false, fclk_fine_grained = false, clock_state_count_equal = false; - - if (!soc_bb || !min_table) - return false; - - if (soc_bb->clk_table.dcfclk.num_clk_values < 2 || soc_bb->clk_table.fclk.num_clk_values < 2) - return false; - - if (soc_bb->clk_table.uclk.num_clk_values > DML_MCG_MAX_CLK_TABLE_SIZE) - return false; - - if (soc_bb->clk_table.dcfclk.num_clk_values == 2) { - dcfclk_fine_grained = true; - } - - if (soc_bb->clk_table.fclk.num_clk_values == 2) { - fclk_fine_grained = true; - } - - if (soc_bb->clk_table.fclk.num_clk_values == soc_bb->clk_table.dcfclk.num_clk_values && - soc_bb->clk_table.fclk.num_clk_values == soc_bb->clk_table.uclk.num_clk_values) - clock_state_count_equal = true; - - min_table->fixed_clocks_khz.amclk = 0; - min_table->fixed_clocks_khz.dprefclk = soc_bb->dprefclk_mhz * 1000; - min_table->fixed_clocks_khz.pcierefclk = soc_bb->pcie_refclk_mhz * 1000; - min_table->fixed_clocks_khz.dchubrefclk = soc_bb->dchub_refclk_mhz * 1000; - min_table->fixed_clocks_khz.xtalclk = soc_bb->xtalclk_mhz * 1000; - - min_table->max_clocks_khz.dispclk = soc_bb->clk_table.dispclk.clk_values_khz[soc_bb->clk_table.dispclk.num_clk_values - 1]; - min_table->max_clocks_khz.dppclk = soc_bb->clk_table.dppclk.clk_values_khz[soc_bb->clk_table.dppclk.num_clk_values - 1]; - min_table->max_clocks_khz.dscclk = soc_bb->clk_table.dscclk.clk_values_khz[soc_bb->clk_table.dscclk.num_clk_values - 1]; - min_table->max_clocks_khz.dtbclk = soc_bb->clk_table.dtbclk.clk_values_khz[soc_bb->clk_table.dtbclk.num_clk_values - 1]; - min_table->max_clocks_khz.phyclk = soc_bb->clk_table.phyclk.clk_values_khz[soc_bb->clk_table.phyclk.num_clk_values - 1]; - - min_table->max_ss_clocks_khz.dispclk = (unsigned int)((double)min_table->max_clocks_khz.dispclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); - min_table->max_ss_clocks_khz.dppclk = (unsigned int)((double)min_table->max_clocks_khz.dppclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); - min_table->max_ss_clocks_khz.dtbclk = (unsigned int)((double)min_table->max_clocks_khz.dtbclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); - - min_table->max_clocks_khz.dcfclk = soc_bb->clk_table.dcfclk.clk_values_khz[soc_bb->clk_table.dcfclk.num_clk_values - 1]; - min_table->max_clocks_khz.fclk = soc_bb->clk_table.fclk.clk_values_khz[soc_bb->clk_table.fclk.num_clk_values - 1]; - - if (dcfclk_fine_grained || fclk_fine_grained || !clock_state_count_equal) - result = build_min_clk_table_fine_grained(soc_bb, min_table); - else - result = build_min_clk_table_coarse_grained(soc_bb, min_table); - - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h deleted file mode 100644 index 02da6f45cbf7..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_MCG_DCN4_H__ -#define __DML2_MCG_DCN4_H__ - -#include "dml2_internal_shared_types.h" - -bool mcg_dcn4_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out); -bool mcg_dcn4_unit_test(void); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c deleted file mode 100644 index c60b8fe90819..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c +++ /dev/null @@ -1,39 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_mcg_factory.h" -#include "dml2_mcg_dcn4.h" -#include "dml2_external_lib_deps.h" - -static bool dummy_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out) -{ - return true; -} - -bool dml2_mcg_create(enum dml2_project_id project_id, struct dml2_mcg_instance *out) -{ - bool result = false; - - if (!out) - return false; - - memset(out, 0, sizeof(struct dml2_mcg_instance)); - - switch (project_id) { - case dml2_project_dcn4x_stage1: - out->build_min_clock_table = &dummy_build_min_clock_table; - result = true; - break; - case dml2_project_dcn4x_stage2: - case dml2_project_dcn4x_stage2_auto_drr_svp: - out->build_min_clock_table = &mcg_dcn4_build_min_clock_table; - result = true; - break; - case dml2_project_invalid: - default: - break; - } - - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h deleted file mode 100644 index ad307deca3b0..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_MCG_FACTORY_H__ -#define __DML2_MCG_FACTORY_H__ - -#include "dml2_internal_shared_types.h" -#include "dml_top_types.h" - -bool dml2_mcg_create(enum dml2_project_id project_id, struct dml2_mcg_instance *out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c deleted file mode 100644 index 1b9579a32ff2..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c +++ /dev/null @@ -1,706 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_pmo_factory.h" -#include "dml2_pmo_dcn3.h" - -static void sort(double *list_a, int list_a_size) -{ - // For all elements b[i] in list_b[] - for (int i = 0; i < list_a_size - 1; i++) { - // Find the first element of list_a that's larger than b[i] - for (int j = i; j < list_a_size - 1; j++) { - if (list_a[j] > list_a[j + 1]) - swap(list_a[j], list_a[j + 1]); - } - } -} - -static double get_max_reserved_time_on_all_planes_with_stream_index(struct display_configuation_with_meta *config, unsigned int stream_index) -{ - struct dml2_plane_parameters *plane_descriptor; - long max_reserved_time_ns = 0; - - for (unsigned int i = 0; i < config->display_config.num_planes; i++) { - plane_descriptor = &config->display_config.plane_descriptors[i]; - - if (plane_descriptor->stream_index == stream_index) - if (plane_descriptor->overrides.reserved_vblank_time_ns > max_reserved_time_ns) - max_reserved_time_ns = plane_descriptor->overrides.reserved_vblank_time_ns; - } - - return (max_reserved_time_ns / 1000.0); -} - - -static void set_reserved_time_on_all_planes_with_stream_index(struct display_configuation_with_meta *config, unsigned int stream_index, double reserved_time_us) -{ - struct dml2_plane_parameters *plane_descriptor; - - for (unsigned int i = 0; i < config->display_config.num_planes; i++) { - plane_descriptor = &config->display_config.plane_descriptors[i]; - - if (plane_descriptor->stream_index == stream_index) - plane_descriptor->overrides.reserved_vblank_time_ns = (long int)(reserved_time_us * 1000); - } -} - -static void remove_duplicates(double *list_a, int *list_a_size) -{ - int j = 0; - - if (*list_a_size == 0) - return; - - for (int i = 1; i < *list_a_size; i++) { - if (list_a[j] != list_a[i]) { - j++; - list_a[j] = list_a[i]; - } - } - - *list_a_size = j + 1; -} - -static bool increase_mpc_combine_factor(unsigned int *mpc_combine_factor, unsigned int limit) -{ - if (*mpc_combine_factor < limit) { - (*mpc_combine_factor)++; - return true; - } - - return false; -} - -static bool optimize_dcc_mcache_no_odm(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out, - int free_pipes) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i; - bool result = true; - - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - // For pipes that failed dcc mcache check, we want to increase the pipe count. - // The logic for doing this depends on how many pipes is already being used, - // and whether it's mpcc or odm combine. - if (!in_out->dcc_mcache_supported[i]) { - // For the general case of "n displays", we can only optimize streams with an ODM combine factor of 1 - if (in_out->cfg_support_info->stream_support_info[in_out->optimized_display_cfg->plane_descriptors[i].stream_index].odms_used == 1) { - in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor = - in_out->cfg_support_info->plane_support_info[i].dpps_used; - // For each plane that is not passing mcache validation, just add another pipe to it, up to the limit. - if (free_pipes > 0) { - if (!increase_mpc_combine_factor(&in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor, - pmo->mpc_combine_limit)) { - // We've reached max pipes allocatable to a single plane, so we fail. - result = false; - break; - } else { - // Successfully added another pipe to this failing plane. - free_pipes--; - } - } else { - // No free pipes to add. - result = false; - break; - } - } else { - // If the stream of this plane needs ODM combine, no further optimization can be done. - result = false; - break; - } - } - } - - return result; -} - -static bool iterate_to_next_candidiate(struct dml2_pmo_instance *pmo, int size) -{ - int borrow_from, i; - bool success = false; - - if (pmo->scratch.pmo_dcn3.current_candidate[0] > 0) { - pmo->scratch.pmo_dcn3.current_candidate[0]--; - success = true; - } else { - for (borrow_from = 1; borrow_from < size && pmo->scratch.pmo_dcn3.current_candidate[borrow_from] == 0; borrow_from++) - ; - - if (borrow_from < size) { - pmo->scratch.pmo_dcn3.current_candidate[borrow_from]--; - for (i = 0; i < borrow_from; i++) { - pmo->scratch.pmo_dcn3.current_candidate[i] = pmo->scratch.pmo_dcn3.reserved_time_candidates_count[i] - 1; - } - - success = true; - } - } - - return success; -} - -static bool increase_odm_combine_factor(enum dml2_odm_mode *odm_mode, int odms_calculated) -{ - bool result = true; - - if (*odm_mode == dml2_odm_mode_auto) { - switch (odms_calculated) { - case 1: - *odm_mode = dml2_odm_mode_bypass; - break; - case 2: - *odm_mode = dml2_odm_mode_combine_2to1; - break; - case 3: - *odm_mode = dml2_odm_mode_combine_3to1; - break; - case 4: - *odm_mode = dml2_odm_mode_combine_4to1; - break; - default: - result = false; - break; - } - } - - if (result) { - if (*odm_mode == dml2_odm_mode_bypass) { - *odm_mode = dml2_odm_mode_combine_2to1; - } else if (*odm_mode == dml2_odm_mode_combine_2to1) { - *odm_mode = dml2_odm_mode_combine_3to1; - } else if (*odm_mode == dml2_odm_mode_combine_3to1) { - *odm_mode = dml2_odm_mode_combine_4to1; - } else { - result = false; - } - } - - return result; -} - -static int count_planes_with_stream_index(const struct dml2_display_cfg *display_cfg, unsigned int stream_index) -{ - unsigned int i, count; - - count = 0; - for (i = 0; i < display_cfg->num_planes; i++) { - if (display_cfg->plane_descriptors[i].stream_index == stream_index) - count++; - } - - return count; -} - -static bool are_timings_trivially_synchronizable(struct display_configuation_with_meta *display_config, int mask) -{ - unsigned int i; - bool identical = true; - bool contains_drr = false; - unsigned int remap_array[DML2_MAX_PLANES]; - unsigned int remap_array_size = 0; - - // Create a remap array to enable simple iteration through only masked stream indicies - for (i = 0; i < display_config->display_config.num_streams; i++) { - if (mask & (0x1 << i)) { - remap_array[remap_array_size++] = i; - } - } - - // 0 or 1 display is always trivially synchronizable - if (remap_array_size <= 1) - return true; - - for (i = 1; i < remap_array_size; i++) { - if (memcmp(&display_config->display_config.stream_descriptors[remap_array[i - 1]].timing, - &display_config->display_config.stream_descriptors[remap_array[i]].timing, - sizeof(struct dml2_timing_cfg))) { - identical = false; - break; - } - } - - for (i = 0; i < remap_array_size; i++) { - if (display_config->display_config.stream_descriptors[remap_array[i]].timing.drr_config.enabled) { - contains_drr = true; - break; - } - } - - return !contains_drr && identical; -} - -bool pmo_dcn3_initialize(struct dml2_pmo_initialize_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - pmo->soc_bb = in_out->soc_bb; - pmo->ip_caps = in_out->ip_caps; - pmo->mpc_combine_limit = 2; - pmo->odm_combine_limit = 4; - pmo->mcg_clock_table_size = in_out->mcg_clock_table_size; - - pmo->options = in_out->options; - - return true; -} - -static bool is_h_timing_divisible_by(const struct dml2_timing_cfg *timing, unsigned char denominator) -{ - /* - * Htotal, Hblank start/end, and Hsync start/end all must be divisible - * in order for the horizontal timing params to be considered divisible - * by 2. Hsync start is always 0. - */ - unsigned long h_blank_start = timing->h_total - timing->h_front_porch; - - return (timing->h_total % denominator == 0) && - (h_blank_start % denominator == 0) && - (timing->h_blank_end % denominator == 0) && - (timing->h_sync_width % denominator == 0); -} - -static bool is_dp_encoder(enum dml2_output_encoder_class encoder_type) -{ - switch (encoder_type) { - case dml2_dp: - case dml2_edp: - case dml2_dp2p0: - case dml2_none: - return true; - case dml2_hdmi: - case dml2_hdmifrl: - default: - return false; - } -} - -bool pmo_dcn3_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out) -{ - unsigned int i; - const struct dml2_display_cfg *display_config = - &in_out->base_display_config->display_config; - const struct dml2_core_mode_support_result *mode_support_result = - &in_out->base_display_config->mode_support_result; - - if (in_out->instance->options->disable_dyn_odm || - (in_out->instance->options->disable_dyn_odm_for_multi_stream && display_config->num_streams > 1)) - return false; - - for (i = 0; i < display_config->num_planes; i++) - /* - * vmin optimization is required to be seamlessly switched off - * at any time when the new configuration is no longer - * supported. However switching from ODM combine to MPC combine - * is not always seamless. When there not enough free pipes, we - * will have to use the same secondary OPP heads as secondary - * DPP pipes in MPC combine in new state. This transition is - * expected to cause glitches. To avoid the transition, we only - * allow vmin optimization if the stream's base configuration - * doesn't require MPC combine. This condition checks if MPC - * combine is enabled. If so do not optimize the stream. - */ - if (mode_support_result->cfg_support_info.plane_support_info[i].dpps_used > 1 && - mode_support_result->cfg_support_info.stream_support_info[display_config->plane_descriptors[i].stream_index].odms_used == 1) - in_out->base_display_config->stage4.unoptimizable_streams[display_config->plane_descriptors[i].stream_index] = true; - - for (i = 0; i < display_config->num_streams; i++) { - if (display_config->stream_descriptors[i].overrides.disable_dynamic_odm) - in_out->base_display_config->stage4.unoptimizable_streams[i] = true; - else if (in_out->base_display_config->stage3.stream_svp_meta[i].valid && - in_out->instance->options->disable_dyn_odm_for_stream_with_svp) - in_out->base_display_config->stage4.unoptimizable_streams[i] = true; - /* - * ODM Combine requires horizontal timing divisible by 2 so each - * ODM segment has the same size. - */ - else if (!is_h_timing_divisible_by(&display_config->stream_descriptors[i].timing, 2)) - in_out->base_display_config->stage4.unoptimizable_streams[i] = true; - /* - * Our hardware support seamless ODM transitions for DP encoders - * only. - */ - else if (!is_dp_encoder(display_config->stream_descriptors[i].output.output_encoder)) - in_out->base_display_config->stage4.unoptimizable_streams[i] = true; - } - - return true; -} - -bool pmo_dcn3_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out) -{ - bool is_vmin = true; - - if (in_out->vmin_limits->dispclk_khz > 0 && - in_out->display_config->mode_support_result.global.dispclk_khz > in_out->vmin_limits->dispclk_khz) - is_vmin = false; - - return is_vmin; -} - -static int find_highest_odm_load_stream_index( - const struct dml2_display_cfg *display_config, - const struct dml2_core_mode_support_result *mode_support_result) -{ - unsigned int i; - int odm_load, highest_odm_load = -1, highest_odm_load_index = -1; - - for (i = 0; i < display_config->num_streams; i++) { - if (mode_support_result->cfg_support_info.stream_support_info[i].odms_used > 0) - odm_load = display_config->stream_descriptors[i].timing.pixel_clock_khz - / mode_support_result->cfg_support_info.stream_support_info[i].odms_used; - else - odm_load = 0; - - if (odm_load > highest_odm_load) { - highest_odm_load_index = i; - highest_odm_load = odm_load; - } - } - - return highest_odm_load_index; -} - -bool pmo_dcn3_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out) -{ - int stream_index; - const struct dml2_display_cfg *display_config = - &in_out->base_display_config->display_config; - const struct dml2_core_mode_support_result *mode_support_result = - &in_out->base_display_config->mode_support_result; - unsigned int odms_used; - struct dml2_stream_parameters *stream_descriptor; - bool optimizable = false; - - /* - * highest odm load stream must be optimizable to continue as dispclk is - * bounded by it. - */ - stream_index = find_highest_odm_load_stream_index(display_config, - mode_support_result); - - if (stream_index < 0 || - in_out->base_display_config->stage4.unoptimizable_streams[stream_index]) - return false; - - odms_used = mode_support_result->cfg_support_info.stream_support_info[stream_index].odms_used; - if ((int)odms_used >= in_out->instance->odm_combine_limit) - return false; - - memcpy(in_out->optimized_display_config, - in_out->base_display_config, - sizeof(struct display_configuation_with_meta)); - - stream_descriptor = &in_out->optimized_display_config->display_config.stream_descriptors[stream_index]; - while (!optimizable && increase_odm_combine_factor( - &stream_descriptor->overrides.odm_mode, - odms_used)) { - switch (stream_descriptor->overrides.odm_mode) { - case dml2_odm_mode_combine_2to1: - optimizable = true; - break; - case dml2_odm_mode_combine_3to1: - /* - * In ODM Combine 3:1 OTG_valid_pixel rate is 1/4 of - * actual pixel rate. Therefore horizontal timing must - * be divisible by 4. - */ - if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { - /* - * DSC h slice count must be divisible - * by 3. - */ - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 3 == 0) - optimizable = true; - } else { - optimizable = true; - } - } - break; - case dml2_odm_mode_combine_4to1: - /* - * In ODM Combine 4:1 OTG_valid_pixel rate is 1/4 of - * actual pixel rate. Therefore horizontal timing must - * be divisible by 4. - */ - if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { - /* - * DSC h slice count must be divisible - * by 4. - */ - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 4 == 0) - optimizable = true; - } else { - optimizable = true; - } - } - break; - case dml2_odm_mode_auto: - case dml2_odm_mode_bypass: - case dml2_odm_mode_split_1to2: - case dml2_odm_mode_mso_1to2: - case dml2_odm_mode_mso_1to4: - default: - break; - } - } - - return optimizable; -} - -bool pmo_dcn3_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i, used_pipes, free_pipes, planes_on_stream; - bool result; - - if (in_out->display_config != in_out->optimized_display_cfg) { - memcpy(in_out->optimized_display_cfg, in_out->display_config, sizeof(struct dml2_display_cfg)); - } - - //Count number of free pipes, and check if any odm combine is in use. - used_pipes = 0; - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - used_pipes += in_out->cfg_support_info->plane_support_info[i].dpps_used; - } - free_pipes = pmo->ip_caps->pipe_count - used_pipes; - - // Optimization loop - // The goal here is to add more pipes to any planes - // which are failing mcache admissibility - result = true; - - // The optimization logic depends on whether ODM combine is enabled, and the stream count. - if (in_out->optimized_display_cfg->num_streams > 1) { - // If there are multiple streams, we are limited to only be able to optimize mcache failures on planes - // which are not ODM combined. - - result = optimize_dcc_mcache_no_odm(in_out, free_pipes); - } else if (in_out->optimized_display_cfg->num_streams == 1) { - // In single stream cases, we still optimize mcache failures when there's ODM combine with some - // additional logic. - - if (in_out->cfg_support_info->stream_support_info[0].odms_used > 1) { - // If ODM combine is enabled, then the logic is to increase ODM combine factor. - - // Optimization for streams with > 1 ODM combine factor is only supported for single display. - planes_on_stream = count_planes_with_stream_index(in_out->optimized_display_cfg, 0); - - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - // For pipes that failed dcc mcache check, we want to increase the pipe count. - // The logic for doing this depends on how many pipes is already being used, - // and whether it's mpcc or odm combine. - if (!in_out->dcc_mcache_supported[i]) { - // Increasing ODM combine factor on a stream requires a free pipe for each plane on the stream. - if (free_pipes >= planes_on_stream) { - if (!increase_odm_combine_factor(&in_out->optimized_display_cfg->stream_descriptors[i].overrides.odm_mode, - in_out->cfg_support_info->plane_support_info[i].dpps_used)) { - result = false; - } else { - break; - } - } else { - result = false; - break; - } - } - } - } else { - // If ODM combine is not enabled, then we can actually use the same logic as before. - - result = optimize_dcc_mcache_no_odm(in_out, free_pipes); - } - } else { - result = true; - } - - return result; -} - -bool pmo_dcn3_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - struct dml2_optimization_stage3_state *state = &in_out->base_display_config->stage3; - const struct dml2_stream_parameters *stream_descriptor; - const struct dml2_plane_parameters *plane_descriptor; - unsigned int stream_index, plane_index, candidate_count; - double min_reserved_vblank_time = 0; - int fclk_twait_needed_mask = 0x0; - int uclk_twait_needed_mask = 0x0; - - state->performed = true; - state->min_clk_index_for_latency = in_out->base_display_config->stage1.min_clk_index_for_latency; - pmo->scratch.pmo_dcn3.min_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; - pmo->scratch.pmo_dcn3.max_latency_index = pmo->mcg_clock_table_size - 1; - pmo->scratch.pmo_dcn3.cur_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; - - pmo->scratch.pmo_dcn3.stream_mask = 0xF; - - for (plane_index = 0; plane_index < in_out->base_display_config->display_config.num_planes; plane_index++) { - plane_descriptor = &in_out->base_display_config->display_config.plane_descriptors[plane_index]; - stream_descriptor = &in_out->base_display_config->display_config.stream_descriptors[plane_descriptor->stream_index]; - - if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[plane_index].active_latency_hiding_us < - in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us && - stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate == dml2_twait_budgeting_setting_if_needed) - uclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); - - if (stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate == dml2_twait_budgeting_setting_try) - uclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); - - if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[plane_index].active_latency_hiding_us < - in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us && - stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate == dml2_twait_budgeting_setting_if_needed) - fclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); - - if (stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate == dml2_twait_budgeting_setting_try) - fclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); - - if (plane_descriptor->overrides.legacy_svp_config != dml2_svp_mode_override_auto) { - pmo->scratch.pmo_dcn3.stream_mask &= ~(0x1 << plane_descriptor->stream_index); - } - } - - for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) { - stream_descriptor = &in_out->base_display_config->display_config.stream_descriptors[stream_index]; - - // The absolute minimum required time is the minimum of all the required budgets - /* - if (stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate - == dml2_twait_budgeting_setting_require) - - if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { - min_reserved_vblank_time = max_double2(min_reserved_vblank_time, - in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us); - } - - if (stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate - == dml2_twait_budgeting_setting_require) { - - if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { - min_reserved_vblank_time = max_double2(min_reserved_vblank_time, - in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us); - } - } - - if (stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit - == dml2_twait_budgeting_setting_require) - min_reserved_vblank_time = max_double2(min_reserved_vblank_time, - in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us); - */ - - min_reserved_vblank_time = get_max_reserved_time_on_all_planes_with_stream_index(in_out->base_display_config, stream_index); - - // Insert the absolute minimum into the array - candidate_count = 1; - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][0] = min_reserved_vblank_time; - pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] = candidate_count; - - if (!(pmo->scratch.pmo_dcn3.stream_mask & (0x1 << stream_index))) - continue; - - // For every optional feature, we create a candidate for it only if it's larger minimum. - if ((fclk_twait_needed_mask & (0x1 << stream_index)) && - in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us > min_reserved_vblank_time) { - - if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = - in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us; - } - } - - if ((uclk_twait_needed_mask & (0x1 << stream_index)) && - in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us > min_reserved_vblank_time) { - - if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = - in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us; - } - } - - if ((stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit == dml2_twait_budgeting_setting_try || - stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit == dml2_twait_budgeting_setting_if_needed) && - in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > min_reserved_vblank_time) { - - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = - in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us; - } - - pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] = candidate_count; - - // Finally sort the array of candidates - sort(pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index], - pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index]); - - remove_duplicates(pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index], - &pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index]); - - pmo->scratch.pmo_dcn3.current_candidate[stream_index] = - pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] - 1; - } - - return true; -} - -bool pmo_dcn3_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i, stream_index; - - for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { - stream_index = in_out->base_display_config->display_config.plane_descriptors[i].stream_index; - - if (in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][pmo->scratch.pmo_dcn3.current_candidate[stream_index]] * 1000) { - return false; - } - } - - return true; -} - -bool pmo_dcn3_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - unsigned int stream_index; - bool success = false; - bool reached_end; - - memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); - - if (in_out->last_candidate_failed) { - if (pmo->scratch.pmo_dcn3.cur_latency_index < pmo->scratch.pmo_dcn3.max_latency_index) { - // If we haven't tried all the clock bounds to support this state, try a higher one - pmo->scratch.pmo_dcn3.cur_latency_index++; - - success = true; - } else { - // If there's nothing higher to try, then we have to have a smaller canadidate - reached_end = !iterate_to_next_candidiate(pmo, in_out->optimized_display_config->display_config.num_streams); - - if (!reached_end) { - pmo->scratch.pmo_dcn3.cur_latency_index = pmo->scratch.pmo_dcn3.min_latency_index; - success = true; - } - } - } else { - success = true; - } - - if (success) { - in_out->optimized_display_config->stage3.min_clk_index_for_latency = pmo->scratch.pmo_dcn3.cur_latency_index; - - for (stream_index = 0; stream_index < in_out->optimized_display_config->display_config.num_streams; stream_index++) { - set_reserved_time_on_all_planes_with_stream_index(in_out->optimized_display_config, stream_index, - pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][pmo->scratch.pmo_dcn3.current_candidate[stream_index]]); - } - } - - return success; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h deleted file mode 100644 index f00bd9e72a86..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h +++ /dev/null @@ -1,22 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_PMO_DCN3_H__ -#define __DML2_PMO_DCN3_H__ - -#include "dml2_internal_shared_types.h" - -bool pmo_dcn3_initialize(struct dml2_pmo_initialize_in_out *in_out); - -bool pmo_dcn3_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); - -bool pmo_dcn3_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out); -bool pmo_dcn3_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out); -bool pmo_dcn3_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out); - -bool pmo_dcn3_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out); -bool pmo_dcn3_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out); -bool pmo_dcn3_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c deleted file mode 100644 index 5769c2638f9a..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c +++ /dev/null @@ -1,2394 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_pmo_factory.h" -#include "dml2_debug.h" -#include "lib_float_math.h" -#include "dml2_pmo_dcn4_fams2.h" - -static const double MIN_VACTIVE_MARGIN_PCT = 0.25; // We need more than non-zero margin because DET buffer granularity can alter vactive latency hiding -static const double MIN_BLANK_STUTTER_FACTOR = 3.0; - -static const struct dml2_pmo_pstate_strategy base_strategy_list_1_display[] = { - // VActive Preferred - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then SVP - { - .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // Then DRR - { - .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Finally VBlank, but allow base clocks for latency to increase - /* - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - */ -}; - -static const int base_strategy_list_1_display_size = sizeof(base_strategy_list_1_display) / sizeof(struct dml2_pmo_pstate_strategy); - -static const struct dml2_pmo_pstate_strategy base_strategy_list_2_display[] = { - // VActive only is preferred - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then VActive + VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // Then VBlank only - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // Then SVP + VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // Then SVP + DRR - { - .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then SVP + SVP - { - .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_fw_svp, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then DRR + VActive - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Then DRR + DRR - { - .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // Finally VBlank, but allow base clocks for latency to increase - /* - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - */ -}; - -static const int base_strategy_list_2_display_size = sizeof(base_strategy_list_2_display) / sizeof(struct dml2_pmo_pstate_strategy); - -static const struct dml2_pmo_pstate_strategy base_strategy_list_3_display[] = { - // All VActive - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // VActive + 1 VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vblank, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // All VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na }, - .allow_state_increase = false, - }, - - // All DRR - { - .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - - // All VBlank, with state increase allowed - /* - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na }, - .allow_state_increase = true, - }, - */ -}; - -static const int base_strategy_list_3_display_size = sizeof(base_strategy_list_3_display) / sizeof(struct dml2_pmo_pstate_strategy); - -static const struct dml2_pmo_pstate_strategy base_strategy_list_4_display[] = { - // All VActive - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive }, - .allow_state_increase = true, - }, - - // VActive + 1 VBlank - { - .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vblank }, - .allow_state_increase = false, - }, - - // All Vblank - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank }, - .allow_state_increase = false, - }, - - // All DRR - { - .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr }, - .allow_state_increase = true, - }, - - // All VBlank, with state increase allowed - /* - { - .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank }, - .allow_state_increase = true, - }, - */ -}; - -static const int base_strategy_list_4_display_size = sizeof(base_strategy_list_4_display) / sizeof(struct dml2_pmo_pstate_strategy); - - -static bool increase_odm_combine_factor(enum dml2_odm_mode *odm_mode, int odms_calculated) -{ - bool result = true; - - if (*odm_mode == dml2_odm_mode_auto) { - switch (odms_calculated) { - case 1: - *odm_mode = dml2_odm_mode_bypass; - break; - case 2: - *odm_mode = dml2_odm_mode_combine_2to1; - break; - case 3: - *odm_mode = dml2_odm_mode_combine_3to1; - break; - case 4: - *odm_mode = dml2_odm_mode_combine_4to1; - break; - default: - result = false; - break; - } - } - - if (result) { - if (*odm_mode == dml2_odm_mode_bypass) { - *odm_mode = dml2_odm_mode_combine_2to1; - } else if (*odm_mode == dml2_odm_mode_combine_2to1) { - *odm_mode = dml2_odm_mode_combine_3to1; - } else if (*odm_mode == dml2_odm_mode_combine_3to1) { - *odm_mode = dml2_odm_mode_combine_4to1; - } else { - result = false; - } - } - - return result; -} - -static bool increase_mpc_combine_factor(unsigned int *mpc_combine_factor, unsigned int limit) -{ - if (*mpc_combine_factor < limit) { - (*mpc_combine_factor)++; - return true; - } - - return false; -} - -static int count_planes_with_stream_index(const struct dml2_display_cfg *display_cfg, unsigned int stream_index) -{ - unsigned int i, count; - - count = 0; - for (i = 0; i < display_cfg->num_planes; i++) { - if (display_cfg->plane_descriptors[i].stream_index == stream_index) - count++; - } - - return count; -} - -static bool optimize_dcc_mcache_no_odm(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out, - int free_pipes) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i; - bool result = true; - - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - // For pipes that failed dcc mcache check, we want to increase the pipe count. - // The logic for doing this depends on how many pipes is already being used, - // and whether it's mpcc or odm combine. - if (!in_out->dcc_mcache_supported[i]) { - // For the general case of "n displays", we can only optimize streams with an ODM combine factor of 1 - if (in_out->cfg_support_info->stream_support_info[in_out->optimized_display_cfg->plane_descriptors[i].stream_index].odms_used == 1) { - in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor = - in_out->cfg_support_info->plane_support_info[i].dpps_used; - // For each plane that is not passing mcache validation, just add another pipe to it, up to the limit. - if (free_pipes > 0) { - if (!increase_mpc_combine_factor(&in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor, - pmo->mpc_combine_limit)) { - // We've reached max pipes allocatable to a single plane, so we fail. - result = false; - break; - } else { - // Successfully added another pipe to this failing plane. - free_pipes--; - } - } else { - // No free pipes to add. - result = false; - break; - } - } else { - // If the stream of this plane needs ODM combine, no further optimization can be done. - result = false; - break; - } - } - } - - return result; -} - -bool pmo_dcn4_fams2_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i, used_pipes, free_pipes, planes_on_stream; - bool result; - - if (in_out->display_config != in_out->optimized_display_cfg) { - memcpy(in_out->optimized_display_cfg, in_out->display_config, sizeof(struct dml2_display_cfg)); - } - - //Count number of free pipes, and check if any odm combine is in use. - used_pipes = 0; - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - used_pipes += in_out->cfg_support_info->plane_support_info[i].dpps_used; - } - free_pipes = pmo->ip_caps->pipe_count - used_pipes; - - // Optimization loop - // The goal here is to add more pipes to any planes - // which are failing mcache admissibility - result = true; - - // The optimization logic depends on whether ODM combine is enabled, and the stream count. - if (in_out->optimized_display_cfg->num_streams > 1 || in_out->instance->options->disable_dyn_odm) { - // If there are multiple streams, we are limited to only be able to optimize mcache failures on planes - // which are not ODM combined. - - result = optimize_dcc_mcache_no_odm(in_out, free_pipes); - } else if (in_out->optimized_display_cfg->num_streams == 1) { - // In single stream cases, we still optimize mcache failures when there's ODM combine with some - // additional logic. - - if (in_out->cfg_support_info->stream_support_info[0].odms_used > 1) { - // If ODM combine is enabled, then the logic is to increase ODM combine factor. - - // Optimization for streams with > 1 ODM combine factor is only supported for single display. - planes_on_stream = count_planes_with_stream_index(in_out->optimized_display_cfg, 0); - - for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { - // For pipes that failed dcc mcache check, we want to increase the pipe count. - // The logic for doing this depends on how many pipes is already being used, - // and whether it's mpcc or odm combine. - if (!in_out->dcc_mcache_supported[i]) { - // Increasing ODM combine factor on a stream requires a free pipe for each plane on the stream. - if (free_pipes >= planes_on_stream) { - if (!increase_odm_combine_factor(&in_out->optimized_display_cfg->stream_descriptors[i].overrides.odm_mode, - in_out->cfg_support_info->plane_support_info[i].dpps_used)) { - result = false; - } else { - break; - } - } else { - result = false; - break; - } - } - } - } else { - // If ODM combine is not enabled, then we can actually use the same logic as before. - - result = optimize_dcc_mcache_no_odm(in_out, free_pipes); - } - } else { - result = true; - } - - return result; -} - -static enum dml2_pstate_method convert_strategy_to_drr_variant(const enum dml2_pstate_method base_strategy) -{ - enum dml2_pstate_method variant_strategy = 0; - - switch (base_strategy) { - case dml2_pstate_method_vactive: - variant_strategy = dml2_pstate_method_fw_vactive_drr; - break; - case dml2_pstate_method_vblank: - variant_strategy = dml2_pstate_method_fw_vblank_drr; - break; - case dml2_pstate_method_fw_svp: - variant_strategy = dml2_pstate_method_fw_svp_drr; - break; - case dml2_pstate_method_fw_vactive_drr: - case dml2_pstate_method_fw_vblank_drr: - case dml2_pstate_method_fw_svp_drr: - case dml2_pstate_method_fw_drr: - case dml2_pstate_method_reserved_hw: - case dml2_pstate_method_reserved_fw: - case dml2_pstate_method_reserved_fw_drr_clamped: - case dml2_pstate_method_reserved_fw_drr_var: - case dml2_pstate_method_count: - case dml2_pstate_method_na: - default: - /* no variant for this mode */ - variant_strategy = base_strategy; - } - - return variant_strategy; -} - -static struct dml2_pmo_pstate_strategy *get_expanded_strategy_list(struct dml2_pmo_init_data *init_data, int stream_count) -{ - struct dml2_pmo_pstate_strategy *expanded_strategy_list = NULL; - - switch (stream_count) { - case 1: - expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_1_display; - break; - case 2: - expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_2_display; - break; - case 3: - expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_3_display; - break; - case 4: - expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_4_display; - break; - default: - break; - } - - return expanded_strategy_list; -} - -static unsigned int get_num_expanded_strategies( - struct dml2_pmo_init_data *init_data, - int stream_count) -{ - return init_data->pmo_dcn4.num_expanded_strategies_per_list[stream_count - 1]; -} - -static void insert_strategy_into_expanded_list( - const struct dml2_pmo_pstate_strategy *per_stream_pstate_strategy, - const int stream_count, - struct dml2_pmo_pstate_strategy *expanded_strategy_list, - unsigned int *num_expanded_strategies) -{ - if (expanded_strategy_list && num_expanded_strategies) { - memcpy(&expanded_strategy_list[*num_expanded_strategies], per_stream_pstate_strategy, sizeof(struct dml2_pmo_pstate_strategy)); - - (*num_expanded_strategies)++; - } -} - -static void expand_base_strategy( - const struct dml2_pmo_pstate_strategy *base_strategy, - const unsigned int stream_count, - struct dml2_pmo_pstate_strategy *expanded_strategy_list, - unsigned int *num_expanded_strategies) -{ - bool skip_to_next_stream; - bool expanded_strategy_added; - bool skip_iteration; - unsigned int i, j; - unsigned int num_streams_per_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; - unsigned int stream_iteration_indices[PMO_DCN4_MAX_DISPLAYS] = { 0 }; - struct dml2_pmo_pstate_strategy cur_strategy_list = { 0 }; - - /* determine number of displays per method */ - for (i = 0; i < stream_count; i++) { - /* increment the count of the earliest index with the same method */ - for (j = 0; j < stream_count; j++) { - if (base_strategy->per_stream_pstate_method[i] == base_strategy->per_stream_pstate_method[j]) { - num_streams_per_method[j] = num_streams_per_method[j] + 1; - break; - } - } - } - - cur_strategy_list.allow_state_increase = base_strategy->allow_state_increase; - - i = 0; - /* uses a while loop instead of recursion to build permutations of base strategy */ - while (stream_iteration_indices[0] < stream_count) { - skip_to_next_stream = false; - expanded_strategy_added = false; - skip_iteration = false; - - /* determine what to do for this iteration */ - if (stream_iteration_indices[i] < stream_count && num_streams_per_method[stream_iteration_indices[i]] != 0) { - /* decrement count and assign method */ - cur_strategy_list.per_stream_pstate_method[i] = base_strategy->per_stream_pstate_method[stream_iteration_indices[i]]; - num_streams_per_method[stream_iteration_indices[i]] -= 1; - - if (i >= stream_count - 1) { - /* insert into strategy list */ - insert_strategy_into_expanded_list(&cur_strategy_list, stream_count, expanded_strategy_list, num_expanded_strategies); - expanded_strategy_added = true; - } else { - /* skip to next stream */ - skip_to_next_stream = true; - } - } else { - skip_iteration = true; - } - - /* prepare for next iteration */ - if (skip_to_next_stream) { - i++; - } else { - /* restore count */ - if (!skip_iteration) { - num_streams_per_method[stream_iteration_indices[i]] += 1; - } - - /* increment iteration count */ - stream_iteration_indices[i]++; - - /* if iterations are complete, or last stream was reached */ - if ((stream_iteration_indices[i] >= stream_count || expanded_strategy_added) && i > 0) { - /* reset per stream index, decrement i */ - stream_iteration_indices[i] = 0; - i--; - - /* restore previous stream's count and increment index */ - num_streams_per_method[stream_iteration_indices[i]] += 1; - stream_iteration_indices[i]++; - } - } - } -} - - -static bool is_variant_method_valid(const struct dml2_pmo_pstate_strategy *base_strategy, - const struct dml2_pmo_pstate_strategy *variant_strategy, - const unsigned int num_streams_per_base_method[PMO_DCN4_MAX_DISPLAYS], - const unsigned int num_streams_per_variant_method[PMO_DCN4_MAX_DISPLAYS], - const unsigned int stream_count) -{ - bool valid = true; - unsigned int i; - - /* check all restrictions are met */ - for (i = 0; i < stream_count; i++) { - /* vblank + vblank_drr variants are invalid */ - if (base_strategy->per_stream_pstate_method[i] == dml2_pstate_method_vblank && - ((num_streams_per_base_method[i] > 0 && num_streams_per_variant_method[i] > 0) || - num_streams_per_variant_method[i] > 1)) { - valid = false; - break; - } - } - - return valid; -} - -static void expand_variant_strategy( - const struct dml2_pmo_pstate_strategy *base_strategy, - const unsigned int stream_count, - const bool should_permute, - struct dml2_pmo_pstate_strategy *expanded_strategy_list, - unsigned int *num_expanded_strategies) -{ - bool variant_found; - unsigned int i, j; - unsigned int method_index; - unsigned int stream_index; - unsigned int num_streams_per_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; - unsigned int num_streams_per_base_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; - unsigned int num_streams_per_variant_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; - enum dml2_pstate_method per_stream_variant_method[DML2_MAX_PLANES]; - struct dml2_pmo_pstate_strategy variant_strategy = { 0 }; - - /* determine number of displays per method */ - for (i = 0; i < stream_count; i++) { - /* increment the count of the earliest index with the same method */ - for (j = 0; j < stream_count; j++) { - if (base_strategy->per_stream_pstate_method[i] == base_strategy->per_stream_pstate_method[j]) { - num_streams_per_method[j] = num_streams_per_method[j] + 1; - break; - } - } - - per_stream_variant_method[i] = convert_strategy_to_drr_variant(base_strategy->per_stream_pstate_method[i]); - } - memcpy(num_streams_per_base_method, num_streams_per_method, sizeof(unsigned int) * PMO_DCN4_MAX_DISPLAYS); - - memcpy(&variant_strategy, base_strategy, sizeof(struct dml2_pmo_pstate_strategy)); - - method_index = 0; - /* uses a while loop instead of recursion to build permutations of base strategy */ - while (num_streams_per_base_method[0] > 0 || method_index != 0) { - if (method_index == stream_count) { - /* construct variant strategy */ - variant_found = false; - stream_index = 0; - - for (i = 0; i < stream_count; i++) { - for (j = 0; j < num_streams_per_base_method[i]; j++) { - variant_strategy.per_stream_pstate_method[stream_index++] = base_strategy->per_stream_pstate_method[i]; - } - - for (j = 0; j < num_streams_per_variant_method[i]; j++) { - variant_strategy.per_stream_pstate_method[stream_index++] = per_stream_variant_method[i]; - if (base_strategy->per_stream_pstate_method[i] != per_stream_variant_method[i]) { - variant_found = true; - } - } - } - - if (variant_found && is_variant_method_valid(base_strategy, &variant_strategy, num_streams_per_base_method, num_streams_per_variant_method, stream_count)) { - if (should_permute) { - /* permutations are permitted, proceed to expand */ - expand_base_strategy(&variant_strategy, stream_count, expanded_strategy_list, num_expanded_strategies); - } else { - /* no permutations allowed, so add to list now */ - insert_strategy_into_expanded_list(&variant_strategy, stream_count, expanded_strategy_list, num_expanded_strategies); - } - } - - /* rollback to earliest method with bases remaining */ - for (method_index = stream_count - 1; method_index > 0; method_index--) { - if (num_streams_per_base_method[method_index]) { - /* bases remaining */ - break; - } else { - /* reset counters */ - num_streams_per_base_method[method_index] = num_streams_per_method[method_index]; - num_streams_per_variant_method[method_index] = 0; - } - } - } - - if (num_streams_per_base_method[method_index]) { - num_streams_per_base_method[method_index]--; - num_streams_per_variant_method[method_index]++; - - method_index++; - } else if (method_index != 0) { - method_index++; - } - } -} - -void pmo_dcn4_fams2_expand_base_pstate_strategies( - const struct dml2_pmo_pstate_strategy *base_strategies_list, - const unsigned int num_base_strategies, - const unsigned int stream_count, - struct dml2_pmo_pstate_strategy *expanded_strategy_list, - unsigned int *num_expanded_strategies) -{ - unsigned int i; - - /* expand every explicit base strategy (except all DRR) */ - for (i = 0; i < num_base_strategies; i++) { - expand_base_strategy(&base_strategies_list[i], stream_count, expanded_strategy_list, num_expanded_strategies); - expand_variant_strategy(&base_strategies_list[i], stream_count, true, expanded_strategy_list, num_expanded_strategies); - } -} - -bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) -{ - int i = 0; - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int base_list_size = 0; - const struct dml2_pmo_pstate_strategy *base_list = NULL; - unsigned int *expanded_list_size = NULL; - struct dml2_pmo_pstate_strategy *expanded_list = NULL; - - pmo->soc_bb = in_out->soc_bb; - pmo->ip_caps = in_out->ip_caps; - pmo->mpc_combine_limit = 2; - pmo->odm_combine_limit = 4; - pmo->mcg_clock_table_size = in_out->mcg_clock_table_size; - - pmo->fams_params.v2.subvp.refresh_rate_limit_max = 175; - pmo->fams_params.v2.subvp.refresh_rate_limit_min = 0; - pmo->fams_params.v2.drr.refresh_rate_limit_max = 1000; - pmo->fams_params.v2.drr.refresh_rate_limit_min = 119; - - pmo->options = in_out->options; - - /* generate permutations of p-state configs from base strategy list */ - for (i = 0; i < PMO_DCN4_MAX_DISPLAYS; i++) { - switch (i+1) { - case 1: - if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { - base_list = pmo->options->override_strategy_lists[i]; - base_list_size = pmo->options->num_override_strategies_per_list[i]; - } else { - base_list = base_strategy_list_1_display; - base_list_size = base_strategy_list_1_display_size; - } - - expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; - expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_1_display; - - break; - case 2: - if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { - base_list = pmo->options->override_strategy_lists[i]; - base_list_size = pmo->options->num_override_strategies_per_list[i]; - } else { - base_list = base_strategy_list_2_display; - base_list_size = base_strategy_list_2_display_size; - } - - expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; - expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_2_display; - - break; - case 3: - if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { - base_list = pmo->options->override_strategy_lists[i]; - base_list_size = pmo->options->num_override_strategies_per_list[i]; - } else { - base_list = base_strategy_list_3_display; - base_list_size = base_strategy_list_3_display_size; - } - - expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; - expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_3_display; - - break; - case 4: - if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { - base_list = pmo->options->override_strategy_lists[i]; - base_list_size = pmo->options->num_override_strategies_per_list[i]; - } else { - base_list = base_strategy_list_4_display; - base_list_size = base_strategy_list_4_display_size; - } - - expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; - expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_4_display; - - break; - } - - DML_ASSERT(base_list_size <= PMO_DCN4_MAX_BASE_STRATEGIES); - - /* populate list */ - pmo_dcn4_fams2_expand_base_pstate_strategies( - base_list, - base_list_size, - i + 1, - expanded_list, - expanded_list_size); - } - - return true; -} - -static bool is_h_timing_divisible_by(const struct dml2_timing_cfg *timing, unsigned char denominator) -{ - /* - * Htotal, Hblank start/end, and Hsync start/end all must be divisible - * in order for the horizontal timing params to be considered divisible - * by 2. Hsync start is always 0. - */ - unsigned long h_blank_start = timing->h_total - timing->h_front_porch; - - return (timing->h_total % denominator == 0) && - (h_blank_start % denominator == 0) && - (timing->h_blank_end % denominator == 0) && - (timing->h_sync_width % denominator == 0); -} - -static bool is_dp_encoder(enum dml2_output_encoder_class encoder_type) -{ - switch (encoder_type) { - case dml2_dp: - case dml2_edp: - case dml2_dp2p0: - case dml2_none: - return true; - case dml2_hdmi: - case dml2_hdmifrl: - default: - return false; - } -} - -bool pmo_dcn4_fams2_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out) -{ - unsigned int i; - const struct dml2_display_cfg *display_config = - &in_out->base_display_config->display_config; - const struct dml2_core_mode_support_result *mode_support_result = - &in_out->base_display_config->mode_support_result; - struct dml2_optimization_stage4_state *state = - &in_out->base_display_config->stage4; - - if (in_out->instance->options->disable_dyn_odm || - (in_out->instance->options->disable_dyn_odm_for_multi_stream && display_config->num_streams > 1)) - return false; - - for (i = 0; i < display_config->num_planes; i++) - /* - * vmin optimization is required to be seamlessly switched off - * at any time when the new configuration is no longer - * supported. However switching from ODM combine to MPC combine - * is not always seamless. When there not enough free pipes, we - * will have to use the same secondary OPP heads as secondary - * DPP pipes in MPC combine in new state. This transition is - * expected to cause glitches. To avoid the transition, we only - * allow vmin optimization if the stream's base configuration - * doesn't require MPC combine. This condition checks if MPC - * combine is enabled. If so do not optimize the stream. - */ - if (mode_support_result->cfg_support_info.plane_support_info[i].dpps_used > 1 && - mode_support_result->cfg_support_info.stream_support_info[display_config->plane_descriptors[i].stream_index].odms_used == 1) - state->unoptimizable_streams[display_config->plane_descriptors[i].stream_index] = true; - - for (i = 0; i < display_config->num_streams; i++) { - if (display_config->stream_descriptors[i].overrides.disable_dynamic_odm) - state->unoptimizable_streams[i] = true; - else if (in_out->base_display_config->stage3.stream_svp_meta[i].valid && - in_out->instance->options->disable_dyn_odm_for_stream_with_svp) - state->unoptimizable_streams[i] = true; - /* - * ODM Combine requires horizontal timing divisible by 2 so each - * ODM segment has the same size. - */ - else if (!is_h_timing_divisible_by(&display_config->stream_descriptors[i].timing, 2)) - state->unoptimizable_streams[i] = true; - /* - * Our hardware support seamless ODM transitions for DP encoders - * only. - */ - else if (!is_dp_encoder(display_config->stream_descriptors[i].output.output_encoder)) - state->unoptimizable_streams[i] = true; - } - - state->performed = true; - - return true; -} - -bool pmo_dcn4_fams2_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out) -{ - bool is_vmin = true; - - if (in_out->vmin_limits->dispclk_khz > 0 && - in_out->display_config->mode_support_result.global.dispclk_khz > in_out->vmin_limits->dispclk_khz) - is_vmin = false; - - return is_vmin; -} - -static int find_highest_odm_load_stream_index( - const struct dml2_display_cfg *display_config, - const struct dml2_core_mode_support_result *mode_support_result) -{ - unsigned int i; - int odm_load, highest_odm_load = -1, highest_odm_load_index = -1; - - for (i = 0; i < display_config->num_streams; i++) { - if (mode_support_result->cfg_support_info.stream_support_info[i].odms_used > 0) - odm_load = display_config->stream_descriptors[i].timing.pixel_clock_khz - / mode_support_result->cfg_support_info.stream_support_info[i].odms_used; - else - odm_load = 0; - - if (odm_load > highest_odm_load) { - highest_odm_load_index = i; - highest_odm_load = odm_load; - } - } - - return highest_odm_load_index; -} - -bool pmo_dcn4_fams2_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out) -{ - int stream_index; - const struct dml2_display_cfg *display_config = - &in_out->base_display_config->display_config; - const struct dml2_core_mode_support_result *mode_support_result = - &in_out->base_display_config->mode_support_result; - unsigned int odms_used; - struct dml2_stream_parameters *stream_descriptor; - bool optimizable = false; - - /* - * highest odm load stream must be optimizable to continue as dispclk is - * bounded by it. - */ - stream_index = find_highest_odm_load_stream_index(display_config, - mode_support_result); - - if (stream_index < 0 || - in_out->base_display_config->stage4.unoptimizable_streams[stream_index]) - return false; - - odms_used = mode_support_result->cfg_support_info.stream_support_info[stream_index].odms_used; - if ((int)odms_used >= in_out->instance->odm_combine_limit) - return false; - - memcpy(in_out->optimized_display_config, - in_out->base_display_config, - sizeof(struct display_configuation_with_meta)); - - stream_descriptor = &in_out->optimized_display_config->display_config.stream_descriptors[stream_index]; - while (!optimizable && increase_odm_combine_factor( - &stream_descriptor->overrides.odm_mode, - odms_used)) { - switch (stream_descriptor->overrides.odm_mode) { - case dml2_odm_mode_combine_2to1: - optimizable = true; - break; - case dml2_odm_mode_combine_3to1: - /* - * In ODM Combine 3:1 OTG_valid_pixel rate is 1/4 of - * actual pixel rate. Therefore horizontal timing must - * be divisible by 4. - */ - if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { - /* - * DSC h slice count must be divisible - * by 3. - */ - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 3 == 0) - optimizable = true; - } else { - optimizable = true; - } - } - break; - case dml2_odm_mode_combine_4to1: - /* - * In ODM Combine 4:1 OTG_valid_pixel rate is 1/4 of - * actual pixel rate. Therefore horizontal timing must - * be divisible by 4. - */ - if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { - /* - * DSC h slice count must be divisible - * by 4. - */ - if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 4 == 0) - optimizable = true; - } else { - optimizable = true; - } - } - break; - case dml2_odm_mode_auto: - case dml2_odm_mode_bypass: - case dml2_odm_mode_split_1to2: - case dml2_odm_mode_mso_1to2: - case dml2_odm_mode_mso_1to4: - default: - break; - } - } - - return optimizable; -} - -static void set_bit_in_bitfield(unsigned int *bit_field, unsigned int bit_offset) -{ - *bit_field = *bit_field | (0x1 << bit_offset); -} - -static bool is_bit_set_in_bitfield(unsigned int bit_field, unsigned int bit_offset) -{ - if (bit_field & (0x1 << bit_offset)) - return true; - - return false; -} - -static void build_synchronized_timing_groups( - struct dml2_pmo_instance *pmo, - struct display_configuation_with_meta *display_config) -{ - unsigned int i, j; - struct dml2_timing_cfg *master_timing; - - unsigned int stream_mapped_mask = 0; - unsigned int num_timing_groups = 0; - unsigned int timing_group_idx = 0; - struct dml2_pmo_scratch *s = &pmo->scratch; - - /* clear all group masks */ - memset(s->pmo_dcn4.synchronized_timing_group_masks, 0, sizeof(s->pmo_dcn4.synchronized_timing_group_masks)); - memset(s->pmo_dcn4.group_is_drr_enabled, 0, sizeof(s->pmo_dcn4.group_is_drr_enabled)); - memset(s->pmo_dcn4.group_is_drr_active, 0, sizeof(s->pmo_dcn4.group_is_drr_active)); - memset(s->pmo_dcn4.group_line_time_us, 0, sizeof(s->pmo_dcn4.group_line_time_us)); - s->pmo_dcn4.num_timing_groups = 0; - - for (i = 0; i < display_config->display_config.num_streams; i++) { - master_timing = &display_config->display_config.stream_descriptors[i].timing; - - /* only need to build group of this stream is not in a group already */ - if (is_bit_set_in_bitfield(stream_mapped_mask, i)) { - continue; - } - set_bit_in_bitfield(&stream_mapped_mask, i); - timing_group_idx = num_timing_groups; - num_timing_groups++; - - /* trivially set default timing group to itself */ - set_bit_in_bitfield(&s->pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], i); - s->pmo_dcn4.group_line_time_us[timing_group_idx] = (double)master_timing->h_total / master_timing->pixel_clock_khz * 1000.0; - - /* if drr is in use, timing is not sychnronizable */ - if (master_timing->drr_config.enabled) { - s->pmo_dcn4.group_is_drr_enabled[timing_group_idx] = true; - s->pmo_dcn4.group_is_drr_active[timing_group_idx] = !master_timing->drr_config.disallowed && - (master_timing->drr_config.drr_active_fixed || master_timing->drr_config.drr_active_variable); - continue; - } - - /* find synchronizable timing groups */ - for (j = i + 1; j < display_config->display_config.num_streams; j++) { - if (memcmp(master_timing, - &display_config->display_config.stream_descriptors[j].timing, - sizeof(struct dml2_timing_cfg)) == 0) { - set_bit_in_bitfield(&pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], j); - set_bit_in_bitfield(&stream_mapped_mask, j); - } - } - } - - s->pmo_dcn4.num_timing_groups = num_timing_groups; -} - -static bool all_timings_support_vactive(const struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_config, - unsigned int mask) -{ - unsigned int i; - bool valid = true; - - // Create a remap array to enable simple iteration through only masked stream indicies - for (i = 0; i < display_config->display_config.num_streams; i++) { - if (is_bit_set_in_bitfield(mask, i)) { - /* check if stream has enough vactive margin */ - valid &= is_bit_set_in_bitfield(pmo->scratch.pmo_dcn4.stream_vactive_capability_mask, i); - } - } - - return valid; -} - -static bool all_timings_support_vblank(const struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_config, - unsigned int mask) -{ - unsigned int i; - - bool synchronizable = true; - - /* find first vblank stream index and compare the timing group mask */ - for (i = 0; i < display_config->display_config.num_streams; i++) { - if (is_bit_set_in_bitfield(mask, i)) { - if (mask != pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[i]) { - /* vblank streams are not synchronizable */ - synchronizable = false; - } - break; - } - } - - return synchronizable; -} - -static unsigned int calc_svp_microschedule(const struct dml2_pstate_meta *pstate_meta) -{ - return pstate_meta->contention_delay_otg_vlines + - pstate_meta->method_subvp.programming_delay_otg_vlines + - pstate_meta->method_subvp.phantom_vtotal + - pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + - pstate_meta->blackout_otg_vlines; -} - -static bool all_timings_support_drr(const struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_config, - unsigned int mask) -{ - unsigned int i; - for (i = 0; i < DML2_MAX_PLANES; i++) { - const struct dml2_stream_parameters *stream_descriptor; - const struct dml2_pstate_meta *stream_pstate_meta; - - if (is_bit_set_in_bitfield(mask, i)) { - stream_descriptor = &display_config->display_config.stream_descriptors[i]; - stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; - - if (!stream_descriptor->timing.drr_config.enabled) - return false; - - /* cannot support required vtotal */ - if (stream_pstate_meta->method_drr.stretched_vtotal > stream_pstate_meta->max_vtotal) { - return false; - } - - /* check rr is within bounds */ - if (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.drr.refresh_rate_limit_min || - stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.drr.refresh_rate_limit_max) { - return false; - } - - /* check required stretch is allowed */ - if (stream_descriptor->timing.drr_config.max_instant_vtotal_delta > 0 && - stream_pstate_meta->method_drr.stretched_vtotal - stream_pstate_meta->nom_vtotal > stream_descriptor->timing.drr_config.max_instant_vtotal_delta) { - return false; - } - } - } - - return true; -} - -static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_config, - unsigned int mask) -{ - const struct dml2_stream_parameters *stream_descriptor; - const struct dml2_plane_parameters *plane_descriptor; - const struct dml2_pstate_meta *stream_pstate_meta; - unsigned int microschedule_vlines; - unsigned int i; - unsigned int mcaches_per_plane; - unsigned int total_mcaches_required = 0; - - unsigned int num_planes_per_stream[DML2_MAX_PLANES] = { 0 }; - - /* confirm timing it is not a centered timing */ - for (i = 0; i < display_config->display_config.num_planes; i++) { - plane_descriptor = &display_config->display_config.plane_descriptors[i]; - mcaches_per_plane = 0; - - if (plane_descriptor->surface.dcc.enable) { - mcaches_per_plane += display_config->stage2.mcache_allocations[i].num_mcaches_plane0 + - display_config->stage2.mcache_allocations[i].num_mcaches_plane1 - - (display_config->stage2.mcache_allocations[i].last_slice_sharing.plane0_plane1 ? 1 : 0); - } - - if (is_bit_set_in_bitfield(mask, (unsigned char)plane_descriptor->stream_index)) { - num_planes_per_stream[plane_descriptor->stream_index]++; - - /* check recout height covers entire otg vactive, and single plane */ - if (num_planes_per_stream[plane_descriptor->stream_index] > 1 || - !plane_descriptor->composition.rect_out_height_spans_vactive || - plane_descriptor->composition.rotation_angle != dml2_rotation_0) { - return false; - } - - /* phantom requires same number of mcaches as main */ - if (plane_descriptor->surface.dcc.enable) { - mcaches_per_plane *= 2; - } - } - total_mcaches_required += mcaches_per_plane; - } - - if (total_mcaches_required > pmo->soc_bb->num_dcc_mcaches) { - /* too many mcaches required */ - return false; - } - - for (i = 0; i < DML2_MAX_PLANES; i++) { - if (is_bit_set_in_bitfield(mask, i)) { - stream_descriptor = &display_config->display_config.stream_descriptors[i]; - stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; - - if (stream_descriptor->overrides.disable_subvp) { - return false; - } - - microschedule_vlines = calc_svp_microschedule(&pmo->scratch.pmo_dcn4.stream_pstate_meta[i]); - - /* block if using an interlaced timing */ - if (stream_descriptor->timing.interlaced) { - return false; - } - - /* 1) svp main stream's vactive must be able to fit the microschedule - * 2) refresh rate must be within the allowed bounds - */ - if (microschedule_vlines >= stream_descriptor->timing.v_active || - (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.subvp.refresh_rate_limit_min || - stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.subvp.refresh_rate_limit_max)) { - return false; - } - } - } - - return true; -} - -static void insert_into_candidate_list(const struct dml2_pmo_pstate_strategy *pstate_strategy, int stream_count, struct dml2_pmo_scratch *scratch) -{ - scratch->pmo_dcn4.pstate_strategy_candidates[scratch->pmo_dcn4.num_pstate_candidates] = *pstate_strategy; - scratch->pmo_dcn4.num_pstate_candidates++; -} - -static enum dml2_pstate_method uclk_pstate_strategy_override_to_pstate_method(const enum dml2_uclk_pstate_change_strategy override_strategy) -{ - enum dml2_pstate_method method = dml2_pstate_method_na; - - switch (override_strategy) { - case dml2_uclk_pstate_change_strategy_force_vactive: - method = dml2_pstate_method_vactive; - break; - case dml2_uclk_pstate_change_strategy_force_vblank: - method = dml2_pstate_method_vblank; - break; - case dml2_uclk_pstate_change_strategy_force_drr: - method = dml2_pstate_method_fw_drr; - break; - case dml2_uclk_pstate_change_strategy_force_mall_svp: - method = dml2_pstate_method_fw_svp; - break; - case dml2_uclk_pstate_change_strategy_force_mall_full_frame: - case dml2_uclk_pstate_change_strategy_auto: - default: - method = dml2_pstate_method_na; - } - - return method; -} - -static enum dml2_uclk_pstate_change_strategy pstate_method_to_uclk_pstate_strategy_override(const enum dml2_pstate_method method) -{ - enum dml2_uclk_pstate_change_strategy override_strategy = dml2_uclk_pstate_change_strategy_auto; - - switch (method) { - case dml2_pstate_method_vactive: - case dml2_pstate_method_fw_vactive_drr: - override_strategy = dml2_uclk_pstate_change_strategy_force_vactive; - break; - case dml2_pstate_method_vblank: - case dml2_pstate_method_fw_vblank_drr: - override_strategy = dml2_uclk_pstate_change_strategy_force_vblank; - break; - case dml2_pstate_method_fw_svp: - case dml2_pstate_method_fw_svp_drr: - override_strategy = dml2_uclk_pstate_change_strategy_force_mall_svp; - break; - case dml2_pstate_method_fw_drr: - override_strategy = dml2_uclk_pstate_change_strategy_force_drr; - break; - case dml2_pstate_method_reserved_hw: - case dml2_pstate_method_reserved_fw: - case dml2_pstate_method_reserved_fw_drr_clamped: - case dml2_pstate_method_reserved_fw_drr_var: - case dml2_pstate_method_count: - case dml2_pstate_method_na: - default: - override_strategy = dml2_uclk_pstate_change_strategy_auto; - } - - return override_strategy; -} - -static bool all_planes_match_method(const struct display_configuation_with_meta *display_cfg, int plane_mask, enum dml2_pstate_method method) -{ - unsigned int i; - - for (i = 0; i < DML2_MAX_PLANES; i++) { - if (is_bit_set_in_bitfield(plane_mask, i)) { - if (display_cfg->display_config.plane_descriptors[i].overrides.uclk_pstate_change_strategy != dml2_uclk_pstate_change_strategy_auto && - display_cfg->display_config.plane_descriptors[i].overrides.uclk_pstate_change_strategy != pstate_method_to_uclk_pstate_strategy_override(method)) - return false; - } - } - - return true; -} - -static void build_method_scheduling_params( - struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta, - struct dml2_pstate_meta *stream_pstate_meta) -{ - stream_method_pstate_meta->allow_time_us = - (double)((int)stream_method_pstate_meta->allow_end_otg_vline - (int)stream_method_pstate_meta->allow_start_otg_vline) * - stream_pstate_meta->otg_vline_time_us; - if (stream_method_pstate_meta->allow_time_us >= stream_method_pstate_meta->period_us) { - /* when allow wave overlaps an entire frame, it is always schedulable (DRR can do this)*/ - stream_method_pstate_meta->disallow_time_us = 0.0; - } else { - stream_method_pstate_meta->disallow_time_us = - stream_method_pstate_meta->period_us - stream_method_pstate_meta->allow_time_us; - } -} - -static struct dml2_pstate_per_method_common_meta *get_per_method_common_meta( - struct dml2_pmo_instance *pmo, - enum dml2_pstate_method stream_pstate_method, - int stream_idx) -{ - struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta = NULL; - - switch (stream_pstate_method) { - case dml2_pstate_method_vactive: - case dml2_pstate_method_fw_vactive_drr: - stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vactive.common; - break; - case dml2_pstate_method_vblank: - case dml2_pstate_method_fw_vblank_drr: - stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vblank.common; - break; - case dml2_pstate_method_fw_svp: - case dml2_pstate_method_fw_svp_drr: - stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_subvp.common; - break; - case dml2_pstate_method_fw_drr: - stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_drr.common; - break; - case dml2_pstate_method_reserved_hw: - case dml2_pstate_method_reserved_fw: - case dml2_pstate_method_reserved_fw_drr_clamped: - case dml2_pstate_method_reserved_fw_drr_var: - case dml2_pstate_method_count: - case dml2_pstate_method_na: - default: - stream_method_pstate_meta = NULL; - } - - return stream_method_pstate_meta; -} - -static bool is_timing_group_schedulable( - struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_cfg, - const struct dml2_pmo_pstate_strategy *pstate_strategy, - const unsigned int timing_group_idx, - struct dml2_pstate_per_method_common_meta *group_pstate_meta) -{ - unsigned int i; - struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta; - - unsigned int base_stream_idx = 0; - struct dml2_pmo_scratch *s = &pmo->scratch; - - /* find base stream idx */ - for (base_stream_idx = 0; base_stream_idx < display_cfg->display_config.num_streams; base_stream_idx++) { - if (is_bit_set_in_bitfield(s->pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], base_stream_idx)) { - /* master stream found */ - break; - } - } - - /* init allow start and end lines for timing group */ - stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[base_stream_idx], base_stream_idx); - if (!stream_method_pstate_meta) - return false; - - group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; - group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; - group_pstate_meta->period_us = stream_method_pstate_meta->period_us; - for (i = base_stream_idx + 1; i < display_cfg->display_config.num_streams; i++) { - if (is_bit_set_in_bitfield(pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], i)) { - stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[i], i); - if (!stream_method_pstate_meta) - continue; - - if (group_pstate_meta->allow_start_otg_vline < stream_method_pstate_meta->allow_start_otg_vline) { - /* set group allow start to larger otg vline */ - group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; - } - - if (group_pstate_meta->allow_end_otg_vline > stream_method_pstate_meta->allow_end_otg_vline) { - /* set group allow end to smaller otg vline */ - group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; - } - - /* check waveform still has positive width */ - if (group_pstate_meta->allow_start_otg_vline >= group_pstate_meta->allow_end_otg_vline) { - /* timing group is not schedulable */ - return false; - } - } - } - - /* calculate the rest of the meta */ - build_method_scheduling_params(group_pstate_meta, &pmo->scratch.pmo_dcn4.stream_pstate_meta[base_stream_idx]); - - return group_pstate_meta->allow_time_us > 0.0 && - group_pstate_meta->disallow_time_us < pmo->ip_caps->fams2.max_allow_delay_us; -} - -static bool is_config_schedulable( - struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_cfg, - const struct dml2_pmo_pstate_strategy *pstate_strategy) -{ - unsigned int i, j; - bool schedulable; - struct dml2_pmo_scratch *s = &pmo->scratch; - - double max_allow_delay_us = 0.0; - - memset(s->pmo_dcn4.group_common_pstate_meta, 0, sizeof(s->pmo_dcn4.group_common_pstate_meta)); - memset(s->pmo_dcn4.sorted_group_gtl_disallow_index, 0, sizeof(unsigned int) * DML2_MAX_PLANES); - - /* search for a general solution to the schedule */ - - /* STAGE 0: Early return for special cases */ - if (display_cfg->display_config.num_streams == 0) { - return true; - } - - /* STAGE 1: confirm allow waves overlap for synchronizable streams */ - schedulable = true; - for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { - s->pmo_dcn4.sorted_group_gtl_disallow_index[i] = i; - s->pmo_dcn4.sorted_group_gtl_period_index[i] = i; - if (!is_timing_group_schedulable(pmo, display_cfg, pstate_strategy, i, &s->pmo_dcn4.group_common_pstate_meta[i])) { - /* synchronized timing group was not schedulable */ - schedulable = false; - break; - } - max_allow_delay_us += s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us; - } - - if ((schedulable && s->pmo_dcn4.num_timing_groups <= 1) || !schedulable) { - /* 1. the only timing group was schedulable, so early pass - * 2. one of the timing groups was not schedulable, so early fail */ - return schedulable; - } - - /* STAGE 2: Check allow can't be masked entirely by other disallows */ - schedulable = true; - - /* sort disallow times from greatest to least */ - for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { - bool swapped = false; - - for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { - double j_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j]].disallow_time_us; - double jp1_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]].disallow_time_us; - if (j_disallow_us < jp1_disallow_us) { - /* swap as A < B */ - swap(s->pmo_dcn4.sorted_group_gtl_disallow_index[j], - s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]); - swapped = true; - } - } - - /* sorted, exit early */ - if (!swapped) - break; - } - - /* Check worst case disallow region occurs in the middle of allow for the - * other display, or when >2 streams continue to halve the remaining allow time. - */ - for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { - if (s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us <= 0.0) { - /* this timing group always allows */ - continue; - } - - double max_allow_time_us = s->pmo_dcn4.group_common_pstate_meta[i].allow_time_us; - for (j = 0; j < s->pmo_dcn4.num_timing_groups; j++) { - unsigned int sorted_j = s->pmo_dcn4.sorted_group_gtl_disallow_index[j]; - /* stream can't overlap itself */ - if (i != sorted_j && s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us > 0.0) { - max_allow_time_us = math_min2( - s->pmo_dcn4.group_common_pstate_meta[sorted_j].allow_time_us, - (max_allow_time_us - s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us) / 2); - - if (max_allow_time_us < 0.0) { - /* failed exit early */ - break; - } - } - } - - if (max_allow_time_us <= 0.0) { - /* not enough time for microschedule in the worst case */ - schedulable = false; - break; - } - } - - if (schedulable && max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { - return true; - } - - /* STAGE 3: check larger allow can fit period of all other streams */ - schedulable = true; - - /* sort periods from greatest to least */ - for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { - bool swapped = false; - - for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { - double j_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j]].period_us; - double jp1_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]].period_us; - if (j_period_us < jp1_period_us) { - /* swap as A < B */ - swap(s->pmo_dcn4.sorted_group_gtl_period_index[j], - s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]); - swapped = true; - } - } - - /* sorted, exit early */ - if (!swapped) - break; - } - - /* check larger allow can fit period of all other streams */ - for (i = 0; i < s->pmo_dcn4.num_timing_groups - 1; i++) { - unsigned int sorted_i = s->pmo_dcn4.sorted_group_gtl_period_index[i]; - unsigned int sorted_ip1 = s->pmo_dcn4.sorted_group_gtl_period_index[i + 1]; - - if (s->pmo_dcn4.group_common_pstate_meta[sorted_i].allow_time_us < s->pmo_dcn4.group_common_pstate_meta[sorted_ip1].period_us || - (s->pmo_dcn4.group_is_drr_enabled[sorted_ip1] && s->pmo_dcn4.group_is_drr_active[sorted_ip1])) { - schedulable = false; - break; - } - } - - if (schedulable && max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { - return true; - } - - /* STAGE 4: When using HW exclusive modes, check disallow alignments are within allowed threshold */ - if (s->pmo_dcn4.num_timing_groups == 2 && - !is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, pstate_strategy->per_stream_pstate_method[0]) && - !is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, pstate_strategy->per_stream_pstate_method[1])) { - double period_ratio; - double max_shift_us; - double shift_per_period; - - /* default period_0 > period_1 */ - unsigned int lrg_idx = 0; - unsigned int sml_idx = 1; - if (s->pmo_dcn4.group_common_pstate_meta[0].period_us < s->pmo_dcn4.group_common_pstate_meta[1].period_us) { - /* period_0 < period_1 */ - lrg_idx = 1; - sml_idx = 0; - } - period_ratio = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us / s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us; - shift_per_period = s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us * (period_ratio - math_floor(period_ratio)); - max_shift_us = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].disallow_time_us - s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us; - max_allow_delay_us = max_shift_us / shift_per_period * s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us; - - if (shift_per_period > 0.0 && - shift_per_period < s->pmo_dcn4.group_common_pstate_meta[lrg_idx].allow_time_us + s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us && - max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { - schedulable = true; - } - } - - return schedulable; -} - -static bool stream_matches_drr_policy(struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_cfg, - const enum dml2_pstate_method stream_pstate_method, - unsigned int stream_index) -{ - const struct dml2_stream_parameters *stream_descriptor = &display_cfg->display_config.stream_descriptors[stream_index]; - bool strategy_matches_drr_requirements = true; - - /* check if strategy is compatible with stream drr capability and strategy */ - if (is_bit_set_in_bitfield(PMO_NO_DRR_STRATEGY_MASK, stream_pstate_method) && - display_cfg->display_config.num_streams > 1 && - stream_descriptor->timing.drr_config.enabled && - (stream_descriptor->timing.drr_config.drr_active_fixed || stream_descriptor->timing.drr_config.drr_active_variable)) { - /* DRR is active, so config may become unschedulable */ - strategy_matches_drr_requirements = false; - } else if (is_bit_set_in_bitfield(PMO_NO_DRR_STRATEGY_MASK, stream_pstate_method) && - is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, stream_pstate_method) && - stream_descriptor->timing.drr_config.enabled && - stream_descriptor->timing.drr_config.drr_active_variable) { - /* DRR is variable, fw exclusive methods require DRR to be clamped */ - strategy_matches_drr_requirements = false; - } else if (is_bit_set_in_bitfield(PMO_DRR_VAR_STRATEGY_MASK, stream_pstate_method) && - pmo->options->disable_drr_var_when_var_active && - stream_descriptor->timing.drr_config.enabled && - stream_descriptor->timing.drr_config.drr_active_variable) { - /* DRR variable is active, but policy blocks DRR for p-state when this happens */ - strategy_matches_drr_requirements = false; - } else if (is_bit_set_in_bitfield(PMO_DRR_VAR_STRATEGY_MASK, stream_pstate_method) && - (pmo->options->disable_drr_var || - !stream_descriptor->timing.drr_config.enabled || - stream_descriptor->timing.drr_config.disallowed)) { - /* DRR variable strategies are disallowed due to settings or policy */ - strategy_matches_drr_requirements = false; - } else if (is_bit_set_in_bitfield(PMO_DRR_CLAMPED_STRATEGY_MASK, stream_pstate_method) && - (pmo->options->disable_drr_clamped || - (!stream_descriptor->timing.drr_config.enabled || - (!stream_descriptor->timing.drr_config.drr_active_fixed && !stream_descriptor->timing.drr_config.drr_active_variable)) || - (pmo->options->disable_drr_clamped_when_var_active && - stream_descriptor->timing.drr_config.enabled && - stream_descriptor->timing.drr_config.drr_active_variable))) { - /* DRR fixed strategies are disallowed due to settings or policy */ - strategy_matches_drr_requirements = false; - } else if (is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, stream_pstate_method) && - pmo->options->disable_fams2) { - /* FW modes require FAMS2 */ - strategy_matches_drr_requirements = false; - } - - return strategy_matches_drr_requirements; -} - -static bool validate_pstate_support_strategy_cofunctionality(struct dml2_pmo_instance *pmo, - const struct display_configuation_with_meta *display_cfg, - const struct dml2_pmo_pstate_strategy *pstate_strategy) -{ - struct dml2_pmo_scratch *s = &pmo->scratch; - - unsigned int stream_index = 0; - - unsigned int svp_count = 0; - unsigned int svp_stream_mask = 0; - unsigned int drr_count = 0; - unsigned int drr_stream_mask = 0; - unsigned int vactive_count = 0; - unsigned int vactive_stream_mask = 0; - unsigned int vblank_count = 0; - unsigned int vblank_stream_mask = 0; - - bool strategy_matches_forced_requirements = true; - bool strategy_matches_drr_requirements = true; - - // Tabulate everything - for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { - - if (!all_planes_match_method(display_cfg, s->pmo_dcn4.stream_plane_mask[stream_index], - pstate_strategy->per_stream_pstate_method[stream_index])) { - strategy_matches_forced_requirements = false; - break; - } - - strategy_matches_drr_requirements &= - stream_matches_drr_policy(pmo, display_cfg, pstate_strategy->per_stream_pstate_method[stream_index], stream_index); - - if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp || - pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { - svp_count++; - set_bit_in_bitfield(&svp_stream_mask, stream_index); - } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { - drr_count++; - set_bit_in_bitfield(&drr_stream_mask, stream_index); - } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive || - pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { - vactive_count++; - set_bit_in_bitfield(&vactive_stream_mask, stream_index); - } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank || - pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { - vblank_count++; - set_bit_in_bitfield(&vblank_stream_mask, stream_index); - } - } - - if (!strategy_matches_forced_requirements || !strategy_matches_drr_requirements) - return false; - - if (vactive_count > 0 && !all_timings_support_vactive(pmo, display_cfg, vactive_stream_mask)) - return false; - - if (vblank_count > 0 && (pmo->options->disable_vblank || !all_timings_support_vblank(pmo, display_cfg, vblank_stream_mask))) - return false; - - if (drr_count > 0 && (pmo->options->disable_drr_var || !all_timings_support_drr(pmo, display_cfg, drr_stream_mask))) - return false; - - if (svp_count > 0 && (pmo->options->disable_svp || !all_timings_support_svp(pmo, display_cfg, svp_stream_mask))) - return false; - - return is_config_schedulable(pmo, display_cfg, pstate_strategy); -} - -static int get_vactive_pstate_margin(const struct display_configuation_with_meta *display_cfg, int plane_mask) -{ - unsigned int i; - int min_vactive_margin_us = 0xFFFFFFF; - - for (i = 0; i < DML2_MAX_PLANES; i++) { - if (is_bit_set_in_bitfield(plane_mask, i)) { - if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_latency_hiding_margin_in_active < min_vactive_margin_us) - min_vactive_margin_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_latency_hiding_margin_in_active; - } - } - - return min_vactive_margin_us; -} - -static unsigned int get_vactive_det_fill_latency_delay_us(const struct display_configuation_with_meta *display_cfg, int plane_mask) -{ - unsigned char i; - unsigned int max_vactive_fill_us = 0; - - for (i = 0; i < DML2_MAX_PLANES; i++) { - if (is_bit_set_in_bitfield(plane_mask, i)) { - if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us > max_vactive_fill_us) - max_vactive_fill_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us; - } - } - - return max_vactive_fill_us; -} - -static void build_pstate_meta_per_stream(struct dml2_pmo_instance *pmo, - struct display_configuation_with_meta *display_config, - int stream_index) -{ - const struct dml2_ip_capabilities *ip_caps = pmo->ip_caps; - const struct dml2_stream_parameters *stream_descriptor = &display_config->display_config.stream_descriptors[stream_index]; - const struct core_stream_support_info *stream_info = &display_config->mode_support_result.cfg_support_info.stream_support_info[stream_index]; - const struct dml2_timing_cfg *timing = &stream_descriptor->timing; - struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; - - /* worst case all other streams require some programming at the same time, 0 if only 1 stream */ - unsigned int contention_delay_us = (ip_caps->fams2.vertical_interrupt_ack_delay_us + - (unsigned int)math_max3(ip_caps->fams2.subvp_programming_delay_us, ip_caps->fams2.drr_programming_delay_us, ip_caps->fams2.allow_programming_delay_us)) * - (display_config->display_config.num_streams - 1); - - /* common */ - stream_pstate_meta->valid = true; - stream_pstate_meta->otg_vline_time_us = (double)timing->h_total / timing->pixel_clock_khz * 1000.0; - stream_pstate_meta->nom_vtotal = stream_descriptor->timing.vblank_nom + stream_descriptor->timing.v_active; - stream_pstate_meta->nom_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / - (stream_pstate_meta->nom_vtotal * timing->h_total); - stream_pstate_meta->nom_frame_time_us = - (double)stream_pstate_meta->nom_vtotal * stream_pstate_meta->otg_vline_time_us; - stream_pstate_meta->vblank_start = timing->v_blank_end + timing->v_active; - - if (stream_descriptor->timing.drr_config.enabled == true) { - if (stream_descriptor->timing.drr_config.min_refresh_uhz != 0.0) { - stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / - ((double)stream_descriptor->timing.drr_config.min_refresh_uhz * stream_descriptor->timing.h_total) * 1e9); - } else { - /* assume min of 48Hz */ - stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / - (48000000.0 * stream_descriptor->timing.h_total) * 1e9); - } - } else { - stream_pstate_meta->max_vtotal = stream_pstate_meta->nom_vtotal; - } - stream_pstate_meta->min_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / - (stream_pstate_meta->max_vtotal * timing->h_total); - stream_pstate_meta->max_frame_time_us = - (double)stream_pstate_meta->max_vtotal * stream_pstate_meta->otg_vline_time_us; - - stream_pstate_meta->scheduling_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.scheduling_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.vertical_interrupt_ack_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->contention_delay_otg_vlines = - (unsigned int)math_ceil(contention_delay_us / stream_pstate_meta->otg_vline_time_us); - /* worst case allow to target needs to account for all streams' allow events overlapping, and 1 line for error */ - stream_pstate_meta->allow_to_target_delay_otg_vlines = - (unsigned int)(math_ceil((ip_caps->fams2.vertical_interrupt_ack_delay_us + contention_delay_us + ip_caps->fams2.allow_programming_delay_us) / stream_pstate_meta->otg_vline_time_us)) + 1; - stream_pstate_meta->min_allow_width_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.min_allow_width_us / stream_pstate_meta->otg_vline_time_us); - /* this value should account for urgent latency */ - stream_pstate_meta->blackout_otg_vlines = - (unsigned int)math_ceil(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us / - stream_pstate_meta->otg_vline_time_us); - - /* scheduling params should be built based on the worst case for allow_time:disallow_time */ - - /* vactive */ - if (display_config->display_config.num_streams == 1) { - /* for single stream, guarantee at least an instant of allow */ - stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = (unsigned int)math_floor( - math_max2(0.0, - timing->v_active - math_max2(1.0, stream_pstate_meta->min_allow_width_otg_vlines) - stream_pstate_meta->blackout_otg_vlines)); - } else { - /* for multi stream, bound to a max fill time defined by IP caps */ - stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = - (unsigned int)math_floor((double)ip_caps->max_vactive_det_fill_delay_us / stream_pstate_meta->otg_vline_time_us); - } - stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us = stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines * stream_pstate_meta->otg_vline_time_us; - - if (stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us > 0.0) { - stream_pstate_meta->method_vactive.common.allow_start_otg_vline = - timing->v_blank_end + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; - stream_pstate_meta->method_vactive.common.allow_end_otg_vline = - stream_pstate_meta->vblank_start - - stream_pstate_meta->blackout_otg_vlines; - } else { - stream_pstate_meta->method_vactive.common.allow_start_otg_vline = 0; - stream_pstate_meta->method_vactive.common.allow_end_otg_vline = 0; - } - stream_pstate_meta->method_vactive.common.period_us = stream_pstate_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_pstate_meta->method_vactive.common, stream_pstate_meta); - - /* vblank */ - stream_pstate_meta->method_vblank.common.allow_start_otg_vline = stream_pstate_meta->vblank_start; - stream_pstate_meta->method_vblank.common.allow_end_otg_vline = - stream_pstate_meta->method_vblank.common.allow_start_otg_vline + 1; - stream_pstate_meta->method_vblank.common.period_us = stream_pstate_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_pstate_meta->method_vblank.common, stream_pstate_meta); - - /* subvp */ - stream_pstate_meta->method_subvp.programming_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_programming_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_df_throttle_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_prefetch_to_mall_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->method_subvp.phantom_vactive = - stream_pstate_meta->allow_to_target_delay_otg_vlines + - stream_pstate_meta->min_allow_width_otg_vlines + - stream_info->phantom_min_v_active; - stream_pstate_meta->method_subvp.phantom_vfp = - stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines; - /* phantom vtotal = v_bp(vstartup) + v_sync(1) + v_fp(throttle_delay) + v_active(allow_to_target + min_allow + min_vactive)*/ - stream_pstate_meta->method_subvp.phantom_vtotal = - stream_info->phantom_v_startup + - stream_pstate_meta->method_subvp.phantom_vfp + - 1 + - stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines + - stream_pstate_meta->method_subvp.phantom_vactive; - stream_pstate_meta->method_subvp.common.allow_start_otg_vline = - stream_descriptor->timing.v_blank_end + - stream_pstate_meta->contention_delay_otg_vlines + - stream_pstate_meta->method_subvp.programming_delay_otg_vlines + - stream_pstate_meta->method_subvp.phantom_vtotal + - stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + - stream_pstate_meta->allow_to_target_delay_otg_vlines; - stream_pstate_meta->method_subvp.common.allow_end_otg_vline = - stream_pstate_meta->vblank_start - - stream_pstate_meta->blackout_otg_vlines; - stream_pstate_meta->method_subvp.common.period_us = stream_pstate_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_pstate_meta->method_subvp.common, stream_pstate_meta); - - /* drr */ - stream_pstate_meta->method_drr.programming_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.drr_programming_delay_us / stream_pstate_meta->otg_vline_time_us); - stream_pstate_meta->method_drr.common.allow_start_otg_vline = - stream_pstate_meta->vblank_start + - stream_pstate_meta->allow_to_target_delay_otg_vlines; - stream_pstate_meta->method_drr.common.period_us = stream_pstate_meta->nom_frame_time_us; - if (display_config->display_config.num_streams <= 1) { - /* only need to stretch vblank for blackout time */ - stream_pstate_meta->method_drr.stretched_vtotal = - stream_pstate_meta->nom_vtotal + - stream_pstate_meta->allow_to_target_delay_otg_vlines + - stream_pstate_meta->min_allow_width_otg_vlines + - stream_pstate_meta->blackout_otg_vlines; - } else { - /* multi display needs to always be schedulable */ - stream_pstate_meta->method_drr.stretched_vtotal = - stream_pstate_meta->nom_vtotal * 2 + - stream_pstate_meta->allow_to_target_delay_otg_vlines + - stream_pstate_meta->min_allow_width_otg_vlines + - stream_pstate_meta->blackout_otg_vlines; - } - stream_pstate_meta->method_drr.common.allow_end_otg_vline = - stream_pstate_meta->method_drr.stretched_vtotal - - stream_pstate_meta->blackout_otg_vlines; - build_method_scheduling_params(&stream_pstate_meta->method_drr.common, stream_pstate_meta); -} - -static void build_subvp_meta_per_stream(struct dml2_pmo_instance *pmo, - struct display_configuation_with_meta *display_config, - int stream_index) -{ - struct dml2_implicit_svp_meta *stream_svp_meta = &pmo->scratch.pmo_dcn4.stream_svp_meta[stream_index]; - struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; - - stream_svp_meta->valid = true; - - /* PMO FAMS2 precaulcates these values */ - stream_svp_meta->v_active = stream_pstate_meta->method_subvp.phantom_vactive; - stream_svp_meta->v_front_porch = stream_pstate_meta->method_subvp.phantom_vfp; - stream_svp_meta->v_total = stream_pstate_meta->method_subvp.phantom_vtotal; -} - -bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out) -{ - struct dml2_pmo_instance *pmo = in_out->instance; - struct dml2_optimization_stage3_state *state = &in_out->base_display_config->stage3; - struct dml2_pmo_scratch *s = &pmo->scratch; - - struct display_configuation_with_meta *display_config; - const struct dml2_plane_parameters *plane_descriptor; - const struct dml2_pmo_pstate_strategy *strategy_list = NULL; - struct dml2_pmo_pstate_strategy override_base_strategy = { 0 }; - unsigned int strategy_list_size = 0; - unsigned int plane_index, stream_index, i; - bool build_override_strategy = true; - - state->performed = true; - in_out->base_display_config->stage3.min_clk_index_for_latency = in_out->base_display_config->stage1.min_clk_index_for_latency; - - display_config = in_out->base_display_config; - display_config->display_config.overrides.enable_subvp_implicit_pmo = true; - - memset(s, 0, sizeof(struct dml2_pmo_scratch)); - - if (display_config->display_config.overrides.all_streams_blanked) { - return true; - } - - pmo->scratch.pmo_dcn4.min_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; - pmo->scratch.pmo_dcn4.max_latency_index = pmo->mcg_clock_table_size; - pmo->scratch.pmo_dcn4.cur_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; - - // First build the stream plane mask (array of bitfields indexed by stream, indicating plane mapping) - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - plane_descriptor = &display_config->display_config.plane_descriptors[plane_index]; - - set_bit_in_bitfield(&s->pmo_dcn4.stream_plane_mask[plane_descriptor->stream_index], plane_index); - - state->pstate_switch_modes[plane_index] = dml2_pstate_method_vactive; - - build_override_strategy &= plane_descriptor->overrides.uclk_pstate_change_strategy != dml2_uclk_pstate_change_strategy_auto; - override_base_strategy.per_stream_pstate_method[plane_descriptor->stream_index] = - uclk_pstate_strategy_override_to_pstate_method(plane_descriptor->overrides.uclk_pstate_change_strategy); - } - - // Figure out which streams can do vactive, and also build up implicit SVP and FAMS2 meta - for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { - if (get_vactive_pstate_margin(display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) >= (int)(MIN_VACTIVE_MARGIN_PCT * pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us)) - set_bit_in_bitfield(&s->pmo_dcn4.stream_vactive_capability_mask, stream_index); - - /* FAMS2 meta */ - build_pstate_meta_per_stream(pmo, display_config, stream_index); - - /* SVP meta */ - build_subvp_meta_per_stream(pmo, display_config, stream_index); - } - - /* get synchronized timing groups */ - build_synchronized_timing_groups(pmo, display_config); - - if (build_override_strategy) { - /* build expanded override strategy list (no permutations) */ - override_base_strategy.allow_state_increase = true; - s->pmo_dcn4.num_expanded_override_strategies = 0; - insert_strategy_into_expanded_list(&override_base_strategy, - display_config->display_config.num_streams, - s->pmo_dcn4.expanded_override_strategy_list, - &s->pmo_dcn4.num_expanded_override_strategies); - expand_variant_strategy(&override_base_strategy, - display_config->display_config.num_streams, - false, - s->pmo_dcn4.expanded_override_strategy_list, - &s->pmo_dcn4.num_expanded_override_strategies); - - /* use override strategy list */ - strategy_list = s->pmo_dcn4.expanded_override_strategy_list; - strategy_list_size = s->pmo_dcn4.num_expanded_override_strategies; - } else { - /* use predefined strategy list */ - strategy_list = get_expanded_strategy_list(&pmo->init_data, display_config->display_config.num_streams); - strategy_list_size = get_num_expanded_strategies(&pmo->init_data, display_config->display_config.num_streams); - } - - if (!strategy_list || strategy_list_size == 0) - return false; - - s->pmo_dcn4.num_pstate_candidates = 0; - - for (i = 0; i < strategy_list_size && s->pmo_dcn4.num_pstate_candidates < DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE; i++) { - if (validate_pstate_support_strategy_cofunctionality(pmo, display_config, &strategy_list[i])) { - insert_into_candidate_list(&strategy_list[i], display_config->display_config.num_streams, s); - } - } - - if (s->pmo_dcn4.num_pstate_candidates > 0) { - s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.num_pstate_candidates-1].allow_state_increase = true; - s->pmo_dcn4.cur_pstate_candidate = -1; - return true; - } else { - return false; - } -} - -static void reset_display_configuration(struct display_configuation_with_meta *display_config) -{ - unsigned int plane_index; - unsigned int stream_index; - struct dml2_plane_parameters *plane; - - for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { - display_config->stage3.stream_svp_meta[stream_index].valid = false; - - display_config->display_config.stream_descriptors[stream_index].overrides.minimize_active_latency_hiding = false; - display_config->display_config.overrides.best_effort_min_active_latency_hiding_us = 0; - } - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - plane = &display_config->display_config.plane_descriptors[plane_index]; - - // Unset SubVP - plane->overrides.legacy_svp_config = dml2_svp_mode_override_auto; - - // Remove reserve time - plane->overrides.reserved_vblank_time_ns = 0; - - // Reset strategy to auto - plane->overrides.uclk_pstate_change_strategy = dml2_uclk_pstate_change_strategy_auto; - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_na; - } -} - -static void setup_planes_for_drr_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - unsigned int plane_index; - struct dml2_plane_parameters *plane; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - plane = &display_config->display_config.plane_descriptors[plane_index]; - - plane->overrides.uclk_pstate_change_strategy = dml2_uclk_pstate_change_strategy_force_drr; - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_drr; - - } - } -} - -static void setup_planes_for_svp_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - struct dml2_pmo_scratch *scratch = &pmo->scratch; - - unsigned int plane_index; - int stream_index = -1; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - stream_index = (char)display_config->display_config.plane_descriptors[plane_index].stream_index; - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_svp; - } - } - - if (stream_index >= 0) { - memcpy(&display_config->stage3.stream_svp_meta[stream_index], - &scratch->pmo_dcn4.stream_svp_meta[stream_index], - sizeof(struct dml2_implicit_svp_meta)); - } -} - -static void setup_planes_for_svp_drr_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - struct dml2_pmo_scratch *scratch = &pmo->scratch; - - unsigned int plane_index; - int stream_index = -1; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - stream_index = (char)display_config->display_config.plane_descriptors[plane_index].stream_index; - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_svp_drr; - } - } - - if (stream_index >= 0) { - memcpy(&display_config->stage3.stream_svp_meta[stream_index], - &scratch->pmo_dcn4.stream_svp_meta[stream_index], - sizeof(struct dml2_implicit_svp_meta)); - } -} - -static void setup_planes_for_vblank_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - unsigned int plane_index; - struct dml2_plane_parameters *plane; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - plane = &display_config->display_config.plane_descriptors[plane_index]; - - plane->overrides.reserved_vblank_time_ns = (long)math_max2(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us * 1000.0, - plane->overrides.reserved_vblank_time_ns); - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vblank; - - } - } -} - -static void setup_planes_for_vblank_drr_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - unsigned int plane_index; - struct dml2_plane_parameters *plane; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - plane = &display_config->display_config.plane_descriptors[plane_index]; - plane->overrides.reserved_vblank_time_ns = (long)(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us * 1000); - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vblank_drr; - } - } -} - -static void setup_planes_for_vactive_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - unsigned int plane_index; - unsigned int stream_index; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - stream_index = display_config->display_config.plane_descriptors[plane_index].stream_index; - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vactive; - - if (!pmo->options->disable_vactive_det_fill_bw_pad) { - display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = - (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); - } - } - } -} - -static void setup_planes_for_vactive_drr_by_mask(struct display_configuation_with_meta *display_config, - struct dml2_pmo_instance *pmo, - int plane_mask) -{ - unsigned int plane_index; - unsigned int stream_index; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - stream_index = display_config->display_config.plane_descriptors[plane_index].stream_index; - - display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vactive_drr; - - if (!pmo->options->disable_vactive_det_fill_bw_pad) { - display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = - (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); - } - } - } -} - -static bool setup_display_config(struct display_configuation_with_meta *display_config, struct dml2_pmo_instance *pmo, int strategy_index) -{ - struct dml2_pmo_scratch *scratch = &pmo->scratch; - - bool fams2_required = false; - bool success = true; - unsigned int stream_index; - - reset_display_configuration(display_config); - - for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { - - if (pmo->scratch.pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_na) { - success = false; - break; - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive) { - setup_planes_for_vactive_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank) { - setup_planes_for_vblank_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp) { - fams2_required = true; - setup_planes_for_svp_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { - fams2_required = true; - setup_planes_for_vactive_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { - fams2_required = true; - setup_planes_for_vblank_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { - fams2_required = true; - setup_planes_for_svp_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { - fams2_required = true; - setup_planes_for_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); - } - } - - /* copy FAMS2 meta */ - if (success) { - display_config->stage3.fams2_required = fams2_required; - memcpy(&display_config->stage3.stream_pstate_meta, - &scratch->pmo_dcn4.stream_pstate_meta, - sizeof(struct dml2_pstate_meta) * DML2_MAX_PLANES); - } - - return success; -} - -static int get_minimum_reserved_time_us_for_planes(struct display_configuation_with_meta *display_config, int plane_mask) -{ - int min_time_us = 0xFFFFFF; - unsigned int plane_index = 0; - - for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { - if (is_bit_set_in_bitfield(plane_mask, plane_index)) { - if (min_time_us > (display_config->display_config.plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000)) - min_time_us = display_config->display_config.plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000; - } - } - return min_time_us; -} - -bool pmo_dcn4_fams2_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out) -{ - bool p_state_supported = true; - unsigned int stream_index; - struct dml2_pmo_scratch *s = &in_out->instance->scratch; - - int MIN_VACTIVE_MARGIN_VBLANK = 0; - int MIN_VACTIVE_MARGIN_DRR = 0; - int REQUIRED_RESERVED_TIME = 0; - - if (in_out->base_display_config->display_config.overrides.all_streams_blanked) { - return true; - } - - MIN_VACTIVE_MARGIN_VBLANK = INT_MIN; - MIN_VACTIVE_MARGIN_DRR = INT_MIN; - REQUIRED_RESERVED_TIME = (int)in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us; - - if (s->pmo_dcn4.cur_pstate_candidate < 0) - return false; - - for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) { - struct dml2_pstate_meta *stream_pstate_meta = &s->pmo_dcn4.stream_pstate_meta[stream_index]; - - if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive || - s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { - if (get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < (MIN_VACTIVE_MARGIN_PCT * in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us) || - get_vactive_det_fill_latency_delay_us(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) > stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us) { - p_state_supported = false; - break; - } - } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank || - s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { - if (get_minimum_reserved_time_us_for_planes(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < - REQUIRED_RESERVED_TIME || - get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < MIN_VACTIVE_MARGIN_VBLANK) { - p_state_supported = false; - break; - } - } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp || - s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { - if (in_out->base_display_config->stage3.stream_svp_meta[stream_index].valid == false) { - p_state_supported = false; - break; - } - } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { - if (!all_planes_match_method(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index], dml2_pstate_method_fw_drr) || - get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < MIN_VACTIVE_MARGIN_DRR) { - p_state_supported = false; - break; - } - } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_na) { - p_state_supported = false; - break; - } - } - - return p_state_supported; -} - -bool pmo_dcn4_fams2_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out) -{ - bool success = false; - struct dml2_pmo_scratch *s = &in_out->instance->scratch; - - memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); - - if (in_out->last_candidate_failed) { - if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].allow_state_increase && - s->pmo_dcn4.cur_latency_index < s->pmo_dcn4.max_latency_index - 1) { - s->pmo_dcn4.cur_latency_index++; - - success = true; - } - } - - if (!success) { - s->pmo_dcn4.cur_latency_index = s->pmo_dcn4.min_latency_index; - s->pmo_dcn4.cur_pstate_candidate++; - - if (s->pmo_dcn4.cur_pstate_candidate < s->pmo_dcn4.num_pstate_candidates) { - success = true; - } - } - - if (success) { - in_out->optimized_display_config->stage3.min_clk_index_for_latency = s->pmo_dcn4.cur_latency_index; - setup_display_config(in_out->optimized_display_config, in_out->instance, in_out->instance->scratch.pmo_dcn4.cur_pstate_candidate); - } - - return success; -} - -bool pmo_dcn4_fams2_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out) -{ - bool success = true; - struct dml2_pmo_instance *pmo = in_out->instance; - bool stutter_period_meets_z8_eco = true; - bool z8_stutter_optimization_too_expensive = false; - bool stutter_optimization_too_expensive = false; - double line_time_us, vblank_nom_time_us; - - unsigned int i; - - if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && - pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && - pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us < pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us) - return false; // Unexpected SoCBB setup - - for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { - if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[i].active_latency_hiding_us < - pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us + pmo->soc_bb->power_management_parameters.z8_min_idle_time) { - stutter_period_meets_z8_eco = false; - break; - } - } - - for (i = 0; i < in_out->base_display_config->display_config.num_streams; i++) { - line_time_us = (double)in_out->base_display_config->display_config.stream_descriptors[i].timing.h_total / (in_out->base_display_config->display_config.stream_descriptors[i].timing.pixel_clock_khz * 1000) * 1000000; - vblank_nom_time_us = line_time_us * in_out->base_display_config->display_config.stream_descriptors[i].timing.vblank_nom; - - if (vblank_nom_time_us < pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us * MIN_BLANK_STUTTER_FACTOR) { - z8_stutter_optimization_too_expensive = true; - break; - } - - if (vblank_nom_time_us < pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us * MIN_BLANK_STUTTER_FACTOR) { - stutter_optimization_too_expensive = true; - break; - } - } - - pmo->scratch.pmo_dcn4.num_stutter_candidates = 0; - pmo->scratch.pmo_dcn4.cur_stutter_candidate = 0; - - if (stutter_period_meets_z8_eco && !z8_stutter_optimization_too_expensive) { - if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0) { - pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.num_stutter_candidates] = (unsigned int)pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us; - pmo->scratch.pmo_dcn4.num_stutter_candidates++; - pmo->scratch.pmo_dcn4.z8_vblank_optimizable = true; - } - } else { - pmo->scratch.pmo_dcn4.z8_vblank_optimizable = false; - } - - if (!stutter_optimization_too_expensive && pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0) { - pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.num_stutter_candidates] = (unsigned int)pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us; - pmo->scratch.pmo_dcn4.num_stutter_candidates++; - } - - if (pmo->scratch.pmo_dcn4.num_stutter_candidates == 0) - success = false; - - return success; -} - -bool pmo_dcn4_fams2_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out) -{ - bool success = true; - struct dml2_pmo_instance *pmo = in_out->instance; - - unsigned int i; - - for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { - if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && - pmo->scratch.pmo_dcn4.z8_vblank_optimizable && - in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < (int)pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us * 1000) { - success = false; - break; - } - if (pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && - in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < (int)pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us * 1000) { - success = false; - break; - } - } - - return success; -} - -bool pmo_dcn4_fams2_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out) -{ - bool success = false; - struct dml2_pmo_instance *pmo = in_out->instance; - unsigned int i; - - memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); - - if (!in_out->last_candidate_failed) { - if (pmo->scratch.pmo_dcn4.cur_stutter_candidate < pmo->scratch.pmo_dcn4.num_stutter_candidates) { - for (i = 0; i < in_out->optimized_display_config->display_config.num_planes; i++) { - /* take the max of the current and the optimal reserved time */ - in_out->optimized_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns = - (long)math_max2(pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.cur_stutter_candidate] * 1000, - in_out->optimized_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns); - } - - success = true; - } - } - - return success; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h deleted file mode 100644 index 6baab7ad6ecc..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h +++ /dev/null @@ -1,33 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_PMO_FAMS2_DCN4_H__ -#define __DML2_PMO_FAMS2_DCN4_H__ - -#include "dml2_internal_shared_types.h" - -bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out); - -bool pmo_dcn4_fams2_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); - -bool pmo_dcn4_fams2_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out); -bool pmo_dcn4_fams2_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out); -bool pmo_dcn4_fams2_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out); - -bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out); -bool pmo_dcn4_fams2_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out); -bool pmo_dcn4_fams2_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); - -bool pmo_dcn4_fams2_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out); -bool pmo_dcn4_fams2_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out); -bool pmo_dcn4_fams2_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out); - -void pmo_dcn4_fams2_expand_base_pstate_strategies( - const struct dml2_pmo_pstate_strategy *base_strategies_list, - const unsigned int num_base_strategies, - const unsigned int stream_count, - struct dml2_pmo_pstate_strategy *expanded_strategy_list, - unsigned int *num_expanded_strategies); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c deleted file mode 100644 index 55d2464365d0..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c +++ /dev/null @@ -1,83 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_pmo_factory.h" -#include "dml2_pmo_dcn4_fams2.h" -#include "dml2_pmo_dcn3.h" -#include "dml2_external_lib_deps.h" - -static bool dummy_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out) -{ - return false; -} - -static bool dummy_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out) -{ - return true; -} - -static bool dummy_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out) -{ - return false; -} - -bool dml2_pmo_create(enum dml2_project_id project_id, struct dml2_pmo_instance *out) -{ - bool result = false; - - if (!out) - return false; - - memset(out, 0, sizeof(struct dml2_pmo_instance)); - - switch (project_id) { - case dml2_project_dcn4x_stage1: - out->initialize = pmo_dcn4_fams2_initialize; - out->optimize_dcc_mcache = pmo_dcn4_fams2_optimize_dcc_mcache; - result = true; - break; - case dml2_project_dcn4x_stage2: - out->initialize = pmo_dcn3_initialize; - - out->optimize_dcc_mcache = pmo_dcn3_optimize_dcc_mcache; - - out->init_for_vmin = pmo_dcn3_init_for_vmin; - out->test_for_vmin = pmo_dcn3_test_for_vmin; - out->optimize_for_vmin = pmo_dcn3_optimize_for_vmin; - - out->init_for_uclk_pstate = pmo_dcn3_init_for_pstate_support; - out->test_for_uclk_pstate = pmo_dcn3_test_for_pstate_support; - out->optimize_for_uclk_pstate = pmo_dcn3_optimize_for_pstate_support; - - out->init_for_stutter = dummy_init_for_stutter; - out->test_for_stutter = dummy_test_for_stutter; - out->optimize_for_stutter = dummy_optimize_for_stutter; - - result = true; - break; - case dml2_project_dcn4x_stage2_auto_drr_svp: - out->initialize = pmo_dcn4_fams2_initialize; - - out->optimize_dcc_mcache = pmo_dcn4_fams2_optimize_dcc_mcache; - - out->init_for_vmin = pmo_dcn4_fams2_init_for_vmin; - out->test_for_vmin = pmo_dcn4_fams2_test_for_vmin; - out->optimize_for_vmin = pmo_dcn4_fams2_optimize_for_vmin; - - out->init_for_uclk_pstate = pmo_dcn4_fams2_init_for_pstate_support; - out->test_for_uclk_pstate = pmo_dcn4_fams2_test_for_pstate_support; - out->optimize_for_uclk_pstate = pmo_dcn4_fams2_optimize_for_pstate_support; - - out->init_for_stutter = pmo_dcn4_fams2_init_for_stutter; - out->test_for_stutter = pmo_dcn4_fams2_test_for_stutter; - out->optimize_for_stutter = pmo_dcn4_fams2_optimize_for_stutter; - result = true; - break; - case dml2_project_invalid: - default: - break; - } - - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h deleted file mode 100644 index 7218de1824cc..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h +++ /dev/null @@ -1,13 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_PMO_FACTORY_H__ -#define __DML2_PMO_FACTORY_H__ - -#include "dml2_internal_shared_types.h" -#include "dml_top_types.h" - -bool dml2_pmo_create(enum dml2_project_id project_id, struct dml2_pmo_instance *out); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c deleted file mode 100644 index e17b5ceba447..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c +++ /dev/null @@ -1,147 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "lib_float_math.h" - -#define ASSERT(condition) - -#define isNaN(number) ((number) != (number)) - - /* - * NOTE: - * This file is gcc-parseable HW gospel, coming straight from HW engineers. - * - * It doesn't adhere to Linux kernel style and sometimes will do things in odd - * ways. Unless there is something clearly wrong with it the code should - * remain as-is as it provides us with a guarantee from HW that it is correct. - */ - -double math_mod(const double arg1, const double arg2) -{ - if (isNaN(arg1)) - return arg2; - if (isNaN(arg2)) - return arg1; - return arg1 - arg1 * ((int)(arg1 / arg2)); -} - -double math_min2(const double arg1, const double arg2) -{ - if (isNaN(arg1)) - return arg2; - if (isNaN(arg2)) - return arg1; - return arg1 < arg2 ? arg1 : arg2; -} - -double math_max2(const double arg1, const double arg2) -{ - if (isNaN(arg1)) - return arg2; - if (isNaN(arg2)) - return arg1; - return arg1 > arg2 ? arg1 : arg2; -} - -double math_floor2(const double arg, const double significance) -{ - ASSERT(significance != 0); - - return ((int)(arg / significance)) * significance; -} - -double math_floor(const double arg) -{ - return ((int)(arg)); -} - -double math_ceil(const double arg) -{ - return (int)(arg + 0.99999); -} - -double math_ceil2(const double arg, const double significance) -{ - return ((int)(arg / significance + 0.99999)) * significance; -} - -double math_max3(double v1, double v2, double v3) -{ - return v3 > math_max2(v1, v2) ? v3 : math_max2(v1, v2); -} - -double math_max4(double v1, double v2, double v3, double v4) -{ - return v4 > math_max3(v1, v2, v3) ? v4 : math_max3(v1, v2, v3); -} - -double math_max5(double v1, double v2, double v3, double v4, double v5) -{ - return math_max3(v1, v2, v3) > math_max2(v4, v5) ? math_max3(v1, v2, v3) : math_max2(v4, v5); -} - -float math_pow(float a, float exp) -{ - double temp; - if ((int)exp == 0) - return 1; - temp = math_pow(a, (float)((int)(exp / 2))); - if (((int)exp % 2) == 0) { - return (float)(temp * temp); - } else { - if ((int)exp > 0) - return (float)(a * temp * temp); - else - return (float)((temp * temp) / a); - } -} - -double math_fabs(double a) -{ - if (a > 0) - return (a); - else - return (-a); -} - -float math_log(float a, float b) -{ - int *const exp_ptr = (int *)(&a); - int x = *exp_ptr; - const int log_2 = ((x >> 23) & 255) - 128; - x &= ~(255 << 23); - x += 127 << 23; - *exp_ptr = x; - - a = ((-1.0f / 3) * a + 2) * a - 2.0f / 3; - - if (b > 2.00001 || b < 1.99999) - return (a + log_2) / math_log(b, 2); - else - return (a + log_2); -} - -float math_log2(float a) -{ - return math_log(a, 2.0); -} - -// approximate log2 value of a input -// - precise if the input pwr of 2, else the approximation will be an integer = floor(actual_log2) -unsigned int math_log2_approx(unsigned int a) -{ - unsigned int log2_val = 0; - while (a > 1) { - a = a >> 1; - log2_val++; - } - return log2_val; -} - -double math_round(double a) -{ - const double round_pt = 0.5; - - return math_floor(a + round_pt); -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h deleted file mode 100644 index e13b0c5939b0..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h +++ /dev/null @@ -1,25 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __LIB_FLOAT_MATH_H__ -#define __LIB_FLOAT_MATH_H__ - -double math_mod(const double arg1, const double arg2); -double math_min2(const double arg1, const double arg2); -double math_max2(const double arg1, const double arg2); -double math_floor2(const double arg, const double significance); -double math_floor(const double arg); -double math_ceil(const double arg); -double math_ceil2(const double arg, const double significance); -double math_max3(double v1, double v2, double v3); -double math_max4(double v1, double v2, double v3, double v4); -double math_max5(double v1, double v2, double v3, double v4, double v5); -float math_pow(float a, float exp); -double math_fabs(double a); -float math_log(float a, float b); -float math_log2(float a); -unsigned int math_log2_approx(unsigned int a); -double math_round(double a); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c deleted file mode 100644 index 5a33e2f357f4..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c +++ /dev/null @@ -1,49 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml_top.h" -#include "dml2_internal_shared_types.h" -#include "dml2_top_soc15.h" - -unsigned int dml2_get_instance_size_bytes(void) -{ - return sizeof(struct dml2_instance); -} - -bool dml2_initialize_instance(struct dml2_initialize_instance_in_out *in_out) -{ - switch (in_out->options.project_id) { - case dml2_project_dcn4x_stage1: - case dml2_project_dcn4x_stage2: - case dml2_project_dcn4x_stage2_auto_drr_svp: - return dml2_top_soc15_initialize_instance(in_out); - case dml2_project_invalid: - default: - return false; - } -} - -bool dml2_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out) -{ - if (!in_out->dml2_instance->funcs.check_mode_supported) - return false; - - return in_out->dml2_instance->funcs.check_mode_supported(in_out); -} - -bool dml2_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out) -{ - if (!in_out->dml2_instance->funcs.build_mode_programming) - return false; - - return in_out->dml2_instance->funcs.build_mode_programming(in_out); -} - -bool dml2_build_mcache_programming(struct dml2_build_mcache_programming_in_out *in_out) -{ - if (!in_out->dml2_instance->funcs.build_mcache_programming) - return false; - - return in_out->dml2_instance->funcs.build_mcache_programming(in_out); -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c deleted file mode 100644 index 5e14d85821e2..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c +++ /dev/null @@ -1,10 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_top_legacy.h" -#include "dml2_top_soc15.h" -#include "dml2_core_factory.h" -#include "dml2_pmo_factory.h" -#include "display_mode_core_structs.h" - diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h deleted file mode 100644 index 14d0ae03dce6..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h +++ /dev/null @@ -1,9 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_TOP_LEGACY_H__ -#define __DML2_TOP_LEGACY_H__ -#include "dml2_internal_shared_types.h" -bool dml2_top_legacy_initialize_instance(struct dml2_initialize_instance_in_out *in_out); -#endif /* __DML2_TOP_LEGACY_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c deleted file mode 100644 index 4a7c4c62111e..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c +++ /dev/null @@ -1,1170 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#include "dml2_top_soc15.h" -#include "dml2_mcg_factory.h" -#include "dml2_dpmm_factory.h" -#include "dml2_core_factory.h" -#include "dml2_pmo_factory.h" -#include "lib_float_math.h" -#include "dml2_debug.h" -static void setup_unoptimized_display_config_with_meta(const struct dml2_instance *dml, struct display_configuation_with_meta *out, const struct dml2_display_cfg *display_config) -{ - memcpy(&out->display_config, display_config, sizeof(struct dml2_display_cfg)); - out->stage1.min_clk_index_for_latency = dml->min_clk_table.dram_bw_table.num_entries - 1; //dml->min_clk_table.clean_me_up.soc_bb.num_states - 1; -} - -static void setup_speculative_display_config_with_meta(const struct dml2_instance *dml, struct display_configuation_with_meta *out, const struct dml2_display_cfg *display_config) -{ - memcpy(&out->display_config, display_config, sizeof(struct dml2_display_cfg)); - out->stage1.min_clk_index_for_latency = 0; -} - -static void copy_display_configuration_with_meta(struct display_configuation_with_meta *dst, const struct display_configuation_with_meta *src) -{ - memcpy(dst, src, sizeof(struct display_configuation_with_meta)); -} - -static bool dml2_top_optimization_init_function_min_clk_for_latency(const struct optimization_init_function_params *params) -{ - struct dml2_optimization_stage1_state *state = ¶ms->display_config->stage1; - - state->performed = true; - - return true; -} - -static bool dml2_top_optimization_test_function_min_clk_for_latency(const struct optimization_test_function_params *params) -{ - struct dml2_optimization_stage1_state *state = ¶ms->display_config->stage1; - - return state->min_clk_index_for_latency == 0; -} - -static bool dml2_top_optimization_optimize_function_min_clk_for_latency(const struct optimization_optimize_function_params *params) -{ - bool result = false; - - if (params->display_config->stage1.min_clk_index_for_latency > 0) { - copy_display_configuration_with_meta(params->optimized_display_config, params->display_config); - params->optimized_display_config->stage1.min_clk_index_for_latency--; - result = true; - } - - return result; -} - -static bool dml2_top_optimization_test_function_mcache(const struct optimization_test_function_params *params) -{ - struct dml2_optimization_test_function_locals *l = params->locals; - bool mcache_success = false; - bool result = false; - - memset(l, 0, sizeof(struct dml2_optimization_test_function_locals)); - - l->test_mcache.calc_mcache_count_params.dml2_instance = params->dml; - l->test_mcache.calc_mcache_count_params.display_config = ¶ms->display_config->display_config; - l->test_mcache.calc_mcache_count_params.mcache_allocations = params->display_config->stage2.mcache_allocations; - - result = dml2_top_mcache_calc_mcache_count_and_offsets(&l->test_mcache.calc_mcache_count_params); // use core to get the basic mcache_allocations - - if (result) { - l->test_mcache.assign_global_mcache_ids_params.allocations = params->display_config->stage2.mcache_allocations; - l->test_mcache.assign_global_mcache_ids_params.num_allocations = params->display_config->display_config.num_planes; - - dml2_top_mcache_assign_global_mcache_ids(&l->test_mcache.assign_global_mcache_ids_params); - - l->test_mcache.validate_admissibility_params.dml2_instance = params->dml; - l->test_mcache.validate_admissibility_params.display_cfg = ¶ms->display_config->display_config; - l->test_mcache.validate_admissibility_params.mcache_allocations = params->display_config->stage2.mcache_allocations; - l->test_mcache.validate_admissibility_params.cfg_support_info = ¶ms->display_config->mode_support_result.cfg_support_info; - - mcache_success = dml2_top_mcache_validate_admissability(&l->test_mcache.validate_admissibility_params); // also find the shift to make mcache allocation works - - memcpy(params->display_config->stage2.per_plane_mcache_support, l->test_mcache.validate_admissibility_params.per_plane_status, sizeof(bool) * DML2_MAX_PLANES); - } - - return mcache_success; -} - -static bool dml2_top_optimization_optimize_function_mcache(const struct optimization_optimize_function_params *params) -{ - struct dml2_optimization_optimize_function_locals *l = params->locals; - bool optimize_success = false; - - if (params->last_candidate_supported == false) - return false; - - copy_display_configuration_with_meta(params->optimized_display_config, params->display_config); - - l->optimize_mcache.optimize_mcache_params.instance = ¶ms->dml->pmo_instance; - l->optimize_mcache.optimize_mcache_params.dcc_mcache_supported = params->display_config->stage2.per_plane_mcache_support; - l->optimize_mcache.optimize_mcache_params.display_config = ¶ms->display_config->display_config; - l->optimize_mcache.optimize_mcache_params.optimized_display_cfg = ¶ms->optimized_display_config->display_config; - l->optimize_mcache.optimize_mcache_params.cfg_support_info = ¶ms->optimized_display_config->mode_support_result.cfg_support_info; - - optimize_success = params->dml->pmo_instance.optimize_dcc_mcache(&l->optimize_mcache.optimize_mcache_params); - - return optimize_success; -} - -static bool dml2_top_optimization_init_function_vmin(const struct optimization_init_function_params *params) -{ - struct dml2_optimization_init_function_locals *l = params->locals; - - l->vmin.init_params.instance = ¶ms->dml->pmo_instance; - l->vmin.init_params.base_display_config = params->display_config; - return params->dml->pmo_instance.init_for_vmin(&l->vmin.init_params); -} - -static bool dml2_top_optimization_test_function_vmin(const struct optimization_test_function_params *params) -{ - struct dml2_optimization_test_function_locals *l = params->locals; - - l->test_vmin.pmo_test_vmin_params.instance = ¶ms->dml->pmo_instance; - l->test_vmin.pmo_test_vmin_params.display_config = params->display_config; - l->test_vmin.pmo_test_vmin_params.vmin_limits = ¶ms->dml->soc_bbox.vmin_limit; - return params->dml->pmo_instance.test_for_vmin(&l->test_vmin.pmo_test_vmin_params); -} - -static bool dml2_top_optimization_optimize_function_vmin(const struct optimization_optimize_function_params *params) -{ - struct dml2_optimization_optimize_function_locals *l = params->locals; - - if (params->last_candidate_supported == false) - return false; - - l->optimize_vmin.pmo_optimize_vmin_params.instance = ¶ms->dml->pmo_instance; - l->optimize_vmin.pmo_optimize_vmin_params.base_display_config = params->display_config; - l->optimize_vmin.pmo_optimize_vmin_params.optimized_display_config = params->optimized_display_config; - return params->dml->pmo_instance.optimize_for_vmin(&l->optimize_vmin.pmo_optimize_vmin_params); -} - -static bool dml2_top_optimization_init_function_uclk_pstate(const struct optimization_init_function_params *params) -{ - struct dml2_optimization_init_function_locals *l = params->locals; - - l->uclk_pstate.init_params.instance = ¶ms->dml->pmo_instance; - l->uclk_pstate.init_params.base_display_config = params->display_config; - - return params->dml->pmo_instance.init_for_uclk_pstate(&l->uclk_pstate.init_params); -} - -static bool dml2_top_optimization_test_function_uclk_pstate(const struct optimization_test_function_params *params) -{ - struct dml2_optimization_test_function_locals *l = params->locals; - - l->uclk_pstate.test_params.instance = ¶ms->dml->pmo_instance; - l->uclk_pstate.test_params.base_display_config = params->display_config; - - return params->dml->pmo_instance.test_for_uclk_pstate(&l->uclk_pstate.test_params); -} - -static bool dml2_top_optimization_optimize_function_uclk_pstate(const struct optimization_optimize_function_params *params) -{ - struct dml2_optimization_optimize_function_locals *l = params->locals; - - l->uclk_pstate.optimize_params.instance = ¶ms->dml->pmo_instance; - l->uclk_pstate.optimize_params.base_display_config = params->display_config; - l->uclk_pstate.optimize_params.optimized_display_config = params->optimized_display_config; - l->uclk_pstate.optimize_params.last_candidate_failed = !params->last_candidate_supported; - - return params->dml->pmo_instance.optimize_for_uclk_pstate(&l->uclk_pstate.optimize_params); -} - -static bool dml2_top_optimization_init_function_stutter(const struct optimization_init_function_params *params) -{ - struct dml2_optimization_init_function_locals *l = params->locals; - - l->uclk_pstate.init_params.instance = ¶ms->dml->pmo_instance; - l->uclk_pstate.init_params.base_display_config = params->display_config; - - return params->dml->pmo_instance.init_for_stutter(&l->stutter.stutter_params); -} - -static bool dml2_top_optimization_test_function_stutter(const struct optimization_test_function_params *params) -{ - struct dml2_optimization_test_function_locals *l = params->locals; - - l->stutter.stutter_params.instance = ¶ms->dml->pmo_instance; - l->stutter.stutter_params.base_display_config = params->display_config; - return params->dml->pmo_instance.test_for_stutter(&l->stutter.stutter_params); -} - -static bool dml2_top_optimization_optimize_function_stutter(const struct optimization_optimize_function_params *params) -{ - struct dml2_optimization_optimize_function_locals *l = params->locals; - - l->stutter.stutter_params.instance = ¶ms->dml->pmo_instance; - l->stutter.stutter_params.base_display_config = params->display_config; - l->stutter.stutter_params.optimized_display_config = params->optimized_display_config; - l->stutter.stutter_params.last_candidate_failed = !params->last_candidate_supported; - return params->dml->pmo_instance.optimize_for_stutter(&l->stutter.stutter_params); -} - -static bool dml2_top_optimization_perform_optimization_phase(struct dml2_optimization_phase_locals *l, const struct optimization_phase_params *params) -{ - bool test_passed = false; - bool optimize_succeeded = true; - bool candidate_validation_passed = true; - struct optimization_init_function_params init_params = { 0 }; - struct optimization_test_function_params test_params = { 0 }; - struct optimization_optimize_function_params optimize_params = { 0 }; - - if (!params->dml || - !params->optimize_function || - !params->test_function || - !params->display_config || - !params->optimized_display_config) - return false; - - copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, params->display_config); - - init_params.locals = &l->init_function_locals; - init_params.dml = params->dml; - init_params.display_config = &l->cur_candidate_display_cfg; - - if (params->init_function && !params->init_function(&init_params)) - return false; - - test_params.locals = &l->test_function_locals; - test_params.dml = params->dml; - test_params.display_config = &l->cur_candidate_display_cfg; - - test_passed = params->test_function(&test_params); - - while (!test_passed && optimize_succeeded) { - memset(&optimize_params, 0, sizeof(struct optimization_optimize_function_params)); - - optimize_params.locals = &l->optimize_function_locals; - optimize_params.dml = params->dml; - optimize_params.display_config = &l->cur_candidate_display_cfg; - optimize_params.optimized_display_config = &l->next_candidate_display_cfg; - optimize_params.last_candidate_supported = candidate_validation_passed; - - optimize_succeeded = params->optimize_function(&optimize_params); - - if (optimize_succeeded) { - l->mode_support_params.instance = ¶ms->dml->core_instance; - l->mode_support_params.display_cfg = &l->next_candidate_display_cfg; - l->mode_support_params.min_clk_table = ¶ms->dml->min_clk_table; - - if (l->next_candidate_display_cfg.stage3.performed) - l->mode_support_params.min_clk_index = l->next_candidate_display_cfg.stage3.min_clk_index_for_latency; - else - l->mode_support_params.min_clk_index = l->next_candidate_display_cfg.stage1.min_clk_index_for_latency; - candidate_validation_passed = params->dml->core_instance.mode_support(&l->mode_support_params); - l->next_candidate_display_cfg.mode_support_result = l->mode_support_params.mode_support_result; - } - - if (optimize_succeeded && candidate_validation_passed) { - memset(&test_params, 0, sizeof(struct optimization_test_function_params)); - test_params.locals = &l->test_function_locals; - test_params.dml = params->dml; - test_params.display_config = &l->next_candidate_display_cfg; - test_passed = params->test_function(&test_params); - - copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, &l->next_candidate_display_cfg); - - // If optimization is not all or nothing, then store partial progress in output - if (!params->all_or_nothing) - copy_display_configuration_with_meta(params->optimized_display_config, &l->next_candidate_display_cfg); - } - } - - if (test_passed) - copy_display_configuration_with_meta(params->optimized_display_config, &l->cur_candidate_display_cfg); - - return test_passed; -} - -static bool dml2_top_optimization_perform_optimization_phase_1(struct dml2_optimization_phase_locals *l, const struct optimization_phase_params *params) -{ - int highest_state, lowest_state, cur_state; - bool supported = false; - - if (!params->dml || - !params->optimize_function || - !params->test_function || - !params->display_config || - !params->optimized_display_config) - return false; - - copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, params->display_config); - highest_state = l->cur_candidate_display_cfg.stage1.min_clk_index_for_latency; - lowest_state = 0; - - while (highest_state > lowest_state) { - cur_state = (highest_state + lowest_state) / 2; - - l->mode_support_params.instance = ¶ms->dml->core_instance; - l->mode_support_params.display_cfg = &l->cur_candidate_display_cfg; - l->mode_support_params.min_clk_table = ¶ms->dml->min_clk_table; - l->mode_support_params.min_clk_index = cur_state; - supported = params->dml->core_instance.mode_support(&l->mode_support_params); - - if (supported) { - l->cur_candidate_display_cfg.mode_support_result = l->mode_support_params.mode_support_result; - highest_state = cur_state; - } else { - lowest_state = cur_state + 1; - } - } - l->cur_candidate_display_cfg.stage1.min_clk_index_for_latency = lowest_state; - - copy_display_configuration_with_meta(params->optimized_display_config, &l->cur_candidate_display_cfg); - - return true; -} - -/* -* Takes an input set of mcache boundaries and finds the appropriate setting of cache programming. -* Returns true if a valid set of programming can be made, and false otherwise. "Valid" means -* that the horizontal viewport does not span more than 2 cache slices. -* -* It optionally also can apply a constant shift to all the cache boundaries. -*/ -static const uint32_t MCACHE_ID_UNASSIGNED = 0xF; -static const uint32_t SPLIT_LOCATION_UNDEFINED = 0xFFFF; - -static bool calculate_first_second_splitting(const int *mcache_boundaries, int num_boundaries, int shift, - int pipe_h_vp_start, int pipe_h_vp_end, int *first_offset, int *second_offset) -{ - const int MAX_VP = 0xFFFFFF; - int left_cache_id; - int right_cache_id; - int range_start; - int range_end; - bool success = false; - - if (num_boundaries <= 1) { - if (first_offset && second_offset) { - *first_offset = 0; - *second_offset = -1; - } - success = true; - return success; - } else { - range_start = 0; - for (left_cache_id = 0; left_cache_id < num_boundaries; left_cache_id++) { - range_end = mcache_boundaries[left_cache_id] - shift - 1; - - if (range_start <= pipe_h_vp_start && pipe_h_vp_start <= range_end) - break; - - range_start = range_end + 1; - } - - range_end = MAX_VP; - for (right_cache_id = num_boundaries - 1; right_cache_id >= -1; right_cache_id--) { - if (right_cache_id >= 0) - range_start = mcache_boundaries[right_cache_id] - shift; - else - range_start = 0; - - if (range_start <= pipe_h_vp_end && pipe_h_vp_end <= range_end) { - break; - } - range_end = range_start - 1; - } - right_cache_id = (right_cache_id + 1) % num_boundaries; - - if (right_cache_id == left_cache_id) { - if (first_offset && second_offset) { - *first_offset = left_cache_id; - *second_offset = -1; - } - success = true; - } else if (right_cache_id == (left_cache_id + 1) % num_boundaries) { - if (first_offset && second_offset) { - *first_offset = left_cache_id; - *second_offset = right_cache_id; - } - success = true; - } - } - - return success; -} - -/* -* For a given set of pipe start/end x positions, checks to see it can support the input mcache splitting. -* It also attempts to "optimize" by finding a shift if the default 0 shift does not work. -*/ -static bool find_shift_for_valid_cache_id_assignment(int *mcache_boundaries, unsigned int num_boundaries, - int *pipe_vp_startx, int *pipe_vp_endx, unsigned int pipe_count, int shift_granularity, int *shift) -{ - int max_shift = 0xFFFF; - unsigned int pipe_index; - unsigned int i, slice_width; - bool success = false; - - for (i = 0; i < num_boundaries; i++) { - if (i == 0) - slice_width = mcache_boundaries[i]; - else - slice_width = mcache_boundaries[i] - mcache_boundaries[i - 1]; - - if (max_shift > (int)slice_width) { - max_shift = slice_width; - } - } - - for (*shift = 0; *shift <= max_shift; *shift += shift_granularity) { - success = true; - for (pipe_index = 0; pipe_index < pipe_count; pipe_index++) { - if (!calculate_first_second_splitting(mcache_boundaries, num_boundaries, *shift, - pipe_vp_startx[pipe_index], pipe_vp_endx[pipe_index], 0, 0)) { - success = false; - break; - } - } - if (success) - break; - } - - return success; -} - -/* -* Counts the number of elements inside input array within the given span length. -* Formally, what is the size of the largest subset of the array where the largest and smallest element -* differ no more than the span. -*/ -static unsigned int count_elements_in_span(int *array, unsigned int array_size, unsigned int span) -{ - unsigned int i; - unsigned int span_start_value; - unsigned int span_start_index; - unsigned int greatest_element_count; - - if (array_size == 0) - return 1; - - if (span == 0) - return array_size > 0 ? 1 : 0; - - span_start_value = 0; - span_start_index = 0; - greatest_element_count = 0; - - while (span_start_index < array_size) { - for (i = span_start_index; i < array_size; i++) { - if (array[i] - span_start_value <= span) { - if (i - span_start_index + 1 > greatest_element_count) { - greatest_element_count = i - span_start_index + 1; - } - } else - break; - } - - span_start_index++; - - if (span_start_index < array_size) { - span_start_value = array[span_start_index - 1] + 1; - } - } - - return greatest_element_count; -} - -static bool calculate_h_split_for_scaling_transform(int full_vp_width, int h_active, int num_pipes, - enum dml2_scaling_transform scaling_transform, int *pipe_vp_x_start, int *pipe_vp_x_end) -{ - int i, slice_width; - const char MAX_SCL_VP_OVERLAP = 3; - bool success = false; - - switch (scaling_transform) { - case dml2_scaling_transform_centered: - case dml2_scaling_transform_aspect_ratio: - case dml2_scaling_transform_fullscreen: - slice_width = full_vp_width / num_pipes; - for (i = 0; i < num_pipes; i++) { - pipe_vp_x_start[i] = i * slice_width; - pipe_vp_x_end[i] = (i + 1) * slice_width - 1; - - if (pipe_vp_x_start[i] < MAX_SCL_VP_OVERLAP) - pipe_vp_x_start[i] = 0; - else - pipe_vp_x_start[i] -= MAX_SCL_VP_OVERLAP; - - if (pipe_vp_x_end[i] > full_vp_width - MAX_SCL_VP_OVERLAP - 1) - pipe_vp_x_end[i] = full_vp_width - 1; - else - pipe_vp_x_end[i] += MAX_SCL_VP_OVERLAP; - } - break; - case dml2_scaling_transform_explicit: - default: - success = false; - break; - } - - return success; -} - -bool dml2_top_mcache_validate_admissability(struct top_mcache_validate_admissability_in_out *params) -{ - struct dml2_instance *dml = (struct dml2_instance *)params->dml2_instance; - struct dml2_top_mcache_validate_admissability_locals *l = &dml->scratch.mcache_validate_admissability_locals; - - const int MAX_PIXEL_OVERLAP = 6; - int max_per_pipe_vp_p0 = 0; - int max_per_pipe_vp_p1 = 0; - int temp, p0shift, p1shift; - unsigned int plane_index = 0; - unsigned int i; - unsigned int odm_combine_factor; - unsigned int mpc_combine_factor; - unsigned int num_dpps; - unsigned int num_boundaries; - enum dml2_scaling_transform scaling_transform; - const struct dml2_plane_parameters *plane; - const struct dml2_stream_parameters *stream; - - bool p0pass = false; - bool p1pass = false; - bool all_pass = true; - - for (plane_index = 0; plane_index < params->display_cfg->num_planes; plane_index++) { - if (!params->display_cfg->plane_descriptors[plane_index].surface.dcc.enable) - continue; - - plane = ¶ms->display_cfg->plane_descriptors[plane_index]; - stream = ¶ms->display_cfg->stream_descriptors[plane->stream_index]; - - num_dpps = odm_combine_factor = params->cfg_support_info->stream_support_info[plane->stream_index].odms_used; - - if (odm_combine_factor == 1) - num_dpps = mpc_combine_factor = (unsigned int)params->cfg_support_info->plane_support_info[plane_index].dpps_used; - else - mpc_combine_factor = 1; - - if (odm_combine_factor > 1) { - max_per_pipe_vp_p0 = plane->surface.plane0.width; - temp = (unsigned int)math_ceil(plane->composition.scaler_info.plane0.h_ratio * stream->timing.h_active / odm_combine_factor); - - if (temp < max_per_pipe_vp_p0) - max_per_pipe_vp_p0 = temp; - - max_per_pipe_vp_p1 = plane->surface.plane1.width; - temp = (unsigned int)math_ceil(plane->composition.scaler_info.plane1.h_ratio * stream->timing.h_active / odm_combine_factor); - - if (temp < max_per_pipe_vp_p1) - max_per_pipe_vp_p1 = temp; - } else { - max_per_pipe_vp_p0 = plane->surface.plane0.width / mpc_combine_factor; - max_per_pipe_vp_p1 = plane->surface.plane1.width / mpc_combine_factor; - } - - max_per_pipe_vp_p0 += 2 * MAX_PIXEL_OVERLAP; - max_per_pipe_vp_p1 += MAX_PIXEL_OVERLAP; - - p0shift = 0; - p1shift = 0; - - // The last element in the unshifted boundary array will always be the first pixel outside the - // plane, which means theres no mcache associated with it, so -1 - num_boundaries = params->mcache_allocations[plane_index].num_mcaches_plane0 == 0 ? 0 : params->mcache_allocations[plane_index].num_mcaches_plane0 - 1; - if ((count_elements_in_span(params->mcache_allocations[plane_index].mcache_x_offsets_plane0, - num_boundaries, max_per_pipe_vp_p0) <= 1) && (num_boundaries <= num_dpps)) { - p0pass = true; - } - num_boundaries = params->mcache_allocations[plane_index].num_mcaches_plane1 == 0 ? 0 : params->mcache_allocations[plane_index].num_mcaches_plane1 - 1; - if ((count_elements_in_span(params->mcache_allocations[plane_index].mcache_x_offsets_plane1, - num_boundaries, max_per_pipe_vp_p1) <= 1) && (num_boundaries <= num_dpps)) { - p1pass = true; - } - - if (!p0pass || !p1pass) { - if (odm_combine_factor > 1) { - num_dpps = odm_combine_factor; - scaling_transform = plane->composition.scaling_transform; - } else { - num_dpps = mpc_combine_factor; - scaling_transform = dml2_scaling_transform_fullscreen; - } - - if (!p0pass) { - if (plane->composition.viewport.stationary) { - calculate_h_split_for_scaling_transform(plane->surface.plane0.width, - stream->timing.h_active, num_dpps, scaling_transform, - &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index]); - p0pass = find_shift_for_valid_cache_id_assignment(params->mcache_allocations[plane_index].mcache_x_offsets_plane0, - params->mcache_allocations[plane_index].num_mcaches_plane0, - &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index], num_dpps, - params->mcache_allocations[plane_index].shift_granularity.p0, &p0shift); - } - } - if (!p1pass) { - if (plane->composition.viewport.stationary) { - calculate_h_split_for_scaling_transform(plane->surface.plane1.width, - stream->timing.h_active, num_dpps, scaling_transform, - &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index]); - p1pass = find_shift_for_valid_cache_id_assignment(params->mcache_allocations[plane_index].mcache_x_offsets_plane1, - params->mcache_allocations[plane_index].num_mcaches_plane1, - &l->plane1.pipe_vp_startx[plane_index], &l->plane1.pipe_vp_endx[plane_index], num_dpps, - params->mcache_allocations[plane_index].shift_granularity.p1, &p1shift); - } - } - } - - if (p0pass && p1pass) { - for (i = 0; i < params->mcache_allocations[plane_index].num_mcaches_plane0; i++) { - params->mcache_allocations[plane_index].mcache_x_offsets_plane0[i] -= p0shift; - } - for (i = 0; i < params->mcache_allocations[plane_index].num_mcaches_plane1; i++) { - params->mcache_allocations[plane_index].mcache_x_offsets_plane1[i] -= p1shift; - } - } - - params->per_plane_status[plane_index] = p0pass && p1pass; - all_pass &= p0pass && p1pass; - } - - return all_pass; -} - -static void reset_mcache_allocations(struct dml2_hubp_pipe_mcache_regs *per_plane_pipe_mcache_regs) -{ - // Initialize all entries to special valid MCache ID and special valid split coordinate - per_plane_pipe_mcache_regs->main.p0.mcache_id_first = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->main.p0.mcache_id_second = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->main.p0.split_location = SPLIT_LOCATION_UNDEFINED; - - per_plane_pipe_mcache_regs->mall.p0.mcache_id_first = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->mall.p0.mcache_id_second = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->mall.p0.split_location = SPLIT_LOCATION_UNDEFINED; - - per_plane_pipe_mcache_regs->main.p1.mcache_id_first = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->main.p1.mcache_id_second = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->main.p1.split_location = SPLIT_LOCATION_UNDEFINED; - - per_plane_pipe_mcache_regs->mall.p1.mcache_id_first = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->mall.p1.mcache_id_second = MCACHE_ID_UNASSIGNED; - per_plane_pipe_mcache_regs->mall.p1.split_location = SPLIT_LOCATION_UNDEFINED; -} - -void dml2_top_mcache_assign_global_mcache_ids(struct top_mcache_assign_global_mcache_ids_in_out *params) -{ - int i; - unsigned int j; - int next_unused_cache_id = 0; - - for (i = 0; i < params->num_allocations; i++) { - if (!params->allocations[i].valid) - continue; - - for (j = 0; j < params->allocations[i].num_mcaches_plane0; j++) { - params->allocations[i].global_mcache_ids_plane0[j] = next_unused_cache_id++; - } - for (j = 0; j < params->allocations[i].num_mcaches_plane1; j++) { - params->allocations[i].global_mcache_ids_plane1[j] = next_unused_cache_id++; - } - - // The "psuedo-last" slice is always wrapped around - params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0] = - params->allocations[i].global_mcache_ids_plane0[0]; - params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1] = - params->allocations[i].global_mcache_ids_plane1[0]; - - // If we need dedicated caches for mall requesting, then we assign them here. - if (params->allocations[i].requires_dedicated_mall_mcache) { - for (j = 0; j < params->allocations[i].num_mcaches_plane0; j++) { - params->allocations[i].global_mcache_ids_mall_plane0[j] = next_unused_cache_id++; - } - for (j = 0; j < params->allocations[i].num_mcaches_plane1; j++) { - params->allocations[i].global_mcache_ids_mall_plane1[j] = next_unused_cache_id++; - } - - // The "psuedo-last" slice is always wrapped around - params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0] = - params->allocations[i].global_mcache_ids_mall_plane0[0]; - params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1] = - params->allocations[i].global_mcache_ids_mall_plane1[0]; - } - - // If P0 and P1 are sharing caches, then it means the largest mcache IDs for p0 and p1 can be the same - // since mcache IDs are always ascending, then it means the largest mcacheID of p1 should be the - // largest mcacheID of P0 - if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].num_mcaches_plane1 > 0 && - params->allocations[i].last_slice_sharing.plane0_plane1) { - params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1 - 1] = - params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0 - 1]; - } - - // If we need dedicated caches handle last slice sharing - if (params->allocations[i].requires_dedicated_mall_mcache) { - if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].num_mcaches_plane1 > 0 && - params->allocations[i].last_slice_sharing.plane0_plane1) { - params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1 - 1] = - params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0 - 1]; - } - // If mall_comb_mcache_l is set then it means that largest mcache ID for MALL p0 can be same as regular read p0 - if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].last_slice_sharing.mall_comb_mcache_p0) { - params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0 - 1] = - params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0 - 1]; - } - // If mall_comb_mcache_c is set then it means that largest mcache ID for MALL p1 can be same as regular - // read p1 (which can be same as regular read p0 if plane0_plane1 is also set) - if (params->allocations[i].num_mcaches_plane1 > 0 && params->allocations[i].last_slice_sharing.mall_comb_mcache_p1) { - params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1 - 1] = - params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1 - 1]; - } - } - - // If you don't need dedicated mall mcaches, the mall mcache assignments are identical to the normal requesting - if (!params->allocations[i].requires_dedicated_mall_mcache) { - memcpy(params->allocations[i].global_mcache_ids_mall_plane0, params->allocations[i].global_mcache_ids_plane0, - sizeof(params->allocations[i].global_mcache_ids_mall_plane0)); - memcpy(params->allocations[i].global_mcache_ids_mall_plane1, params->allocations[i].global_mcache_ids_plane1, - sizeof(params->allocations[i].global_mcache_ids_mall_plane1)); - } - } -} - -bool dml2_top_mcache_calc_mcache_count_and_offsets(struct top_mcache_calc_mcache_count_and_offsets_in_out *params) -{ - struct dml2_instance *dml = (struct dml2_instance *)params->dml2_instance; - struct dml2_top_mcache_verify_mcache_size_locals *l = &dml->scratch.mcache_verify_mcache_size_locals; - - unsigned int total_mcaches_required; - unsigned int i; - bool result = false; - - if (dml->soc_bbox.num_dcc_mcaches == 0) { - return true; - } - - total_mcaches_required = 0; - l->calc_mcache_params.instance = &dml->core_instance; - for (i = 0; i < params->display_config->num_planes; i++) { - if (!params->display_config->plane_descriptors[i].surface.dcc.enable) { - memset(¶ms->mcache_allocations[i], 0, sizeof(struct dml2_mcache_surface_allocation)); - continue; - } - - l->calc_mcache_params.plane_descriptor = ¶ms->display_config->plane_descriptors[i]; - l->calc_mcache_params.mcache_allocation = ¶ms->mcache_allocations[i]; - l->calc_mcache_params.plane_index = i; - - if (!dml->core_instance.calculate_mcache_allocation(&l->calc_mcache_params)) { - result = false; - break; - } - - if (params->mcache_allocations[i].valid) { - total_mcaches_required += params->mcache_allocations[i].num_mcaches_plane0 + params->mcache_allocations[i].num_mcaches_plane1; - if (params->mcache_allocations[i].last_slice_sharing.plane0_plane1) - total_mcaches_required--; - } - } - DML_LOG_VERBOSE("DML_CORE_DCN3::%s: plane_%d, total_mcaches_required=%d\n", __func__, i, total_mcaches_required); - - if (total_mcaches_required > dml->soc_bbox.num_dcc_mcaches) { - result = false; - } else { - result = true; - } - - return result; -} - -static bool dml2_top_soc15_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out) -{ - struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; - struct dml2_check_mode_supported_locals *l = &dml->scratch.check_mode_supported_locals; - struct dml2_display_cfg_programming *dpmm_programming = &dml->dpmm_instance.dpmm_scratch.programming; - - bool result = false; - bool mcache_success = false; - memset(dpmm_programming, 0, sizeof(struct dml2_display_cfg_programming)); - - setup_unoptimized_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); - - l->mode_support_params.instance = &dml->core_instance; - l->mode_support_params.display_cfg = &l->base_display_config_with_meta; - l->mode_support_params.min_clk_table = &dml->min_clk_table; - l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; - result = dml->core_instance.mode_support(&l->mode_support_params); - l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; - - if (result) { - struct optimization_phase_params mcache_phase = { - .dml = dml, - .display_config = &l->base_display_config_with_meta, - .test_function = dml2_top_optimization_test_function_mcache, - .optimize_function = dml2_top_optimization_optimize_function_mcache, - .optimized_display_config = &l->optimized_display_config_with_meta, - .all_or_nothing = false, - }; - mcache_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &mcache_phase); - } - - /* - * Call DPMM to map all requirements to minimum clock state - */ - if (result) { - l->dppm_map_mode_params.min_clk_table = &dml->min_clk_table; - l->dppm_map_mode_params.display_cfg = &l->base_display_config_with_meta; - l->dppm_map_mode_params.programming = dpmm_programming; - l->dppm_map_mode_params.soc_bb = &dml->soc_bbox; - l->dppm_map_mode_params.ip = &dml->core_instance.clean_me_up.mode_lib.ip; - result = dml->dpmm_instance.map_mode_to_soc_dpm(&l->dppm_map_mode_params); - } - - in_out->is_supported = mcache_success; - result = result && in_out->is_supported; - - return result; -} - -static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out) -{ - struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; - struct dml2_build_mode_programming_locals *l = &dml->scratch.build_mode_programming_locals; - - bool result = false; - bool mcache_success = false; - bool uclk_pstate_success = false; - bool vmin_success = false; - bool stutter_success = false; - - memset(l, 0, sizeof(struct dml2_build_mode_programming_locals)); - memset(in_out->programming, 0, sizeof(struct dml2_display_cfg_programming)); - - memcpy(&in_out->programming->display_config, in_out->display_config, sizeof(struct dml2_display_cfg)); - - setup_speculative_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); - - l->mode_support_params.instance = &dml->core_instance; - l->mode_support_params.display_cfg = &l->base_display_config_with_meta; - l->mode_support_params.min_clk_table = &dml->min_clk_table; - l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; - result = dml->core_instance.mode_support(&l->mode_support_params); - - l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; - - if (!result) { - setup_unoptimized_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); - - l->mode_support_params.instance = &dml->core_instance; - l->mode_support_params.display_cfg = &l->base_display_config_with_meta; - l->mode_support_params.min_clk_table = &dml->min_clk_table; - l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; - result = dml->core_instance.mode_support(&l->mode_support_params); - l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; - - if (!result) { - l->informative_params.instance = &dml->core_instance; - l->informative_params.programming = in_out->programming; - l->informative_params.mode_is_supported = false; - dml->core_instance.populate_informative(&l->informative_params); - - return false; - } - - /* - * Phase 1: Determine minimum clocks to satisfy latency requirements for this mode - */ - memset(&l->min_clock_for_latency_phase, 0, sizeof(struct optimization_phase_params)); - l->min_clock_for_latency_phase.dml = dml; - l->min_clock_for_latency_phase.display_config = &l->base_display_config_with_meta; - l->min_clock_for_latency_phase.init_function = dml2_top_optimization_init_function_min_clk_for_latency; - l->min_clock_for_latency_phase.test_function = dml2_top_optimization_test_function_min_clk_for_latency; - l->min_clock_for_latency_phase.optimize_function = dml2_top_optimization_optimize_function_min_clk_for_latency; - l->min_clock_for_latency_phase.optimized_display_config = &l->optimized_display_config_with_meta; - l->min_clock_for_latency_phase.all_or_nothing = false; - - dml2_top_optimization_perform_optimization_phase_1(&l->optimization_phase_locals, &l->min_clock_for_latency_phase); - - memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); - } - - /* - * Phase 2: Satisfy DCC mcache requirements - */ - memset(&l->mcache_phase, 0, sizeof(struct optimization_phase_params)); - l->mcache_phase.dml = dml; - l->mcache_phase.display_config = &l->base_display_config_with_meta; - l->mcache_phase.test_function = dml2_top_optimization_test_function_mcache; - l->mcache_phase.optimize_function = dml2_top_optimization_optimize_function_mcache; - l->mcache_phase.optimized_display_config = &l->optimized_display_config_with_meta; - l->mcache_phase.all_or_nothing = true; - - mcache_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->mcache_phase); - - if (!mcache_success) { - l->informative_params.instance = &dml->core_instance; - l->informative_params.programming = in_out->programming; - l->informative_params.mode_is_supported = false; - - dml->core_instance.populate_informative(&l->informative_params); - - in_out->programming->informative.failed_mcache_validation = true; - return false; - } - - memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); - - /* - * Phase 3: Optimize for Pstate - */ - memset(&l->uclk_pstate_phase, 0, sizeof(struct optimization_phase_params)); - l->uclk_pstate_phase.dml = dml; - l->uclk_pstate_phase.display_config = &l->base_display_config_with_meta; - l->uclk_pstate_phase.init_function = dml2_top_optimization_init_function_uclk_pstate; - l->uclk_pstate_phase.test_function = dml2_top_optimization_test_function_uclk_pstate; - l->uclk_pstate_phase.optimize_function = dml2_top_optimization_optimize_function_uclk_pstate; - l->uclk_pstate_phase.optimized_display_config = &l->optimized_display_config_with_meta; - l->uclk_pstate_phase.all_or_nothing = true; - - uclk_pstate_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->uclk_pstate_phase); - - if (uclk_pstate_success) { - memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); - l->base_display_config_with_meta.stage3.success = true; - } - - /* - * Phase 4: Optimize for Vmin - */ - memset(&l->vmin_phase, 0, sizeof(struct optimization_phase_params)); - l->vmin_phase.dml = dml; - l->vmin_phase.display_config = &l->base_display_config_with_meta; - l->vmin_phase.init_function = dml2_top_optimization_init_function_vmin; - l->vmin_phase.test_function = dml2_top_optimization_test_function_vmin; - l->vmin_phase.optimize_function = dml2_top_optimization_optimize_function_vmin; - l->vmin_phase.optimized_display_config = &l->optimized_display_config_with_meta; - l->vmin_phase.all_or_nothing = false; - - vmin_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->vmin_phase); - - if (l->optimized_display_config_with_meta.stage4.performed) { - /* - * when performed is true, optimization has applied to - * optimized_display_config_with_meta and it has passed mode - * support. However it may or may not pass the test function to - * reach actual Vmin. As long as voltage is optimized even if it - * doesn't reach Vmin level, there is still power benefit so in - * this case we will still copy this optimization into base - * display config. - */ - memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); - l->base_display_config_with_meta.stage4.success = vmin_success; - } - - /* - * Phase 5: Optimize for Stutter - */ - memset(&l->stutter_phase, 0, sizeof(struct optimization_phase_params)); - l->stutter_phase.dml = dml; - l->stutter_phase.display_config = &l->base_display_config_with_meta; - l->stutter_phase.init_function = dml2_top_optimization_init_function_stutter; - l->stutter_phase.test_function = dml2_top_optimization_test_function_stutter; - l->stutter_phase.optimize_function = dml2_top_optimization_optimize_function_stutter; - l->stutter_phase.optimized_display_config = &l->optimized_display_config_with_meta; - l->stutter_phase.all_or_nothing = true; - - stutter_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->stutter_phase); - - if (stutter_success) { - memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); - l->base_display_config_with_meta.stage5.success = true; - } - - /* - * Call DPMM to map all requirements to minimum clock state - */ - if (result) { - l->dppm_map_mode_params.min_clk_table = &dml->min_clk_table; - l->dppm_map_mode_params.display_cfg = &l->base_display_config_with_meta; - l->dppm_map_mode_params.programming = in_out->programming; - l->dppm_map_mode_params.soc_bb = &dml->soc_bbox; - l->dppm_map_mode_params.ip = &dml->core_instance.clean_me_up.mode_lib.ip; - result = dml->dpmm_instance.map_mode_to_soc_dpm(&l->dppm_map_mode_params); - if (!result) - in_out->programming->informative.failed_dpmm = true; - } - - if (result) { - l->mode_programming_params.instance = &dml->core_instance; - l->mode_programming_params.display_cfg = &l->base_display_config_with_meta; - l->mode_programming_params.cfg_support_info = &l->base_display_config_with_meta.mode_support_result.cfg_support_info; - l->mode_programming_params.programming = in_out->programming; - result = dml->core_instance.mode_programming(&l->mode_programming_params); - if (!result) - in_out->programming->informative.failed_mode_programming = true; - } - - if (result) { - l->dppm_map_watermarks_params.core = &dml->core_instance; - l->dppm_map_watermarks_params.display_cfg = &l->base_display_config_with_meta; - l->dppm_map_watermarks_params.programming = in_out->programming; - result = dml->dpmm_instance.map_watermarks(&l->dppm_map_watermarks_params); - } - - l->informative_params.instance = &dml->core_instance; - l->informative_params.programming = in_out->programming; - l->informative_params.mode_is_supported = result; - - dml->core_instance.populate_informative(&l->informative_params); - - return result; -} - -bool dml2_top_soc15_build_mcache_programming(struct dml2_build_mcache_programming_in_out *params) -{ - bool success = true; - int config_index, pipe_index; - int first_offset, second_offset; - int free_per_plane_reg_index = 0; - - memset(params->per_plane_pipe_mcache_regs, 0, DML2_MAX_PLANES * DML2_MAX_DCN_PIPES * sizeof(struct dml2_hubp_pipe_mcache_regs *)); - - for (config_index = 0; config_index < params->num_configurations; config_index++) { - for (pipe_index = 0; pipe_index < params->mcache_configurations[config_index].num_pipes; pipe_index++) { - // Allocate storage for the mcache regs - params->per_plane_pipe_mcache_regs[config_index][pipe_index] = ¶ms->mcache_regs_set[free_per_plane_reg_index++]; - - reset_mcache_allocations(params->per_plane_pipe_mcache_regs[config_index][pipe_index]); - - if (params->mcache_configurations[config_index].plane_descriptor->surface.dcc.enable) { - // P0 always enabled - if (!calculate_first_second_splitting(params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0, - params->mcache_configurations[config_index].mcache_allocation->num_mcaches_plane0, - 0, - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_x_start, - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_x_start + - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_width - 1, - &first_offset, &second_offset)) { - success = false; - break; - } - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.mcache_id_first = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane0[first_offset]; - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.mcache_id_first = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane0[first_offset]; - - if (second_offset >= 0) { - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.mcache_id_second = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane0[second_offset]; - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.split_location = - params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0[first_offset] - 1; - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.mcache_id_second = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane0[second_offset]; - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.split_location = - params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0[first_offset] - 1; - } - - // Populate P1 if enabled - if (params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1_enabled) { - if (!calculate_first_second_splitting(params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1, - params->mcache_configurations[config_index].mcache_allocation->num_mcaches_plane1, - 0, - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_x_start, - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_x_start + - params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_width - 1, - &first_offset, &second_offset)) { - success = false; - break; - } - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.mcache_id_first = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane1[first_offset]; - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.mcache_id_first = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane1[first_offset]; - - if (second_offset >= 0) { - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.mcache_id_second = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane1[second_offset]; - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.split_location = - params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1[first_offset] - 1; - - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.mcache_id_second = - params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane1[second_offset]; - params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.split_location = - params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1[first_offset] - 1; - } - } - } - } - } - - return success; -} - -static const struct dml2_top_funcs soc15_funcs = { - .check_mode_supported = dml2_top_soc15_check_mode_supported, - .build_mode_programming = dml2_top_soc15_build_mode_programming, - .build_mcache_programming = dml2_top_soc15_build_mcache_programming, -}; - -bool dml2_top_soc15_initialize_instance(struct dml2_initialize_instance_in_out *in_out) -{ - struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; - struct dml2_initialize_instance_locals *l = &dml->scratch.initialize_instance_locals; - struct dml2_core_initialize_in_out core_init_params = { 0 }; - struct dml2_mcg_build_min_clock_table_params_in_out mcg_build_min_clk_params = { 0 }; - struct dml2_pmo_initialize_in_out pmo_init_params = { 0 }; - bool result = false; - - memset(l, 0, sizeof(struct dml2_initialize_instance_locals)); - memset(dml, 0, sizeof(struct dml2_instance)); - - memcpy(&dml->ip_caps, &in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); - memcpy(&dml->soc_bbox, &in_out->soc_bb, sizeof(struct dml2_soc_bb)); - - dml->project_id = in_out->options.project_id; - dml->pmo_options = in_out->options.pmo_options; - - // Initialize All Components - result = dml2_mcg_create(in_out->options.project_id, &dml->mcg_instance); - - if (result) - result = dml2_dpmm_create(in_out->options.project_id, &dml->dpmm_instance); - - if (result) - result = dml2_core_create(in_out->options.project_id, &dml->core_instance); - - if (result) { - mcg_build_min_clk_params.soc_bb = &in_out->soc_bb; - mcg_build_min_clk_params.min_clk_table = &dml->min_clk_table; - result = dml->mcg_instance.build_min_clock_table(&mcg_build_min_clk_params); - } - - if (result) { - core_init_params.project_id = in_out->options.project_id; - core_init_params.instance = &dml->core_instance; - core_init_params.minimum_clock_table = &dml->min_clk_table; - core_init_params.explicit_ip_bb = in_out->overrides.explicit_ip_bb; - core_init_params.explicit_ip_bb_size = in_out->overrides.explicit_ip_bb_size; - core_init_params.ip_caps = &in_out->ip_caps; - core_init_params.soc_bb = &in_out->soc_bb; - result = dml->core_instance.initialize(&core_init_params); - - if (core_init_params.explicit_ip_bb && core_init_params.explicit_ip_bb_size > 0) { - memcpy(&dml->ip_caps, &in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); - } - } - - if (result) - result = dml2_pmo_create(in_out->options.project_id, &dml->pmo_instance); - - if (result) { - pmo_init_params.instance = &dml->pmo_instance; - pmo_init_params.soc_bb = &dml->soc_bbox; - pmo_init_params.ip_caps = &dml->ip_caps; - pmo_init_params.mcg_clock_table_size = dml->min_clk_table.dram_bw_table.num_entries; - pmo_init_params.options = &dml->pmo_options; - dml->pmo_instance.initialize(&pmo_init_params); - } - dml->funcs = soc15_funcs; - return result; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h deleted file mode 100644 index 53bd8602f9ef..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h +++ /dev/null @@ -1,14 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_TOP_SOC15_H__ -#define __DML2_TOP_SOC15_H__ -#include "dml2_internal_shared_types.h" -bool dml2_top_soc15_initialize_instance(struct dml2_initialize_instance_in_out *in_out); - -bool dml2_top_mcache_calc_mcache_count_and_offsets(struct top_mcache_calc_mcache_count_and_offsets_in_out *params); -void dml2_top_mcache_assign_global_mcache_ids(struct top_mcache_assign_global_mcache_ids_in_out *params); -bool dml2_top_mcache_validate_admissability(struct top_mcache_validate_admissability_in_out *params); -bool dml2_top_soc15_build_mcache_programming(struct dml2_build_mcache_programming_in_out *params); -#endif /* __DML2_TOP_SOC15_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h deleted file mode 100644 index 611c80f4f1bf..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h +++ /dev/null @@ -1,189 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_DEBUG_H__ -#define __DML2_DEBUG_H__ - -#include "os_types.h" -#define DML_ASSERT(condition) ASSERT(condition) -#define DML_LOG_LEVEL_DEFAULT DML_LOG_LEVEL_WARN -#define DML_LOG_INTERNAL(fmt, ...) dm_output_to_console(fmt, ## __VA_ARGS__) - -/* private helper macros */ -#define _BOOL_FORMAT(field) "%s", field ? "true" : "false" -#define _UINT_FORMAT(field) "%u", field -#define _INT_FORMAT(field) "%d", field -#define _DOUBLE_FORMAT(field) "%lf", field -#define _ELEMENT_FUNC "function" -#define _ELEMENT_COMP_IF "component_interface" -#define _ELEMENT_TOP_IF "top_interface" -#define _LOG_ENTRY(element) do { \ - DML_LOG_INTERNAL("<"element" name=\""); \ - DML_LOG_INTERNAL(__func__); \ - DML_LOG_INTERNAL("\">\n"); \ -} while (0) -#define _LOG_EXIT(element) DML_LOG_INTERNAL("\n") -#define _LOG_SCALAR(field, format) do { \ - DML_LOG_INTERNAL(#field" = "format(field)); \ - DML_LOG_INTERNAL("\n"); \ -} while (0) -#define _LOG_ARRAY(field, size, format) do { \ - DML_LOG_INTERNAL(#field " = ["); \ - for (int _i = 0; _i < (int) size; _i++) { \ - DML_LOG_INTERNAL(format(field[_i])); \ - if (_i + 1 == (int) size) \ - DML_LOG_INTERNAL("]\n"); \ - else \ - DML_LOG_INTERNAL(", "); \ -}} while (0) -#define _LOG_2D_ARRAY(field, size0, size1, format) do { \ - DML_LOG_INTERNAL(#field" = ["); \ - for (int _i = 0; _i < (int) size0; _i++) { \ - DML_LOG_INTERNAL("\n\t["); \ - for (int _j = 0; _j < (int) size1; _j++) { \ - DML_LOG_INTERNAL(format(field[_i][_j])); \ - if (_j + 1 == (int) size1) \ - DML_LOG_INTERNAL("]"); \ - else \ - DML_LOG_INTERNAL(", "); \ - } \ - if (_i + 1 == (int) size0) \ - DML_LOG_INTERNAL("]\n"); \ - else \ - DML_LOG_INTERNAL(", "); \ - } \ -} while (0) -#define _LOG_3D_ARRAY(field, size0, size1, size2, format) do { \ - DML_LOG_INTERNAL(#field" = ["); \ - for (int _i = 0; _i < (int) size0; _i++) { \ - DML_LOG_INTERNAL("\n\t["); \ - for (int _j = 0; _j < (int) size1; _j++) { \ - DML_LOG_INTERNAL("["); \ - for (int _k = 0; _k < (int) size2; _k++) { \ - DML_LOG_INTERNAL(format(field[_i][_j][_k])); \ - if (_k + 1 == (int) size2) \ - DML_LOG_INTERNAL("]"); \ - else \ - DML_LOG_INTERNAL(", "); \ - } \ - if (_j + 1 == (int) size1) \ - DML_LOG_INTERNAL("]"); \ - else \ - DML_LOG_INTERNAL(", "); \ - } \ - if (_i + 1 == (int) size0) \ - DML_LOG_INTERNAL("]\n"); \ - else \ - DML_LOG_INTERNAL(", "); \ - } \ -} while (0) - -/* fatal errors for unrecoverable DML states until a full reset */ -#define DML_LOG_LEVEL_FATAL 0 -/* unexpected but recoverable failures inside DML */ -#define DML_LOG_LEVEL_ERROR 1 -/* unexpected inputs or events to DML */ -#define DML_LOG_LEVEL_WARN 2 -/* high level tracing of DML interfaces */ -#define DML_LOG_LEVEL_INFO 3 -/* tracing of DML internal executions */ -#define DML_LOG_LEVEL_DEBUG 4 -/* detailed tracing of DML calculation procedure */ -#define DML_LOG_LEVEL_VERBOSE 5 - -#ifndef DML_LOG_LEVEL -#define DML_LOG_LEVEL DML_LOG_LEVEL_DEFAULT -#endif /* #ifndef DML_LOG_LEVEL */ - -/* public macros for DML_LOG_LEVEL_FATAL and up */ -#define DML_LOG_FATAL(fmt, ...) DML_LOG_INTERNAL("[DML FATAL] " fmt, ## __VA_ARGS__) - -/* public macros for DML_LOG_LEVEL_ERROR and up */ -#if DML_LOG_LEVEL >= DML_LOG_LEVEL_ERROR -#define DML_LOG_ERROR(fmt, ...) DML_LOG_INTERNAL("[DML ERROR] "fmt, ## __VA_ARGS__) -#define DML_ASSERT_MSG(condition, fmt, ...) \ - do { \ - if (!(condition)) { \ - DML_LOG_ERROR("ASSERT hit in %s line %d\n", __func__, __LINE__); \ - DML_LOG_ERROR(fmt, ## __VA_ARGS__); \ - DML_ASSERT(condition); \ - } \ - } while (0) -#else -#define DML_LOG_ERROR(fmt, ...) ((void)0) -#define DML_ASSERT_MSG(condition, fmt, ...) ((void)0) -#endif - -/* public macros for DML_LOG_LEVEL_WARN and up */ -#if DML_LOG_LEVEL >= DML_LOG_LEVEL_WARN -#define DML_LOG_WARN(fmt, ...) DML_LOG_INTERNAL("[DML WARN] "fmt, ## __VA_ARGS__) -#else -#define DML_LOG_WARN(fmt, ...) ((void)0) -#endif - -/* public macros for DML_LOG_LEVEL_INFO and up */ -#if DML_LOG_LEVEL >= DML_LOG_LEVEL_INFO -#define DML_LOG_INFO(fmt, ...) DML_LOG_INTERNAL("[DML INFO] "fmt, ## __VA_ARGS__) -#define DML_LOG_TOP_IF_ENTER() _LOG_ENTRY(_ELEMENT_TOP_IF) -#define DML_LOG_TOP_IF_EXIT() _LOG_EXIT(_ELEMENT_TOP_IF) -#else -#define DML_LOG_INFO(fmt, ...) ((void)0) -#define DML_LOG_TOP_IF_ENTER() ((void)0) -#define DML_LOG_TOP_IF_EXIT() ((void)0) -#endif - -/* public macros for DML_LOG_LEVEL_DEBUG and up */ -#if DML_LOG_LEVEL >= DML_LOG_LEVEL_DEBUG -#define DML_LOG_DEBUG(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) -#define DML_LOG_COMP_IF_ENTER() _LOG_ENTRY(_ELEMENT_COMP_IF) -#define DML_LOG_COMP_IF_EXIT() _LOG_EXIT(_ELEMENT_COMP_IF) -#define DML_LOG_FUNC_ENTER() _LOG_ENTRY(_ELEMENT_FUNC) -#define DML_LOG_FUNC_EXIT() _LOG_EXIT(_ELEMENT_FUNC) -#define DML_LOG_DEBUG_BOOL(field) _LOG_SCALAR(field, _BOOL_FORMAT) -#define DML_LOG_DEBUG_UINT(field) _LOG_SCALAR(field, _UINT_FORMAT) -#define DML_LOG_DEBUG_INT(field) _LOG_SCALAR(field, _INT_FORMAT) -#define DML_LOG_DEBUG_DOUBLE(field) _LOG_SCALAR(field, _DOUBLE_FORMAT) -#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) _LOG_ARRAY(field, size, _BOOL_FORMAT) -#define DML_LOG_DEBUG_ARRAY_UINT(field, size) _LOG_ARRAY(field, size, _UINT_FORMAT) -#define DML_LOG_DEBUG_ARRAY_INT(field, size) _LOG_ARRAY(field, size, _INT_FORMAT) -#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) _LOG_ARRAY(field, size, _DOUBLE_FORMAT) -#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _BOOL_FORMAT) -#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _UINT_FORMAT) -#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _INT_FORMAT) -#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _DOUBLE_FORMAT) -#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _BOOL_FORMAT) -#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _UINT_FORMAT) -#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _INT_FORMAT) -#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _DOUBLE_FORMAT) -#else -#define DML_LOG_DEBUG(fmt, ...) ((void)0) -#define DML_LOG_COMP_IF_ENTER() ((void)0) -#define DML_LOG_COMP_IF_EXIT() ((void)0) -#define DML_LOG_FUNC_ENTER() ((void)0) -#define DML_LOG_FUNC_EXIT() ((void)0) -#define DML_LOG_DEBUG_BOOL(field) ((void)0) -#define DML_LOG_DEBUG_UINT(field) ((void)0) -#define DML_LOG_DEBUG_INT(field) ((void)0) -#define DML_LOG_DEBUG_DOUBLE(field) ((void)0) -#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) ((void)0) -#define DML_LOG_DEBUG_ARRAY_UINT(field, size) ((void)0) -#define DML_LOG_DEBUG_ARRAY_INT(field, size) ((void)0) -#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) ((void)0) -#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) ((void)0) -#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) ((void)0) -#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) ((void)0) -#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) ((void)0) -#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) ((void)0) -#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) ((void)0) -#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) ((void)0) -#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) ((void)0) -#endif - -/* public macros for DML_LOG_LEVEL_VERBOSE */ -#if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE -#define DML_LOG_VERBOSE(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) -#else -#define DML_LOG_VERBOSE(fmt, ...) ((void)0) -#endif /* #if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE */ -#endif /* __DML2_DEBUG_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h deleted file mode 100644 index 9f562f0c4797..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h +++ /dev/null @@ -1,1004 +0,0 @@ -// SPDX-License-Identifier: MIT -// -// Copyright 2024 Advanced Micro Devices, Inc. - -#ifndef __DML2_INTERNAL_SHARED_TYPES_H__ -#define __DML2_INTERNAL_SHARED_TYPES_H__ - -#include "dml2_external_lib_deps.h" -#include "dml_top_types.h" -#include "dml2_core_shared_types.h" -/* -* DML2 MCG Types and Interfaces -*/ - -#define DML_MCG_MAX_CLK_TABLE_SIZE 20 - -struct dram_bw_to_min_clk_table_entry { - unsigned long long pre_derate_dram_bw_kbps; - unsigned long min_fclk_khz; - unsigned long min_dcfclk_khz; -}; - -struct dml2_mcg_dram_bw_to_min_clk_table { - struct dram_bw_to_min_clk_table_entry entries[DML_MCG_MAX_CLK_TABLE_SIZE]; - - unsigned int num_entries; -}; - -struct dml2_mcg_min_clock_table { - struct { - unsigned int dispclk; - unsigned int dppclk; - unsigned int dscclk; - unsigned int dtbclk; - unsigned int phyclk; - unsigned int fclk; - unsigned int dcfclk; - } max_clocks_khz; - - struct { - unsigned int dispclk; - unsigned int dppclk; - unsigned int dtbclk; - } max_ss_clocks_khz; - - struct { - unsigned int dprefclk; - unsigned int xtalclk; - unsigned int pcierefclk; - unsigned int dchubrefclk; - unsigned int amclk; - } fixed_clocks_khz; - - struct dml2_mcg_dram_bw_to_min_clk_table dram_bw_table; -}; - -struct dml2_mcg_build_min_clock_table_params_in_out { - /* - * Input - */ - struct dml2_soc_bb *soc_bb; - struct { - bool perform_pseudo_build; - } clean_me_up; - - /* - * Output - */ - struct dml2_mcg_min_clock_table *min_clk_table; -}; -struct dml2_mcg_instance { - bool (*build_min_clock_table)(struct dml2_mcg_build_min_clock_table_params_in_out *in_out); -}; - -/* -* DML2 DPMM Types and Interfaces -*/ - -struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out { - /* - * Input - */ - struct dml2_core_ip_params *ip; - struct dml2_soc_bb *soc_bb; - struct dml2_mcg_min_clock_table *min_clk_table; - const struct display_configuation_with_meta *display_cfg; - struct { - bool perform_pseudo_map; - struct dml2_core_internal_soc_bb *soc_bb; - } clean_me_up; - - /* - * Output - */ - struct dml2_display_cfg_programming *programming; -}; - -struct dml2_dpmm_map_watermarks_params_in_out { - /* - * Input - */ - const struct display_configuation_with_meta *display_cfg; - const struct dml2_core_instance *core; - - /* - * Output - */ - struct dml2_display_cfg_programming *programming; -}; - -struct dml2_dpmm_scratch { - struct dml2_display_cfg_programming programming; -}; - -struct dml2_dpmm_instance { - bool (*map_mode_to_soc_dpm)(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); - bool (*map_watermarks)(struct dml2_dpmm_map_watermarks_params_in_out *in_out); - - struct dml2_dpmm_scratch dpmm_scratch; -}; - -/* -* DML2 Core Types and Interfaces -*/ - -struct dml2_core_initialize_in_out { - enum dml2_project_id project_id; - struct dml2_core_instance *instance; - struct dml2_soc_bb *soc_bb; - struct dml2_ip_capabilities *ip_caps; - - struct dml2_mcg_min_clock_table *minimum_clock_table; - - void *explicit_ip_bb; - unsigned int explicit_ip_bb_size; - - // FIXME_STAGE2 can remove but dcn3 version still need this - struct { - struct soc_bounding_box_st *soc_bb; - struct soc_states_st *soc_states; - } legacy; -}; - -struct core_bandwidth_requirements { - int urgent_bandwidth_kbytes_per_sec; - int average_bandwidth_kbytes_per_sec; -}; - -struct core_plane_support_info { - int dpps_used; - int dram_change_latency_hiding_margin_in_active; - int active_latency_hiding_us; - int mall_svp_size_requirement_ways; - int nominal_vblank_pstate_latency_hiding_us; - unsigned int dram_change_vactive_det_fill_delay_us; -}; - -struct core_stream_support_info { - unsigned int odms_used; - unsigned int num_odm_output_segments; // for odm split mode (e.g. a value of 2 for odm_mode_mso_1to2) - - /* FAMS2 SubVP support info */ - unsigned int phantom_min_v_active; - unsigned int phantom_v_startup; - - unsigned int phantom_v_active; - unsigned int phantom_v_total; - int vblank_reserved_time_us; - int num_dsc_slices; - bool dsc_enable; -}; - -struct core_display_cfg_support_info { - bool is_supported; - - struct core_stream_support_info stream_support_info[DML2_MAX_PLANES]; - struct core_plane_support_info plane_support_info[DML2_MAX_PLANES]; - - struct { - struct dml2_core_internal_mode_support_info support_info; - } clean_me_up; -}; - -struct dml2_core_mode_support_result { - struct { - struct { - unsigned long urgent_bw_sdp_kbps; - unsigned long average_bw_sdp_kbps; - unsigned long urgent_bw_dram_kbps; - unsigned long average_bw_dram_kbps; - unsigned long dcfclk_khz; - unsigned long fclk_khz; - } svp_prefetch; - - struct { - unsigned long urgent_bw_sdp_kbps; - unsigned long average_bw_sdp_kbps; - unsigned long urgent_bw_dram_kbps; - unsigned long average_bw_dram_kbps; - unsigned long dcfclk_khz; - unsigned long fclk_khz; - } active; - - unsigned int dispclk_khz; - unsigned int dpprefclk_khz; - unsigned int dtbrefclk_khz; - unsigned int dcfclk_deepsleep_khz; - unsigned int socclk_khz; - - unsigned int uclk_pstate_supported; - unsigned int fclk_pstate_supported; - } global; - - struct { - unsigned int dscclk_khz; - unsigned int dtbclk_khz; - unsigned int phyclk_khz; - } per_stream[DML2_MAX_PLANES]; - - struct { - unsigned int dppclk_khz; - unsigned int mall_svp_allocation_mblks; - unsigned int mall_full_frame_allocation_mblks; - } per_plane[DML2_MAX_PLANES]; - - struct core_display_cfg_support_info cfg_support_info; -}; - -struct dml2_optimization_stage1_state { - bool performed; - bool success; - - int min_clk_index_for_latency; -}; - -struct dml2_optimization_stage2_state { - bool performed; - bool success; - - // Whether or not each plane supports mcache - // The number of valid elements == display_cfg.num_planes - // The indexing of pstate_switch_modes matches plane_descriptors[] - bool per_plane_mcache_support[DML2_MAX_PLANES]; - struct dml2_mcache_surface_allocation mcache_allocations[DML2_MAX_PLANES]; -}; - -#define DML2_PMO_LEGACY_PREFETCH_MAX_TWAIT_OPTIONS 8 -#define DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE 10 -#define DML2_PMO_STUTTER_CANDIDATE_LIST_SIZE 3 - -struct dml2_implicit_svp_meta { - bool valid; - unsigned long v_active; - unsigned long v_total; - unsigned long v_front_porch; -}; - -struct dml2_pstate_per_method_common_meta { - /* generic params */ - unsigned int allow_start_otg_vline; - unsigned int allow_end_otg_vline; - /* scheduling params */ - double allow_time_us; - double disallow_time_us; - double period_us; -}; - -struct dml2_pstate_meta { - bool valid; - double otg_vline_time_us; - unsigned int scheduling_delay_otg_vlines; - unsigned int vertical_interrupt_ack_delay_otg_vlines; - unsigned int allow_to_target_delay_otg_vlines; - unsigned int contention_delay_otg_vlines; - unsigned int min_allow_width_otg_vlines; - unsigned int nom_vtotal; - unsigned int vblank_start; - double nom_refresh_rate_hz; - double nom_frame_time_us; - unsigned int max_vtotal; - double min_refresh_rate_hz; - double max_frame_time_us; - unsigned int blackout_otg_vlines; - struct { - double max_vactive_det_fill_delay_us; - unsigned int max_vactive_det_fill_delay_otg_vlines; - struct dml2_pstate_per_method_common_meta common; - } method_vactive; - struct { - struct dml2_pstate_per_method_common_meta common; - } method_vblank; - struct { - unsigned int programming_delay_otg_vlines; - unsigned int df_throttle_delay_otg_vlines; - unsigned int prefetch_to_mall_delay_otg_vlines; - unsigned long phantom_vactive; - unsigned long phantom_vfp; - unsigned long phantom_vtotal; - struct dml2_pstate_per_method_common_meta common; - } method_subvp; - struct { - unsigned int programming_delay_otg_vlines; - unsigned int stretched_vtotal; - struct dml2_pstate_per_method_common_meta common; - } method_drr; -}; - -/* mask of synchronized timings by stream index */ -struct dml2_pmo_synchronized_timing_groups { - unsigned int num_timing_groups; - unsigned int synchronized_timing_group_masks[DML2_MAX_PLANES]; - bool group_is_drr_enabled[DML2_MAX_PLANES]; - bool group_is_drr_active[DML2_MAX_PLANES]; - double group_line_time_us[DML2_MAX_PLANES]; -}; - -struct dml2_optimization_stage3_state { - bool performed; - bool success; - - // The pstate support mode for each plane - // The number of valid elements == display_cfg.num_planes - // The indexing of pstate_switch_modes matches plane_descriptors[] - enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; - - // Meta-data for implicit SVP generation, indexed by stream index - struct dml2_implicit_svp_meta stream_svp_meta[DML2_MAX_PLANES]; - - // Meta-data for FAMS2 - bool fams2_required; - struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; - - int min_clk_index_for_latency; -}; - -struct dml2_optimization_stage4_state { - bool performed; - bool success; - bool unoptimizable_streams[DML2_MAX_DCN_PIPES]; -}; - -struct dml2_optimization_stage5_state { - bool performed; - bool success; - - bool optimal_reserved_time_in_vblank_us; - bool vblank_includes_z8_optimization; -}; - -struct display_configuation_with_meta { - struct dml2_display_cfg display_config; - - struct dml2_core_mode_support_result mode_support_result; - - // Stage 1 = Min Clocks for Latency - struct dml2_optimization_stage1_state stage1; - - // Stage 2 = MCache - struct dml2_optimization_stage2_state stage2; - - // Stage 3 = UCLK PState - struct dml2_optimization_stage3_state stage3; - - // Stage 4 = Vmin - struct dml2_optimization_stage4_state stage4; - - // Stage 5 = Stutter - struct dml2_optimization_stage5_state stage5; -}; - -struct dml2_pmo_pstate_strategy { - enum dml2_pstate_method per_stream_pstate_method[DML2_MAX_PLANES]; - bool allow_state_increase; -}; -struct dml2_core_mode_support_in_out { - /* - * Inputs - */ - struct dml2_core_instance *instance; - const struct display_configuation_with_meta *display_cfg; - - struct dml2_mcg_min_clock_table *min_clk_table; - int min_clk_index; - /* - * Outputs - */ - struct dml2_core_mode_support_result mode_support_result; - - struct { - // Inputs - struct dml_display_cfg_st *display_cfg; - - // Outputs - struct dml_mode_support_info_st *support_info; - unsigned int out_lowest_state_idx; - unsigned int min_fclk_khz; - unsigned int min_dcfclk_khz; - unsigned int min_dram_speed_mts; - unsigned int min_socclk_khz; - unsigned int min_dscclk_khz; - unsigned int min_dtbclk_khz; - unsigned int min_phyclk_khz; - } legacy; -}; - -struct dml2_core_mode_programming_in_out { - /* - * Inputs - */ - struct dml2_core_instance *instance; - const struct display_configuation_with_meta *display_cfg; - const struct core_display_cfg_support_info *cfg_support_info; - /* - * Outputs (also Input the clk freq are also from programming struct) - */ - struct dml2_display_cfg_programming *programming; - -}; - -struct dml2_core_populate_informative_in_out { - /* - * Inputs - */ - struct dml2_core_instance *instance; - - // If this is set, then the mode was supported, and mode programming - // was successfully run. - // Otherwise, mode programming was not run, because mode support failed. - bool mode_is_supported; - - /* - * Outputs - */ - struct dml2_display_cfg_programming *programming; -}; - -struct dml2_calculate_mcache_allocation_in_out { - /* - * Inputs - */ - struct dml2_core_instance *instance; - const struct dml2_plane_parameters *plane_descriptor; - unsigned int plane_index; - - /* - * Outputs - */ - struct dml2_mcache_surface_allocation *mcache_allocation; -}; - -struct dml2_core_internal_state_inputs { - unsigned int dummy; -}; - -struct dml2_core_internal_state_intermediates { - unsigned int dummy; -}; - -struct dml2_core_mode_support_locals { - union { - struct dml2_core_calcs_mode_support_ex mode_support_ex_params; - }; - struct dml2_display_cfg svp_expanded_display_cfg; - struct dml2_calculate_mcache_allocation_in_out calc_mcache_allocation_params; -}; - -struct dml2_core_mode_programming_locals { - union { - struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; - }; - struct dml2_display_cfg svp_expanded_display_cfg; -}; - -struct dml2_core_scratch { - struct dml2_core_mode_support_locals mode_support_locals; - struct dml2_core_mode_programming_locals mode_programming_locals; - int main_stream_index_from_svp_stream_index[DML2_MAX_PLANES]; - int svp_stream_index_from_main_stream_index[DML2_MAX_PLANES]; - int main_plane_index_to_phantom_plane_index[DML2_MAX_PLANES]; - int phantom_plane_index_to_main_plane_index[DML2_MAX_PLANES]; -}; - -struct dml2_core_instance { - enum dml2_project_id project_id; - struct dml2_mcg_min_clock_table *minimum_clock_table; - struct dml2_core_internal_state_inputs inputs; - struct dml2_core_internal_state_intermediates intermediates; - - struct dml2_core_scratch scratch; - - bool (*initialize)(struct dml2_core_initialize_in_out *in_out); - bool (*mode_support)(struct dml2_core_mode_support_in_out *in_out); - bool (*mode_programming)(struct dml2_core_mode_programming_in_out *in_out); - bool (*populate_informative)(struct dml2_core_populate_informative_in_out *in_out); - bool (*calculate_mcache_allocation)(struct dml2_calculate_mcache_allocation_in_out *in_out); - - struct { - struct dml2_core_internal_display_mode_lib mode_lib; - } clean_me_up; -}; - -/* -* DML2 PMO Types and Interfaces -*/ - -struct dml2_pmo_initialize_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct dml2_soc_bb *soc_bb; - struct dml2_ip_capabilities *ip_caps; - struct dml2_pmo_options *options; - int mcg_clock_table_size; -}; - -struct dml2_pmo_optimize_dcc_mcache_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - const struct dml2_display_cfg *display_config; - bool *dcc_mcache_supported; - struct core_display_cfg_support_info *cfg_support_info; - - /* - * Output - */ - struct dml2_display_cfg *optimized_display_cfg; -}; - -struct dml2_pmo_init_for_vmin_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; -}; - -struct dml2_pmo_test_for_vmin_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - const struct display_configuation_with_meta *display_config; - const struct dml2_soc_vmin_clock_limits *vmin_limits; -}; - -struct dml2_pmo_optimize_for_vmin_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; - - /* - * Output - */ - struct display_configuation_with_meta *optimized_display_config; -}; - -struct dml2_pmo_init_for_pstate_support_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; -}; - -struct dml2_pmo_test_for_pstate_support_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; -}; - -struct dml2_pmo_optimize_for_pstate_support_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; - bool last_candidate_failed; - - /* - * Output - */ - struct display_configuation_with_meta *optimized_display_config; -}; - -struct dml2_pmo_init_for_stutter_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; -}; - -struct dml2_pmo_test_for_stutter_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; -}; - -struct dml2_pmo_optimize_for_stutter_in_out { - /* - * Input - */ - struct dml2_pmo_instance *instance; - struct display_configuation_with_meta *base_display_config; - bool last_candidate_failed; - - /* - * Output - */ - struct display_configuation_with_meta *optimized_display_config; -}; - -#define PMO_NO_DRR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw - dml2_pstate_method_na + 1)) - 1) << dml2_pstate_method_na) -#define PMO_DRR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_vactive_drr + 1)) - 1) << dml2_pstate_method_fw_vactive_drr) -#define PMO_DRR_CLAMPED_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_clamped - dml2_pstate_method_fw_vactive_drr + 1)) - 1) << dml2_pstate_method_fw_vactive_drr) -#define PMO_DRR_VAR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_drr + 1)) - 1) << dml2_pstate_method_fw_drr) -#define PMO_FW_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_svp + 1)) - 1) << dml2_pstate_method_fw_svp) - -#define PMO_DCN4_MAX_DISPLAYS 4 -#define PMO_DCN4_MAX_NUM_VARIANTS 2 -#define PMO_DCN4_MAX_BASE_STRATEGIES 10 - -struct dml2_scheduling_check_locals { - struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; - unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; - unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; -}; - -struct dml2_pmo_scratch { - union { - struct { - double reserved_time_candidates[DML2_MAX_PLANES][DML2_PMO_LEGACY_PREFETCH_MAX_TWAIT_OPTIONS]; - int reserved_time_candidates_count[DML2_MAX_PLANES]; - int current_candidate[DML2_MAX_PLANES]; - int min_latency_index; - int max_latency_index; - int cur_latency_index; - int stream_mask; - } pmo_dcn3; - struct { - struct dml2_pmo_pstate_strategy expanded_override_strategy_list[2 * 2 * 2 * 2]; - unsigned int num_expanded_override_strategies; - struct dml2_pmo_pstate_strategy pstate_strategy_candidates[DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE]; - int num_pstate_candidates; - int cur_pstate_candidate; - - unsigned int stream_plane_mask[DML2_MAX_PLANES]; - - unsigned int stream_vactive_capability_mask; - - int min_latency_index; - int max_latency_index; - int cur_latency_index; - - // Stores all the implicit SVP meta information indexed by stream index of the display - // configuration under inspection, built at optimization stage init - struct dml2_implicit_svp_meta stream_svp_meta[DML2_MAX_PLANES]; - struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; - - unsigned int optimal_vblank_reserved_time_for_stutter_us[DML2_PMO_STUTTER_CANDIDATE_LIST_SIZE]; - unsigned int num_stutter_candidates; - unsigned int cur_stutter_candidate; - bool z8_vblank_optimizable; - - /* mask of synchronized timings by stream index */ - unsigned int num_timing_groups; - unsigned int synchronized_timing_group_masks[DML2_MAX_PLANES]; - bool group_is_drr_enabled[DML2_MAX_PLANES]; - bool group_is_drr_active[DML2_MAX_PLANES]; - double group_line_time_us[DML2_MAX_PLANES]; - - /* scheduling check locals */ - struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; - unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; - unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; - double group_phase_offset[DML2_MAX_PLANES]; - } pmo_dcn4; - }; -}; - -struct dml2_pmo_init_data { - union { - struct { - /* populated once during initialization */ - struct dml2_pmo_pstate_strategy expanded_strategy_list_1_display[PMO_DCN4_MAX_BASE_STRATEGIES * 2]; - struct dml2_pmo_pstate_strategy expanded_strategy_list_2_display[PMO_DCN4_MAX_BASE_STRATEGIES * 4 * 4]; - struct dml2_pmo_pstate_strategy expanded_strategy_list_3_display[PMO_DCN4_MAX_BASE_STRATEGIES * 6 * 6 * 6]; - struct dml2_pmo_pstate_strategy expanded_strategy_list_4_display[PMO_DCN4_MAX_BASE_STRATEGIES * 8 * 8 * 8 * 8]; - unsigned int num_expanded_strategies_per_list[PMO_DCN4_MAX_DISPLAYS]; - } pmo_dcn4; - }; -}; - -struct dml2_pmo_instance { - struct dml2_soc_bb *soc_bb; - struct dml2_ip_capabilities *ip_caps; - - struct dml2_pmo_options *options; - - int disp_clk_vmin_threshold; - int mpc_combine_limit; - int odm_combine_limit; - int mcg_clock_table_size; - union { - struct { - struct { - int prefetch_end_to_mall_start_us; - int fw_processing_delay_us; - int refresh_rate_limit_min; - int refresh_rate_limit_max; - } subvp; - } v1; - struct { - struct { - int refresh_rate_limit_min; - int refresh_rate_limit_max; - } subvp; - struct { - int refresh_rate_limit_min; - int refresh_rate_limit_max; - } drr; - } v2; - } fams_params; - - bool (*initialize)(struct dml2_pmo_initialize_in_out *in_out); - bool (*optimize_dcc_mcache)(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); - - bool (*init_for_vmin)(struct dml2_pmo_init_for_vmin_in_out *in_out); - bool (*test_for_vmin)(struct dml2_pmo_test_for_vmin_in_out *in_out); - bool (*optimize_for_vmin)(struct dml2_pmo_optimize_for_vmin_in_out *in_out); - - bool (*init_for_uclk_pstate)(struct dml2_pmo_init_for_pstate_support_in_out *in_out); - bool (*test_for_uclk_pstate)(struct dml2_pmo_test_for_pstate_support_in_out *in_out); - bool (*optimize_for_uclk_pstate)(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); - - bool (*init_for_stutter)(struct dml2_pmo_init_for_stutter_in_out *in_out); - bool (*test_for_stutter)(struct dml2_pmo_test_for_stutter_in_out *in_out); - bool (*optimize_for_stutter)(struct dml2_pmo_optimize_for_stutter_in_out *in_out); - - struct dml2_pmo_init_data init_data; - struct dml2_pmo_scratch scratch; -}; - -/* -* DML2 MCache Types -*/ - -struct top_mcache_validate_admissability_in_out { - struct dml2_instance *dml2_instance; - - const struct dml2_display_cfg *display_cfg; - const struct core_display_cfg_support_info *cfg_support_info; - struct dml2_mcache_surface_allocation *mcache_allocations; - - bool per_plane_status[DML2_MAX_PLANES]; - - struct { - const struct dml_mode_support_info_st *mode_support_info; - } legacy; -}; - -struct top_mcache_assign_ids_in_out { - /* - * Input - */ - const struct dml2_mcache_surface_allocation *mcache_allocations; - int plane_count; - - int per_pipe_viewport_x_start[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; - int per_pipe_viewport_x_end[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; - int pipe_count_per_plane[DML2_MAX_PLANES]; - - struct dml2_display_mcache_regs *current_mcache_regs[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; //One set per pipe/hubp - - /* - * Output - */ - struct dml2_display_mcache_regs mcache_regs[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; //One set per pipe/hubp - struct dml2_build_mcache_programming_in_out *mcache_programming; -}; - -struct top_mcache_calc_mcache_count_and_offsets_in_out { - /* - * Inputs - */ - struct dml2_instance *dml2_instance; - const struct dml2_display_cfg *display_config; - - /* - * Outputs - */ - struct dml2_mcache_surface_allocation *mcache_allocations; -}; - -struct top_mcache_assign_global_mcache_ids_in_out { - /* - * Inputs/Outputs - */ - struct dml2_mcache_surface_allocation *allocations; - int num_allocations; -}; - -/* -* DML2 Top Types -*/ - -struct dml2_initialize_instance_locals { - int dummy; -}; - -struct dml2_optimization_init_function_locals { - union { - struct { - struct dml2_pmo_init_for_pstate_support_in_out init_params; - } uclk_pstate; - struct { - struct dml2_pmo_init_for_stutter_in_out stutter_params; - } stutter; - struct { - struct dml2_pmo_init_for_vmin_in_out init_params; - } vmin; - }; -}; - -struct dml2_optimization_test_function_locals { - union { - struct { - struct top_mcache_calc_mcache_count_and_offsets_in_out calc_mcache_count_params; - struct top_mcache_assign_global_mcache_ids_in_out assign_global_mcache_ids_params; - struct top_mcache_validate_admissability_in_out validate_admissibility_params; - } test_mcache; - struct { - struct dml2_pmo_test_for_vmin_in_out pmo_test_vmin_params; - } test_vmin; - struct { - struct dml2_pmo_test_for_pstate_support_in_out test_params; - } uclk_pstate; - struct { - struct dml2_pmo_test_for_stutter_in_out stutter_params; - } stutter; - }; -}; - -struct dml2_optimization_optimize_function_locals { - union { - struct { - struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; - } optimize_mcache; - struct { - struct dml2_pmo_optimize_for_vmin_in_out pmo_optimize_vmin_params; - } optimize_vmin; - struct { - struct dml2_pmo_optimize_for_pstate_support_in_out optimize_params; - } uclk_pstate; - struct { - struct dml2_pmo_optimize_for_stutter_in_out stutter_params; - } stutter; - }; -}; - -struct dml2_optimization_phase_locals { - struct display_configuation_with_meta cur_candidate_display_cfg; - struct display_configuation_with_meta next_candidate_display_cfg; - struct dml2_core_mode_support_in_out mode_support_params; - struct dml2_optimization_init_function_locals init_function_locals; - struct dml2_optimization_test_function_locals test_function_locals; - struct dml2_optimization_optimize_function_locals optimize_function_locals; -}; - -struct dml2_check_mode_supported_locals { - struct dml2_display_cfg display_cfg_working_copy; - struct dml2_core_mode_support_in_out mode_support_params; - struct dml2_optimization_phase_locals optimization_phase_locals; - struct display_configuation_with_meta base_display_config_with_meta; - struct display_configuation_with_meta optimized_display_config_with_meta; - struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out dppm_map_mode_params; -}; - -struct optimization_init_function_params { - struct dml2_optimization_init_function_locals *locals; - struct dml2_instance *dml; - struct display_configuation_with_meta *display_config; -}; - -struct optimization_test_function_params { - struct dml2_optimization_test_function_locals *locals; - struct dml2_instance *dml; - struct display_configuation_with_meta *display_config; -}; - -struct optimization_optimize_function_params { - bool last_candidate_supported; - struct dml2_optimization_optimize_function_locals *locals; - struct dml2_instance *dml; - struct display_configuation_with_meta *display_config; - struct display_configuation_with_meta *optimized_display_config; -}; - -struct optimization_phase_params { - struct dml2_instance *dml; - const struct display_configuation_with_meta *display_config; // Initial Display Configuration - bool (*init_function)(const struct optimization_init_function_params *params); // Test function to determine optimization is complete - bool (*test_function)(const struct optimization_test_function_params *params); // Test function to determine optimization is complete - bool (*optimize_function)(const struct optimization_optimize_function_params *params); // Function which produces a more optimized display configuration - struct display_configuation_with_meta *optimized_display_config; // The optimized display configuration - - bool all_or_nothing; -}; - -struct dml2_build_mode_programming_locals { - struct dml2_core_mode_support_in_out mode_support_params; - struct dml2_core_mode_programming_in_out mode_programming_params; - struct dml2_core_populate_informative_in_out informative_params; - struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; - struct display_configuation_with_meta base_display_config_with_meta; - struct display_configuation_with_meta optimized_display_config_with_meta; - struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out dppm_map_mode_params; - struct dml2_dpmm_map_watermarks_params_in_out dppm_map_watermarks_params; - struct dml2_optimization_phase_locals optimization_phase_locals; - struct optimization_phase_params min_clock_for_latency_phase; - struct optimization_phase_params mcache_phase; - struct optimization_phase_params uclk_pstate_phase; - struct optimization_phase_params vmin_phase; - struct optimization_phase_params stutter_phase; -}; - -struct dml2_legacy_core_build_mode_programming_wrapper_locals { - struct dml2_core_mode_support_in_out mode_support_params; - struct dml2_core_mode_programming_in_out mode_programming_params; - struct dml2_core_populate_informative_in_out informative_params; - struct top_mcache_calc_mcache_count_and_offsets_in_out calc_mcache_count_params; - struct top_mcache_validate_admissability_in_out validate_admissibility_params; - struct dml2_mcache_surface_allocation mcache_allocations[DML2_MAX_PLANES]; - struct top_mcache_assign_global_mcache_ids_in_out assign_global_mcache_ids_params; - struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; - struct dml2_display_cfg optimized_display_cfg; - struct core_display_cfg_support_info core_support_info; -}; - -struct dml2_top_mcache_verify_mcache_size_locals { - struct dml2_calculate_mcache_allocation_in_out calc_mcache_params; -}; - -struct dml2_top_mcache_validate_admissability_locals { - struct { - int pipe_vp_startx[DML2_MAX_DCN_PIPES]; - int pipe_vp_endx[DML2_MAX_DCN_PIPES]; - } plane0; - struct { - int pipe_vp_startx[DML2_MAX_DCN_PIPES]; - int pipe_vp_endx[DML2_MAX_DCN_PIPES]; - } plane1; -}; - -struct dml2_top_display_cfg_support_info { - const struct dml2_display_cfg *display_config; - struct core_display_cfg_support_info core_info; -}; - -struct dml2_top_funcs { - bool (*check_mode_supported)(struct dml2_check_mode_supported_in_out *in_out); - bool (*build_mode_programming)(struct dml2_build_mode_programming_in_out *in_out); - bool (*build_mcache_programming)(struct dml2_build_mcache_programming_in_out *in_out); -}; - -struct dml2_instance { - enum dml2_project_id project_id; - - struct dml2_core_instance core_instance; - struct dml2_mcg_instance mcg_instance; - struct dml2_dpmm_instance dpmm_instance; - struct dml2_pmo_instance pmo_instance; - - struct dml2_soc_bb soc_bbox; - struct dml2_ip_capabilities ip_caps; - - struct dml2_mcg_min_clock_table min_clk_table; - struct dml2_pmo_options pmo_options; - struct dml2_top_funcs funcs; - - struct { - struct dml2_initialize_instance_locals initialize_instance_locals; - struct dml2_top_mcache_verify_mcache_size_locals mcache_verify_mcache_size_locals; - struct dml2_top_mcache_validate_admissability_locals mcache_validate_admissability_locals; - struct dml2_check_mode_supported_locals check_mode_supported_locals; - struct dml2_build_mode_programming_locals build_mode_programming_locals; - } scratch; - - struct { - struct { - struct dml2_legacy_core_build_mode_programming_wrapper_locals legacy_core_build_mode_programming_wrapper_locals; - } scratch; - } legacy; -}; -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c deleted file mode 100644 index 4cfe64aa8492..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c +++ /dev/null @@ -1,1174 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "dml2_mall_phantom.h" - -#include "dml2_dc_types.h" -#include "dml2_internal_types.h" -#include "dml2_utils.h" -#include "dml2_dc_resource_mgmt.h" - -#define MAX_ODM_FACTOR 4 -#define MAX_MPCC_FACTOR 4 - -struct dc_plane_pipe_pool { - int pipes_assigned_to_plane[MAX_ODM_FACTOR][MAX_MPCC_FACTOR]; - bool pipe_used[MAX_ODM_FACTOR][MAX_MPCC_FACTOR]; - int num_pipes_assigned_to_plane_for_mpcc_combine; - int num_pipes_assigned_to_plane_for_odm_combine; -}; - -struct dc_pipe_mapping_scratch { - struct { - unsigned int odm_factor; - unsigned int odm_slice_end_x[MAX_PIPES]; - struct pipe_ctx *next_higher_pipe_for_odm_slice[MAX_PIPES]; - } odm_info; - struct { - unsigned int mpc_factor; - struct pipe_ctx *prev_odm_pipe; - } mpc_info; - - struct dc_plane_pipe_pool pipe_pool; -}; - -static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *state, const struct dc_plane_state *plane, - unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) -{ - int i, j; - bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; - - if (!plane_id) - return false; - - for (i = 0; i < state->stream_count; i++) { - if (state->streams[i]->stream_id == stream_id) { - for (j = 0; j < state->stream_status[i].plane_count; j++) { - if (state->stream_status[i].plane_states[j] == plane && - (!is_plane_duplicate || (j == plane_index))) { - *plane_id = (i << 16) | j; - return true; - } - } - } - } - - return false; -} - -static int find_disp_cfg_idx_by_plane_id(struct dml2_dml_to_dc_pipe_mapping *mapping, unsigned int plane_id) -{ - int i; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (mapping->disp_cfg_to_plane_id_valid[i] && mapping->disp_cfg_to_plane_id[i] == plane_id) - return i; - } - - ASSERT(false); - return __DML2_WRAPPER_MAX_STREAMS_PLANES__; -} - -static int find_disp_cfg_idx_by_stream_id(struct dml2_dml_to_dc_pipe_mapping *mapping, unsigned int stream_id) -{ - int i; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (mapping->disp_cfg_to_stream_id_valid[i] && mapping->disp_cfg_to_stream_id[i] == stream_id) - return i; - } - - ASSERT(false); - return __DML2_WRAPPER_MAX_STREAMS_PLANES__; -} - -// The master pipe of a stream is defined as the top pipe in odm slice 0 -static struct pipe_ctx *find_master_pipe_of_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int stream_id) -{ - int i; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - if (state->res_ctx.pipe_ctx[i].stream && state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id) { - if (!state->res_ctx.pipe_ctx[i].prev_odm_pipe && !state->res_ctx.pipe_ctx[i].top_pipe) - return &state->res_ctx.pipe_ctx[i]; - } - } - - return NULL; -} - -static struct pipe_ctx *find_master_pipe_of_plane(struct dml2_context *ctx, - struct dc_state *state, unsigned int plane_id) -{ - int i; - unsigned int plane_id_assigned_to_pipe; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - if (state->res_ctx.pipe_ctx[i].plane_state && get_plane_id(ctx, state, state->res_ctx.pipe_ctx[i].plane_state, - state->res_ctx.pipe_ctx[i].stream->stream_id, - ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[state->res_ctx.pipe_ctx[i].pipe_idx], &plane_id_assigned_to_pipe)) { - if (plane_id_assigned_to_pipe == plane_id) - return &state->res_ctx.pipe_ctx[i]; - } - } - - return NULL; -} - -static unsigned int find_pipes_assigned_to_plane(struct dml2_context *ctx, - struct dc_state *state, unsigned int plane_id, unsigned int *pipes) -{ - int i; - unsigned int num_found = 0; - unsigned int plane_id_assigned_to_pipe = -1; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - if (!pipe->plane_state || !pipe->stream) - continue; - - get_plane_id(ctx, state, pipe->plane_state, pipe->stream->stream_id, - ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[pipe->pipe_idx], - &plane_id_assigned_to_pipe); - if (plane_id_assigned_to_pipe == plane_id && !pipe->prev_odm_pipe - && (!pipe->top_pipe || pipe->top_pipe->plane_state != pipe->plane_state)) { - while (pipe) { - struct pipe_ctx *mpc_pipe = pipe; - - while (mpc_pipe) { - pipes[num_found++] = mpc_pipe->pipe_idx; - mpc_pipe = mpc_pipe->bottom_pipe; - if (!mpc_pipe) - break; - if (mpc_pipe->plane_state != pipe->plane_state) - mpc_pipe = NULL; - } - pipe = pipe->next_odm_pipe; - } - break; - } - } - - return num_found; -} - -static bool validate_pipe_assignment(const struct dml2_context *ctx, const struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, const struct dml2_dml_to_dc_pipe_mapping *mapping) -{ -// int i, j, k; -// -// unsigned int plane_id; -// -// unsigned int disp_cfg_index; -// -// unsigned int pipes_assigned_to_plane[MAX_PIPES]; -// unsigned int num_pipes_assigned_to_plane; -// -// struct pipe_ctx *top_pipe; -// -// for (i = 0; i < state->stream_count; i++) { -// for (j = 0; j < state->stream_status[i]->plane_count; j++) { -// if (get_plane_id(state, state->stream_status.plane_states[j], &plane_id)) { -// disp_cfg_index = find_disp_cfg_idx_by_plane_id(mapping, plane_id); -// num_pipes_assigned_to_plane = find_pipes_assigned_to_plane(ctx, state, plane_id, pipes_assigned_to_plane); -// -// if (disp_cfg_index >= 0 && num_pipes_assigned_to_plane > 0) { -// // Verify the number of pipes assigned matches -// if (disp_cfg->hw.DPPPerSurface != num_pipes_assigned_to_plane) -// return false; -// -// top_pipe = find_top_pipe_in_tree(state->res_ctx.pipe_ctx[pipes_assigned_to_plane[0]]); -// -// // Verify MPC and ODM combine -// if (disp_cfg->hw.ODMMode == dml_odm_mode_bypass) { -// verify_combine_tree(top_pipe, state->streams[i]->stream_id, plane_id, state, false); -// } else { -// verify_combine_tree(top_pipe, state->streams[i]->stream_id, plane_id, state, true); -// } -// -// // TODO: could also do additional verification that the pipes in tree are the same as -// // pipes_assigned_to_plane -// } else { -// ASSERT(false); -// return false; -// } -// } else { -// ASSERT(false); -// return false; -// } -// } -// } - return true; -} - -static bool is_plane_using_pipe(const struct pipe_ctx *pipe) -{ - if (pipe->plane_state) - return true; - - return false; -} - -static bool is_pipe_free(const struct pipe_ctx *pipe) -{ - if (!pipe->plane_state && !pipe->stream) - return true; - - return false; -} - -static unsigned int find_preferred_pipe_candidates(const struct dc_state *existing_state, - const int pipe_count, - const unsigned int stream_id, - unsigned int *preferred_pipe_candidates) -{ - unsigned int num_preferred_candidates = 0; - int i; - - /* There is only one case which we consider for adding a pipe to the preferred - * pipe candidate array: - * - * 1. If the existing stream id of the pipe is equivalent to the stream id - * of the stream we are trying to achieve MPC/ODM combine for. This allows - * us to minimize the changes in pipe topology during the transition. - * - * However this condition comes with a caveat. We need to ignore pipes that will - * require a change in OPP but still have the same stream id. For example during - * an MPC to ODM transiton. - * - * Adding check to avoid pipe select on the head pipe by utilizing dc resource - * helper function resource_get_primary_dpp_pipe and comparing the pipe index. - */ - if (existing_state) { - for (i = 0; i < pipe_count; i++) { - if (existing_state->res_ctx.pipe_ctx[i].stream && existing_state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id) { - struct pipe_ctx *head_pipe = - resource_is_pipe_type(&existing_state->res_ctx.pipe_ctx[i], DPP_PIPE) ? - resource_get_primary_dpp_pipe(&existing_state->res_ctx.pipe_ctx[i]) : - NULL; - - // we should always respect the head pipe from selection - if (head_pipe && head_pipe->pipe_idx == i) - continue; - if (existing_state->res_ctx.pipe_ctx[i].plane_res.hubp && - existing_state->res_ctx.pipe_ctx[i].plane_res.hubp->opp_id != i && - (existing_state->res_ctx.pipe_ctx[i].prev_odm_pipe || - existing_state->res_ctx.pipe_ctx[i].next_odm_pipe)) - continue; - - preferred_pipe_candidates[num_preferred_candidates++] = i; - } - } - } - - return num_preferred_candidates; -} - -static unsigned int find_last_resort_pipe_candidates(const struct dc_state *existing_state, - const int pipe_count, - const unsigned int stream_id, - unsigned int *last_resort_pipe_candidates) -{ - unsigned int num_last_resort_candidates = 0; - int i; - - /* There are two cases where we would like to add a given pipe into the last - * candidate array: - * - * 1. If the pipe requires a change in OPP, for example during an MPC - * to ODM transiton. - * - * 2. If the pipe already has an enabled OTG. - */ - if (existing_state) { - for (i = 0; i < pipe_count; i++) { - struct pipe_ctx *head_pipe = - resource_is_pipe_type(&existing_state->res_ctx.pipe_ctx[i], DPP_PIPE) ? - resource_get_primary_dpp_pipe(&existing_state->res_ctx.pipe_ctx[i]) : - NULL; - - // we should always respect the head pipe from selection - if (head_pipe && head_pipe->pipe_idx == i) - continue; - if ((existing_state->res_ctx.pipe_ctx[i].plane_res.hubp && - existing_state->res_ctx.pipe_ctx[i].plane_res.hubp->opp_id != i) || - existing_state->res_ctx.pipe_ctx[i].stream_res.tg) - last_resort_pipe_candidates[num_last_resort_candidates++] = i; - } - } - - return num_last_resort_candidates; -} - -static bool is_pipe_in_candidate_array(const unsigned int pipe_idx, - const unsigned int *candidate_array, - const unsigned int candidate_array_size) -{ - int i; - - for (i = 0; i < candidate_array_size; i++) { - if (candidate_array[i] == pipe_idx) - return true; - } - - return false; -} - -static bool find_more_pipes_for_stream(struct dml2_context *ctx, - struct dc_state *state, // The state we want to find a free mapping in - unsigned int stream_id, // The stream we want this pipe to drive - int *assigned_pipes, - int *assigned_pipe_count, - int pipes_needed, - const struct dc_state *existing_state) // The state (optional) that we want to minimize remapping relative to -{ - struct pipe_ctx *pipe = NULL; - unsigned int preferred_pipe_candidates[MAX_PIPES] = {0}; - unsigned int last_resort_pipe_candidates[MAX_PIPES] = {0}; - unsigned int num_preferred_candidates = 0; - unsigned int num_last_resort_candidates = 0; - int i; - - if (existing_state) { - num_preferred_candidates = - find_preferred_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, preferred_pipe_candidates); - - num_last_resort_candidates = - find_last_resort_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, last_resort_pipe_candidates); - } - - // First see if any of the preferred are unmapped, and choose those instead - for (i = 0; pipes_needed > 0 && i < num_preferred_candidates; i++) { - pipe = &state->res_ctx.pipe_ctx[preferred_pipe_candidates[i]]; - if (!is_plane_using_pipe(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = preferred_pipe_candidates[i]; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - // We like to pair pipes starting from the higher order indicies for combining - for (i = ctx->config.dcn_pipe_count - 1; pipes_needed > 0 && i >= 0; i--) { - // Ignore any pipes that are the preferred or last resort candidate - if (is_pipe_in_candidate_array(i, preferred_pipe_candidates, num_preferred_candidates) || - is_pipe_in_candidate_array(i, last_resort_pipe_candidates, num_last_resort_candidates)) - continue; - - pipe = &state->res_ctx.pipe_ctx[i]; - if (!is_plane_using_pipe(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = i; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - // Only use the last resort pipe candidates as a last resort - for (i = 0; pipes_needed > 0 && i < num_last_resort_candidates; i++) { - pipe = &state->res_ctx.pipe_ctx[last_resort_pipe_candidates[i]]; - if (!is_plane_using_pipe(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = last_resort_pipe_candidates[i]; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - ASSERT(pipes_needed <= 0); // Validation should prevent us from building a pipe context that exceeds the number of HW resoruces available - - return pipes_needed <= 0; -} - -static bool find_more_free_pipes(struct dml2_context *ctx, - struct dc_state *state, // The state we want to find a free mapping in - unsigned int stream_id, // The stream we want this pipe to drive - int *assigned_pipes, - int *assigned_pipe_count, - int pipes_needed, - const struct dc_state *existing_state) // The state (optional) that we want to minimize remapping relative to -{ - struct pipe_ctx *pipe = NULL; - unsigned int preferred_pipe_candidates[MAX_PIPES] = {0}; - unsigned int last_resort_pipe_candidates[MAX_PIPES] = {0}; - unsigned int num_preferred_candidates = 0; - unsigned int num_last_resort_candidates = 0; - int i; - - if (existing_state) { - num_preferred_candidates = - find_preferred_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, preferred_pipe_candidates); - - num_last_resort_candidates = - find_last_resort_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, last_resort_pipe_candidates); - } - - // First see if any of the preferred are unmapped, and choose those instead - for (i = 0; pipes_needed > 0 && i < num_preferred_candidates; i++) { - pipe = &state->res_ctx.pipe_ctx[preferred_pipe_candidates[i]]; - if (is_pipe_free(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = preferred_pipe_candidates[i]; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - // We like to pair pipes starting from the higher order indicies for combining - for (i = ctx->config.dcn_pipe_count - 1; pipes_needed > 0 && i >= 0; i--) { - // Ignore any pipes that are the preferred or last resort candidate - if (is_pipe_in_candidate_array(i, preferred_pipe_candidates, num_preferred_candidates) || - is_pipe_in_candidate_array(i, last_resort_pipe_candidates, num_last_resort_candidates)) - continue; - - pipe = &state->res_ctx.pipe_ctx[i]; - if (is_pipe_free(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = i; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - // Only use the last resort pipe candidates as a last resort - for (i = 0; pipes_needed > 0 && i < num_last_resort_candidates; i++) { - pipe = &state->res_ctx.pipe_ctx[last_resort_pipe_candidates[i]]; - if (is_pipe_free(pipe)) { - pipes_needed--; - // TODO: This doens't make sense really, pipe_idx should always be valid - pipe->pipe_idx = last_resort_pipe_candidates[i]; - assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; - } - } - - ASSERT(pipes_needed == 0); // Validation should prevent us from building a pipe context that exceeds the number of HW resoruces available - - return pipes_needed == 0; -} - -static void sort_pipes_for_splitting(struct dc_plane_pipe_pool *pipes) -{ - bool sorted, swapped; - unsigned int cur_index; - int odm_slice_index; - - for (odm_slice_index = 0; odm_slice_index < pipes->num_pipes_assigned_to_plane_for_odm_combine; odm_slice_index++) { - // Sort each MPCC set - //Un-optimized bubble sort, but that's okay for array sizes <= 6 - - if (pipes->num_pipes_assigned_to_plane_for_mpcc_combine <= 1) - sorted = true; - else - sorted = false; - - cur_index = 0; - swapped = false; - while (!sorted) { - if (pipes->pipes_assigned_to_plane[odm_slice_index][cur_index] > pipes->pipes_assigned_to_plane[odm_slice_index][cur_index + 1]) { - swap(pipes->pipes_assigned_to_plane[odm_slice_index][cur_index + 1], - pipes->pipes_assigned_to_plane[odm_slice_index][cur_index]); - - swapped = true; - } - - cur_index++; - - if (cur_index == pipes->num_pipes_assigned_to_plane_for_mpcc_combine - 1) { - cur_index = 0; - - if (swapped) - sorted = false; - else - sorted = true; - - swapped = false; - } - - } - } -} - -// For example, 3840 x 2160, ODM2:1 has a slice array of [1919, 3839], meaning, slice0 spans h_pixels 0->1919, and slice1 spans 1920->3840 -static void calculate_odm_slices(const struct dc_stream_state *stream, unsigned int odm_factor, unsigned int *odm_slice_end_x) -{ - unsigned int slice_size = 0; - int i; - - if (odm_factor < 1 || odm_factor > 4) { - ASSERT(false); - return; - } - - slice_size = stream->src.width / odm_factor; - - for (i = 0; i < odm_factor; i++) - odm_slice_end_x[i] = (slice_size * (i + 1)) - 1; - - odm_slice_end_x[odm_factor - 1] = stream->src.width - 1; -} - -static void add_odm_slice_to_odm_tree(struct dml2_context *ctx, - struct dc_state *state, - struct dc_pipe_mapping_scratch *scratch, - unsigned int odm_slice_index) -{ - struct pipe_ctx *pipe = NULL; - int i; - - // MPCC Combine + ODM Combine is not supported, so there should never be a case where the current plane - // has more than 1 pipe mapped to it for a given slice. - ASSERT(scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine == 1 || scratch->pipe_pool.num_pipes_assigned_to_plane_for_odm_combine == 1); - - for (i = 0; i < scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine; i++) { - pipe = &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][i]]; - - if (scratch->mpc_info.prev_odm_pipe) - scratch->mpc_info.prev_odm_pipe->next_odm_pipe = pipe; - - pipe->prev_odm_pipe = scratch->mpc_info.prev_odm_pipe; - pipe->next_odm_pipe = NULL; - } - scratch->mpc_info.prev_odm_pipe = pipe; -} - -static struct pipe_ctx *add_plane_to_blend_tree(struct dml2_context *ctx, - struct dc_state *state, - const struct dc_plane_state *plane, - struct dc_plane_pipe_pool *pipe_pool, - unsigned int odm_slice, - struct pipe_ctx *top_pipe) -{ - int i; - - for (i = 0; i < pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine; i++) { - if (top_pipe) - top_pipe->bottom_pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]]; - - pipe_pool->pipe_used[odm_slice][i] = true; - - state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]].top_pipe = top_pipe; - state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]].bottom_pipe = NULL; - - top_pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]]; - } - - // After running the above loop, the top pipe actually ends up pointing to the bottom of this MPCC combine tree, so we are actually - // returning the bottom pipe here - return top_pipe; -} - -static unsigned int find_pipes_assigned_to_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int stream_id, unsigned int *pipes) -{ - int i; - unsigned int num_found = 0; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - if (pipe->stream && pipe->stream->stream_id == stream_id && !pipe->top_pipe && !pipe->prev_odm_pipe) { - while (pipe) { - pipes[num_found++] = pipe->pipe_idx; - pipe = pipe->next_odm_pipe; - } - break; - } - } - - return num_found; -} - -static struct pipe_ctx *assign_pipes_to_stream(struct dml2_context *ctx, struct dc_state *state, - const struct dc_stream_state *stream, - int odm_factor, - struct dc_plane_pipe_pool *pipe_pool, - const struct dc_state *existing_state) -{ - struct pipe_ctx *master_pipe; - unsigned int pipes_needed; - unsigned int pipes_assigned; - unsigned int pipes[MAX_PIPES] = {0}; - unsigned int next_pipe_to_assign; - int odm_slice; - - pipes_needed = odm_factor; - - master_pipe = find_master_pipe_of_stream(ctx, state, stream->stream_id); - ASSERT(master_pipe); - - pipes_assigned = find_pipes_assigned_to_stream(ctx, state, stream->stream_id, pipes); - - find_more_free_pipes(ctx, state, stream->stream_id, pipes, &pipes_assigned, pipes_needed - pipes_assigned, existing_state); - - ASSERT(pipes_assigned == pipes_needed); - - next_pipe_to_assign = 0; - for (odm_slice = 0; odm_slice < odm_factor; odm_slice++) - pipe_pool->pipes_assigned_to_plane[odm_slice][0] = pipes[next_pipe_to_assign++]; - - pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine = 1; - pipe_pool->num_pipes_assigned_to_plane_for_odm_combine = odm_factor; - - return master_pipe; -} - -static struct pipe_ctx *assign_pipes_to_plane(struct dml2_context *ctx, struct dc_state *state, - const struct dc_stream_state *stream, - const struct dc_plane_state *plane, - int odm_factor, - int mpc_factor, - int plane_index, - struct dc_plane_pipe_pool *pipe_pool, - const struct dc_state *existing_state) -{ - struct pipe_ctx *master_pipe = NULL; - unsigned int plane_id; - unsigned int pipes_needed; - unsigned int pipes_assigned; - unsigned int pipes[MAX_PIPES] = {0}; - unsigned int next_pipe_to_assign; - int odm_slice, mpc_slice; - - if (!get_plane_id(ctx, state, plane, stream->stream_id, plane_index, &plane_id)) { - ASSERT(false); - return master_pipe; - } - - pipes_needed = mpc_factor * odm_factor; - - master_pipe = find_master_pipe_of_plane(ctx, state, plane_id); - ASSERT(master_pipe); - - pipes_assigned = find_pipes_assigned_to_plane(ctx, state, plane_id, pipes); - - find_more_pipes_for_stream(ctx, state, stream->stream_id, pipes, &pipes_assigned, pipes_needed - pipes_assigned, existing_state); - - ASSERT(pipes_assigned >= pipes_needed); - - next_pipe_to_assign = 0; - for (odm_slice = 0; odm_slice < odm_factor; odm_slice++) - for (mpc_slice = 0; mpc_slice < mpc_factor; mpc_slice++) - pipe_pool->pipes_assigned_to_plane[odm_slice][mpc_slice] = pipes[next_pipe_to_assign++]; - - pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine = mpc_factor; - pipe_pool->num_pipes_assigned_to_plane_for_odm_combine = odm_factor; - - return master_pipe; -} - -static bool is_pipe_used(const struct dc_plane_pipe_pool *pool, unsigned int pipe_idx) -{ - int i, j; - - for (i = 0; i < pool->num_pipes_assigned_to_plane_for_odm_combine; i++) { - for (j = 0; j < pool->num_pipes_assigned_to_plane_for_mpcc_combine; j++) { - if (pool->pipes_assigned_to_plane[i][j] == pipe_idx && pool->pipe_used[i][j]) - return true; - } - } - - return false; -} - -static void free_pipe(struct pipe_ctx *pipe) -{ - memset(pipe, 0, sizeof(struct pipe_ctx)); -} - -static void free_unused_pipes_for_plane(struct dml2_context *ctx, struct dc_state *state, - const struct dc_plane_state *plane, const struct dc_plane_pipe_pool *pool, unsigned int stream_id, int plane_index) -{ - int i; - bool is_plane_duplicate = ctx->v20.scratch.plane_duplicate_exists; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - if (state->res_ctx.pipe_ctx[i].plane_state == plane && - state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id && - (!is_plane_duplicate || - ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[state->res_ctx.pipe_ctx[i].pipe_idx] == plane_index) && - !is_pipe_used(pool, state->res_ctx.pipe_ctx[i].pipe_idx)) { - free_pipe(&state->res_ctx.pipe_ctx[i]); - } - } -} - -static void remove_pipes_from_blend_trees(struct dml2_context *ctx, struct dc_state *state, struct dc_plane_pipe_pool *pipe_pool, unsigned int odm_slice) -{ - struct pipe_ctx *pipe; - int i; - - for (i = 0; i < pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine; i++) { - pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][0]]; - if (pipe->top_pipe) - pipe->top_pipe->bottom_pipe = pipe->bottom_pipe; - - if (pipe->bottom_pipe) - pipe->bottom_pipe = pipe->top_pipe; - - pipe_pool->pipe_used[odm_slice][i] = true; - } -} - -static void map_pipes_for_stream(struct dml2_context *ctx, struct dc_state *state, const struct dc_stream_state *stream, - struct dc_pipe_mapping_scratch *scratch, const struct dc_state *existing_state) -{ - int odm_slice_index; - struct pipe_ctx *master_pipe = NULL; - - - master_pipe = assign_pipes_to_stream(ctx, state, stream, scratch->odm_info.odm_factor, &scratch->pipe_pool, existing_state); - sort_pipes_for_splitting(&scratch->pipe_pool); - - for (odm_slice_index = 0; odm_slice_index < scratch->odm_info.odm_factor; odm_slice_index++) { - remove_pipes_from_blend_trees(ctx, state, &scratch->pipe_pool, odm_slice_index); - - add_odm_slice_to_odm_tree(ctx, state, scratch, odm_slice_index); - - ctx->config.callbacks.acquire_secondary_pipe_for_mpc_odm(ctx->config.callbacks.dc, state, - master_pipe, &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][0]], true); - } -} - -static void map_pipes_for_plane(struct dml2_context *ctx, struct dc_state *state, const struct dc_stream_state *stream, const struct dc_plane_state *plane, - int plane_index, struct dc_pipe_mapping_scratch *scratch, const struct dc_state *existing_state) -{ - int odm_slice_index; - unsigned int plane_id; - struct pipe_ctx *master_pipe = NULL; - int i; - - if (!get_plane_id(ctx, state, plane, stream->stream_id, plane_index, &plane_id)) { - ASSERT(false); - return; - } - - master_pipe = assign_pipes_to_plane(ctx, state, stream, plane, scratch->odm_info.odm_factor, - scratch->mpc_info.mpc_factor, plane_index, &scratch->pipe_pool, existing_state); - sort_pipes_for_splitting(&scratch->pipe_pool); - - for (odm_slice_index = 0; odm_slice_index < scratch->odm_info.odm_factor; odm_slice_index++) { - // Now we have a list of all pipes to be used for this plane/stream, now setup the tree. - scratch->odm_info.next_higher_pipe_for_odm_slice[odm_slice_index] = add_plane_to_blend_tree(ctx, state, - plane, - &scratch->pipe_pool, - odm_slice_index, - scratch->odm_info.next_higher_pipe_for_odm_slice[odm_slice_index]); - - add_odm_slice_to_odm_tree(ctx, state, scratch, odm_slice_index); - - for (i = 0; i < scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine; i++) { - - ctx->config.callbacks.acquire_secondary_pipe_for_mpc_odm(ctx->config.callbacks.dc, state, - master_pipe, &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][i]], true); - } - } - - free_unused_pipes_for_plane(ctx, state, plane, &scratch->pipe_pool, stream->stream_id, plane_index); -} - -static unsigned int get_target_mpc_factor(struct dml2_context *ctx, - struct dc_state *state, - const struct dml_display_cfg_st *disp_cfg, - struct dml2_dml_to_dc_pipe_mapping *mapping, - const struct dc_stream_status *status, - const struct dc_stream_state *stream, - int plane_idx) -{ - unsigned int plane_id; - unsigned int cfg_idx; - unsigned int mpc_factor; - - if (ctx->architecture == dml2_architecture_20) { - get_plane_id(ctx, state, status->plane_states[plane_idx], - stream->stream_id, plane_idx, &plane_id); - cfg_idx = find_disp_cfg_idx_by_plane_id(mapping, plane_id); - mpc_factor = (unsigned int)disp_cfg->hw.DPPPerSurface[cfg_idx]; - } else if (ctx->architecture == dml2_architecture_21) { - if (ctx->config.svp_pstate.callbacks.get_stream_subvp_type(state, stream) == SUBVP_PHANTOM) { - struct dc_stream_state *main_stream; - struct dc_stream_status *main_stream_status; - - /* get stream id of main stream */ - main_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(state, stream); - if (!main_stream) { - ASSERT(false); - return 1; - } - - main_stream_status = ctx->config.callbacks.get_stream_status(state, main_stream); - if (!main_stream_status) { - ASSERT(false); - return 1; - } - - /* get plane id for associated main plane */ - get_plane_id(ctx, state, main_stream_status->plane_states[plane_idx], - main_stream->stream_id, plane_idx, &plane_id); - } else { - get_plane_id(ctx, state, status->plane_states[plane_idx], - stream->stream_id, plane_idx, &plane_id); - } - - cfg_idx = find_disp_cfg_idx_by_plane_id(mapping, plane_id); - mpc_factor = ctx->v21.mode_programming.programming->plane_programming[cfg_idx].num_dpps_required; - } else { - mpc_factor = 1; - ASSERT(false); - } - - /* For stereo timings, we need to pipe split */ - if (dml2_is_stereo_timing(stream)) - mpc_factor = 2; - - return mpc_factor; -} - -static unsigned int get_target_odm_factor( - const struct dml2_context *ctx, - struct dc_state *state, - const struct dml_display_cfg_st *disp_cfg, - struct dml2_dml_to_dc_pipe_mapping *mapping, - const struct dc_stream_state *stream) -{ - unsigned int cfg_idx; - - if (ctx->architecture == dml2_architecture_20) { - cfg_idx = find_disp_cfg_idx_by_stream_id( - mapping, stream->stream_id); - switch (disp_cfg->hw.ODMMode[cfg_idx]) { - case dml_odm_mode_bypass: - return 1; - case dml_odm_mode_combine_2to1: - return 2; - case dml_odm_mode_combine_4to1: - return 4; - default: - break; - } - } else if (ctx->architecture == dml2_architecture_21) { - if (ctx->config.svp_pstate.callbacks.get_stream_subvp_type(state, stream) == SUBVP_PHANTOM) { - struct dc_stream_state *main_stream; - - /* get stream id of main stream */ - main_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(state, stream); - if (!main_stream) - goto failed; - - /* get cfg idx for associated main stream */ - cfg_idx = find_disp_cfg_idx_by_stream_id( - mapping, main_stream->stream_id); - } else { - cfg_idx = find_disp_cfg_idx_by_stream_id( - mapping, stream->stream_id); - } - - return ctx->v21.mode_programming.programming->stream_programming[cfg_idx].num_odms_required; - } - -failed: - ASSERT(false); - return 1; -} - -static unsigned int get_source_odm_factor(const struct dml2_context *ctx, - struct dc_state *state, - const struct dc_stream_state *stream) -{ - struct pipe_ctx *otg_master = ctx->config.callbacks.get_otg_master_for_stream(&state->res_ctx, stream); - - if (!otg_master) - return 0; - - return ctx->config.callbacks.get_odm_slice_count(otg_master); -} - -static unsigned int get_source_mpc_factor(const struct dml2_context *ctx, - struct dc_state *state, - const struct dc_plane_state *plane) -{ - struct pipe_ctx *dpp_pipes[MAX_PIPES] = {0}; - int dpp_pipe_count = ctx->config.callbacks.get_dpp_pipes_for_plane(plane, - &state->res_ctx, dpp_pipes); - - ASSERT(dpp_pipe_count > 0); - return ctx->config.callbacks.get_mpc_slice_count(dpp_pipes[0]); -} - - -static void populate_mpc_factors_for_stream( - struct dml2_context *ctx, - const struct dml_display_cfg_st *disp_cfg, - struct dml2_dml_to_dc_pipe_mapping *mapping, - struct dc_state *state, - unsigned int stream_idx, - struct dml2_pipe_combine_factor odm_factor, - struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) -{ - const struct dc_stream_status *status = &state->stream_status[stream_idx]; - int i; - - for (i = 0; i < status->plane_count; i++) { - mpc_factors[i].source = get_source_mpc_factor(ctx, state, status->plane_states[i]); - mpc_factors[i].target = (odm_factor.target == 1) ? - get_target_mpc_factor(ctx, state, disp_cfg, mapping, status, state->streams[stream_idx], i) : 1; - } -} - -static void populate_odm_factors(const struct dml2_context *ctx, - const struct dml_display_cfg_st *disp_cfg, - struct dml2_dml_to_dc_pipe_mapping *mapping, - struct dc_state *state, - struct dml2_pipe_combine_factor odm_factors[MAX_PIPES]) -{ - int i; - - for (i = 0; i < state->stream_count; i++) { - odm_factors[i].source = get_source_odm_factor(ctx, state, state->streams[i]); - odm_factors[i].target = get_target_odm_factor( - ctx, state, disp_cfg, mapping, state->streams[i]); - } -} - -static bool unmap_dc_pipes_for_stream(struct dml2_context *ctx, - struct dc_state *state, - const struct dc_state *existing_state, - const struct dc_stream_state *stream, - const struct dc_stream_status *status, - struct dml2_pipe_combine_factor odm_factor, - struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) -{ - int plane_idx; - bool result = true; - - for (plane_idx = 0; plane_idx < status->plane_count; plane_idx++) - if (mpc_factors[plane_idx].target < mpc_factors[plane_idx].source) - result &= ctx->config.callbacks.update_pipes_for_plane_with_slice_count( - state, - existing_state, - ctx->config.callbacks.dc->res_pool, - status->plane_states[plane_idx], - mpc_factors[plane_idx].target); - if (odm_factor.target < odm_factor.source) - result &= ctx->config.callbacks.update_pipes_for_stream_with_slice_count( - state, - existing_state, - ctx->config.callbacks.dc->res_pool, - stream, - odm_factor.target); - return result; -} - -static bool map_dc_pipes_for_stream(struct dml2_context *ctx, - struct dc_state *state, - const struct dc_state *existing_state, - const struct dc_stream_state *stream, - const struct dc_stream_status *status, - struct dml2_pipe_combine_factor odm_factor, - struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) -{ - int plane_idx; - bool result = true; - - for (plane_idx = 0; plane_idx < status->plane_count; plane_idx++) - if (mpc_factors[plane_idx].target > mpc_factors[plane_idx].source) - result &= ctx->config.callbacks.update_pipes_for_plane_with_slice_count( - state, - existing_state, - ctx->config.callbacks.dc->res_pool, - status->plane_states[plane_idx], - mpc_factors[plane_idx].target); - if (odm_factor.target > odm_factor.source) - result &= ctx->config.callbacks.update_pipes_for_stream_with_slice_count( - state, - existing_state, - ctx->config.callbacks.dc->res_pool, - stream, - odm_factor.target); - return result; -} - -static bool map_dc_pipes_with_callbacks(struct dml2_context *ctx, - struct dc_state *state, - const struct dml_display_cfg_st *disp_cfg, - struct dml2_dml_to_dc_pipe_mapping *mapping, - const struct dc_state *existing_state) -{ - int i; - bool result = true; - - populate_odm_factors(ctx, disp_cfg, mapping, state, ctx->pipe_combine_scratch.odm_factors); - for (i = 0; i < state->stream_count; i++) - populate_mpc_factors_for_stream(ctx, disp_cfg, mapping, state, - i, ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); - for (i = 0; i < state->stream_count; i++) - result &= unmap_dc_pipes_for_stream(ctx, state, existing_state, state->streams[i], - &state->stream_status[i], ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); - for (i = 0; i < state->stream_count; i++) - result &= map_dc_pipes_for_stream(ctx, state, existing_state, state->streams[i], - &state->stream_status[i], ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); - - return result; -} - -bool dml2_map_dc_pipes(struct dml2_context *ctx, struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, struct dml2_dml_to_dc_pipe_mapping *mapping, const struct dc_state *existing_state) -{ - int stream_index, plane_index, i; - - unsigned int stream_disp_cfg_index; - unsigned int plane_disp_cfg_index; - unsigned int disp_cfg_index_max; - - unsigned int plane_id; - unsigned int stream_id; - - const unsigned int *ODMMode, *DPPPerSurface; - unsigned int odm_mode_array[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}, dpp_per_surface_array[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; - struct dc_pipe_mapping_scratch scratch; - - if (ctx->config.map_dc_pipes_with_callbacks) - return map_dc_pipes_with_callbacks( - ctx, state, disp_cfg, mapping, existing_state); - - if (ctx->architecture == dml2_architecture_21) { - /* - * Extract ODM and DPP outputs from DML2.1 and map them in an array as required for pipe mapping in dml2_map_dc_pipes. - * As data cannot be directly extracted in const pointers, assign these arrays to const pointers before proceeding to - * maximize the reuse of existing code. Const pointers are required because dml2.0 dml_display_cfg_st is const. - * - */ - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - odm_mode_array[i] = ctx->v21.mode_programming.programming->stream_programming[i].num_odms_required; - dpp_per_surface_array[i] = ctx->v21.mode_programming.programming->plane_programming[i].num_dpps_required; - } - - ODMMode = (const unsigned int *)odm_mode_array; - DPPPerSurface = (const unsigned int *)dpp_per_surface_array; - disp_cfg_index_max = __DML2_WRAPPER_MAX_STREAMS_PLANES__; - } else { - ODMMode = (unsigned int *)disp_cfg->hw.ODMMode; - DPPPerSurface = disp_cfg->hw.DPPPerSurface; - disp_cfg_index_max = __DML_NUM_PLANES__; - } - - for (stream_index = 0; stream_index < state->stream_count; stream_index++) { - memset(&scratch, 0, sizeof(struct dc_pipe_mapping_scratch)); - - stream_id = state->streams[stream_index]->stream_id; - stream_disp_cfg_index = find_disp_cfg_idx_by_stream_id(mapping, stream_id); - if (stream_disp_cfg_index >= disp_cfg_index_max) - continue; - - if (ctx->architecture == dml2_architecture_20) { - if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_bypass) { - scratch.odm_info.odm_factor = 1; - } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_2to1) { - scratch.odm_info.odm_factor = 2; - } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_4to1) { - scratch.odm_info.odm_factor = 4; - } else { - ASSERT(false); - scratch.odm_info.odm_factor = 1; - } - } else if (ctx->architecture == dml2_architecture_21) { - /* After DML2.1 update, ODM interpretation needs to change and is no longer same as for DML2.0. - * This is not an issue with new resource management logic. This block ensure backcompat - * with legacy pipe management with updated DML. - * */ - if (ODMMode[stream_disp_cfg_index] == 1) { - scratch.odm_info.odm_factor = 1; - } else if (ODMMode[stream_disp_cfg_index] == 2) { - scratch.odm_info.odm_factor = 2; - } else if (ODMMode[stream_disp_cfg_index] == 4) { - scratch.odm_info.odm_factor = 4; - } else { - ASSERT(false); - scratch.odm_info.odm_factor = 1; - } - } - calculate_odm_slices(state->streams[stream_index], scratch.odm_info.odm_factor, scratch.odm_info.odm_slice_end_x); - - // If there are no planes, you still want to setup ODM... - if (state->stream_status[stream_index].plane_count == 0) { - map_pipes_for_stream(ctx, state, state->streams[stream_index], &scratch, existing_state); - } - - for (plane_index = 0; plane_index < state->stream_status[stream_index].plane_count; plane_index++) { - // Planes are ordered top to bottom. - if (get_plane_id(ctx, state, state->stream_status[stream_index].plane_states[plane_index], - stream_id, plane_index, &plane_id)) { - plane_disp_cfg_index = find_disp_cfg_idx_by_plane_id(mapping, plane_id); - - // Setup mpc_info for this plane - scratch.mpc_info.prev_odm_pipe = NULL; - if (scratch.odm_info.odm_factor == 1 && plane_disp_cfg_index < disp_cfg_index_max) { - // If ODM combine is not inuse, then the number of pipes - // per plane is determined by MPC combine factor - scratch.mpc_info.mpc_factor = DPPPerSurface[plane_disp_cfg_index]; - - //For stereo timings, we need to pipe split - if (dml2_is_stereo_timing(state->streams[stream_index])) - scratch.mpc_info.mpc_factor = 2; - } else { - // If ODM combine is enabled, then we use at most 1 pipe per - // odm slice per plane, i.e. MPC combine is never used - scratch.mpc_info.mpc_factor = 1; - } - - ASSERT(scratch.odm_info.odm_factor * scratch.mpc_info.mpc_factor > 0); - - // Clear the pool assignment scratch (which is per plane) - memset(&scratch.pipe_pool, 0, sizeof(struct dc_plane_pipe_pool)); - - map_pipes_for_plane(ctx, state, state->streams[stream_index], - state->stream_status[stream_index].plane_states[plane_index], plane_index, &scratch, existing_state); - } else { - // Plane ID cannot be generated, therefore no DML mapping can be performed. - ASSERT(false); - } - } - - } - - if (!validate_pipe_assignment(ctx, state, disp_cfg, mapping)) - ASSERT(false); - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state) { - if (!ctx->config.callbacks.build_scaling_params(pipe)) { - ASSERT(false); - } - } - - if (ctx->config.callbacks.build_test_pattern_params && - pipe->stream && - pipe->prev_odm_pipe == NULL && - pipe->top_pipe == NULL) - ctx->config.callbacks.build_test_pattern_params(&state->res_ctx, pipe); - } - - return true; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h deleted file mode 100644 index 1538b708d8be..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h +++ /dev/null @@ -1,52 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DML2_DC_RESOURCE_MGMT_H__ -#define __DML2_DC_RESOURCE_MGMT_H__ - -#include "dml2_dc_types.h" - -struct dml2_context; -struct dml2_dml_to_dc_pipe_mapping; -struct dml_display_cfg_st; - -/* - * dml2_map_dc_pipes - Creates a pipe linkage in dc_state based on current display config. - * @ctx: Input dml2 context - * @state: Current dc_state to be updated. - * @disp_cfg: Current display config. - * @mapping: Pipe mapping logic structure to keep a track of pipes to be used. - * - * Based on ODM and DPPPersurface outputs calculated by the DML for the current display - * config, create a pipe linkage in dc_state which is then used by DC core. - * Make this function generic to be used by multiple DML versions. - * - * Return: True if pipe mapping and linking is successful, false otherwise. - */ - -bool dml2_map_dc_pipes(struct dml2_context *ctx, struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, struct dml2_dml_to_dc_pipe_mapping *mapping, const struct dc_state *existing_state); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h deleted file mode 100644 index 7ca7f2a743c2..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h +++ /dev/null @@ -1,43 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - - -/* - * Wrapper header for externally defined types from DC. These types come from - * dc headers when building DML2 as part of DC, but are defined here when building - * DML2 as a standalone library (such as for unit testing). - */ - -#ifndef __DML2_DC_TYPES_H__ -#define __DML2_DC_TYPES_H__ - -#include "resource.h" -#include "core_types.h" -#include "dsc.h" -#include "clk_mgr.h" -#include "dc_state_priv.h" - -#endif //__DML2_DC_TYPES_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h deleted file mode 100644 index 140ec01545db..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h +++ /dev/null @@ -1,157 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DML2_INTERNAL_TYPES_H__ -#define __DML2_INTERNAL_TYPES_H__ - -#include "dml2_dc_types.h" -#include "display_mode_core.h" -#include "dml2_wrapper.h" -#include "dml2_policy.h" - -#include "dml_top.h" -#include "dml21_wrapper.h" - -struct dml2_wrapper_optimize_configuration_params { - struct display_mode_lib_st *dml_core_ctx; - struct dml2_configuration_options *config; - struct ip_params_st *ip_params; - struct dml_display_cfg_st *cur_display_config; - struct dml_display_cfg_st *new_display_config; - const struct dml_mode_support_info_st *cur_mode_support_info; - struct dml_mode_eval_policy_st *cur_policy; - struct dml_mode_eval_policy_st *new_policy; -}; - -struct dml2_calculate_lowest_supported_state_for_temp_read_scratch { - struct dml_mode_support_info_st evaluation_info; - dml_float_t uclk_change_latencies[__DML_MAX_STATE_ARRAY_SIZE__]; - struct dml_display_cfg_st cur_display_config; - struct dml_display_cfg_st new_display_config; - struct dml_mode_eval_policy_st new_policy; - struct dml_mode_eval_policy_st cur_policy; -}; - -struct dml2_create_scratch { - struct dml2_policy_build_synthetic_soc_states_scratch build_synthetic_socbb_scratch; - struct soc_states_st in_states; -}; - -struct dml2_calculate_rq_and_dlg_params_scratch { - struct _vcs_dpi_dml_display_rq_regs_st rq_regs; - struct _vcs_dpi_dml_display_dlg_regs_st disp_dlg_regs; - struct _vcs_dpi_dml_display_ttu_regs_st disp_ttu_regs; -}; - -#define __DML2_WRAPPER_MAX_STREAMS_PLANES__ 6 - -struct dml2_dml_to_dc_pipe_mapping { - unsigned int disp_cfg_to_stream_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - bool disp_cfg_to_stream_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - unsigned int disp_cfg_to_plane_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - bool disp_cfg_to_plane_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - unsigned int dml_pipe_idx_to_stream_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - bool dml_pipe_idx_to_stream_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - unsigned int dml_pipe_idx_to_plane_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - bool dml_pipe_idx_to_plane_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - unsigned int dml_pipe_idx_to_plane_index[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; - bool dml_pipe_idx_to_plane_index_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; -}; - -struct dml2_wrapper_scratch { - struct dml_display_cfg_st cur_display_config; - struct dml_display_cfg_st new_display_config; - struct dml_mode_eval_policy_st new_policy; - struct dml_mode_eval_policy_st cur_policy; - struct dml_mode_support_info_st mode_support_info; - struct dml_mode_support_ex_params_st mode_support_params; - - struct dummy_pstate_entry dummy_pstate_table[4]; - - struct dml2_create_scratch create_scratch; - struct dml2_calculate_lowest_supported_state_for_temp_read_scratch dml2_calculate_lowest_supported_state_for_temp_read_scratch; - struct dml2_calculate_rq_and_dlg_params_scratch calculate_rq_and_dlg_params_scratch; - - struct dml2_wrapper_optimize_configuration_params optimize_configuration_params; - struct dml2_policy_build_synthetic_soc_states_params build_synthetic_socbb_params; - - struct dml2_dml_to_dc_pipe_mapping dml_to_dc_pipe_mapping; - bool enable_flexible_pipe_mapping; - bool plane_duplicate_exists; - int hpo_stream_to_link_encoder_mapping[MAX_HPO_DP2_ENCODERS]; -}; - -struct dml2_helper_det_policy_scratch { - int dpps_per_surface[MAX_PLANES]; -}; - -enum dml2_architecture { - dml2_architecture_20, - dml2_architecture_21 -}; - -struct prepare_mcache_programming_locals { - struct dml2_build_mcache_programming_in_out build_mcache_programming_params; -}; - -struct dml21_wrapper_scratch { - struct prepare_mcache_programming_locals prepare_mcache_locals; - struct pipe_ctx temp_pipe; -}; - -struct dml2_pipe_combine_factor { - unsigned int source; - unsigned int target; -}; - -struct dml2_pipe_combine_scratch { - struct dml2_pipe_combine_factor odm_factors[MAX_PIPES]; - struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES][MAX_PIPES]; -}; - -struct dml2_context { - enum dml2_architecture architecture; - struct dml2_configuration_options config; - struct dml2_helper_det_policy_scratch det_helper_scratch; - struct dml2_pipe_combine_scratch pipe_combine_scratch; - union { - struct { - struct display_mode_lib_st dml_core_ctx; - struct dml2_wrapper_scratch scratch; - struct dcn_watermarks g6_temp_read_watermark_set; - } v20; - struct { - struct dml21_wrapper_scratch scratch; - struct dml2_initialize_instance_in_out dml_init; - struct dml2_display_cfg display_config; - struct dml2_check_mode_supported_in_out mode_support; - struct dml2_build_mode_programming_in_out mode_programming; - struct dml2_dml_to_dc_pipe_mapping dml_to_dc_pipe_mapping; - } v21; - }; -}; - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c deleted file mode 100644 index c59f825cfae9..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c +++ /dev/null @@ -1,910 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "dml2_dc_types.h" -#include "dml2_internal_types.h" -#include "dml2_utils.h" -#include "dml2_mall_phantom.h" - -unsigned int dml2_helper_calculate_num_ways_for_subvp(struct dml2_context *ctx, struct dc_state *context) -{ - uint32_t num_ways = 0; - uint32_t bytes_per_pixel = 0; - uint32_t cache_lines_used = 0; - uint32_t lines_per_way = 0; - uint32_t total_cache_lines = 0; - uint32_t bytes_in_mall = 0; - uint32_t num_mblks = 0; - uint32_t cache_lines_per_plane = 0; - uint32_t i = 0; - uint32_t mblk_width = 0; - uint32_t mblk_height = 0; - uint32_t full_vp_width_blk_aligned = 0; - uint32_t mall_alloc_width_blk_aligned = 0; - uint32_t mall_alloc_height_blk_aligned = 0; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - // Find the phantom pipes - if (pipe->stream && pipe->plane_state && !pipe->top_pipe && !pipe->prev_odm_pipe && - ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { - bytes_per_pixel = pipe->plane_state->format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 ? 8 : 4; - mblk_width = ctx->config.mall_cfg.mblk_width_pixels; - mblk_height = bytes_per_pixel == 4 ? mblk_width = ctx->config.mall_cfg.mblk_height_4bpe_pixels : ctx->config.mall_cfg.mblk_height_8bpe_pixels; - - /* full_vp_width_blk_aligned = FLOOR(vp_x_start + full_vp_width + blk_width - 1, blk_width) - - * FLOOR(vp_x_start, blk_width) - */ - full_vp_width_blk_aligned = ((pipe->plane_res.scl_data.viewport.x + - pipe->plane_res.scl_data.viewport.width + mblk_width - 1) / mblk_width * mblk_width) + - (pipe->plane_res.scl_data.viewport.x / mblk_width * mblk_width); - - /* mall_alloc_width_blk_aligned_l/c = full_vp_width_blk_aligned_l/c */ - mall_alloc_width_blk_aligned = full_vp_width_blk_aligned; - - /* mall_alloc_height_blk_aligned_l/c = CEILING(sub_vp_height_l/c - 1, blk_height_l/c) + blk_height_l/c */ - mall_alloc_height_blk_aligned = (pipe->stream->timing.v_addressable - 1 + mblk_height - 1) / - mblk_height * mblk_height + mblk_height; - - /* full_mblk_width_ub_l/c = malldml2_mall_phantom.c_alloc_width_blk_aligned_l/c; - * full_mblk_height_ub_l/c = mall_alloc_height_blk_aligned_l/c; - * num_mblk_l/c = (full_mblk_width_ub_l/c / mblk_width_l/c) * (full_mblk_height_ub_l/c / mblk_height_l/c); - * (Should be divisible, but round up if not) - */ - num_mblks = ((mall_alloc_width_blk_aligned + mblk_width - 1) / mblk_width) * - ((mall_alloc_height_blk_aligned + mblk_height - 1) / mblk_height); - bytes_in_mall = num_mblks * ctx->config.mall_cfg.mblk_size_bytes; - // cache lines used is total bytes / cache_line size. Add +2 for worst case alignment - // (MALL is 64-byte aligned) - cache_lines_per_plane = bytes_in_mall / ctx->config.mall_cfg.cache_line_size_bytes + 2; - - // For DCC we must cache the meat surface, so double cache lines required - if (pipe->plane_state->dcc.enable) - cache_lines_per_plane *= 2; - cache_lines_used += cache_lines_per_plane; - } - } - - total_cache_lines = ctx->config.mall_cfg.max_cab_allocation_bytes / ctx->config.mall_cfg.cache_line_size_bytes; - lines_per_way = total_cache_lines / ctx->config.mall_cfg.cache_num_ways; - num_ways = cache_lines_used / lines_per_way; - if (cache_lines_used % lines_per_way > 0) - num_ways++; - - return num_ways; -} - -static void merge_pipes_for_subvp(struct dml2_context *ctx, struct dc_state *context) -{ - int i; - - /* merge pipes if necessary */ - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - // For now merge all pipes for SubVP since pipe split case isn't supported yet - - /* if ODM merge we ignore mpc tree, mpo pipes will have their own flags */ - if (pipe->prev_odm_pipe) { - /*split off odm pipe*/ - pipe->prev_odm_pipe->next_odm_pipe = pipe->next_odm_pipe; - if (pipe->next_odm_pipe) - pipe->next_odm_pipe->prev_odm_pipe = pipe->prev_odm_pipe; - - pipe->bottom_pipe = NULL; - pipe->next_odm_pipe = NULL; - pipe->plane_state = NULL; - pipe->stream = NULL; - pipe->top_pipe = NULL; - pipe->prev_odm_pipe = NULL; - if (pipe->stream_res.dsc) - ctx->config.svp_pstate.callbacks.release_dsc(&context->res_ctx, ctx->config.svp_pstate.callbacks.dc->res_pool, &pipe->stream_res.dsc); - memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); - memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); - } else if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state) { - struct pipe_ctx *top_pipe = pipe->top_pipe; - struct pipe_ctx *bottom_pipe = pipe->bottom_pipe; - - top_pipe->bottom_pipe = bottom_pipe; - if (bottom_pipe) - bottom_pipe->top_pipe = top_pipe; - - pipe->top_pipe = NULL; - pipe->bottom_pipe = NULL; - pipe->plane_state = NULL; - pipe->stream = NULL; - memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); - memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); - } - } -} - -static bool all_pipes_have_stream_and_plane(struct dml2_context *ctx, const struct dc_state *context) -{ - int i; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (!pipe->stream) - continue; - - if (!pipe->plane_state) - return false; - } - return true; -} - -static bool mpo_in_use(const struct dc_state *context) -{ - int i; - - for (i = 0; i < context->stream_count; i++) { - if (context->stream_status[i].plane_count > 1) - return true; - } - return false; -} - -/* - * dcn32_get_num_free_pipes: Calculate number of free pipes - * - * This function assumes that a "used" pipe is a pipe that has - * both a stream and a plane assigned to it. - * - * @dc: current dc state - * @context: new dc state - * - * Return: - * Number of free pipes available in the context - */ -static unsigned int get_num_free_pipes(struct dml2_context *ctx, struct dc_state *state) -{ - unsigned int i; - unsigned int free_pipes = 0; - unsigned int num_pipes = 0; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - if (pipe->stream && !pipe->top_pipe) { - while (pipe) { - num_pipes++; - pipe = pipe->bottom_pipe; - } - } - } - - free_pipes = ctx->config.dcn_pipe_count - num_pipes; - return free_pipes; -} - -/* - * dcn32_assign_subvp_pipe: Function to decide which pipe will use Sub-VP. - * - * We enter this function if we are Sub-VP capable (i.e. enough pipes available) - * and regular P-State switching (i.e. VACTIVE/VBLANK) is not supported, or if - * we are forcing SubVP P-State switching on the current config. - * - * The number of pipes used for the chosen surface must be less than or equal to the - * number of free pipes available. - * - * In general we choose surfaces with the longest frame time first (better for SubVP + VBLANK). - * For multi-display cases the ActiveDRAMClockChangeMargin doesn't provide enough info on its own - * for determining which should be the SubVP pipe (need a way to determine if a pipe / plane doesn't - * support MCLK switching naturally [i.e. ACTIVE or VBLANK]). - * - * @param dc: current dc state - * @param context: new dc state - * @param index: [out] dc pipe index for the pipe chosen to have phantom pipes assigned - * - * Return: - * True if a valid pipe assignment was found for Sub-VP. Otherwise false. - */ -static bool assign_subvp_pipe(struct dml2_context *ctx, struct dc_state *context, unsigned int *index) -{ - unsigned int i, pipe_idx; - unsigned int max_frame_time = 0; - bool valid_assignment_found = false; - unsigned int free_pipes = 2; //dcn32_get_num_free_pipes(dc, context); - bool current_assignment_freesync = false; - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - - for (i = 0, pipe_idx = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - unsigned int num_pipes = 0; - unsigned int refresh_rate = 0; - - if (!pipe->stream) - continue; - - // Round up - refresh_rate = (pipe->stream->timing.pix_clk_100hz * 100 + - pipe->stream->timing.v_total * pipe->stream->timing.h_total - 1) - / (double)(pipe->stream->timing.v_total * pipe->stream->timing.h_total); - /* SubVP pipe candidate requirements: - * - Refresh rate < 120hz - * - Not able to switch in vactive naturally (switching in active means the - * DET provides enough buffer to hide the P-State switch latency -- trying - * to combine this with SubVP can cause issues with the scheduling). - */ - if (pipe->plane_state && !pipe->top_pipe && - ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_NONE && refresh_rate < 120 && - vba->ActiveDRAMClockChangeLatencyMarginPerState[vba->VoltageLevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]] <= 0) { - while (pipe) { - num_pipes++; - pipe = pipe->bottom_pipe; - } - - pipe = &context->res_ctx.pipe_ctx[i]; - if (num_pipes <= free_pipes) { - struct dc_stream_state *stream = pipe->stream; - unsigned int frame_us = (stream->timing.v_total * stream->timing.h_total / - (double)(stream->timing.pix_clk_100hz * 100)) * 1000000; - if (frame_us > max_frame_time && !stream->ignore_msa_timing_param) { - *index = i; - max_frame_time = frame_us; - valid_assignment_found = true; - current_assignment_freesync = false; - /* For the 2-Freesync display case, still choose the one with the - * longest frame time - */ - } else if (stream->ignore_msa_timing_param && (!valid_assignment_found || - (current_assignment_freesync && frame_us > max_frame_time))) { - *index = i; - valid_assignment_found = true; - current_assignment_freesync = true; - } - } - } - pipe_idx++; - } - return valid_assignment_found; -} - -/* - * enough_pipes_for_subvp: Function to check if there are "enough" pipes for SubVP. - * - * This function returns true if there are enough free pipes - * to create the required phantom pipes for any given stream - * (that does not already have phantom pipe assigned). - * - * e.g. For a 2 stream config where the first stream uses one - * pipe and the second stream uses 2 pipes (i.e. pipe split), - * this function will return true because there is 1 remaining - * pipe which can be used as the phantom pipe for the non pipe - * split pipe. - * - * @dc: current dc state - * @context: new dc state - * - * Return: - * True if there are enough free pipes to assign phantom pipes to at least one - * stream that does not already have phantom pipes assigned. Otherwise false. - */ -static bool enough_pipes_for_subvp(struct dml2_context *ctx, struct dc_state *state) -{ - unsigned int i, split_cnt, free_pipes; - unsigned int min_pipe_split = ctx->config.dcn_pipe_count + 1; // init as max number of pipes + 1 - bool subvp_possible = false; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - // Find the minimum pipe split count for non SubVP pipes - if (pipe->stream && !pipe->top_pipe && - ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_NONE) { - split_cnt = 0; - while (pipe) { - split_cnt++; - pipe = pipe->bottom_pipe; - } - - if (split_cnt < min_pipe_split) - min_pipe_split = split_cnt; - } - } - - free_pipes = get_num_free_pipes(ctx, state); - - // SubVP only possible if at least one pipe is being used (i.e. free_pipes - // should not equal to the pipe_count) - if (free_pipes >= min_pipe_split && free_pipes < ctx->config.dcn_pipe_count) - subvp_possible = true; - - return subvp_possible; -} - -/* - * subvp_subvp_schedulable: Determine if SubVP + SubVP config is schedulable - * - * High level algorithm: - * 1. Find longest microschedule length (in us) between the two SubVP pipes - * 2. Check if the worst case overlap (VBLANK in middle of ACTIVE) for both - * pipes still allows for the maximum microschedule to fit in the active - * region for both pipes. - * - * @dc: current dc state - * @context: new dc state - * - * Return: - * bool - True if the SubVP + SubVP config is schedulable, false otherwise - */ -static bool subvp_subvp_schedulable(struct dml2_context *ctx, struct dc_state *context) -{ - struct pipe_ctx *subvp_pipes[2]; - struct dc_stream_state *phantom = NULL; - uint32_t microschedule_lines = 0; - uint32_t index = 0; - uint32_t i; - uint32_t max_microschedule_us = 0; - int32_t vactive1_us, vactive2_us, vblank1_us, vblank2_us; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - uint32_t time_us = 0; - - /* Loop to calculate the maximum microschedule time between the two SubVP pipes, - * and also to store the two main SubVP pipe pointers in subvp_pipes[2]. - */ - if (pipe->stream && pipe->plane_state && !pipe->top_pipe && - ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) { - phantom = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, pipe->stream); - microschedule_lines = (phantom->timing.v_total - phantom->timing.v_front_porch) + - phantom->timing.v_addressable; - - // Round up when calculating microschedule time (+ 1 at the end) - time_us = (microschedule_lines * phantom->timing.h_total) / - (double)(phantom->timing.pix_clk_100hz * 100) * 1000000 + - ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us + - ctx->config.svp_pstate.subvp_fw_processing_delay_us + 1; - if (time_us > max_microschedule_us) - max_microschedule_us = time_us; - - subvp_pipes[index] = pipe; - index++; - - // Maximum 2 SubVP pipes - if (index == 2) - break; - } - } - vactive1_us = ((subvp_pipes[0]->stream->timing.v_addressable * subvp_pipes[0]->stream->timing.h_total) / - (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vactive2_us = ((subvp_pipes[1]->stream->timing.v_addressable * subvp_pipes[1]->stream->timing.h_total) / - (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vblank1_us = (((subvp_pipes[0]->stream->timing.v_total - subvp_pipes[0]->stream->timing.v_addressable) * - subvp_pipes[0]->stream->timing.h_total) / - (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; - vblank2_us = (((subvp_pipes[1]->stream->timing.v_total - subvp_pipes[1]->stream->timing.v_addressable) * - subvp_pipes[1]->stream->timing.h_total) / - (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; - - if ((vactive1_us - vblank2_us) / 2 > max_microschedule_us && - (vactive2_us - vblank1_us) / 2 > max_microschedule_us) - return true; - - return false; -} - -/* - * dml2_svp_drr_schedulable: Determine if SubVP + DRR config is schedulable - * - * High level algorithm: - * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe - * 2. Determine the frame time for the DRR display when adding required margin for MCLK switching - * (the margin is equal to the MALL region + DRR margin (500us)) - * 3.If (SubVP Active - Prefetch > Stretched DRR frame + max(MALL region, Stretched DRR frame)) - * then report the configuration as supported - * - * @dc: current dc state - * @context: new dc state - * @drr_pipe: DRR pipe_ctx for the SubVP + DRR config - * - * Return: - * bool - True if the SubVP + DRR config is schedulable, false otherwise - */ -bool dml2_svp_drr_schedulable(struct dml2_context *ctx, struct dc_state *context, struct dc_crtc_timing *drr_timing) -{ - bool schedulable = false; - uint32_t i; - struct pipe_ctx *pipe = NULL; - struct dc_crtc_timing *main_timing = NULL; - struct dc_crtc_timing *phantom_timing = NULL; - struct dc_stream_state *phantom_stream; - int16_t prefetch_us = 0; - int16_t mall_region_us = 0; - int16_t drr_frame_us = 0; // nominal frame time - int16_t subvp_active_us = 0; - int16_t stretched_drr_us = 0; - int16_t drr_stretched_vblank_us = 0; - int16_t max_vblank_mallregion = 0; - - // Find SubVP pipe - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - - // We check for master pipe, but it shouldn't matter since we only need - // the pipe for timing info (stream should be same for any pipe splits) - if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) - continue; - - // Find the SubVP pipe - if (ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) - break; - } - - phantom_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, pipe->stream); - main_timing = &pipe->stream->timing; - phantom_timing = &phantom_stream->timing; - prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + - ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us; - subvp_active_us = main_timing->v_addressable * main_timing->h_total / - (double)(main_timing->pix_clk_100hz * 100) * 1000000; - drr_frame_us = drr_timing->v_total * drr_timing->h_total / - (double)(drr_timing->pix_clk_100hz * 100) * 1000000; - // P-State allow width and FW delays already included phantom_timing->v_addressable - mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; - stretched_drr_us = drr_frame_us + mall_region_us + SUBVP_DRR_MARGIN_US; - drr_stretched_vblank_us = (drr_timing->v_total - drr_timing->v_addressable) * drr_timing->h_total / - (double)(drr_timing->pix_clk_100hz * 100) * 1000000 + (stretched_drr_us - drr_frame_us); - max_vblank_mallregion = drr_stretched_vblank_us > mall_region_us ? drr_stretched_vblank_us : mall_region_us; - - /* We consider SubVP + DRR schedulable if the stretched frame duration of the DRR display (i.e. the - * highest refresh rate + margin that can support UCLK P-State switch) passes the static analysis - * for VBLANK: (VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, - * and the max of (VBLANK blanking time, MALL region)). - */ - if (stretched_drr_us < (1 / (double)drr_timing->min_refresh_in_uhz) * 1000000 * 1000000 && - subvp_active_us - prefetch_us - stretched_drr_us - max_vblank_mallregion > 0) - schedulable = true; - - return schedulable; -} - - -/* - * subvp_vblank_schedulable: Determine if SubVP + VBLANK config is schedulable - * - * High level algorithm: - * 1. Get timing for SubVP pipe, phantom pipe, and VBLANK pipe - * 2. If (SubVP Active - Prefetch > Vblank Frame Time + max(MALL region, Vblank blanking time)) - * then report the configuration as supported - * 3. If the VBLANK display is DRR, then take the DRR static schedulability path - * - * @dc: current dc state - * @context: new dc state - * - * Return: - * bool - True if the SubVP + VBLANK/DRR config is schedulable, false otherwise - */ -static bool subvp_vblank_schedulable(struct dml2_context *ctx, struct dc_state *context) -{ - struct pipe_ctx *pipe = NULL; - struct pipe_ctx *subvp_pipe = NULL; - bool found = false; - bool schedulable = false; - uint32_t i = 0; - uint8_t vblank_index = 0; - uint16_t prefetch_us = 0; - uint16_t mall_region_us = 0; - uint16_t vblank_frame_us = 0; - uint16_t subvp_active_us = 0; - uint16_t vblank_blank_us = 0; - uint16_t max_vblank_mallregion = 0; - struct dc_crtc_timing *main_timing = NULL; - struct dc_crtc_timing *phantom_timing = NULL; - struct dc_crtc_timing *vblank_timing = NULL; - struct dc_stream_state *phantom_stream; - enum mall_stream_type pipe_mall_type; - - /* For SubVP + VBLANK/DRR cases, we assume there can only be - * a single VBLANK/DRR display. If DML outputs SubVP + VBLANK - * is supported, it is either a single VBLANK case or two VBLANK - * displays which are synchronized (in which case they have identical - * timings). - */ - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - pipe = &context->res_ctx.pipe_ctx[i]; - pipe_mall_type = ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe); - - // We check for master pipe, but it shouldn't matter since we only need - // the pipe for timing info (stream should be same for any pipe splits) - if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) - continue; - - if (!found && pipe_mall_type == SUBVP_NONE) { - // Found pipe which is not SubVP or Phantom (i.e. the VBLANK pipe). - vblank_index = i; - found = true; - } - - if (!subvp_pipe && pipe_mall_type == SUBVP_MAIN) - subvp_pipe = pipe; - } - // Use ignore_msa_timing_param flag to identify as DRR - if (found && context->res_ctx.pipe_ctx[vblank_index].stream->ignore_msa_timing_param) { - // SUBVP + DRR case - schedulable = dml2_svp_drr_schedulable(ctx, context, &context->res_ctx.pipe_ctx[vblank_index].stream->timing); - } else if (found) { - phantom_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, subvp_pipe->stream); - main_timing = &subvp_pipe->stream->timing; - phantom_timing = &phantom_stream->timing; - vblank_timing = &context->res_ctx.pipe_ctx[vblank_index].stream->timing; - // Prefetch time is equal to VACTIVE + BP + VSYNC of the phantom pipe - // Also include the prefetch end to mallstart delay time - prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + - ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us; - // P-State allow width and FW delays already included phantom_timing->v_addressable - mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / - (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; - vblank_frame_us = vblank_timing->v_total * vblank_timing->h_total / - (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; - vblank_blank_us = (vblank_timing->v_total - vblank_timing->v_addressable) * vblank_timing->h_total / - (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; - subvp_active_us = main_timing->v_addressable * main_timing->h_total / - (double)(main_timing->pix_clk_100hz * 100) * 1000000; - max_vblank_mallregion = vblank_blank_us > mall_region_us ? vblank_blank_us : mall_region_us; - - // Schedulable if VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, - // and the max of (VBLANK blanking time, MALL region) - // TODO: Possibly add some margin (i.e. the below conditions should be [...] > X instead of [...] > 0) - if (subvp_active_us - prefetch_us - vblank_frame_us - max_vblank_mallregion > 0) - schedulable = true; - } - return schedulable; -} - -/* - * subvp_validate_static_schedulability: Check which SubVP case is calculated and handle - * static analysis based on the case. - * - * Three cases: - * 1. SubVP + SubVP - * 2. SubVP + VBLANK (DRR checked internally) - * 3. SubVP + VACTIVE (currently unsupported) - * - * @dc: current dc state - * @context: new dc state - * @vlevel: Voltage level calculated by DML - * - * Return: - * bool - True if statically schedulable, false otherwise - */ -bool dml2_svp_validate_static_schedulability(struct dml2_context *ctx, struct dc_state *context, enum dml_dram_clock_change_support pstate_change_type) -{ - bool schedulable = true; // true by default for single display case - struct vba_vars_st *vba = &context->bw_ctx.dml.vba; - uint32_t i, pipe_idx; - uint8_t subvp_count = 0; - uint8_t vactive_count = 0; - - for (i = 0, pipe_idx = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - enum mall_stream_type pipe_mall_type = ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe); - - if (!pipe->stream) - continue; - - if (pipe->plane_state && !pipe->top_pipe && - pipe_mall_type == SUBVP_MAIN) - subvp_count++; - - // Count how many planes that aren't SubVP/phantom are capable of VACTIVE - // switching (SubVP + VACTIVE unsupported). In situations where we force - // SubVP for a VACTIVE plane, we don't want to increment the vactive_count. - if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] > 0 && - pipe_mall_type == SUBVP_NONE) { - vactive_count++; - } - pipe_idx++; - } - - if (subvp_count == 2) { - // Static schedulability check for SubVP + SubVP case - schedulable = subvp_subvp_schedulable(ctx, context); - } else if (pstate_change_type == dml_dram_clock_change_vblank_w_mall_sub_vp) { - // Static schedulability check for SubVP + VBLANK case. Also handle the case where - // DML outputs SubVP + VBLANK + VACTIVE (DML will report as SubVP + VBLANK) - if (vactive_count > 0) - schedulable = false; - else - schedulable = subvp_vblank_schedulable(ctx, context); - } else if (pstate_change_type == dml_dram_clock_change_vactive_w_mall_sub_vp && - vactive_count > 0) { - // For single display SubVP cases, DML will output dm_dram_clock_change_vactive_w_mall_sub_vp by default. - // We tell the difference between SubVP vs. SubVP + VACTIVE by checking the vactive_count. - // SubVP + VACTIVE currently unsupported - schedulable = false; - } - return schedulable; -} - -static void set_phantom_stream_timing(struct dml2_context *ctx, struct dc_state *state, - struct pipe_ctx *ref_pipe, - struct dc_stream_state *phantom_stream, - unsigned int dc_pipe_idx, - unsigned int svp_height, - unsigned int svp_vstartup) -{ - unsigned int i; - double line_time, fp_and_sync_width_time; - struct pipe_ctx *pipe; - uint32_t phantom_vactive, phantom_bp, pstate_width_fw_delay_lines; - static const double cvt_rb_vblank_max = ((double) 460 / (1000 * 1000)); - - // Find DML pipe index (pipe_idx) using dc_pipe_idx - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - pipe = &state->res_ctx.pipe_ctx[i]; - - if (!pipe->stream) - continue; - - if (i == dc_pipe_idx) - break; - } - - // Calculate lines required for pstate allow width and FW processing delays - pstate_width_fw_delay_lines = ((double)(ctx->config.svp_pstate.subvp_fw_processing_delay_us + - ctx->config.svp_pstate.subvp_pstate_allow_width_us) / 1000000) * - (ref_pipe->stream->timing.pix_clk_100hz * 100) / - (double)ref_pipe->stream->timing.h_total; - - // DML calculation for MALL region doesn't take into account FW delay - // and required pstate allow width for multi-display cases - /* Add 16 lines margin to the MALL REGION because SUB_VP_START_LINE must be aligned - * to 2 swaths (i.e. 16 lines) - */ - phantom_vactive = svp_height + pstate_width_fw_delay_lines + ctx->config.svp_pstate.subvp_swath_height_margin_lines; - - phantom_stream->timing.v_front_porch = 1; - - line_time = phantom_stream->timing.h_total / ((double)phantom_stream->timing.pix_clk_100hz * 100); - fp_and_sync_width_time = (phantom_stream->timing.v_front_porch + phantom_stream->timing.v_sync_width) * line_time; - - if ((svp_vstartup * line_time) + fp_and_sync_width_time > cvt_rb_vblank_max) { - svp_vstartup = (cvt_rb_vblank_max - fp_and_sync_width_time) / line_time; - } - - // For backporch of phantom pipe, use vstartup of the main pipe - phantom_bp = svp_vstartup; - - phantom_stream->dst.y = 0; - phantom_stream->dst.height = phantom_vactive; - phantom_stream->src.y = 0; - phantom_stream->src.height = phantom_vactive; - - phantom_stream->timing.v_addressable = phantom_vactive; - - phantom_stream->timing.v_total = phantom_stream->timing.v_addressable + - phantom_stream->timing.v_front_porch + - phantom_stream->timing.v_sync_width + - phantom_bp; - phantom_stream->timing.flags.DSC = 0; // Don't need DSC for phantom timing -} - -static struct dc_stream_state *enable_phantom_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int dc_pipe_idx, unsigned int svp_height, unsigned int vstartup) -{ - struct pipe_ctx *ref_pipe = &state->res_ctx.pipe_ctx[dc_pipe_idx]; - struct dc_stream_state *phantom_stream = ctx->config.svp_pstate.callbacks.create_phantom_stream( - ctx->config.svp_pstate.callbacks.dc, - state, - ref_pipe->stream); - - /* stream has limited viewport and small timing */ - memcpy(&phantom_stream->timing, &ref_pipe->stream->timing, sizeof(phantom_stream->timing)); - memcpy(&phantom_stream->src, &ref_pipe->stream->src, sizeof(phantom_stream->src)); - memcpy(&phantom_stream->dst, &ref_pipe->stream->dst, sizeof(phantom_stream->dst)); - set_phantom_stream_timing(ctx, state, ref_pipe, phantom_stream, dc_pipe_idx, svp_height, vstartup); - - ctx->config.svp_pstate.callbacks.add_phantom_stream(ctx->config.svp_pstate.callbacks.dc, - state, - phantom_stream, - ref_pipe->stream); - return phantom_stream; -} - -static void enable_phantom_plane(struct dml2_context *ctx, - struct dc_state *state, - struct dc_stream_state *phantom_stream, - unsigned int dc_pipe_idx) -{ - struct dc_plane_state *phantom_plane = NULL; - struct dc_plane_state *prev_phantom_plane = NULL; - struct pipe_ctx *curr_pipe = &state->res_ctx.pipe_ctx[dc_pipe_idx]; - - while (curr_pipe) { - if (curr_pipe->top_pipe && curr_pipe->top_pipe->plane_state == curr_pipe->plane_state) { - phantom_plane = prev_phantom_plane; - } else { - phantom_plane = ctx->config.svp_pstate.callbacks.create_phantom_plane( - ctx->config.svp_pstate.callbacks.dc, - state, - curr_pipe->plane_state); - if (!phantom_plane) - return; - } - - memcpy(&phantom_plane->address, &curr_pipe->plane_state->address, sizeof(phantom_plane->address)); - memcpy(&phantom_plane->scaling_quality, &curr_pipe->plane_state->scaling_quality, - sizeof(phantom_plane->scaling_quality)); - memcpy(&phantom_plane->src_rect, &curr_pipe->plane_state->src_rect, sizeof(phantom_plane->src_rect)); - memcpy(&phantom_plane->dst_rect, &curr_pipe->plane_state->dst_rect, sizeof(phantom_plane->dst_rect)); - memcpy(&phantom_plane->clip_rect, &curr_pipe->plane_state->clip_rect, sizeof(phantom_plane->clip_rect)); - memcpy(&phantom_plane->plane_size, &curr_pipe->plane_state->plane_size, - sizeof(phantom_plane->plane_size)); - memcpy(&phantom_plane->tiling_info, &curr_pipe->plane_state->tiling_info, - sizeof(phantom_plane->tiling_info)); - memcpy(&phantom_plane->dcc, &curr_pipe->plane_state->dcc, sizeof(phantom_plane->dcc)); - //phantom_plane->tiling_info.gfx10compatible.compat_level = curr_pipe->plane_state->tiling_info.gfx10compatible.compat_level; - phantom_plane->format = curr_pipe->plane_state->format; - phantom_plane->rotation = curr_pipe->plane_state->rotation; - phantom_plane->visible = curr_pipe->plane_state->visible; - - /* Shadow pipe has small viewport. */ - phantom_plane->clip_rect.y = 0; - phantom_plane->clip_rect.height = phantom_stream->timing.v_addressable; - - ctx->config.svp_pstate.callbacks.add_phantom_plane(ctx->config.svp_pstate.callbacks.dc, phantom_stream, phantom_plane, state); - - curr_pipe = curr_pipe->bottom_pipe; - prev_phantom_plane = phantom_plane; - } -} - -static void add_phantom_pipes_for_main_pipe(struct dml2_context *ctx, struct dc_state *state, unsigned int main_pipe_idx, unsigned int svp_height, unsigned int vstartup) -{ - struct dc_stream_state *phantom_stream = NULL; - unsigned int i; - - // The index of the DC pipe passed into this function is guarenteed to - // be a valid candidate for SubVP (i.e. has a plane, stream, doesn't - // already have phantom pipe assigned, etc.) by previous checks. - phantom_stream = enable_phantom_stream(ctx, state, main_pipe_idx, svp_height, vstartup); - enable_phantom_plane(ctx, state, phantom_stream, main_pipe_idx); - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - // Build scaling params for phantom pipes which were newly added. - // We determine which phantom pipes were added by comparing with - // the phantom stream. - if (pipe->plane_state && pipe->stream && pipe->stream == phantom_stream && - ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_PHANTOM) { - pipe->stream->use_dynamic_meta = false; - pipe->plane_state->flip_immediate = false; - if (!ctx->config.svp_pstate.callbacks.build_scaling_params(pipe)) { - // Log / remove phantom pipes since failed to build scaling params - } - } - } -} - -static bool remove_all_phantom_planes_for_stream(struct dml2_context *ctx, struct dc_stream_state *stream, struct dc_state *context) -{ - int i, old_plane_count; - struct dc_stream_status *stream_status = NULL; - struct dc_plane_state *del_planes[MAX_SURFACES] = { 0 }; - - for (i = 0; i < context->stream_count; i++) - if (context->streams[i] == stream) { - stream_status = &context->stream_status[i]; - break; - } - - if (stream_status == NULL) { - return false; - } - - old_plane_count = stream_status->plane_count; - - for (i = 0; i < old_plane_count; i++) - del_planes[i] = stream_status->plane_states[i]; - - for (i = 0; i < old_plane_count; i++) { - if (!ctx->config.svp_pstate.callbacks.remove_phantom_plane(ctx->config.svp_pstate.callbacks.dc, stream, del_planes[i], context)) - return false; - ctx->config.svp_pstate.callbacks.release_phantom_plane(ctx->config.svp_pstate.callbacks.dc, context, del_planes[i]); - } - - return true; -} - -bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state *state) -{ - int i; - bool removed_pipe = false; - struct dc_stream_state *phantom_stream = NULL; - - for (i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - // build scaling params for phantom pipes - if (pipe->plane_state && pipe->stream && ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_PHANTOM) { - phantom_stream = pipe->stream; - - remove_all_phantom_planes_for_stream(ctx, phantom_stream, state); - ctx->config.svp_pstate.callbacks.remove_phantom_stream(ctx->config.svp_pstate.callbacks.dc, state, phantom_stream); - ctx->config.svp_pstate.callbacks.release_phantom_stream(ctx->config.svp_pstate.callbacks.dc, state, phantom_stream); - - removed_pipe = true; - } - - if (pipe->plane_state) { - pipe->plane_state->is_phantom = false; - } - } - return removed_pipe; -} - - -/* Conditions for setting up phantom pipes for SubVP: - * 1. Not force disable SubVP - * 2. Full update (i.e. DC_VALIDATE_MODE_AND_PROGRAMMING) - * 3. Enough pipes are available to support SubVP (TODO: Which pipes will use VACTIVE / VBLANK / SUBVP?) - * 4. Display configuration passes validation - * 5. (Config doesn't support MCLK in VACTIVE/VBLANK || dc->debug.force_subvp_mclk_switch) - */ -bool dml2_svp_add_phantom_pipe_to_dc_state(struct dml2_context *ctx, struct dc_state *state, struct dml_mode_support_info_st *mode_support_info) -{ - unsigned int dc_pipe_idx, dml_pipe_idx; - unsigned int svp_height, vstartup; - - if (ctx->config.svp_pstate.force_disable_subvp) - return false; - - if (!all_pipes_have_stream_and_plane(ctx, state)) - return false; - - if (mpo_in_use(state)) - return false; - - merge_pipes_for_subvp(ctx, state); - // to re-initialize viewport after the pipe merge - for (int i = 0; i < ctx->config.dcn_pipe_count; i++) { - struct pipe_ctx *pipe_ctx = &state->res_ctx.pipe_ctx[i]; - - if (!pipe_ctx->plane_state || !pipe_ctx->stream) - continue; - - ctx->config.svp_pstate.callbacks.build_scaling_params(pipe_ctx); - } - - if (enough_pipes_for_subvp(ctx, state) && assign_subvp_pipe(ctx, state, &dc_pipe_idx)) { - dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(ctx, state->res_ctx.pipe_ctx[dc_pipe_idx].stream->stream_id); - svp_height = mode_support_info->SubViewportLinesNeededInMALL[dml_pipe_idx]; - vstartup = dml_get_vstartup_calculated(&ctx->v20.dml_core_ctx, dml_pipe_idx); - - add_phantom_pipes_for_main_pipe(ctx, state, dc_pipe_idx, svp_height, vstartup); - - return true; - } - - return false; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h deleted file mode 100644 index 9d64851f54e7..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h +++ /dev/null @@ -1,52 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DML2_MALL_PHANTOM_H__ -#define __DML2_MALL_PHANTOM_H__ - -#include "dml2_dc_types.h" -#include "display_mode_core_structs.h" - -struct dml2_svp_helper_select_best_svp_candidate_params { - const struct dml_display_cfg_st *dml_config; - const struct dml_mode_support_info_st *mode_support_info; - const unsigned int blacklist; - unsigned int *candidate_index; -}; - -struct dml2_context; - -unsigned int dml2_helper_calculate_num_ways_for_subvp(struct dml2_context *ctx, struct dc_state *context); - -bool dml2_svp_add_phantom_pipe_to_dc_state(struct dml2_context *ctx, struct dc_state *state, struct dml_mode_support_info_st *mode_support_info); - -bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state *state); - -bool dml2_svp_validate_static_schedulability(struct dml2_context *ctx, struct dc_state *context, enum dml_dram_clock_change_support pstate_change_type); - -bool dml2_svp_drr_schedulable(struct dml2_context *ctx, struct dc_state *context, struct dc_crtc_timing *drr_timing); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c deleted file mode 100644 index ef693f608d59..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c +++ /dev/null @@ -1,311 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "dml2_policy.h" - -static void get_optimal_ntuple( - const struct soc_bounding_box_st *socbb, - struct soc_state_bounding_box_st *entry) -{ - if (entry->dcfclk_mhz > 0) { - float bw_on_sdp = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); - - entry->fabricclk_mhz = bw_on_sdp / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); - entry->dram_speed_mts = bw_on_sdp / (socbb->num_chans * - socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); - } else if (entry->fabricclk_mhz > 0) { - float bw_on_fabric = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); - - entry->dcfclk_mhz = bw_on_fabric / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); - entry->dram_speed_mts = bw_on_fabric / (socbb->num_chans * - socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); - } else if (entry->dram_speed_mts > 0) { - float bw_on_dram = (float)(entry->dram_speed_mts * socbb->num_chans * - socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); - - entry->fabricclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); - entry->dcfclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); - } -} - -static float calculate_net_bw_in_mbytes_sec(const struct soc_bounding_box_st *socbb, - struct soc_state_bounding_box_st *entry) -{ - float memory_bw_mbytes_sec = (float)(entry->dram_speed_mts * socbb->num_chans * - socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); - - float fabric_bw_mbytes_sec = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); - - float sdp_bw_mbytes_sec = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); - - float limiting_bw_mbytes_sec = memory_bw_mbytes_sec; - - if (fabric_bw_mbytes_sec < limiting_bw_mbytes_sec) - limiting_bw_mbytes_sec = fabric_bw_mbytes_sec; - - if (sdp_bw_mbytes_sec < limiting_bw_mbytes_sec) - limiting_bw_mbytes_sec = sdp_bw_mbytes_sec; - - return limiting_bw_mbytes_sec; -} - -static void insert_entry_into_table_sorted(const struct soc_bounding_box_st *socbb, - struct soc_states_st *table, - struct soc_state_bounding_box_st *entry) -{ - int index = 0; - int i = 0; - float net_bw_of_new_state = 0; - - get_optimal_ntuple(socbb, entry); - - if (table->num_states == 0) { - index = 0; - } else { - net_bw_of_new_state = calculate_net_bw_in_mbytes_sec(socbb, entry); - while (net_bw_of_new_state > calculate_net_bw_in_mbytes_sec(socbb, &table->state_array[index])) { - index++; - if (index >= (int) table->num_states) - break; - } - - for (i = table->num_states; i > index; i--) { - table->state_array[i] = table->state_array[i - 1]; - } - //ASSERT(index < MAX_CLK_TABLE_SIZE); - } - - table->state_array[index] = *entry; - table->state_array[index].dcfclk_mhz = (int)entry->dcfclk_mhz; - table->state_array[index].fabricclk_mhz = (int)entry->fabricclk_mhz; - table->state_array[index].dram_speed_mts = (int)entry->dram_speed_mts; - table->num_states++; -} - -static void remove_entry_from_table_at_index(struct soc_states_st *table, - unsigned int index) -{ - int i; - - if (table->num_states == 0) - return; - - for (i = index; i < (int) table->num_states - 1; i++) { - table->state_array[i] = table->state_array[i + 1]; - } - memset(&table->state_array[--table->num_states], 0, sizeof(struct soc_state_bounding_box_st)); -} - -int dml2_policy_build_synthetic_soc_states(struct dml2_policy_build_synthetic_soc_states_scratch *s, - struct dml2_policy_build_synthetic_soc_states_params *p) -{ - int i, j; - unsigned int min_fclk_mhz = p->in_states->state_array[0].fabricclk_mhz; - unsigned int min_dcfclk_mhz = p->in_states->state_array[0].dcfclk_mhz; - unsigned int min_socclk_mhz = p->in_states->state_array[0].socclk_mhz; - - int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, - max_phyclk_mhz = 0, max_dtbclk_mhz = 0, max_fclk_mhz = 0, - max_uclk_mhz = 0, max_socclk_mhz = 0; - - int num_uclk_dpms = 0, num_fclk_dpms = 0; - - for (i = 0; i < __DML_MAX_STATE_ARRAY_SIZE__; i++) { - if (p->in_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz) - max_dcfclk_mhz = (int) p->in_states->state_array[i].dcfclk_mhz; - if (p->in_states->state_array[i].fabricclk_mhz > max_fclk_mhz) - max_fclk_mhz = (int) p->in_states->state_array[i].fabricclk_mhz; - if (p->in_states->state_array[i].socclk_mhz > max_socclk_mhz) - max_socclk_mhz = (int) p->in_states->state_array[i].socclk_mhz; - if (p->in_states->state_array[i].dram_speed_mts > max_uclk_mhz) - max_uclk_mhz = (int) p->in_states->state_array[i].dram_speed_mts; - if (p->in_states->state_array[i].dispclk_mhz > max_dispclk_mhz) - max_dispclk_mhz = (int) p->in_states->state_array[i].dispclk_mhz; - if (p->in_states->state_array[i].dppclk_mhz > max_dppclk_mhz) - max_dppclk_mhz = (int) p->in_states->state_array[i].dppclk_mhz; - if (p->in_states->state_array[i].phyclk_mhz > max_phyclk_mhz) - max_phyclk_mhz = (int)p->in_states->state_array[i].phyclk_mhz; - if (p->in_states->state_array[i].dtbclk_mhz > max_dtbclk_mhz) - max_dtbclk_mhz = (int)p->in_states->state_array[i].dtbclk_mhz; - - if (p->in_states->state_array[i].fabricclk_mhz > 0) - num_fclk_dpms++; - if (p->in_states->state_array[i].dram_speed_mts > 0) - num_uclk_dpms++; - } - - if (!max_dcfclk_mhz || !max_dispclk_mhz || !max_dppclk_mhz || !max_phyclk_mhz || !max_dtbclk_mhz) - return -1; - - p->out_states->num_states = 0; - - s->entry = p->in_states->state_array[0]; - - s->entry.dispclk_mhz = max_dispclk_mhz; - s->entry.dppclk_mhz = max_dppclk_mhz; - s->entry.dtbclk_mhz = max_dtbclk_mhz; - s->entry.phyclk_mhz = max_phyclk_mhz; - - s->entry.dscclk_mhz = max_dispclk_mhz / 3; - s->entry.phyclk_mhz = max_phyclk_mhz; - s->entry.dtbclk_mhz = max_dtbclk_mhz; - - // Insert all the DCFCLK STAs first - for (i = 0; i < p->num_dcfclk_stas; i++) { - s->entry.dcfclk_mhz = p->dcfclk_stas_mhz[i]; - s->entry.fabricclk_mhz = 0; - s->entry.dram_speed_mts = 0; - if (i > 0) - s->entry.socclk_mhz = max_socclk_mhz; - - insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); - } - - // Insert the UCLK DPMS - for (i = 0; i < num_uclk_dpms; i++) { - s->entry.dcfclk_mhz = 0; - s->entry.fabricclk_mhz = 0; - s->entry.dram_speed_mts = p->in_states->state_array[i].dram_speed_mts; - if (i == 0) { - s->entry.socclk_mhz = min_socclk_mhz; - } else { - s->entry.socclk_mhz = max_socclk_mhz; - } - - insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); - } - - // Insert FCLK DPMs (if present) - if (num_fclk_dpms > 2) { - for (i = 0; i < num_fclk_dpms; i++) { - s->entry.dcfclk_mhz = 0; - s->entry.fabricclk_mhz = p->in_states->state_array[i].fabricclk_mhz; - s->entry.dram_speed_mts = 0; - - insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); - } - } - // Add max FCLK - else { - s->entry.dcfclk_mhz = 0; - s->entry.fabricclk_mhz = p->in_states->state_array[num_fclk_dpms - 1].fabricclk_mhz; - s->entry.dram_speed_mts = 0; - - insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); - } - - // Remove states that require higher clocks than are supported - for (i = p->out_states->num_states - 1; i >= 0; i--) { - if (p->out_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz || - p->out_states->state_array[i].fabricclk_mhz > max_fclk_mhz || - p->out_states->state_array[i].dram_speed_mts > max_uclk_mhz) - remove_entry_from_table_at_index(p->out_states, i); - } - - // At this point, the table contains all "points of interest" based on - // DPMs from PMFW, and STAs. Table is sorted by BW, and all clock - // ratios (by derate, are exact). - - // Round up UCLK to DPMs - for (i = p->out_states->num_states - 1; i >= 0; i--) { - for (j = 0; j < num_uclk_dpms; j++) { - if (p->in_states->state_array[j].dram_speed_mts >= p->out_states->state_array[i].dram_speed_mts) { - p->out_states->state_array[i].dram_speed_mts = p->in_states->state_array[j].dram_speed_mts; - break; - } - } - } - - // If FCLK is coarse grained, round up to next DPMs - if (num_fclk_dpms > 2) { - for (i = p->out_states->num_states - 1; i >= 0; i--) { - for (j = 0; j < num_fclk_dpms; j++) { - if (p->in_states->state_array[j].fabricclk_mhz >= p->out_states->state_array[i].fabricclk_mhz) { - p->out_states->state_array[i].fabricclk_mhz = p->in_states->state_array[j].fabricclk_mhz; - break; - } - } - } - } - - // Clamp to min FCLK/DCFCLK - for (i = p->out_states->num_states - 1; i >= 0; i--) { - if (p->out_states->state_array[i].fabricclk_mhz < min_fclk_mhz) { - p->out_states->state_array[i].fabricclk_mhz = min_fclk_mhz; - } - if (p->out_states->state_array[i].dcfclk_mhz < min_dcfclk_mhz) { - p->out_states->state_array[i].dcfclk_mhz = min_dcfclk_mhz; - } - } - - // Remove duplicate states, note duplicate states are always neighbouring since table is sorted. - i = 0; - while (i < (int) p->out_states->num_states - 1) { - if (p->out_states->state_array[i].dcfclk_mhz == p->out_states->state_array[i + 1].dcfclk_mhz && - p->out_states->state_array[i].fabricclk_mhz == p->out_states->state_array[i + 1].fabricclk_mhz && - p->out_states->state_array[i].dram_speed_mts == p->out_states->state_array[i + 1].dram_speed_mts) - remove_entry_from_table_at_index(p->out_states, i); - else - i++; - } - - return 0; -} - -void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_mode_eval_policy_st *policy) -{ - for (int i = 0; i < __DML_NUM_PLANES__; i++) { - policy->MPCCombineUse[i] = dml_mpc_as_needed_for_voltage; // TOREVIEW: Is this still needed? When is MPCC useful for pstate given CRB? - policy->ODMUse[i] = dml_odm_use_policy_combine_as_needed; - policy->ImmediateFlipRequirement[i] = dml_immediate_flip_required; - policy->AllowForPStateChangeOrStutterInVBlank[i] = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; - } - - /* Change the default policy initializations as per spreadsheet. We might need to - * review and change them later on as per Jun's earlier comments. - */ - policy->UseUnboundedRequesting = dml_unbounded_requesting_enable; - policy->UseMinimumRequiredDCFCLK = false; - policy->DRAMClockChangeRequirementFinal = true; // TOREVIEW: What does this mean? - policy->FCLKChangeRequirementFinal = true; // TOREVIEW: What does this mean? - policy->USRRetrainingRequiredFinal = true; - policy->EnhancedPrefetchScheduleAccelerationFinal = true; // TOREVIEW: What does this mean? - policy->NomDETInKByteOverrideEnable = false; - policy->NomDETInKByteOverrideValue = 0; - policy->DCCProgrammingAssumesScanDirectionUnknownFinal = true; - policy->SynchronizeTimingsFinal = true; - policy->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = true; - policy->AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported = true; // TOREVIEW: What does this mean? - policy->AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported = true; // TOREVIEW: What does this mean? - if (project == dml_project_dcn35 || - project == dml_project_dcn36 || - project == dml_project_dcn351) { - policy->DCCProgrammingAssumesScanDirectionUnknownFinal = false; - policy->EnhancedPrefetchScheduleAccelerationFinal = 0; - policy->AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; /*new*/ - policy->UseOnlyMaxPrefetchModes = 1; - } -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h deleted file mode 100644 index e83e05248592..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h +++ /dev/null @@ -1,47 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef __DML2_POLICY_H__ -#define __DML2_POLICY_H__ - -#include "display_mode_core_structs.h" - -struct dml2_policy_build_synthetic_soc_states_params { - const struct soc_bounding_box_st *in_bbox; - struct soc_states_st *in_states; - struct soc_states_st *out_states; - int *dcfclk_stas_mhz; - int num_dcfclk_stas; -}; - -struct dml2_policy_build_synthetic_soc_states_scratch { - struct soc_state_bounding_box_st entry; -}; - -int dml2_policy_build_synthetic_soc_states(struct dml2_policy_build_synthetic_soc_states_scratch *s, - struct dml2_policy_build_synthetic_soc_states_params *p); - -void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_mode_eval_policy_st *policy); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c deleted file mode 100644 index 3b866e876bf4..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c +++ /dev/null @@ -1,1525 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "display_mode_core.h" -#include "dml2_internal_types.h" -#include "dml2_translation_helper.h" - -#define NUM_DCFCLK_STAS 5 -#define NUM_DCFCLK_STAS_NEW 8 - -void dml2_init_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out) -{ - switch (dml2->v20.dml_core_ctx.project) { - case dml_project_dcn32: - case dml_project_dcn321: - default: - // Hardcoded values for DCN32x - out->vblank_nom_default_us = 600; - out->rob_buffer_size_kbytes = 128; - out->config_return_buffer_size_in_kbytes = 1280; - out->config_return_buffer_segment_size_in_kbytes = 64; - out->compressed_buffer_segment_size_in_kbytes = 64; - out->meta_fifo_size_in_kentries = 22; - out->zero_size_buffer_entries = 512; - out->dpte_buffer_size_in_pte_reqs_luma = 68; - out->dpte_buffer_size_in_pte_reqs_chroma = 36; - out->dcc_meta_buffer_size_bytes = 6272; - out->gpuvm_max_page_table_levels = 4; - out->hostvm_max_page_table_levels = 0; - out->pixel_chunk_size_kbytes = 8; - //out->alpha_pixel_chunk_size_kbytes; - out->min_pixel_chunk_size_bytes = 1024; - out->meta_chunk_size_kbytes = 2; - out->min_meta_chunk_size_bytes = 256; - out->writeback_chunk_size_kbytes = 8; - out->line_buffer_size_bits = 1171920; - out->max_line_buffer_lines = 32; - out->writeback_interface_buffer_size_kbytes = 90; - //Number of pipes after DCN Pipe harvesting - out->max_num_dpp = dml2->config.dcn_pipe_count; - out->max_num_otg = dml2->config.dcn_pipe_count; - out->max_num_wb = 1; - out->max_dchub_pscl_bw_pix_per_clk = 4; - out->max_pscl_lb_bw_pix_per_clk = 2; - out->max_lb_vscl_bw_pix_per_clk = 4; - out->max_vscl_hscl_bw_pix_per_clk = 4; - out->max_hscl_ratio = 6; - out->max_vscl_ratio = 6; - out->max_hscl_taps = 8; - out->max_vscl_taps = 8; - out->dispclk_ramp_margin_percent = 1; - out->dppclk_delay_subtotal = 47; - out->dppclk_delay_scl = 50; - out->dppclk_delay_scl_lb_only = 16; - out->dppclk_delay_cnvc_formatter = 28; - out->dppclk_delay_cnvc_cursor = 6; - out->cursor_buffer_size = 16; - out->cursor_chunk_size = 2; - out->dispclk_delay_subtotal = 125; - out->max_inter_dcn_tile_repeaters = 8; - out->writeback_max_hscl_ratio = 1; - out->writeback_max_vscl_ratio = 1; - out->writeback_min_hscl_ratio = 1; - out->writeback_min_vscl_ratio = 1; - out->writeback_max_hscl_taps = 1; - out->writeback_max_vscl_taps = 1; - out->writeback_line_buffer_buffer_size = 0; - out->num_dsc = 4; - out->maximum_dsc_bits_per_component = 12; - out->maximum_pixels_per_line_per_dsc_unit = 6016; - out->dsc422_native_support = true; - out->dcc_supported = true; - out->ptoi_supported = false; - - out->gpuvm_enable = false; - out->hostvm_enable = false; - out->cursor_64bpp_support = false; - out->dynamic_metadata_vm_enabled = false; - - out->max_num_hdmi_frl_outputs = 1; - out->max_num_dp2p0_outputs = 2; - out->max_num_dp2p0_streams = 4; - break; - - case dml_project_dcn35: - case dml_project_dcn351: - case dml_project_dcn36: - out->rob_buffer_size_kbytes = 64; - out->config_return_buffer_size_in_kbytes = 1792; - out->compressed_buffer_segment_size_in_kbytes = 64; - out->meta_fifo_size_in_kentries = 32; - out->zero_size_buffer_entries = 512; - out->pixel_chunk_size_kbytes = 8; - out->alpha_pixel_chunk_size_kbytes = 4; - out->min_pixel_chunk_size_bytes = 1024; - out->meta_chunk_size_kbytes = 2; - out->min_meta_chunk_size_bytes = 256; - out->writeback_chunk_size_kbytes = 8; - out->dpte_buffer_size_in_pte_reqs_luma = 68; - out->dpte_buffer_size_in_pte_reqs_chroma = 36; - out->dcc_meta_buffer_size_bytes = 6272; - out->gpuvm_enable = 1; - out->hostvm_enable = 1; - out->gpuvm_max_page_table_levels = 1; - out->hostvm_max_page_table_levels = 2; - out->num_dsc = 4; - out->maximum_dsc_bits_per_component = 12; - out->maximum_pixels_per_line_per_dsc_unit = 6016; - out->dsc422_native_support = 1; - out->line_buffer_size_bits = 986880; - out->dcc_supported = 1; - out->max_line_buffer_lines = 32; - out->writeback_interface_buffer_size_kbytes = 90; - out->max_num_dpp = 4; - out->max_num_otg = 4; - out->max_num_hdmi_frl_outputs = 1; - out->max_num_dp2p0_outputs = 2; - out->max_num_dp2p0_streams = 4; - out->max_num_wb = 1; - - out->max_dchub_pscl_bw_pix_per_clk = 4; - out->max_pscl_lb_bw_pix_per_clk = 2; - out->max_lb_vscl_bw_pix_per_clk = 4; - out->max_vscl_hscl_bw_pix_per_clk = 4; - out->max_hscl_ratio = 6; - out->max_vscl_ratio = 6; - out->max_hscl_taps = 8; - out->max_vscl_taps = 8; - out->dispclk_ramp_margin_percent = 1.11; - - out->dppclk_delay_subtotal = 47; - out->dppclk_delay_scl = 50; - out->dppclk_delay_scl_lb_only = 16; - out->dppclk_delay_cnvc_formatter = 28; - out->dppclk_delay_cnvc_cursor = 6; - out->dispclk_delay_subtotal = 125; - - out->dynamic_metadata_vm_enabled = false; - out->max_inter_dcn_tile_repeaters = 8; - out->cursor_buffer_size = 16; // kBytes - out->cursor_chunk_size = 2; // kBytes - - out->writeback_line_buffer_buffer_size = 0; - out->writeback_max_hscl_ratio = 1; - out->writeback_max_vscl_ratio = 1; - out->writeback_min_hscl_ratio = 1; - out->writeback_min_vscl_ratio = 1; - out->writeback_max_hscl_taps = 1; - out->writeback_max_vscl_taps = 1; - out->ptoi_supported = 0; - - out->vblank_nom_default_us = 668; /*not in dml, but in programming guide, hard coded in dml2_translate_ip_params*/ - out->config_return_buffer_segment_size_in_kbytes = 64; /*required, but not exist,, hard coded in dml2_translate_ip_params*/ - break; - - case dml_project_dcn401: - // Hardcoded values for DCN4m - out->vblank_nom_default_us = 668; //600; - out->rob_buffer_size_kbytes = 192; //128; - out->config_return_buffer_size_in_kbytes = 1344; //1280; - out->config_return_buffer_segment_size_in_kbytes = 64; - out->compressed_buffer_segment_size_in_kbytes = 64; - out->meta_fifo_size_in_kentries = 22; - out->dpte_buffer_size_in_pte_reqs_luma = 68; - out->dpte_buffer_size_in_pte_reqs_chroma = 36; - out->gpuvm_max_page_table_levels = 4; - out->pixel_chunk_size_kbytes = 8; - out->alpha_pixel_chunk_size_kbytes = 4; - out->min_pixel_chunk_size_bytes = 1024; - out->writeback_chunk_size_kbytes = 8; - out->line_buffer_size_bits = 1171920; - out->max_line_buffer_lines = 32; - out->writeback_interface_buffer_size_kbytes = 90; - //Number of pipes after DCN Pipe harvesting - out->max_num_dpp = dml2->config.dcn_pipe_count; - out->max_num_otg = dml2->config.dcn_pipe_count; - out->max_num_wb = 1; - out->max_dchub_pscl_bw_pix_per_clk = 4; - out->max_pscl_lb_bw_pix_per_clk = 2; - out->max_lb_vscl_bw_pix_per_clk = 4; - out->max_vscl_hscl_bw_pix_per_clk = 4; - out->max_hscl_ratio = 6; - out->max_vscl_ratio = 6; - out->max_hscl_taps = 8; - out->max_vscl_taps = 8; - out->dispclk_ramp_margin_percent = 1; - out->dppclk_delay_subtotal = 47; - out->dppclk_delay_scl = 50; - out->dppclk_delay_scl_lb_only = 16; - out->dppclk_delay_cnvc_formatter = 28; - out->dppclk_delay_cnvc_cursor = 6; - out->dispclk_delay_subtotal = 125; - out->cursor_buffer_size = 24; //16 - out->cursor_chunk_size = 2; - out->max_inter_dcn_tile_repeaters = 8; - out->writeback_max_hscl_ratio = 1; - out->writeback_max_vscl_ratio = 1; - out->writeback_min_hscl_ratio = 1; - out->writeback_min_vscl_ratio = 1; - out->writeback_max_hscl_taps = 1; - out->writeback_max_vscl_taps = 1; - out->writeback_line_buffer_buffer_size = 0; - out->num_dsc = 4; - out->maximum_dsc_bits_per_component = 12; - out->maximum_pixels_per_line_per_dsc_unit = 5760; - out->dsc422_native_support = true; - out->dcc_supported = true; - out->ptoi_supported = false; - - out->gpuvm_enable = false; - out->hostvm_enable = false; - out->cursor_64bpp_support = true; //false; - out->dynamic_metadata_vm_enabled = false; - - out->max_num_hdmi_frl_outputs = 1; - out->max_num_dp2p0_outputs = 4; //2; - out->max_num_dp2p0_streams = 4; - break; - } -} - -void dml2_init_socbb_params(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out) -{ - out->dprefclk_mhz = dml2->config.bbox_overrides.dprefclk_mhz; - out->xtalclk_mhz = dml2->config.bbox_overrides.xtalclk_mhz; - out->pcierefclk_mhz = 100; - out->refclk_mhz = dml2->config.bbox_overrides.dchub_refclk_mhz; - - out->max_outstanding_reqs = 512; - out->pct_ideal_sdp_bw_after_urgent = 100; - out->pct_ideal_fabric_bw_after_urgent = 67; - out->pct_ideal_dram_bw_after_urgent_pixel_only = 20; - out->pct_ideal_dram_bw_after_urgent_pixel_and_vm = 60; - out->pct_ideal_dram_bw_after_urgent_vm_only = 30; - out->pct_ideal_dram_bw_after_urgent_strobe = 67; - out->max_avg_sdp_bw_use_normal_percent = 80; - out->max_avg_fabric_bw_use_normal_percent = 60; - out->max_avg_dram_bw_use_normal_percent = 15; - out->max_avg_dram_bw_use_normal_strobe_percent = 50; - - out->urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096; - out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096; - out->urgent_out_of_order_return_per_channel_vm_only_bytes = 4096; - out->return_bus_width_bytes = 64; - out->dram_channel_width_bytes = 2; - out->fabric_datapath_to_dcn_data_return_bytes = 64; - out->hostvm_min_page_size_kbytes = 0; - out->gpuvm_min_page_size_kbytes = 256; - out->phy_downspread_percent = 0.38; - out->dcn_downspread_percent = 0.5; - out->dispclk_dppclk_vco_speed_mhz = dml2->config.bbox_overrides.disp_pll_vco_speed_mhz; - out->mall_allocated_for_dcn_mbytes = dml2->config.mall_cfg.max_cab_allocation_bytes / 1048576; // 64 or 32 MB; - - out->do_urgent_latency_adjustment = true; - - switch (dml2->v20.dml_core_ctx.project) { - - case dml_project_dcn32: - default: - out->num_chans = 24; - out->round_trip_ping_latency_dcfclk_cycles = 263; - out->smn_latency_us = 2; - break; - - case dml_project_dcn321: - out->num_chans = 8; - out->round_trip_ping_latency_dcfclk_cycles = 207; - out->smn_latency_us = 0; - break; - - case dml_project_dcn35: - case dml_project_dcn351: - case dml_project_dcn36: - out->num_chans = 4; - out->round_trip_ping_latency_dcfclk_cycles = 106; - out->smn_latency_us = 2; - out->dispclk_dppclk_vco_speed_mhz = 3600; - out->pct_ideal_dram_bw_after_urgent_pixel_only = 65.0; - break; - - case dml_project_dcn401: - out->pct_ideal_fabric_bw_after_urgent = 76; //67; - out->max_avg_sdp_bw_use_normal_percent = 75; //80; - out->max_avg_fabric_bw_use_normal_percent = 57; //60; - - out->urgent_out_of_order_return_per_channel_pixel_only_bytes = 0; //4096; - out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 0; //4096; - out->urgent_out_of_order_return_per_channel_vm_only_bytes = 0; //4096; - - out->num_chans = 16; - out->round_trip_ping_latency_dcfclk_cycles = 1000; //263; - out->smn_latency_us = 0; //2 us - out->mall_allocated_for_dcn_mbytes = dml2->config.mall_cfg.max_cab_allocation_bytes / 1048576; // 64; - break; - } - /* ---Overrides if available--- */ - if (dml2->config.bbox_overrides.dram_num_chan) - out->num_chans = dml2->config.bbox_overrides.dram_num_chan; - - if (dml2->config.bbox_overrides.dram_chanel_width_bytes) - out->dram_channel_width_bytes = dml2->config.bbox_overrides.dram_chanel_width_bytes; -} - -void dml2_init_soc_states(struct dml2_context *dml2, const struct dc *in_dc, - const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out) -{ - struct dml2_policy_build_synthetic_soc_states_scratch *s = &dml2->v20.scratch.create_scratch.build_synthetic_socbb_scratch; - struct dml2_policy_build_synthetic_soc_states_params *p = &dml2->v20.scratch.build_synthetic_socbb_params; - unsigned int dcfclk_stas_mhz[NUM_DCFCLK_STAS] = {0}; - unsigned int dcfclk_stas_mhz_new[NUM_DCFCLK_STAS_NEW] = {0}; - unsigned int dml_project = dml2->v20.dml_core_ctx.project; - - unsigned int i = 0; - unsigned int transactions_per_mem_clock = 16; // project specific, depends on used Memory type - - if (dml_project == dml_project_dcn351) { - p->dcfclk_stas_mhz = dcfclk_stas_mhz_new; - p->num_dcfclk_stas = NUM_DCFCLK_STAS_NEW; - } else { - p->dcfclk_stas_mhz = dcfclk_stas_mhz; - p->num_dcfclk_stas = NUM_DCFCLK_STAS; - } - - p->in_bbox = in_bbox; - p->out_states = out; - p->in_states = &dml2->v20.scratch.create_scratch.in_states; - - - /* Initial hardcoded values */ - switch (dml2->v20.dml_core_ctx.project) { - - case dml_project_dcn32: - default: - p->in_states->num_states = 2; - transactions_per_mem_clock = 16; - p->in_states->state_array[0].socclk_mhz = 620.0; - p->in_states->state_array[0].dscclk_mhz = 716.667; - p->in_states->state_array[0].phyclk_mhz = 810; - p->in_states->state_array[0].phyclk_d18_mhz = 667; - p->in_states->state_array[0].phyclk_d32_mhz = 625; - p->in_states->state_array[0].dtbclk_mhz = 1564.0; - p->in_states->state_array[0].fabricclk_mhz = 450.0; - p->in_states->state_array[0].dcfclk_mhz = 300.0; - p->in_states->state_array[0].dispclk_mhz = 2150.0; - p->in_states->state_array[0].dppclk_mhz = 2150.0; - p->in_states->state_array[0].dram_speed_mts = 100 * transactions_per_mem_clock; - - p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; - p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; - p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; - p->in_states->state_array[0].writeback_latency_us = 12; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 3000; - p->in_states->state_array[0].sr_exit_z8_time_us = 0; - p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; - p->in_states->state_array[0].dram_clock_change_latency_us = 400; - p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; - p->in_states->state_array[0].sr_exit_time_us = 42.97; - p->in_states->state_array[0].sr_enter_plus_exit_time_us = 49.94; - p->in_states->state_array[0].fclk_change_latency_us = 20; - p->in_states->state_array[0].usr_retraining_latency_us = 2; - - p->in_states->state_array[1].socclk_mhz = 1200.0; - p->in_states->state_array[1].fabricclk_mhz = 2500.0; - p->in_states->state_array[1].dcfclk_mhz = 1564.0; - p->in_states->state_array[1].dram_speed_mts = 1125 * transactions_per_mem_clock; - break; - - case dml_project_dcn321: - p->in_states->num_states = 2; - transactions_per_mem_clock = 16; - p->in_states->state_array[0].socclk_mhz = 582.0; - p->in_states->state_array[0].dscclk_mhz = 573.333; - p->in_states->state_array[0].phyclk_mhz = 810; - p->in_states->state_array[0].phyclk_d18_mhz = 667; - p->in_states->state_array[0].phyclk_d32_mhz = 313; - p->in_states->state_array[0].dtbclk_mhz = 1564.0; - p->in_states->state_array[0].fabricclk_mhz = 450.0; - p->in_states->state_array[0].dcfclk_mhz = 300.0; - p->in_states->state_array[0].dispclk_mhz = 1720.0; - p->in_states->state_array[0].dppclk_mhz = 1720.0; - p->in_states->state_array[0].dram_speed_mts = 100 * transactions_per_mem_clock; - - p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; - p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; - p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; - p->in_states->state_array[0].writeback_latency_us = 12; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 3000; - p->in_states->state_array[0].sr_exit_z8_time_us = 0; - p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; - p->in_states->state_array[0].dram_clock_change_latency_us = 400; - p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; - p->in_states->state_array[0].sr_exit_time_us = 19.95; - p->in_states->state_array[0].sr_enter_plus_exit_time_us = 24.36; - p->in_states->state_array[0].fclk_change_latency_us = 7; - p->in_states->state_array[0].usr_retraining_latency_us = 0; - - p->in_states->state_array[1].socclk_mhz = 1200.0; - p->in_states->state_array[1].fabricclk_mhz = 2250.0; - p->in_states->state_array[1].dcfclk_mhz = 1434.0; - p->in_states->state_array[1].dram_speed_mts = 1000 * transactions_per_mem_clock; - break; - case dml_project_dcn401: - p->in_states->num_states = 2; - transactions_per_mem_clock = 16; - p->in_states->state_array[0].socclk_mhz = 300; //620.0; - p->in_states->state_array[0].dscclk_mhz = 666.667; //716.667; - p->in_states->state_array[0].phyclk_mhz = 810; - p->in_states->state_array[0].phyclk_d18_mhz = 667; - p->in_states->state_array[0].phyclk_d32_mhz = 625; - p->in_states->state_array[0].dtbclk_mhz = 2000; //1564.0; - p->in_states->state_array[0].fabricclk_mhz = 300; //450.0; - p->in_states->state_array[0].dcfclk_mhz = 200; //300.0; - p->in_states->state_array[0].dispclk_mhz = 2000; //2150.0; - p->in_states->state_array[0].dppclk_mhz = 2000; //2150.0; - p->in_states->state_array[0].dram_speed_mts = 97 * transactions_per_mem_clock; //100 * - - p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; - p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; - p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; - p->in_states->state_array[0].writeback_latency_us = 12; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; - p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 1000; //3000; - p->in_states->state_array[0].sr_exit_z8_time_us = 0; - p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; - p->in_states->state_array[0].dram_clock_change_latency_us = 400; - p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; - p->in_states->state_array[0].sr_exit_time_us = 15.70; //42.97; - p->in_states->state_array[0].sr_enter_plus_exit_time_us = 20.20; //49.94; - p->in_states->state_array[0].fclk_change_latency_us = 0; //20; - p->in_states->state_array[0].usr_retraining_latency_us = 0; //2; - - p->in_states->state_array[1].socclk_mhz = 1600; //1200.0; - p->in_states->state_array[1].fabricclk_mhz = 2500; //2500.0; - p->in_states->state_array[1].dcfclk_mhz = 1800; //1564.0; - p->in_states->state_array[1].dram_speed_mts = 1125 * transactions_per_mem_clock; - break; - } - - /* Override from passed values, if available */ - for (i = 0; i < p->in_states->num_states; i++) { - if (dml2->config.bbox_overrides.sr_exit_latency_us) { - p->in_states->state_array[i].sr_exit_time_us = - dml2->config.bbox_overrides.sr_exit_latency_us; - } - - if (dml2->config.bbox_overrides.sr_enter_plus_exit_latency_us) { - p->in_states->state_array[i].sr_enter_plus_exit_time_us = - dml2->config.bbox_overrides.sr_enter_plus_exit_latency_us; - } - - if (dml2->config.bbox_overrides.sr_exit_z8_time_us) { - p->in_states->state_array[i].sr_exit_z8_time_us = - dml2->config.bbox_overrides.sr_exit_z8_time_us; - } - - if (dml2->config.bbox_overrides.sr_enter_plus_exit_z8_time_us) { - p->in_states->state_array[i].sr_enter_plus_exit_z8_time_us = - dml2->config.bbox_overrides.sr_enter_plus_exit_z8_time_us; - } - - if (dml2->config.bbox_overrides.urgent_latency_us) { - p->in_states->state_array[i].urgent_latency_pixel_data_only_us = - dml2->config.bbox_overrides.urgent_latency_us; - } - - if (dml2->config.bbox_overrides.dram_clock_change_latency_us) { - p->in_states->state_array[i].dram_clock_change_latency_us = - dml2->config.bbox_overrides.dram_clock_change_latency_us; - } - - if (dml2->config.bbox_overrides.fclk_change_latency_us) { - p->in_states->state_array[i].fclk_change_latency_us = - dml2->config.bbox_overrides.fclk_change_latency_us; - } - } - - /* DCFCLK stas values are project specific */ - if ((dml2->v20.dml_core_ctx.project == dml_project_dcn32) || - (dml2->v20.dml_core_ctx.project == dml_project_dcn321)) { - p->dcfclk_stas_mhz[0] = p->in_states->state_array[0].dcfclk_mhz; - p->dcfclk_stas_mhz[1] = 615; - p->dcfclk_stas_mhz[2] = 906; - p->dcfclk_stas_mhz[3] = 1324; - p->dcfclk_stas_mhz[4] = p->in_states->state_array[1].dcfclk_mhz; - } else if (dml2->v20.dml_core_ctx.project != dml_project_dcn35 && - dml2->v20.dml_core_ctx.project != dml_project_dcn36 && - dml2->v20.dml_core_ctx.project != dml_project_dcn351) { - p->dcfclk_stas_mhz[0] = 300; - p->dcfclk_stas_mhz[1] = 615; - p->dcfclk_stas_mhz[2] = 906; - p->dcfclk_stas_mhz[3] = 1324; - p->dcfclk_stas_mhz[4] = 1500; - } - /* Copy clocks tables entries, if available */ - if (dml2->config.bbox_overrides.clks_table.num_states) { - p->in_states->num_states = dml2->config.bbox_overrides.clks_table.num_states; - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dcfclk_levels; i++) { - p->in_states->state_array[i].dcfclk_mhz = dml2->config.bbox_overrides.clks_table.clk_entries[i].dcfclk_mhz; - } - - p->dcfclk_stas_mhz[0] = dml2->config.bbox_overrides.clks_table.clk_entries[0].dcfclk_mhz; - if (i > 1) - p->dcfclk_stas_mhz[4] = dml2->config.bbox_overrides.clks_table.clk_entries[i-1].dcfclk_mhz; - - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_fclk_levels; i++) { - p->in_states->state_array[i].fabricclk_mhz = - dml2->config.bbox_overrides.clks_table.clk_entries[i].fclk_mhz; - } - - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_memclk_levels; i++) { - p->in_states->state_array[i].dram_speed_mts = - dml2->config.bbox_overrides.clks_table.clk_entries[i].memclk_mhz * transactions_per_mem_clock; - } - - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_socclk_levels; i++) { - p->in_states->state_array[i].socclk_mhz = - dml2->config.bbox_overrides.clks_table.clk_entries[i].socclk_mhz; - } - - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dtbclk_levels; i++) { - if (dml2->config.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz > 0) - p->in_states->state_array[i].dtbclk_mhz = - dml2->config.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz; - } - - for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dispclk_levels; i++) { - p->in_states->state_array[i].dispclk_mhz = - dml2->config.bbox_overrides.clks_table.clk_entries[i].dispclk_mhz; - p->in_states->state_array[i].dppclk_mhz = - dml2->config.bbox_overrides.clks_table.clk_entries[i].dppclk_mhz; - } - } - - if (dml2->v20.dml_core_ctx.project == dml_project_dcn35 || - dml2->v20.dml_core_ctx.project == dml_project_dcn36 || - dml2->v20.dml_core_ctx.project == dml_project_dcn351) { - int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, max_phyclk_mhz = 0, - max_dtbclk_mhz = 0, max_fclk_mhz = 0, max_uclk_mhz = 0, max_socclk_mhz = 0; - - for (i = 0; i < p->in_states->num_states; i++) { - if (p->in_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz) - max_dcfclk_mhz = (int)p->in_states->state_array[i].dcfclk_mhz; - if (p->in_states->state_array[i].fabricclk_mhz > max_fclk_mhz) - max_fclk_mhz = (int)p->in_states->state_array[i].fabricclk_mhz; - if (p->in_states->state_array[i].socclk_mhz > max_socclk_mhz) - max_socclk_mhz = (int)p->in_states->state_array[i].socclk_mhz; - if (p->in_states->state_array[i].dram_speed_mts > max_uclk_mhz) - max_uclk_mhz = (int)p->in_states->state_array[i].dram_speed_mts; - if (p->in_states->state_array[i].dispclk_mhz > max_dispclk_mhz) - max_dispclk_mhz = (int)p->in_states->state_array[i].dispclk_mhz; - if (p->in_states->state_array[i].dppclk_mhz > max_dppclk_mhz) - max_dppclk_mhz = (int)p->in_states->state_array[i].dppclk_mhz; - if (p->in_states->state_array[i].phyclk_mhz > max_phyclk_mhz) - max_phyclk_mhz = (int)p->in_states->state_array[i].phyclk_mhz; - if (p->in_states->state_array[i].dtbclk_mhz > max_dtbclk_mhz) - max_dtbclk_mhz = (int)p->in_states->state_array[i].dtbclk_mhz; - } - - for (i = 0; i < p->in_states->num_states; i++) { - /* Independent states - including base (unlisted) parameters from state 0. */ - p->out_states->state_array[i] = p->in_states->state_array[0]; - - p->out_states->state_array[i].dispclk_mhz = max_dispclk_mhz; - p->out_states->state_array[i].dppclk_mhz = max_dppclk_mhz; - p->out_states->state_array[i].dtbclk_mhz = max_dtbclk_mhz; - p->out_states->state_array[i].phyclk_mhz = max_phyclk_mhz; - - p->out_states->state_array[i].dscclk_mhz = max_dispclk_mhz / 3.0; - p->out_states->state_array[i].phyclk_mhz = max_phyclk_mhz; - p->out_states->state_array[i].dtbclk_mhz = max_dtbclk_mhz; - - /* Dependent states. */ - p->out_states->state_array[i].dram_speed_mts = p->in_states->state_array[i].dram_speed_mts; - p->out_states->state_array[i].fabricclk_mhz = p->in_states->state_array[i].fabricclk_mhz; - p->out_states->state_array[i].socclk_mhz = p->in_states->state_array[i].socclk_mhz; - p->out_states->state_array[i].dcfclk_mhz = p->in_states->state_array[i].dcfclk_mhz; - } - - p->out_states->num_states = p->in_states->num_states; - } else { - dml2_policy_build_synthetic_soc_states(s, p); - } -} - -void dml2_translate_ip_params(const struct dc *in, struct ip_params_st *out) -{ - const struct _vcs_dpi_ip_params_st *in_ip_params = &in->dml.ip; - /* Copy over the IP params tp dml2_ctx */ - out->compressed_buffer_segment_size_in_kbytes = in_ip_params->compressed_buffer_segment_size_in_kbytes; - out->config_return_buffer_size_in_kbytes = in_ip_params->config_return_buffer_size_in_kbytes; - out->cursor_buffer_size = in_ip_params->cursor_buffer_size; - out->cursor_chunk_size = in_ip_params->cursor_chunk_size; - out->dcc_meta_buffer_size_bytes = in_ip_params->dcc_meta_buffer_size_bytes; - out->dcc_supported = in_ip_params->dcc_supported; - out->dispclk_delay_subtotal = in_ip_params->dispclk_delay_subtotal; - out->dispclk_ramp_margin_percent = in_ip_params->dispclk_ramp_margin_percent; - out->dppclk_delay_cnvc_cursor = in_ip_params->dppclk_delay_cnvc_cursor; - out->dppclk_delay_cnvc_formatter = in_ip_params->dppclk_delay_cnvc_formatter; - out->dppclk_delay_scl = in_ip_params->dppclk_delay_scl; - out->dppclk_delay_scl_lb_only = in_ip_params->dppclk_delay_scl_lb_only; - out->dppclk_delay_subtotal = in_ip_params->dppclk_delay_subtotal; - out->dpte_buffer_size_in_pte_reqs_chroma = in_ip_params->dpte_buffer_size_in_pte_reqs_chroma; - out->dpte_buffer_size_in_pte_reqs_luma = in_ip_params->dpte_buffer_size_in_pte_reqs_luma; - out->dsc422_native_support = in_ip_params->dsc422_native_support; - out->dynamic_metadata_vm_enabled = in_ip_params->dynamic_metadata_vm_enabled; - out->gpuvm_enable = in_ip_params->gpuvm_enable; - out->gpuvm_max_page_table_levels = in_ip_params->gpuvm_max_page_table_levels; - out->hostvm_enable = in_ip_params->hostvm_enable; - out->hostvm_max_page_table_levels = in_ip_params->hostvm_max_page_table_levels; - out->line_buffer_size_bits = in_ip_params->line_buffer_size_bits; - out->maximum_dsc_bits_per_component = in_ip_params->maximum_dsc_bits_per_component; - out->maximum_pixels_per_line_per_dsc_unit = in_ip_params->maximum_pixels_per_line_per_dsc_unit; - out->max_dchub_pscl_bw_pix_per_clk = in_ip_params->max_dchub_pscl_bw_pix_per_clk; - out->max_hscl_ratio = in_ip_params->max_hscl_ratio; - out->max_hscl_taps = in_ip_params->max_hscl_taps; - out->max_inter_dcn_tile_repeaters = in_ip_params->max_inter_dcn_tile_repeaters; - out->max_lb_vscl_bw_pix_per_clk = in_ip_params->max_lb_vscl_bw_pix_per_clk; - out->max_line_buffer_lines = in_ip_params->max_line_buffer_lines; - out->max_num_dp2p0_outputs = in_ip_params->max_num_dp2p0_outputs; - out->max_num_dp2p0_streams = in_ip_params->max_num_dp2p0_streams; - out->max_num_dpp = in_ip_params->max_num_dpp; - out->max_num_hdmi_frl_outputs = in_ip_params->max_num_hdmi_frl_outputs; - out->max_num_otg = in_ip_params->max_num_otg; - out->max_num_wb = in_ip_params->max_num_wb; - out->max_pscl_lb_bw_pix_per_clk = in_ip_params->max_pscl_lb_bw_pix_per_clk; - out->max_vscl_hscl_bw_pix_per_clk = in_ip_params->max_vscl_hscl_bw_pix_per_clk; - out->max_vscl_ratio = in_ip_params->max_vscl_ratio; - out->max_vscl_taps = in_ip_params->max_vscl_taps; - out->meta_chunk_size_kbytes = in_ip_params->meta_chunk_size_kbytes; - out->meta_fifo_size_in_kentries = in_ip_params->meta_fifo_size_in_kentries; - out->min_meta_chunk_size_bytes = in_ip_params->min_meta_chunk_size_bytes; - out->min_pixel_chunk_size_bytes = in_ip_params->min_pixel_chunk_size_bytes; - out->num_dsc = in_ip_params->num_dsc; - out->pixel_chunk_size_kbytes = in_ip_params->pixel_chunk_size_kbytes; - out->ptoi_supported = in_ip_params->ptoi_supported; - out->rob_buffer_size_kbytes = in_ip_params->rob_buffer_size_kbytes; - out->writeback_chunk_size_kbytes = in_ip_params->writeback_chunk_size_kbytes; - out->writeback_interface_buffer_size_kbytes = in_ip_params->writeback_interface_buffer_size_kbytes; - out->writeback_line_buffer_buffer_size = in_ip_params->writeback_line_buffer_buffer_size; - out->writeback_max_hscl_ratio = in_ip_params->writeback_max_hscl_ratio; - out->writeback_max_hscl_taps = in_ip_params->writeback_max_hscl_taps; - out->writeback_max_vscl_ratio = in_ip_params->writeback_max_vscl_ratio; - out->writeback_max_vscl_taps = in_ip_params->writeback_max_vscl_taps; - out->writeback_min_hscl_ratio = in_ip_params->writeback_min_hscl_ratio; - out->writeback_min_vscl_ratio = in_ip_params->writeback_min_vscl_ratio; - out->zero_size_buffer_entries = in_ip_params->zero_size_buffer_entries; - - /* As per hardcoded reference / discussions */ - out->config_return_buffer_segment_size_in_kbytes = 64; - //out->vblank_nom_default_us = 600; - out->vblank_nom_default_us = in_ip_params->VBlankNomDefaultUS; -} - -void dml2_translate_socbb_params(const struct dc *in, struct soc_bounding_box_st *out) -{ - const struct _vcs_dpi_soc_bounding_box_st *in_soc_params = &in->dml.soc; - /* Copy over the SOCBB params to dml2_ctx */ - out->dispclk_dppclk_vco_speed_mhz = in_soc_params->dispclk_dppclk_vco_speed_mhz; - out->do_urgent_latency_adjustment = in_soc_params->do_urgent_latency_adjustment; - out->dram_channel_width_bytes = (dml_uint_t)in_soc_params->dram_channel_width_bytes; - out->fabric_datapath_to_dcn_data_return_bytes = (dml_uint_t)in_soc_params->fabric_datapath_to_dcn_data_return_bytes; - out->gpuvm_min_page_size_kbytes = in_soc_params->gpuvm_min_page_size_bytes / 1024; - out->hostvm_min_page_size_kbytes = in_soc_params->hostvm_min_page_size_bytes / 1024; - out->mall_allocated_for_dcn_mbytes = (dml_uint_t)in_soc_params->mall_allocated_for_dcn_mbytes; - out->max_avg_dram_bw_use_normal_percent = in_soc_params->max_avg_dram_bw_use_normal_percent; - out->max_avg_fabric_bw_use_normal_percent = in_soc_params->max_avg_fabric_bw_use_normal_percent; - out->max_avg_dram_bw_use_normal_strobe_percent = in_soc_params->max_avg_dram_bw_use_normal_strobe_percent; - out->max_avg_sdp_bw_use_normal_percent = in_soc_params->max_avg_sdp_bw_use_normal_percent; - out->max_outstanding_reqs = in_soc_params->max_request_size_bytes; - out->num_chans = in_soc_params->num_chans; - out->pct_ideal_dram_bw_after_urgent_strobe = in_soc_params->pct_ideal_dram_bw_after_urgent_strobe; - out->pct_ideal_dram_bw_after_urgent_vm_only = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_vm_only; - out->pct_ideal_fabric_bw_after_urgent = in_soc_params->pct_ideal_fabric_bw_after_urgent; - out->pct_ideal_sdp_bw_after_urgent = in_soc_params->pct_ideal_sdp_bw_after_urgent; - out->phy_downspread_percent = in_soc_params->downspread_percent; - out->refclk_mhz = 50; // As per hardcoded reference. - out->return_bus_width_bytes = in_soc_params->return_bus_width_bytes; - out->round_trip_ping_latency_dcfclk_cycles = in_soc_params->round_trip_ping_latency_dcfclk_cycles; - out->smn_latency_us = in_soc_params->smn_latency_us; - out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = in_soc_params->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; - out->urgent_out_of_order_return_per_channel_pixel_only_bytes = in_soc_params->urgent_out_of_order_return_per_channel_pixel_only_bytes; - out->urgent_out_of_order_return_per_channel_vm_only_bytes = in_soc_params->urgent_out_of_order_return_per_channel_vm_only_bytes; - out->pct_ideal_dram_bw_after_urgent_pixel_and_vm = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm; - out->pct_ideal_dram_bw_after_urgent_pixel_only = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_pixel_only; - out->dcn_downspread_percent = in_soc_params->dcn_downspread_percent; -} - -void dml2_translate_soc_states(const struct dc *dc, struct soc_states_st *out, int num_states) -{ - unsigned int i = 0; - out->num_states = num_states; - - for (i = 0; i < out->num_states; i++) { - out->state_array[i].dcfclk_mhz = dc->dml.soc.clock_limits[i].dcfclk_mhz; - out->state_array[i].dispclk_mhz = dc->dml.soc.clock_limits[i].dispclk_mhz; - out->state_array[i].dppclk_mhz = dc->dml.soc.clock_limits[i].dppclk_mhz; - out->state_array[i].dram_speed_mts = dc->dml.soc.clock_limits[i].dram_speed_mts; - out->state_array[i].dtbclk_mhz = dc->dml.soc.clock_limits[i].dtbclk_mhz; - out->state_array[i].socclk_mhz = dc->dml.soc.clock_limits[i].socclk_mhz; - out->state_array[i].fabricclk_mhz = dc->dml.soc.clock_limits[i].fabricclk_mhz; - out->state_array[i].dscclk_mhz = dc->dml.soc.clock_limits[i].dscclk_mhz; - out->state_array[i].phyclk_d18_mhz = dc->dml.soc.clock_limits[i].phyclk_d18_mhz; - out->state_array[i].phyclk_d32_mhz = dc->dml.soc.clock_limits[i].phyclk_d32_mhz; - out->state_array[i].phyclk_mhz = dc->dml.soc.clock_limits[i].phyclk_mhz; - out->state_array[i].sr_enter_plus_exit_time_us = dc->dml.soc.sr_enter_plus_exit_time_us; - out->state_array[i].sr_exit_time_us = dc->dml.soc.sr_exit_time_us; - out->state_array[i].fclk_change_latency_us = dc->dml.soc.fclk_change_latency_us; - out->state_array[i].dram_clock_change_latency_us = dc->dml.soc.dram_clock_change_latency_us; - out->state_array[i].usr_retraining_latency_us = dc->dml.soc.usr_retraining_latency_us; - out->state_array[i].writeback_latency_us = dc->dml.soc.writeback_latency_us; - /* Driver initialized values for these are different than the spreadsheet. Use the - * spreadsheet ones for now. We need to decided which ones to use. - */ - out->state_array[i].sr_exit_z8_time_us = dc->dml.soc.sr_exit_z8_time_us; - out->state_array[i].sr_enter_plus_exit_z8_time_us = dc->dml.soc.sr_enter_plus_exit_z8_time_us; - //out->state_array[i].sr_exit_z8_time_us = 5.20; - //out->state_array[i].sr_enter_plus_exit_z8_time_us = 9.60; - out->state_array[i].use_ideal_dram_bw_strobe = true; - out->state_array[i].urgent_latency_pixel_data_only_us = dc->dml.soc.urgent_latency_pixel_data_only_us; - out->state_array[i].urgent_latency_pixel_mixed_with_vm_data_us = dc->dml.soc.urgent_latency_pixel_mixed_with_vm_data_us; - out->state_array[i].urgent_latency_vm_data_only_us = dc->dml.soc.urgent_latency_vm_data_only_us; - out->state_array[i].urgent_latency_adjustment_fabric_clock_component_us = dc->dml.soc.urgent_latency_adjustment_fabric_clock_component_us; - out->state_array[i].urgent_latency_adjustment_fabric_clock_reference_mhz = dc->dml.soc.urgent_latency_adjustment_fabric_clock_reference_mhz; - } -} - -static void populate_dml_timing_cfg_from_stream_state(struct dml_timing_cfg_st *out, unsigned int location, const struct dc_stream_state *in) -{ - dml_uint_t hblank_start, vblank_start; - - out->HActive[location] = in->timing.h_addressable + in->timing.h_border_left + in->timing.h_border_right; - out->VActive[location] = in->timing.v_addressable + in->timing.v_border_bottom + in->timing.v_border_top; - out->RefreshRate[location] = ((in->timing.pix_clk_100hz * 100) / in->timing.h_total) / in->timing.v_total; - out->VFrontPorch[location] = in->timing.v_front_porch; - out->PixelClock[location] = in->timing.pix_clk_100hz / 10000.00; - if (in->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING) - out->PixelClock[location] *= 2; - out->HTotal[location] = in->timing.h_total; - out->VTotal[location] = in->timing.v_total; - out->Interlace[location] = in->timing.flags.INTERLACE; - hblank_start = in->timing.h_total - in->timing.h_front_porch; - out->HBlankEnd[location] = hblank_start - - in->timing.h_addressable - - in->timing.h_border_left - - in->timing.h_border_right; - vblank_start = in->timing.v_total - in->timing.v_front_porch; - out->VBlankEnd[location] = vblank_start - - in->timing.v_addressable - - in->timing.v_border_top - - in->timing.v_border_bottom; - out->DRRDisplay[location] = false; -} - -static void populate_dml_output_cfg_from_stream_state(struct dml_output_cfg_st *out, unsigned int location, - const struct dc_stream_state *in, const struct pipe_ctx *pipe, struct dml2_context *dml2) -{ - unsigned int output_bpc; - - out->DSCEnable[location] = (enum dml_dsc_enable)in->timing.flags.DSC; - out->OutputLinkDPLanes[location] = 4; // As per code in dcn20_resource.c - out->DSCInputBitPerComponent[location] = 12; // As per code in dcn20_resource.c - out->DSCSlices[location] = in->timing.dsc_cfg.num_slices_h; - - switch (in->signal) { - case SIGNAL_TYPE_DISPLAY_PORT_MST: - case SIGNAL_TYPE_DISPLAY_PORT: - out->OutputEncoder[location] = dml_dp; - if (location < MAX_HPO_DP2_ENCODERS && dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[location] != -1) - out->OutputEncoder[dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[location]] = dml_dp2p0; - break; - case SIGNAL_TYPE_EDP: - out->OutputEncoder[location] = dml_edp; - break; - case SIGNAL_TYPE_HDMI_TYPE_A: - case SIGNAL_TYPE_DVI_SINGLE_LINK: - case SIGNAL_TYPE_DVI_DUAL_LINK: - out->OutputEncoder[location] = dml_hdmi; - break; - default: - out->OutputEncoder[location] = dml_dp; - } - - switch (in->timing.display_color_depth) { - case COLOR_DEPTH_666: - output_bpc = 6; - break; - case COLOR_DEPTH_888: - output_bpc = 8; - break; - case COLOR_DEPTH_101010: - output_bpc = 10; - break; - case COLOR_DEPTH_121212: - output_bpc = 12; - break; - case COLOR_DEPTH_141414: - output_bpc = 14; - break; - case COLOR_DEPTH_161616: - output_bpc = 16; - break; - case COLOR_DEPTH_999: - output_bpc = 9; - break; - case COLOR_DEPTH_111111: - output_bpc = 11; - break; - default: - output_bpc = 8; - break; - } - - switch (in->timing.pixel_encoding) { - case PIXEL_ENCODING_RGB: - case PIXEL_ENCODING_YCBCR444: - out->OutputFormat[location] = dml_444; - out->OutputBpp[location] = (dml_float_t)output_bpc * 3; - break; - case PIXEL_ENCODING_YCBCR420: - out->OutputFormat[location] = dml_420; - out->OutputBpp[location] = (output_bpc * 3.0) / 2; - break; - case PIXEL_ENCODING_YCBCR422: - if (in->timing.flags.DSC && !in->timing.dsc_cfg.ycbcr422_simple) - out->OutputFormat[location] = dml_n422; - else - out->OutputFormat[location] = dml_s422; - out->OutputBpp[location] = (dml_float_t)output_bpc * 2; - break; - default: - out->OutputFormat[location] = dml_444; - out->OutputBpp[location] = (dml_float_t)output_bpc * 3; - break; - } - - if (in->timing.flags.DSC) { - out->OutputBpp[location] = in->timing.dsc_cfg.bits_per_pixel / 16.0; - } - - // This has been false throughout DCN32x development. If needed we can change this later on. - out->OutputMultistreamEn[location] = false; - - switch (in->signal) { - case SIGNAL_TYPE_NONE: - case SIGNAL_TYPE_DVI_SINGLE_LINK: - case SIGNAL_TYPE_DVI_DUAL_LINK: - case SIGNAL_TYPE_HDMI_TYPE_A: - case SIGNAL_TYPE_LVDS: - case SIGNAL_TYPE_RGB: - case SIGNAL_TYPE_DISPLAY_PORT: - case SIGNAL_TYPE_DISPLAY_PORT_MST: - case SIGNAL_TYPE_EDP: - case SIGNAL_TYPE_VIRTUAL: - default: - out->OutputLinkDPRate[location] = dml_dp_rate_na; - break; - } - - out->PixelClockBackEnd[location] = in->timing.pix_clk_100hz / 10000.00; - - out->AudioSampleLayout[location] = in->audio_info.modes->sample_size; - out->AudioSampleRate[location] = in->audio_info.modes->max_bit_rate; - - out->OutputDisabled[location] = true; -} - -static void populate_dummy_dml_surface_cfg(struct dml_surface_cfg_st *out, unsigned int location, const struct dc_stream_state *in) -{ - out->SurfaceWidthY[location] = in->timing.h_addressable; - out->SurfaceHeightY[location] = in->timing.v_addressable; - out->SurfaceWidthC[location] = in->timing.h_addressable; - out->SurfaceHeightC[location] = in->timing.v_addressable; - out->PitchY[location] = ((out->SurfaceWidthY[location] + 127) / 128) * 128; - out->PitchC[location] = 1; - out->DCCEnable[location] = false; - out->DCCMetaPitchY[location] = 0; - out->DCCMetaPitchC[location] = 0; - out->DCCRateLuma[location] = 1.0; - out->DCCRateChroma[location] = 1.0; - out->DCCFractionOfZeroSizeRequestsLuma[location] = 0; - out->DCCFractionOfZeroSizeRequestsChroma[location] = 0; - out->SurfaceTiling[location] = dml_sw_64kb_r_x; - out->SourcePixelFormat[location] = dml_444_32; -} - -static void populate_dml_surface_cfg_from_plane_state(enum dml_project_id dml2_project, struct dml_surface_cfg_st *out, unsigned int location, const struct dc_plane_state *in) -{ - out->PitchY[location] = in->plane_size.surface_pitch; - out->SurfaceHeightY[location] = in->plane_size.surface_size.height; - out->SurfaceWidthY[location] = in->plane_size.surface_size.width; - out->SurfaceHeightC[location] = in->plane_size.chroma_size.height; - out->SurfaceWidthC[location] = in->plane_size.chroma_size.width; - out->PitchC[location] = in->plane_size.chroma_pitch; - out->DCCEnable[location] = in->dcc.enable; - out->DCCMetaPitchY[location] = in->dcc.meta_pitch; - out->DCCMetaPitchC[location] = in->dcc.meta_pitch_c; - out->DCCRateLuma[location] = 1.0; - out->DCCRateChroma[location] = 1.0; - out->DCCFractionOfZeroSizeRequestsLuma[location] = in->dcc.independent_64b_blks; - out->DCCFractionOfZeroSizeRequestsChroma[location] = in->dcc.independent_64b_blks_c; - - switch (dml2_project) { - default: - out->SurfaceTiling[location] = (enum dml_swizzle_mode)in->tiling_info.gfx9.swizzle; - break; - case dml_project_dcn401: - // Temporary use gfx11 swizzle in dml, until proper dml for DCN4x is integrated/implemented - switch (in->tiling_info.gfx_addr3.swizzle) { - case DC_ADDR3_SW_4KB_2D: - case DC_ADDR3_SW_64KB_2D: - case DC_ADDR3_SW_256KB_2D: - default: - out->SurfaceTiling[location] = dml_sw_64kb_r_x; - break; - case DC_ADDR3_SW_LINEAR: - out->SurfaceTiling[location] = dml_sw_linear; - break; - } - } - - switch (in->format) { - case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr: - case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: - out->SourcePixelFormat[location] = dml_420_8; - break; - case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: - case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: - out->SourcePixelFormat[location] = dml_420_10; - break; - case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: - case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: - case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: - case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: - out->SourcePixelFormat[location] = dml_444_64; - break; - case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555: - case SURFACE_PIXEL_FORMAT_GRPH_RGB565: - out->SourcePixelFormat[location] = dml_444_16; - break; - case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS: - out->SourcePixelFormat[location] = dml_444_8; - break; - case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA: - out->SourcePixelFormat[location] = dml_rgbe_alpha; - break; - default: - out->SourcePixelFormat[location] = dml_444_32; - break; - } -} - -static struct scaler_data *get_scaler_data_for_plane( - const struct dc_plane_state *in, - struct dc_state *context) -{ - int i; - struct pipe_ctx *temp_pipe = &context->res_ctx.temp_pipe; - - memset(temp_pipe, 0, sizeof(struct pipe_ctx)); - - for (i = 0; i < MAX_PIPES; i++) { - const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state == in && !pipe->prev_odm_pipe) { - temp_pipe->stream = pipe->stream; - temp_pipe->plane_state = pipe->plane_state; - temp_pipe->plane_res.scl_data.taps = pipe->plane_res.scl_data.taps; - temp_pipe->stream_res = pipe->stream_res; - resource_build_scaling_params(temp_pipe); - break; - } - } - - ASSERT(i < MAX_PIPES); - return &temp_pipe->plane_res.scl_data; -} - -static void populate_dummy_dml_plane_cfg(struct dml_plane_cfg_st *out, unsigned int location, - const struct dc_stream_state *in, - const struct soc_bounding_box_st *soc) -{ - dml_uint_t width, height; - - if (in->timing.h_addressable > 3840) - width = 3840; - else - width = in->timing.h_addressable; // 4K max - - if (in->timing.v_addressable > 2160) - height = 2160; - else - height = in->timing.v_addressable; // 4K max - - out->CursorBPP[location] = dml_cur_32bit; - out->CursorWidth[location] = 256; - - out->GPUVMMinPageSizeKBytes[location] = soc->gpuvm_min_page_size_kbytes; - - out->ViewportWidth[location] = width; - out->ViewportHeight[location] = height; - out->ViewportStationary[location] = false; - out->ViewportWidthChroma[location] = 0; - out->ViewportHeightChroma[location] = 0; - out->ViewportXStart[location] = 0; - out->ViewportXStartC[location] = 0; - out->ViewportYStart[location] = 0; - out->ViewportYStartC[location] = 0; - - out->ScalerEnabled[location] = false; - out->HRatio[location] = 1.0; - out->VRatio[location] = 1.0; - out->HRatioChroma[location] = 0; - out->VRatioChroma[location] = 0; - out->HTaps[location] = 1; - out->VTaps[location] = 1; - out->HTapsChroma[location] = 0; - out->VTapsChroma[location] = 0; - out->SourceScan[location] = dml_rotation_0; - out->ScalerRecoutWidth[location] = width; - - out->LBBitPerPixel[location] = 57; - - out->DynamicMetadataEnable[location] = false; - - out->NumberOfCursors[location] = 1; - out->UseMALLForStaticScreen[location] = dml_use_mall_static_screen_disable; - out->UseMALLForPStateChange[location] = dml_use_mall_pstate_change_disable; - - out->DETSizeOverride[location] = 256; - - out->ScalerEnabled[location] = false; -} - -static void populate_dml_plane_cfg_from_plane_state(struct dml_plane_cfg_st *out, unsigned int location, - const struct dc_plane_state *in, struct dc_state *context, - const struct soc_bounding_box_st *soc) -{ - struct scaler_data *scaler_data = get_scaler_data_for_plane(in, context); - - out->CursorBPP[location] = dml_cur_32bit; - out->CursorWidth[location] = 256; - - out->GPUVMMinPageSizeKBytes[location] = soc->gpuvm_min_page_size_kbytes; - - out->ViewportWidth[location] = scaler_data->viewport.width; - out->ViewportHeight[location] = scaler_data->viewport.height; - out->ViewportWidthChroma[location] = scaler_data->viewport_c.width; - out->ViewportHeightChroma[location] = scaler_data->viewport_c.height; - out->ViewportXStart[location] = scaler_data->viewport.x; - out->ViewportYStart[location] = scaler_data->viewport.y; - out->ViewportXStartC[location] = scaler_data->viewport_c.x; - out->ViewportYStartC[location] = scaler_data->viewport_c.y; - out->ViewportStationary[location] = false; - - out->ScalerEnabled[location] = scaler_data->ratios.horz.value != dc_fixpt_one.value || - scaler_data->ratios.horz_c.value != dc_fixpt_one.value || - scaler_data->ratios.vert.value != dc_fixpt_one.value || - scaler_data->ratios.vert_c.value != dc_fixpt_one.value; - - /* Current driver code base uses LBBitPerPixel as 57. There is a discrepancy - * from the HW/DML teams about this value. Initialize LBBitPerPixel with the - * value current used in Navi3x . - */ - - out->LBBitPerPixel[location] = 57; - - if (out->ScalerEnabled[location] == false) { - out->HRatio[location] = 1; - out->HRatioChroma[location] = 1; - out->VRatio[location] = 1; - out->VRatioChroma[location] = 1; - } else { - /* Follow the original dml_wrapper.c code direction to fix scaling issues */ - out->HRatio[location] = (dml_float_t)scaler_data->ratios.horz.value / (1ULL << 32); - out->HRatioChroma[location] = (dml_float_t)scaler_data->ratios.horz_c.value / (1ULL << 32); - out->VRatio[location] = (dml_float_t)scaler_data->ratios.vert.value / (1ULL << 32); - out->VRatioChroma[location] = (dml_float_t)scaler_data->ratios.vert_c.value / (1ULL << 32); - } - - if (!scaler_data->taps.h_taps) { - out->HTaps[location] = 1; - out->HTapsChroma[location] = 1; - } else { - out->HTaps[location] = scaler_data->taps.h_taps; - out->HTapsChroma[location] = scaler_data->taps.h_taps_c; - } - if (!scaler_data->taps.v_taps) { - out->VTaps[location] = 1; - out->VTapsChroma[location] = 1; - } else { - out->VTaps[location] = scaler_data->taps.v_taps; - out->VTapsChroma[location] = scaler_data->taps.v_taps_c; - } - - out->SourceScan[location] = (enum dml_rotation_angle)in->rotation; - out->ScalerRecoutWidth[location] = in->dst_rect.width; - - out->DynamicMetadataEnable[location] = false; - out->DynamicMetadataLinesBeforeActiveRequired[location] = 0; - out->DynamicMetadataTransmittedBytes[location] = 0; - - out->NumberOfCursors[location] = 1; -} - -static unsigned int map_stream_to_dml_display_cfg(const struct dml2_context *dml2, - const struct dc_stream_state *stream, const struct dml_display_cfg_st *dml_dispcfg) -{ - int i = 0; - int location = -1; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] && dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == stream->stream_id) { - location = i; - break; - } - } - - return location; -} - -static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *context, const struct dc_plane_state *plane, - unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) -{ - int i, j; - bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; - - if (!plane_id) - return false; - - for (i = 0; i < context->stream_count; i++) { - if (context->streams[i]->stream_id == stream_id) { - for (j = 0; j < context->stream_status[i].plane_count; j++) { - if (context->stream_status[i].plane_states[j] == plane && - (!is_plane_duplicate || (j == plane_index))) { - *plane_id = (i << 16) | j; - return true; - } - } - } - } - - return false; -} - -static unsigned int map_plane_to_dml_display_cfg(const struct dml2_context *dml2, const struct dc_plane_state *plane, - const struct dc_state *context, const struct dml_display_cfg_st *dml_dispcfg, unsigned int stream_id, int plane_index) -{ - unsigned int plane_id; - int i = 0; - int location = -1; - - if (!get_plane_id(context->bw_ctx.dml2, context, plane, stream_id, plane_index, &plane_id)) { - ASSERT(false); - return -1; - } - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] && dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i] == plane_id) { - location = i; - break; - } - } - - return location; -} - -static void dml2_populate_pipe_to_plane_index_mapping(struct dml2_context *dml2, struct dc_state *state) -{ - unsigned int i; - unsigned int pipe_index = 0; - unsigned int plane_index = 0; - struct dml2_dml_to_dc_pipe_mapping *dml_to_dc_pipe_mapping = &dml2->v20.scratch.dml_to_dc_pipe_mapping; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[i] = false; - dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index[i] = 0; - } - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; - - if (!pipe || !pipe->stream || !pipe->plane_state) - continue; - - while (pipe) { - pipe_index = pipe->pipe_idx; - - if (pipe->stream && dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[pipe_index] == false) { - dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index[pipe_index] = plane_index; - plane_index++; - dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[pipe_index] = true; - } - - pipe = pipe->bottom_pipe; - } - - plane_index = 0; - } -} - -static void populate_dml_writeback_cfg_from_stream_state(struct dml_writeback_cfg_st *out, - unsigned int location, const struct dc_stream_state *in) -{ - if (in->num_wb_info > 0) { - for (int i = 0; i < __DML_NUM_DMB__; i++) { - const struct dc_writeback_info *wb_info = &in->writeback_info[i]; - /*current dml support 1 dwb per stream, limitation*/ - if (wb_info->wb_enabled) { - out->WritebackEnable[location] = wb_info->wb_enabled; - out->ActiveWritebacksPerSurface[location] = wb_info->dwb_params.cnv_params.src_width; - out->WritebackDestinationWidth[location] = wb_info->dwb_params.dest_width; - out->WritebackDestinationHeight[location] = wb_info->dwb_params.dest_height; - - out->WritebackSourceWidth[location] = wb_info->dwb_params.cnv_params.crop_en ? - wb_info->dwb_params.cnv_params.crop_width : - wb_info->dwb_params.cnv_params.src_width; - - out->WritebackSourceHeight[location] = wb_info->dwb_params.cnv_params.crop_en ? - wb_info->dwb_params.cnv_params.crop_height : - wb_info->dwb_params.cnv_params.src_height; - /*current design does not have chroma scaling, need to follow up*/ - out->WritebackHTaps[location] = wb_info->dwb_params.scaler_taps.h_taps > 0 ? - wb_info->dwb_params.scaler_taps.h_taps : 1; - out->WritebackVTaps[location] = wb_info->dwb_params.scaler_taps.v_taps > 0 ? - wb_info->dwb_params.scaler_taps.v_taps : 1; - - out->WritebackHRatio[location] = wb_info->dwb_params.cnv_params.crop_en ? - (double)wb_info->dwb_params.cnv_params.crop_width / - (double)wb_info->dwb_params.dest_width : - (double)wb_info->dwb_params.cnv_params.src_width / - (double)wb_info->dwb_params.dest_width; - out->WritebackVRatio[location] = wb_info->dwb_params.cnv_params.crop_en ? - (double)wb_info->dwb_params.cnv_params.crop_height / - (double)wb_info->dwb_params.dest_height : - (double)wb_info->dwb_params.cnv_params.src_height / - (double)wb_info->dwb_params.dest_height; - } - } - } -} - -static void dml2_map_hpo_stream_encoder_to_hpo_link_encoder_index(struct dml2_context *dml2, struct dc_state *context) -{ - int i; - struct pipe_ctx *current_pipe_context; - - /* Scratch gets reset to zero in dml, but link encoder instance can be zero, so reset to -1 */ - for (i = 0; i < MAX_HPO_DP2_ENCODERS; i++) { - dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[i] = -1; - } - - /* If an HPO stream encoder is allocated to a pipe, get the instance of it's allocated HPO Link encoder */ - for (i = 0; i < MAX_PIPES; i++) { - current_pipe_context = &context->res_ctx.pipe_ctx[i]; - if (current_pipe_context->stream && - current_pipe_context->stream_res.hpo_dp_stream_enc && - current_pipe_context->link_res.hpo_dp_link_enc && - dc_is_dp_signal(current_pipe_context->stream->signal)) { - dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[current_pipe_context->stream_res.hpo_dp_stream_enc->inst] = - current_pipe_context->link_res.hpo_dp_link_enc->inst; - } - } -} - -void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_state *context, struct dml_display_cfg_st *dml_dispcfg) -{ - int i = 0, j = 0, k = 0; - int disp_cfg_stream_location, disp_cfg_plane_location; - enum mall_stream_type stream_mall_type; - struct pipe_ctx *current_pipe_context; - - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] = false; - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] = false; - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] = false; - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] = false; - } - - //Generally these are set by referencing our latest BB/IP params in dcn32_resource.c file - dml_dispcfg->plane.GPUVMEnable = dml2->v20.dml_core_ctx.ip.gpuvm_enable; - dml_dispcfg->plane.GPUVMMaxPageTableLevels = dml2->v20.dml_core_ctx.ip.gpuvm_max_page_table_levels; - dml_dispcfg->plane.HostVMEnable = dml2->v20.dml_core_ctx.ip.hostvm_enable; - dml_dispcfg->plane.HostVMMaxPageTableLevels = dml2->v20.dml_core_ctx.ip.hostvm_max_page_table_levels; - if (dml2->v20.dml_core_ctx.ip.hostvm_enable) - dml2->v20.dml_core_ctx.policy.AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter; - - dml2_populate_pipe_to_plane_index_mapping(dml2, context); - dml2_map_hpo_stream_encoder_to_hpo_link_encoder_index(dml2, context); - - for (i = 0; i < context->stream_count; i++) { - current_pipe_context = NULL; - for (k = 0; k < MAX_PIPES; k++) { - /* find one pipe allocated to this stream for the purpose of getting - info about the link later */ - if (context->streams[i] == context->res_ctx.pipe_ctx[k].stream) { - current_pipe_context = &context->res_ctx.pipe_ctx[k]; - break; - } - } - disp_cfg_stream_location = map_stream_to_dml_display_cfg(dml2, context->streams[i], dml_dispcfg); - stream_mall_type = dc_state_get_stream_subvp_type(context, context->streams[i]); - - if (disp_cfg_stream_location < 0) - disp_cfg_stream_location = dml_dispcfg->num_timings++; - - ASSERT(disp_cfg_stream_location >= 0 && disp_cfg_stream_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); - - populate_dml_timing_cfg_from_stream_state(&dml_dispcfg->timing, disp_cfg_stream_location, context->streams[i]); - populate_dml_output_cfg_from_stream_state(&dml_dispcfg->output, disp_cfg_stream_location, context->streams[i], current_pipe_context, dml2); - /*Call site for populate_dml_writeback_cfg_from_stream_state*/ - populate_dml_writeback_cfg_from_stream_state(&dml_dispcfg->writeback, - disp_cfg_stream_location, context->streams[i]); - - switch (context->streams[i]->debug.force_odm_combine_segments) { - case 2: - dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_stream_location] = dml_odm_use_policy_combine_2to1; - break; - case 4: - dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_stream_location] = dml_odm_use_policy_combine_4to1; - break; - default: - break; - } - - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_stream_location] = context->streams[i]->stream_id; - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_stream_location] = true; - - if (context->stream_status[i].plane_count == 0) { - disp_cfg_plane_location = dml_dispcfg->num_surfaces++; - - populate_dummy_dml_surface_cfg(&dml_dispcfg->surface, disp_cfg_plane_location, context->streams[i]); - populate_dummy_dml_plane_cfg(&dml_dispcfg->plane, disp_cfg_plane_location, - context->streams[i], &dml2->v20.dml_core_ctx.soc); - - dml_dispcfg->plane.BlendingAndTiming[disp_cfg_plane_location] = disp_cfg_stream_location; - - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; - } else { - for (j = 0; j < context->stream_status[i].plane_count; j++) { - disp_cfg_plane_location = map_plane_to_dml_display_cfg(dml2, - context->stream_status[i].plane_states[j], context, dml_dispcfg, context->streams[i]->stream_id, j); - - if (disp_cfg_plane_location < 0) - disp_cfg_plane_location = dml_dispcfg->num_surfaces++; - - ASSERT(disp_cfg_plane_location >= 0 && disp_cfg_plane_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); - - populate_dml_surface_cfg_from_plane_state(dml2->v20.dml_core_ctx.project, &dml_dispcfg->surface, disp_cfg_plane_location, context->stream_status[i].plane_states[j]); - populate_dml_plane_cfg_from_plane_state( - &dml_dispcfg->plane, disp_cfg_plane_location, - context->stream_status[i].plane_states[j], context, - &dml2->v20.dml_core_ctx.soc); - - if (stream_mall_type == SUBVP_MAIN) { - dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_sub_viewport; - dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_optimize; - } else if (stream_mall_type == SUBVP_PHANTOM) { - dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_phantom_pipe; - dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_disable; - dml2->v20.dml_core_ctx.policy.ImmediateFlipRequirement[disp_cfg_plane_location] = dml_immediate_flip_not_required; - } else { - dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_disable; - dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_optimize; - } - - dml_dispcfg->plane.BlendingAndTiming[disp_cfg_plane_location] = disp_cfg_stream_location; - - if (get_plane_id(dml2, context, context->stream_status[i].plane_states[j], context->streams[i]->stream_id, j, - &dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; - - if (j >= 1) { - populate_dml_timing_cfg_from_stream_state(&dml_dispcfg->timing, disp_cfg_plane_location, context->streams[i]); - populate_dml_output_cfg_from_stream_state(&dml_dispcfg->output, disp_cfg_plane_location, context->streams[i], current_pipe_context, dml2); - switch (context->streams[i]->debug.force_odm_combine_segments) { - case 2: - dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_plane_location] = dml_odm_use_policy_combine_2to1; - break; - case 4: - dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_plane_location] = dml_odm_use_policy_combine_4to1; - break; - default: - break; - } - - if (stream_mall_type == SUBVP_MAIN) - dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_sub_viewport; - else if (stream_mall_type == SUBVP_PHANTOM) - dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_phantom_pipe; - - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_plane_location] = context->streams[i]->stream_id; - dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_plane_location] = true; - - dml_dispcfg->num_timings++; - } - } - } - } -} - -void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, - struct _vcs_dpi_dml_display_dlg_regs_st *disp_dlg_regs, - struct _vcs_dpi_dml_display_ttu_regs_st *disp_ttu_regs, - struct pipe_ctx *out) -{ - memset(&out->rq_regs, 0, sizeof(out->rq_regs)); - out->rq_regs.rq_regs_l.chunk_size = rq_regs->rq_regs_l.chunk_size; - out->rq_regs.rq_regs_l.min_chunk_size = rq_regs->rq_regs_l.min_chunk_size; - out->rq_regs.rq_regs_l.meta_chunk_size = rq_regs->rq_regs_l.meta_chunk_size; - out->rq_regs.rq_regs_l.min_meta_chunk_size = rq_regs->rq_regs_l.min_meta_chunk_size; - out->rq_regs.rq_regs_l.dpte_group_size = rq_regs->rq_regs_l.dpte_group_size; - out->rq_regs.rq_regs_l.mpte_group_size = rq_regs->rq_regs_l.mpte_group_size; - out->rq_regs.rq_regs_l.swath_height = rq_regs->rq_regs_l.swath_height; - out->rq_regs.rq_regs_l.pte_row_height_linear = rq_regs->rq_regs_l.pte_row_height_linear; - - out->rq_regs.rq_regs_c.chunk_size = rq_regs->rq_regs_c.chunk_size; - out->rq_regs.rq_regs_c.min_chunk_size = rq_regs->rq_regs_c.min_chunk_size; - out->rq_regs.rq_regs_c.meta_chunk_size = rq_regs->rq_regs_c.meta_chunk_size; - out->rq_regs.rq_regs_c.min_meta_chunk_size = rq_regs->rq_regs_c.min_meta_chunk_size; - out->rq_regs.rq_regs_c.dpte_group_size = rq_regs->rq_regs_c.dpte_group_size; - out->rq_regs.rq_regs_c.mpte_group_size = rq_regs->rq_regs_c.mpte_group_size; - out->rq_regs.rq_regs_c.swath_height = rq_regs->rq_regs_c.swath_height; - out->rq_regs.rq_regs_c.pte_row_height_linear = rq_regs->rq_regs_c.pte_row_height_linear; - - out->rq_regs.drq_expansion_mode = rq_regs->drq_expansion_mode; - out->rq_regs.prq_expansion_mode = rq_regs->prq_expansion_mode; - out->rq_regs.mrq_expansion_mode = rq_regs->mrq_expansion_mode; - out->rq_regs.crq_expansion_mode = rq_regs->crq_expansion_mode; - out->rq_regs.plane1_base_address = rq_regs->plane1_base_address; - - memset(&out->dlg_regs, 0, sizeof(out->dlg_regs)); - out->dlg_regs.refcyc_h_blank_end = disp_dlg_regs->refcyc_h_blank_end; - out->dlg_regs.dlg_vblank_end = disp_dlg_regs->dlg_vblank_end; - out->dlg_regs.min_dst_y_next_start = disp_dlg_regs->min_dst_y_next_start; - out->dlg_regs.refcyc_per_htotal = disp_dlg_regs->refcyc_per_htotal; - out->dlg_regs.refcyc_x_after_scaler = disp_dlg_regs->refcyc_x_after_scaler; - out->dlg_regs.dst_y_after_scaler = disp_dlg_regs->dst_y_after_scaler; - out->dlg_regs.dst_y_prefetch = disp_dlg_regs->dst_y_prefetch; - out->dlg_regs.dst_y_per_vm_vblank = disp_dlg_regs->dst_y_per_vm_vblank; - out->dlg_regs.dst_y_per_row_vblank = disp_dlg_regs->dst_y_per_row_vblank; - out->dlg_regs.dst_y_per_vm_flip = disp_dlg_regs->dst_y_per_vm_flip; - out->dlg_regs.dst_y_per_row_flip = disp_dlg_regs->dst_y_per_row_flip; - out->dlg_regs.ref_freq_to_pix_freq = disp_dlg_regs->ref_freq_to_pix_freq; - out->dlg_regs.vratio_prefetch = disp_dlg_regs->vratio_prefetch; - out->dlg_regs.vratio_prefetch_c = disp_dlg_regs->vratio_prefetch_c; - out->dlg_regs.refcyc_per_pte_group_vblank_l = disp_dlg_regs->refcyc_per_pte_group_vblank_l; - out->dlg_regs.refcyc_per_pte_group_vblank_c = disp_dlg_regs->refcyc_per_pte_group_vblank_c; - out->dlg_regs.refcyc_per_meta_chunk_vblank_l = disp_dlg_regs->refcyc_per_meta_chunk_vblank_l; - out->dlg_regs.refcyc_per_meta_chunk_vblank_c = disp_dlg_regs->refcyc_per_meta_chunk_vblank_c; - out->dlg_regs.refcyc_per_pte_group_flip_l = disp_dlg_regs->refcyc_per_pte_group_flip_l; - out->dlg_regs.refcyc_per_pte_group_flip_c = disp_dlg_regs->refcyc_per_pte_group_flip_c; - out->dlg_regs.refcyc_per_meta_chunk_flip_l = disp_dlg_regs->refcyc_per_meta_chunk_flip_l; - out->dlg_regs.refcyc_per_meta_chunk_flip_c = disp_dlg_regs->refcyc_per_meta_chunk_flip_c; - out->dlg_regs.dst_y_per_pte_row_nom_l = disp_dlg_regs->dst_y_per_pte_row_nom_l; - out->dlg_regs.dst_y_per_pte_row_nom_c = disp_dlg_regs->dst_y_per_pte_row_nom_c; - out->dlg_regs.refcyc_per_pte_group_nom_l = disp_dlg_regs->refcyc_per_pte_group_nom_l; - out->dlg_regs.refcyc_per_pte_group_nom_c = disp_dlg_regs->refcyc_per_pte_group_nom_c; - out->dlg_regs.dst_y_per_meta_row_nom_l = disp_dlg_regs->dst_y_per_meta_row_nom_l; - out->dlg_regs.dst_y_per_meta_row_nom_c = disp_dlg_regs->dst_y_per_meta_row_nom_c; - out->dlg_regs.refcyc_per_meta_chunk_nom_l = disp_dlg_regs->refcyc_per_meta_chunk_nom_l; - out->dlg_regs.refcyc_per_meta_chunk_nom_c = disp_dlg_regs->refcyc_per_meta_chunk_nom_c; - out->dlg_regs.refcyc_per_line_delivery_pre_l = disp_dlg_regs->refcyc_per_line_delivery_pre_l; - out->dlg_regs.refcyc_per_line_delivery_pre_c = disp_dlg_regs->refcyc_per_line_delivery_pre_c; - out->dlg_regs.refcyc_per_line_delivery_l = disp_dlg_regs->refcyc_per_line_delivery_l; - out->dlg_regs.refcyc_per_line_delivery_c = disp_dlg_regs->refcyc_per_line_delivery_c; - out->dlg_regs.refcyc_per_vm_group_vblank = disp_dlg_regs->refcyc_per_vm_group_vblank; - out->dlg_regs.refcyc_per_vm_group_flip = disp_dlg_regs->refcyc_per_vm_group_flip; - out->dlg_regs.refcyc_per_vm_req_vblank = disp_dlg_regs->refcyc_per_vm_req_vblank; - out->dlg_regs.refcyc_per_vm_req_flip = disp_dlg_regs->refcyc_per_vm_req_flip; - out->dlg_regs.dst_y_offset_cur0 = disp_dlg_regs->dst_y_offset_cur0; - out->dlg_regs.chunk_hdl_adjust_cur0 = disp_dlg_regs->chunk_hdl_adjust_cur0; - out->dlg_regs.dst_y_offset_cur1 = disp_dlg_regs->dst_y_offset_cur1; - out->dlg_regs.chunk_hdl_adjust_cur1 = disp_dlg_regs->chunk_hdl_adjust_cur1; - out->dlg_regs.vready_after_vcount0 = disp_dlg_regs->vready_after_vcount0; - out->dlg_regs.dst_y_delta_drq_limit = disp_dlg_regs->dst_y_delta_drq_limit; - out->dlg_regs.refcyc_per_vm_dmdata = disp_dlg_regs->refcyc_per_vm_dmdata; - out->dlg_regs.dmdata_dl_delta = disp_dlg_regs->dmdata_dl_delta; - - memset(&out->ttu_regs, 0, sizeof(out->ttu_regs)); - out->ttu_regs.qos_level_low_wm = disp_ttu_regs->qos_level_low_wm; - out->ttu_regs.qos_level_high_wm = disp_ttu_regs->qos_level_high_wm; - out->ttu_regs.min_ttu_vblank = disp_ttu_regs->min_ttu_vblank; - out->ttu_regs.qos_level_flip = disp_ttu_regs->qos_level_flip; - out->ttu_regs.refcyc_per_req_delivery_l = disp_ttu_regs->refcyc_per_req_delivery_l; - out->ttu_regs.refcyc_per_req_delivery_c = disp_ttu_regs->refcyc_per_req_delivery_c; - out->ttu_regs.refcyc_per_req_delivery_cur0 = disp_ttu_regs->refcyc_per_req_delivery_cur0; - out->ttu_regs.refcyc_per_req_delivery_cur1 = disp_ttu_regs->refcyc_per_req_delivery_cur1; - out->ttu_regs.refcyc_per_req_delivery_pre_l = disp_ttu_regs->refcyc_per_req_delivery_pre_l; - out->ttu_regs.refcyc_per_req_delivery_pre_c = disp_ttu_regs->refcyc_per_req_delivery_pre_c; - out->ttu_regs.refcyc_per_req_delivery_pre_cur0 = disp_ttu_regs->refcyc_per_req_delivery_pre_cur0; - out->ttu_regs.refcyc_per_req_delivery_pre_cur1 = disp_ttu_regs->refcyc_per_req_delivery_pre_cur1; - out->ttu_regs.qos_level_fixed_l = disp_ttu_regs->qos_level_fixed_l; - out->ttu_regs.qos_level_fixed_c = disp_ttu_regs->qos_level_fixed_c; - out->ttu_regs.qos_level_fixed_cur0 = disp_ttu_regs->qos_level_fixed_cur0; - out->ttu_regs.qos_level_fixed_cur1 = disp_ttu_regs->qos_level_fixed_cur1; - out->ttu_regs.qos_ramp_disable_l = disp_ttu_regs->qos_ramp_disable_l; - out->ttu_regs.qos_ramp_disable_c = disp_ttu_regs->qos_ramp_disable_c; - out->ttu_regs.qos_ramp_disable_cur0 = disp_ttu_regs->qos_ramp_disable_cur0; - out->ttu_regs.qos_ramp_disable_cur1 = disp_ttu_regs->qos_ramp_disable_cur1; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h deleted file mode 100644 index d764773938f4..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h +++ /dev/null @@ -1,41 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DML2_TRANSLATION_HELPER_H__ -#define __DML2_TRANSLATION_HELPER_H__ - -void dml2_init_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out); -void dml2_init_socbb_params(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out); -void dml2_init_soc_states(struct dml2_context *dml2, const struct dc *in_dc, - const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out); -void dml2_translate_ip_params(const struct dc *in_dc, struct ip_params_st *out); -void dml2_translate_socbb_params(const struct dc *in_dc, struct soc_bounding_box_st *out); -void dml2_translate_soc_states(const struct dc *in_dc, struct soc_states_st *out, int num_states); -void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_state *context, struct dml_display_cfg_st *dml_dispcfg); -void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, struct _vcs_dpi_dml_display_dlg_regs_st *disp_dlg_regs, struct _vcs_dpi_dml_display_ttu_regs_st *disp_ttu_regs, struct pipe_ctx *out); -bool is_dp2p0_output_encoder(const struct pipe_ctx *pipe); - -#endif //__DML2_TRANSLATION_HELPER_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c deleted file mode 100644 index 9a33158b63bf..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c +++ /dev/null @@ -1,560 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -//#include "dml2_utils.h" -#include "display_mode_core.h" -#include "dml_display_rq_dlg_calc.h" -#include "dml2_internal_types.h" -#include "dml2_translation_helper.h" -#include "dml2_utils.h" - -void dml2_util_copy_dml_timing(struct dml_timing_cfg_st *dml_timing_array, unsigned int dst_index, unsigned int src_index) -{ - dml_timing_array->HTotal[dst_index] = dml_timing_array->HTotal[src_index]; - dml_timing_array->VTotal[dst_index] = dml_timing_array->VTotal[src_index]; - dml_timing_array->HBlankEnd[dst_index] = dml_timing_array->HBlankEnd[src_index]; - dml_timing_array->VBlankEnd[dst_index] = dml_timing_array->VBlankEnd[src_index]; - dml_timing_array->RefreshRate[dst_index] = dml_timing_array->RefreshRate[src_index]; - dml_timing_array->VFrontPorch[dst_index] = dml_timing_array->VFrontPorch[src_index]; - dml_timing_array->PixelClock[dst_index] = dml_timing_array->PixelClock[src_index]; - dml_timing_array->HActive[dst_index] = dml_timing_array->HActive[src_index]; - dml_timing_array->VActive[dst_index] = dml_timing_array->VActive[src_index]; - dml_timing_array->Interlace[dst_index] = dml_timing_array->Interlace[src_index]; - dml_timing_array->DRRDisplay[dst_index] = dml_timing_array->DRRDisplay[src_index]; - dml_timing_array->VBlankNom[dst_index] = dml_timing_array->VBlankNom[src_index]; -} - -void dml2_util_copy_dml_plane(struct dml_plane_cfg_st *dml_plane_array, unsigned int dst_index, unsigned int src_index) -{ - dml_plane_array->GPUVMMinPageSizeKBytes[dst_index] = dml_plane_array->GPUVMMinPageSizeKBytes[src_index]; - dml_plane_array->ForceOneRowForFrame[dst_index] = dml_plane_array->ForceOneRowForFrame[src_index]; - dml_plane_array->PTEBufferModeOverrideEn[dst_index] = dml_plane_array->PTEBufferModeOverrideEn[src_index]; - dml_plane_array->PTEBufferMode[dst_index] = dml_plane_array->PTEBufferMode[src_index]; - dml_plane_array->ViewportWidth[dst_index] = dml_plane_array->ViewportWidth[src_index]; - dml_plane_array->ViewportHeight[dst_index] = dml_plane_array->ViewportHeight[src_index]; - dml_plane_array->ViewportWidthChroma[dst_index] = dml_plane_array->ViewportWidthChroma[src_index]; - dml_plane_array->ViewportHeightChroma[dst_index] = dml_plane_array->ViewportHeightChroma[src_index]; - dml_plane_array->ViewportXStart[dst_index] = dml_plane_array->ViewportXStart[src_index]; - dml_plane_array->ViewportXStartC[dst_index] = dml_plane_array->ViewportXStartC[src_index]; - dml_plane_array->ViewportYStart[dst_index] = dml_plane_array->ViewportYStart[src_index]; - dml_plane_array->ViewportYStartC[dst_index] = dml_plane_array->ViewportYStartC[src_index]; - dml_plane_array->ViewportStationary[dst_index] = dml_plane_array->ViewportStationary[src_index]; - - dml_plane_array->ScalerEnabled[dst_index] = dml_plane_array->ScalerEnabled[src_index]; - dml_plane_array->HRatio[dst_index] = dml_plane_array->HRatio[src_index]; - dml_plane_array->VRatio[dst_index] = dml_plane_array->VRatio[src_index]; - dml_plane_array->HRatioChroma[dst_index] = dml_plane_array->HRatioChroma[src_index]; - dml_plane_array->VRatioChroma[dst_index] = dml_plane_array->VRatioChroma[src_index]; - dml_plane_array->HTaps[dst_index] = dml_plane_array->HTaps[src_index]; - dml_plane_array->VTaps[dst_index] = dml_plane_array->VTaps[src_index]; - dml_plane_array->HTapsChroma[dst_index] = dml_plane_array->HTapsChroma[src_index]; - dml_plane_array->VTapsChroma[dst_index] = dml_plane_array->VTapsChroma[src_index]; - dml_plane_array->LBBitPerPixel[dst_index] = dml_plane_array->LBBitPerPixel[src_index]; - - dml_plane_array->SourceScan[dst_index] = dml_plane_array->SourceScan[src_index]; - dml_plane_array->ScalerRecoutWidth[dst_index] = dml_plane_array->ScalerRecoutWidth[src_index]; - - dml_plane_array->DynamicMetadataEnable[dst_index] = dml_plane_array->DynamicMetadataEnable[src_index]; - dml_plane_array->DynamicMetadataLinesBeforeActiveRequired[dst_index] = dml_plane_array->DynamicMetadataLinesBeforeActiveRequired[src_index]; - dml_plane_array->DynamicMetadataTransmittedBytes[dst_index] = dml_plane_array->DynamicMetadataTransmittedBytes[src_index]; - dml_plane_array->DETSizeOverride[dst_index] = dml_plane_array->DETSizeOverride[src_index]; - - dml_plane_array->NumberOfCursors[dst_index] = dml_plane_array->NumberOfCursors[src_index]; - dml_plane_array->CursorWidth[dst_index] = dml_plane_array->CursorWidth[src_index]; - dml_plane_array->CursorBPP[dst_index] = dml_plane_array->CursorBPP[src_index]; - - dml_plane_array->UseMALLForStaticScreen[dst_index] = dml_plane_array->UseMALLForStaticScreen[src_index]; - dml_plane_array->UseMALLForPStateChange[dst_index] = dml_plane_array->UseMALLForPStateChange[src_index]; - - dml_plane_array->BlendingAndTiming[dst_index] = dml_plane_array->BlendingAndTiming[src_index]; -} - -void dml2_util_copy_dml_surface(struct dml_surface_cfg_st *dml_surface_array, unsigned int dst_index, unsigned int src_index) -{ - dml_surface_array->SurfaceTiling[dst_index] = dml_surface_array->SurfaceTiling[src_index]; - dml_surface_array->SourcePixelFormat[dst_index] = dml_surface_array->SourcePixelFormat[src_index]; - dml_surface_array->PitchY[dst_index] = dml_surface_array->PitchY[src_index]; - dml_surface_array->SurfaceWidthY[dst_index] = dml_surface_array->SurfaceWidthY[src_index]; - dml_surface_array->SurfaceHeightY[dst_index] = dml_surface_array->SurfaceHeightY[src_index]; - dml_surface_array->PitchC[dst_index] = dml_surface_array->PitchC[src_index]; - dml_surface_array->SurfaceWidthC[dst_index] = dml_surface_array->SurfaceWidthC[src_index]; - dml_surface_array->SurfaceHeightC[dst_index] = dml_surface_array->SurfaceHeightC[src_index]; - - dml_surface_array->DCCEnable[dst_index] = dml_surface_array->DCCEnable[src_index]; - dml_surface_array->DCCMetaPitchY[dst_index] = dml_surface_array->DCCMetaPitchY[src_index]; - dml_surface_array->DCCMetaPitchC[dst_index] = dml_surface_array->DCCMetaPitchC[src_index]; - - dml_surface_array->DCCRateLuma[dst_index] = dml_surface_array->DCCRateLuma[src_index]; - dml_surface_array->DCCRateChroma[dst_index] = dml_surface_array->DCCRateChroma[src_index]; - dml_surface_array->DCCFractionOfZeroSizeRequestsLuma[dst_index] = dml_surface_array->DCCFractionOfZeroSizeRequestsLuma[src_index]; - dml_surface_array->DCCFractionOfZeroSizeRequestsChroma[dst_index] = dml_surface_array->DCCFractionOfZeroSizeRequestsChroma[src_index]; -} - -void dml2_util_copy_dml_output(struct dml_output_cfg_st *dml_output_array, unsigned int dst_index, unsigned int src_index) -{ - dml_output_array->DSCInputBitPerComponent[dst_index] = dml_output_array->DSCInputBitPerComponent[src_index]; - dml_output_array->OutputFormat[dst_index] = dml_output_array->OutputFormat[src_index]; - dml_output_array->OutputEncoder[dst_index] = dml_output_array->OutputEncoder[src_index]; - dml_output_array->OutputMultistreamId[dst_index] = dml_output_array->OutputMultistreamId[src_index]; - dml_output_array->OutputMultistreamEn[dst_index] = dml_output_array->OutputMultistreamEn[src_index]; - dml_output_array->OutputBpp[dst_index] = dml_output_array->OutputBpp[src_index]; - dml_output_array->PixelClockBackEnd[dst_index] = dml_output_array->PixelClockBackEnd[src_index]; - dml_output_array->DSCEnable[dst_index] = dml_output_array->DSCEnable[src_index]; - dml_output_array->OutputLinkDPLanes[dst_index] = dml_output_array->OutputLinkDPLanes[src_index]; - dml_output_array->OutputLinkDPRate[dst_index] = dml_output_array->OutputLinkDPRate[src_index]; - dml_output_array->ForcedOutputLinkBPP[dst_index] = dml_output_array->ForcedOutputLinkBPP[src_index]; - dml_output_array->AudioSampleRate[dst_index] = dml_output_array->AudioSampleRate[src_index]; - dml_output_array->AudioSampleLayout[dst_index] = dml_output_array->AudioSampleLayout[src_index]; -} - -unsigned int dml2_util_get_maximum_odm_combine_for_output(bool force_odm_4to1, enum dml_output_encoder_class encoder, bool dsc_enabled) -{ - switch (encoder) { - case dml_dp: - case dml_edp: - return 2; - case dml_dp2p0: - if (dsc_enabled || force_odm_4to1) - return 4; - else - return 2; - case dml_hdmi: - return 1; - case dml_hdmifrl: - if (force_odm_4to1) - return 4; - else - return 2; - default: - return 1; - } -} - -bool is_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx) -{ - if (pipe_ctx == NULL || pipe_ctx->stream == NULL) - return false; - - /* If this assert is hit then we have a link encoder dynamic management issue */ - ASSERT(pipe_ctx->stream_res.hpo_dp_stream_enc ? pipe_ctx->link_res.hpo_dp_link_enc != NULL : true); - - return (pipe_ctx->stream_res.hpo_dp_stream_enc && - pipe_ctx->link_res.hpo_dp_link_enc && - dc_is_dp_signal(pipe_ctx->stream->signal)); -} - -bool is_dtbclk_required(const struct dc *dc, struct dc_state *context) -{ - int i; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (!context->res_ctx.pipe_ctx[i].stream) - continue; - if (is_dp2p0_output_encoder(&context->res_ctx.pipe_ctx[i])) - return true; - } - return false; -} - -void dml2_copy_clocks_to_dc_state(struct dml2_dcn_clocks *out_clks, struct dc_state *context) -{ - context->bw_ctx.bw.dcn.clk.dispclk_khz = out_clks->dispclk_khz; - context->bw_ctx.bw.dcn.clk.dcfclk_khz = out_clks->dcfclk_khz; - context->bw_ctx.bw.dcn.clk.dramclk_khz = out_clks->uclk_mts / 16; - context->bw_ctx.bw.dcn.clk.fclk_khz = out_clks->fclk_khz; - context->bw_ctx.bw.dcn.clk.phyclk_khz = out_clks->phyclk_khz; - context->bw_ctx.bw.dcn.clk.socclk_khz = out_clks->socclk_khz; - context->bw_ctx.bw.dcn.clk.ref_dtbclk_khz = out_clks->ref_dtbclk_khz; - context->bw_ctx.bw.dcn.clk.p_state_change_support = out_clks->p_state_supported; -} - -int dml2_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id) -{ - int i; - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] && ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] == stream_id) - return i; - } - - return -1; -} - -static int find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id) -{ - int i; - for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { - if (ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] && ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] == plane_id) - return i; - } - - return -1; -} - -static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *state, const struct dc_plane_state *plane, - unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) -{ - unsigned int i, j; - bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; - - if (!plane_id) - return false; - - for (i = 0; i < state->stream_count; i++) { - if (state->streams[i]->stream_id == stream_id) { - for (j = 0; j < state->stream_status[i].plane_count; j++) { - if (state->stream_status[i].plane_states[j] == plane && - (!is_plane_duplicate || (j == plane_index))) { - *plane_id = (i << 16) | j; - return true; - } - } - } - } - - return false; -} - -static void populate_pipe_ctx_dlg_params_from_dml(struct pipe_ctx *pipe_ctx, struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) -{ - unsigned int hactive, vactive, hblank_start, vblank_start, hblank_end, vblank_end; - struct dc_crtc_timing *timing = &pipe_ctx->stream->timing; - - hactive = timing->h_addressable + timing->h_border_left + timing->h_border_right; - vactive = timing->v_addressable + timing->v_border_bottom + timing->v_border_top; - hblank_start = pipe_ctx->stream->timing.h_total - pipe_ctx->stream->timing.h_front_porch; - vblank_start = pipe_ctx->stream->timing.v_total - pipe_ctx->stream->timing.v_front_porch; - - hblank_end = hblank_start - timing->h_addressable - timing->h_border_left - timing->h_border_right; - vblank_end = vblank_start - timing->v_addressable - timing->v_border_top - timing->v_border_bottom; - - pipe_ctx->pipe_dlg_param.vstartup_start = dml_get_vstartup_calculated(mode_lib, pipe_idx); - pipe_ctx->pipe_dlg_param.vupdate_offset = dml_get_vupdate_offset(mode_lib, pipe_idx); - pipe_ctx->pipe_dlg_param.vupdate_width = dml_get_vupdate_width(mode_lib, pipe_idx); - pipe_ctx->pipe_dlg_param.vready_offset = dml_get_vready_offset(mode_lib, pipe_idx); - - pipe_ctx->pipe_dlg_param.otg_inst = pipe_ctx->stream_res.tg->inst; - - pipe_ctx->pipe_dlg_param.hactive = hactive; - pipe_ctx->pipe_dlg_param.vactive = vactive; - pipe_ctx->pipe_dlg_param.htotal = pipe_ctx->stream->timing.h_total; - pipe_ctx->pipe_dlg_param.vtotal = pipe_ctx->stream->timing.v_total; - pipe_ctx->pipe_dlg_param.hblank_end = hblank_end; - pipe_ctx->pipe_dlg_param.vblank_end = vblank_end; - pipe_ctx->pipe_dlg_param.hblank_start = hblank_start; - pipe_ctx->pipe_dlg_param.vblank_start = vblank_start; - pipe_ctx->pipe_dlg_param.vfront_porch = pipe_ctx->stream->timing.v_front_porch; - pipe_ctx->pipe_dlg_param.pixel_rate_mhz = pipe_ctx->stream->timing.pix_clk_100hz / 10000.00; - pipe_ctx->pipe_dlg_param.refresh_rate = ((timing->pix_clk_100hz * 100) / timing->h_total) / timing->v_total; - pipe_ctx->pipe_dlg_param.vtotal_max = pipe_ctx->stream->adjust.v_total_max; - pipe_ctx->pipe_dlg_param.vtotal_min = pipe_ctx->stream->adjust.v_total_min; - pipe_ctx->pipe_dlg_param.recout_height = pipe_ctx->plane_res.scl_data.recout.height; - pipe_ctx->pipe_dlg_param.recout_width = pipe_ctx->plane_res.scl_data.recout.width; - pipe_ctx->pipe_dlg_param.full_recout_height = pipe_ctx->plane_res.scl_data.recout.height; - pipe_ctx->pipe_dlg_param.full_recout_width = pipe_ctx->plane_res.scl_data.recout.width; -} - -void dml2_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, struct dml2_context *in_ctx, unsigned int pipe_cnt) -{ - unsigned int dc_pipe_ctx_index, dml_pipe_idx, plane_id; - enum mall_stream_type pipe_mall_type; - struct dml2_calculate_rq_and_dlg_params_scratch *s = &in_ctx->v20.scratch.calculate_rq_and_dlg_params_scratch; - - context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = (unsigned int)in_ctx->v20.dml_core_ctx.mp.DCFCLKDeepSleep * 1000; - context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; - - if (in_ctx->v20.dml_core_ctx.ms.support.FCLKChangeSupport[0] == dml_fclock_change_unsupported) - context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = false; - else - context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = true; - - if (context->bw_ctx.bw.dcn.clk.dispclk_khz < dc->debug.min_disp_clk_khz) - context->bw_ctx.bw.dcn.clk.dispclk_khz = dc->debug.min_disp_clk_khz; - - context->bw_ctx.bw.dcn.compbuf_size_kb = in_ctx->v20.dml_core_ctx.ip.config_return_buffer_size_in_kbytes; - - for (dc_pipe_ctx_index = 0; dc_pipe_ctx_index < pipe_cnt; dc_pipe_ctx_index++) { - if (!context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream) - continue; - /* The DML2 and the DC logic of determining pipe indices are different from each other so - * there is a need to know which DML pipe index maps to which DC pipe. The code below - * finds a dml_pipe_index from the plane id if a plane is valid. If a plane is not valid then - * it finds a dml_pipe_index from the stream id. */ - if (get_plane_id(in_ctx, context, context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state, - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id, - in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[context->res_ctx.pipe_ctx[dc_pipe_ctx_index].pipe_idx], &plane_id)) { - dml_pipe_idx = find_dml_pipe_idx_by_plane_id(in_ctx, plane_id); - } else { - dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(in_ctx, context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id); - } - - if (dml_pipe_idx == 0xFFFFFFFF) - continue; - ASSERT(in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[dml_pipe_idx]); - ASSERT(in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_pipe_idx] == context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id); - - /* Use the dml_pipe_index here for the getters to fetch the correct values and dc_pipe_index in the pipe_ctx to populate them - * at the right locations. - */ - populate_pipe_ctx_dlg_params_from_dml(&context->res_ctx.pipe_ctx[dc_pipe_ctx_index], &context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); - - pipe_mall_type = dc_state_get_pipe_subvp_type(context, &context->res_ctx.pipe_ctx[dc_pipe_ctx_index]); - if (pipe_mall_type == SUBVP_PHANTOM) { - // Phantom pipe requires that DET_SIZE = 0 and no unbounded requests - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb = 0; - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].unbounded_req = false; - } else { - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb = dml_get_det_buffer_size_kbytes(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); - // Unbounded requesting should not ever be used when more than 1 pipe is enabled. - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].unbounded_req = in_ctx->v20.dml_core_ctx.ms.UnboundedRequestEnabledThisState; - } - - context->bw_ctx.bw.dcn.compbuf_size_kb -= context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb; - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz = dml_get_dppclk_calculated(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx) * 1000; - if (context->bw_ctx.bw.dcn.clk.dppclk_khz < context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz) - context->bw_ctx.bw.dcn.clk.dppclk_khz = context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz; - - dml_rq_dlg_get_rq_reg(&s->rq_regs, &in_ctx->v20.dml_core_ctx, dml_pipe_idx); - dml_rq_dlg_get_dlg_reg(&s->disp_dlg_regs, &s->disp_ttu_regs, &in_ctx->v20.dml_core_ctx, dml_pipe_idx); - dml2_update_pipe_ctx_dchub_regs(&s->rq_regs, &s->disp_dlg_regs, &s->disp_ttu_regs, &out_new_hw_state->pipe_ctx[dc_pipe_ctx_index]); - - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes = dml_get_surface_size_for_mall(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); - - /* Reuse MALL Allocation Sizes logic from dcn32_fpu.c */ - /* Count from active, top pipes per plane only. Only add mall_ss_size_bytes for each unique plane. */ - if (context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream && context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state && - (context->res_ctx.pipe_ctx[dc_pipe_ctx_index].top_pipe == NULL || - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state != context->res_ctx.pipe_ctx[dc_pipe_ctx_index].top_pipe->plane_state) && - context->res_ctx.pipe_ctx[dc_pipe_ctx_index].prev_odm_pipe == NULL) { - /* SS: all active surfaces stored in MALL */ - if (pipe_mall_type != SUBVP_PHANTOM) { - context->bw_ctx.bw.dcn.mall_ss_size_bytes += context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes; - } else { - /* SUBVP: phantom surfaces only stored in MALL */ - context->bw_ctx.bw.dcn.mall_subvp_size_bytes += context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes; - } - } - } - - context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; - context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; - - context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = in_ctx->v20.dml_core_ctx.states.state_array[in_ctx->v20.scratch.mode_support_params.out_lowest_state_idx].dppclk_mhz - * 1000; - context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = in_ctx->v20.dml_core_ctx.states.state_array[in_ctx->v20.scratch.mode_support_params.out_lowest_state_idx].dispclk_mhz - * 1000; - - if (dc->config.forced_clocks || dc->debug.max_disp_clk) { - context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz; - context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz ; - } -} - -void dml2_extract_watermark_set(struct dcn_watermarks *watermark, struct display_mode_lib_st *dml_core_ctx) -{ - watermark->urgent_ns = dml_get_wm_urgent(dml_core_ctx) * 1000; - watermark->cstate_pstate.cstate_enter_plus_exit_ns = dml_get_wm_stutter_enter_exit(dml_core_ctx) * 1000; - watermark->cstate_pstate.cstate_exit_ns = dml_get_wm_stutter_exit(dml_core_ctx) * 1000; - watermark->cstate_pstate.pstate_change_ns = dml_get_wm_dram_clock_change(dml_core_ctx) * 1000; - watermark->pte_meta_urgent_ns = dml_get_wm_memory_trip(dml_core_ctx) * 1000; - watermark->frac_urg_bw_nom = dml_get_fraction_of_urgent_bandwidth(dml_core_ctx) * 1000; - watermark->frac_urg_bw_flip = dml_get_fraction_of_urgent_bandwidth_imm_flip(dml_core_ctx) * 1000; - watermark->urgent_latency_ns = dml_get_urgent_latency(dml_core_ctx) * 1000; - watermark->cstate_pstate.fclk_pstate_change_ns = dml_get_wm_fclk_change(dml_core_ctx) * 1000; - watermark->usr_retraining_ns = dml_get_wm_usr_retraining(dml_core_ctx) * 1000; - watermark->cstate_pstate.cstate_enter_plus_exit_z8_ns = dml_get_wm_z8_stutter_enter_exit(dml_core_ctx) * 1000; - watermark->cstate_pstate.cstate_exit_z8_ns = dml_get_wm_z8_stutter(dml_core_ctx) * 1000; -} - -unsigned int dml2_calc_max_scaled_time( - unsigned int time_per_pixel, - enum mmhubbub_wbif_mode mode, - unsigned int urgent_watermark) -{ - unsigned int time_per_byte = 0; - unsigned int total_free_entry = 0xb40; - unsigned int buf_lh_capability; - unsigned int max_scaled_time; - - if (mode == PACKED_444) /* packed mode 32 bpp */ - time_per_byte = time_per_pixel/4; - else if (mode == PACKED_444_FP16) /* packed mode 64 bpp */ - time_per_byte = time_per_pixel/8; - - if (time_per_byte == 0) - time_per_byte = 1; - - buf_lh_capability = (total_free_entry*time_per_byte*32) >> 6; /* time_per_byte is in u6.6*/ - max_scaled_time = buf_lh_capability - urgent_watermark; - return max_scaled_time; -} - -void dml2_extract_writeback_wm(struct dc_state *context, struct display_mode_lib_st *dml_core_ctx) -{ - int i, j = 0; - struct mcif_arb_params *wb_arb_params = NULL; - struct dcn_bw_writeback *bw_writeback = NULL; - enum mmhubbub_wbif_mode wbif_mode = PACKED_444_FP16; /*for now*/ - - if (context->stream_count != 0) { - for (i = 0; i < context->stream_count; i++) { - if (context->streams[i]->num_wb_info != 0) - j++; - } - } - if (j == 0) /*no dwb */ - return; - for (i = 0; i < __DML_NUM_DMB__; i++) { - bw_writeback = &context->bw_ctx.bw.dcn.bw_writeback; - wb_arb_params = &context->bw_ctx.bw.dcn.bw_writeback.mcif_wb_arb[i]; - - for (j = 0 ; j < 4; j++) { - /*current dml only has one set of watermark, need to follow up*/ - bw_writeback->mcif_wb_arb[i].cli_watermark[j] = - dml_get_wm_writeback_urgent(dml_core_ctx) * 1000; - bw_writeback->mcif_wb_arb[i].pstate_watermark[j] = - dml_get_wm_writeback_dram_clock_change(dml_core_ctx) * 1000; - } - if (context->res_ctx.pipe_ctx[i].stream->phy_pix_clk != 0) { - /* time_per_pixel should be in u6.6 format */ - bw_writeback->mcif_wb_arb[i].time_per_pixel = - (1000000 << 6) / context->res_ctx.pipe_ctx[i].stream->phy_pix_clk; - } - bw_writeback->mcif_wb_arb[i].slice_lines = 32; - bw_writeback->mcif_wb_arb[i].arbitration_slice = 2; - bw_writeback->mcif_wb_arb[i].max_scaled_time = - dml2_calc_max_scaled_time(wb_arb_params->time_per_pixel, - wbif_mode, wb_arb_params->cli_watermark[0]); - /*not required any more*/ - bw_writeback->mcif_wb_arb[i].dram_speed_change_duration = - dml_get_wm_writeback_dram_clock_change(dml_core_ctx) * 1000; - - } -} -void dml2_initialize_det_scratch(struct dml2_context *in_ctx) -{ - int i; - - for (i = 0; i < MAX_PLANES; i++) { - in_ctx->det_helper_scratch.dpps_per_surface[i] = 1; - } -} - -static unsigned int find_planes_per_stream_and_stream_count(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg, int *num_of_planes_per_stream) -{ - unsigned int plane_index, stream_index = 0, num_of_streams; - - for (plane_index = 0; plane_index < dml_dispcfg->num_surfaces; plane_index++) { - /* Number of planes per stream */ - num_of_planes_per_stream[stream_index] += 1; - - if (plane_index + 1 < dml_dispcfg->num_surfaces && dml_dispcfg->plane.BlendingAndTiming[plane_index] != dml_dispcfg->plane.BlendingAndTiming[plane_index + 1]) - stream_index++; - } - - num_of_streams = stream_index + 1; - - return num_of_streams; -} - -void dml2_apply_det_buffer_allocation_policy(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg) -{ - unsigned int num_of_streams = 0, plane_index = 0, max_det_size, stream_index = 0; - int num_of_planes_per_stream[__DML_NUM_PLANES__] = { 0 }; - - max_det_size = in_ctx->config.det_segment_size * in_ctx->config.max_segments_per_hubp; - - num_of_streams = find_planes_per_stream_and_stream_count(in_ctx, dml_dispcfg, num_of_planes_per_stream); - - for (plane_index = 0; plane_index < dml_dispcfg->num_surfaces; plane_index++) { - - if (in_ctx->config.override_det_buffer_size_kbytes) - dml_dispcfg->plane.DETSizeOverride[plane_index] = max_det_size / in_ctx->config.dcn_pipe_count; - else { - dml_dispcfg->plane.DETSizeOverride[plane_index] = ((max_det_size / num_of_streams) / num_of_planes_per_stream[stream_index] / in_ctx->det_helper_scratch.dpps_per_surface[plane_index]); - - /* If the override size is not divisible by det_segment_size then round off to nearest number divisible by det_segment_size as - * this is a requirement. - */ - if (dml_dispcfg->plane.DETSizeOverride[plane_index] % in_ctx->config.det_segment_size != 0) { - dml_dispcfg->plane.DETSizeOverride[plane_index] = dml_dispcfg->plane.DETSizeOverride[plane_index] & ~0x3F; - } - - if (plane_index + 1 < dml_dispcfg->num_surfaces && dml_dispcfg->plane.BlendingAndTiming[plane_index] != dml_dispcfg->plane.BlendingAndTiming[plane_index + 1]) - stream_index++; - } - } -} - -bool dml2_verify_det_buffer_configuration(struct dml2_context *in_ctx, struct dc_state *display_state, struct dml2_helper_det_policy_scratch *det_scratch) -{ - unsigned int i = 0, dml_pipe_idx = 0, plane_id = 0; - unsigned int max_det_size, total_det_allocated = 0; - bool need_recalculation = false; - - max_det_size = in_ctx->config.det_segment_size * in_ctx->config.max_segments_per_hubp; - - for (i = 0; i < MAX_PIPES; i++) { - if (!display_state->res_ctx.pipe_ctx[i].stream) - continue; - if (get_plane_id(in_ctx, display_state, display_state->res_ctx.pipe_ctx[i].plane_state, - display_state->res_ctx.pipe_ctx[i].stream->stream_id, - in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[display_state->res_ctx.pipe_ctx[i].pipe_idx], &plane_id)) - dml_pipe_idx = find_dml_pipe_idx_by_plane_id(in_ctx, plane_id); - else - dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(in_ctx, display_state->res_ctx.pipe_ctx[i].stream->stream_id); - - if (dml_pipe_idx == 0xFFFFFFFF) - continue; - total_det_allocated += dml_get_det_buffer_size_kbytes(&in_ctx->v20.dml_core_ctx, dml_pipe_idx); - if (total_det_allocated > max_det_size) { - need_recalculation = true; - } - } - - /* Store the DPPPerSurface for correctly determining the number of planes in the next call. */ - for (i = 0; i < MAX_PLANES; i++) { - det_scratch->dpps_per_surface[i] = in_ctx->v20.scratch.cur_display_config.hw.DPPPerSurface[i]; - } - - return need_recalculation; -} - -bool dml2_is_stereo_timing(const struct dc_stream_state *stream) -{ - bool is_stereo = false; - - if ((stream->view_format == - VIEW_3D_FORMAT_SIDE_BY_SIDE || - stream->view_format == - VIEW_3D_FORMAT_TOP_AND_BOTTOM) && - (stream->timing.timing_3d_format == - TIMING_3D_FORMAT_TOP_AND_BOTTOM || - stream->timing.timing_3d_format == - TIMING_3D_FORMAT_SIDE_BY_SIDE)) - is_stereo = true; - - return is_stereo; -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h deleted file mode 100644 index 04fcfe637119..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h +++ /dev/null @@ -1,149 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef _DML2_UTILS_H_ -#define _DML2_UTILS_H_ - -#include "os_types.h" -#include "dml2_dc_types.h" - -struct dc; -struct dml_timing_cfg_st; -struct dml2_dcn_clocks; -struct dc_state; - -void dml2_util_copy_dml_timing(struct dml_timing_cfg_st *dml_timing_array, unsigned int dst_index, unsigned int src_index); -void dml2_util_copy_dml_plane(struct dml_plane_cfg_st *dml_plane_array, unsigned int dst_index, unsigned int src_index); -void dml2_util_copy_dml_surface(struct dml_surface_cfg_st *dml_surface_array, unsigned int dst_index, unsigned int src_index); -void dml2_util_copy_dml_output(struct dml_output_cfg_st *dml_output_array, unsigned int dst_index, unsigned int src_index); -unsigned int dml2_util_get_maximum_odm_combine_for_output(bool force_odm_4to1, enum dml_output_encoder_class encoder, bool dsc_enabled); -void dml2_copy_clocks_to_dc_state(struct dml2_dcn_clocks *out_clks, struct dc_state *context); -void dml2_extract_watermark_set(struct dcn_watermarks *watermark, struct display_mode_lib_st *dml_core_ctx); -void dml2_extract_writeback_wm(struct dc_state *context, struct display_mode_lib_st *dml_core_ctx); -int dml2_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id); -bool is_dtbclk_required(const struct dc *dc, struct dc_state *context); -bool dml2_is_stereo_timing(const struct dc_stream_state *stream); -unsigned int dml2_calc_max_scaled_time( - unsigned int time_per_pixel, - enum mmhubbub_wbif_mode mode, - unsigned int urgent_watermark); - -/* - * dml2_dc_construct_pipes - This function will determine if we need additional pipes based - * on the DML calculated outputs for MPC, ODM and allocate them as necessary. This function - * could be called after in dml_validate_build_resource after dml_mode_pragramming like : - * { - * ... - * map_hw_resources(&s->cur_display_config, &s->mode_support_info); - * result = dml_mode_programming(&in_ctx->dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); - * dml2_dc_construct_pipes(in_display_state, s->mode_support_info, out_hw_context); - * ... - * } - * - * @context: To obtain res_ctx and read other information like stream ID etc. - * @dml_mode_support_st : To get the ODM, MPC outputs as determined by the DML. - * @out_hw_context : Handle to the new hardware context. - * - * - * Return: None. - */ -void dml2_dc_construct_pipes(struct dc_state *context, struct dml_mode_support_info_st *dml_mode_support_st, - struct resource_context *out_hw_context); - -/* - * dml2_predict_pipe_split - This function is the dml2 version of predict split pipe. It predicts a - * if pipe split is required or not and returns the output as a bool. - * @context : dc_state. - * @pipe : old_index is the index of the pipe as derived from pipe_idx. - * @index : index of the pipe - * - * - * Return: Returns the result in boolean. - */ -bool dml2_predict_pipe_split(struct dc_state *context, display_pipe_params_st pipe, int index); - -/* - * dml2_build_mapped_resource - This function is the dml2 version of build_mapped_resource. - * In case of ODM, we need to build pipe hardware params again as done in dcn20_build_mapped_resource. - * @dc : struct dc - * @context : struct dc_state. - * @stream : stream whoose corresponding pipe params need to be modified. - * - * - * Return: Returns DC_OK if successful. - */ -enum dc_status dml2_build_mapped_resource(const struct dc *dc, struct dc_state *context, struct dc_stream_state *stream); - -/* - * dml2_extract_rq_regs - This function will extract information needed for struct _vcs_dpi_display_rq_regs_st - * and populate it. - * @context: To obtain and populate the res_ctx->pipe_ctx->rq_regs with DML outputs. - * @support : This structure has the DML intermediate outputs required to populate rq_regs. - * - * - * Return: None. - */ - - /* - * dml2_calculate_rq_and_dlg_params - This function will call into DML2 functions needed - * for populating rq, ttu and dlg param structures and populate it. - * @dc : struct dc - * @context : dc_state provides a handle to selectively populate pipe_ctx - * @out_new_hw_state: To obtain and populate the rq, dlg and ttu regs in - * out_new_hw_state->pipe_ctx with DML outputs. - * @in_ctx : This structure has the pointer to display_mode_lib_st. - * @pipe_cnt : DML functions to obtain RQ, TTu and DLG params need a pipe_index. - * This helps provide pipe_index in the pipe_cnt loop. - * - * - * Return: None. - */ -void dml2_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, struct dml2_context *in_ctx, unsigned int pipe_cnt); - -/* - * dml2_apply_det_buffer_allocation_policy - This function will determine the DET Buffer size - * and return the number of streams. - * @dml2 : Handle for dml2 context - * @dml_dispcfg : dml_dispcfg is the DML2 struct representing the current display config - * Return : None. - */ -void dml2_apply_det_buffer_allocation_policy(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg); - -/* - * dml2_verify_det_buffer_configuration - This function will verify if the allocated DET buffer exceeds - * the total available DET size available and outputs a boolean to indicate if recalulation is needed. - * @dml2 : Handle for dml2 context - * @dml_dispcfg : dml_dispcfg is the DML2 struct representing the current display config - * @struct dml2_helper_det_policy_scratch : Pointer to DET helper scratch - * Return : returns true if recalculation is required, false otherwise. - */ -bool dml2_verify_det_buffer_configuration(struct dml2_context *in_ctx, struct dc_state *display_state, struct dml2_helper_det_policy_scratch *det_scratch); - -/* - * dml2_initialize_det_scratch - This function will initialize the DET scratch space as per requirements. - * @dml2 : Handle for dml2 context - * Return : None - */ -void dml2_initialize_det_scratch(struct dml2_context *in_ctx); -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c deleted file mode 100644 index 9deb03a18ccc..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c +++ /dev/null @@ -1,704 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#include "display_mode_core.h" -#include "dml2_internal_types.h" -#include "dml2_utils.h" -#include "dml2_policy.h" -#include "dml2_translation_helper.h" -#include "dml2_mall_phantom.h" -#include "dml2_dc_resource_mgmt.h" -#include "dml21_wrapper.h" - -static void initialize_dml2_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out) -{ - if (dml2->config.use_native_soc_bb_construction) - dml2_init_ip_params(dml2, in_dc, out); - else - dml2_translate_ip_params(in_dc, out); -} - -static void initialize_dml2_soc_bbox(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out) -{ - if (dml2->config.use_native_soc_bb_construction) - dml2_init_socbb_params(dml2, in_dc, out); - else - dml2_translate_socbb_params(in_dc, out); -} - -static void initialize_dml2_soc_states(struct dml2_context *dml2, - const struct dc *in_dc, const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out) -{ - if (dml2->config.use_native_soc_bb_construction) - dml2_init_soc_states(dml2, in_dc, in_bbox, out); - else - dml2_translate_soc_states(in_dc, out, in_dc->dml.soc.num_states); -} - -static void map_hw_resources(struct dml2_context *dml2, - struct dml_display_cfg_st *in_out_display_cfg, struct dml_mode_support_info_st *mode_support_info) -{ - unsigned int num_pipes = 0; - int i, j; - - for (i = 0; i < __DML_NUM_PLANES__; i++) { - in_out_display_cfg->hw.ODMMode[i] = mode_support_info->ODMMode[i]; - in_out_display_cfg->hw.DPPPerSurface[i] = mode_support_info->DPPPerSurface[i]; - in_out_display_cfg->hw.DSCEnabled[i] = mode_support_info->DSCEnabled[i]; - in_out_display_cfg->hw.NumberOfDSCSlices[i] = mode_support_info->NumberOfDSCSlices[i]; - in_out_display_cfg->hw.DLGRefClkFreqMHz = 24; - if (dml2->v20.dml_core_ctx.project != dml_project_dcn35 && - dml2->v20.dml_core_ctx.project != dml_project_dcn36 && - dml2->v20.dml_core_ctx.project != dml_project_dcn351) { - /*dGPU default as 50Mhz*/ - in_out_display_cfg->hw.DLGRefClkFreqMHz = 50; - } - for (j = 0; j < mode_support_info->DPPPerSurface[i]; j++) { - if (i >= __DML2_WRAPPER_MAX_STREAMS_PLANES__) { - dml_print("DML::%s: Index out of bounds: i=%d, __DML2_WRAPPER_MAX_STREAMS_PLANES__=%d\n", - __func__, i, __DML2_WRAPPER_MAX_STREAMS_PLANES__); - break; - } - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[num_pipes] = dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i]; - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[num_pipes] = true; - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[num_pipes] = dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i]; - dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[num_pipes] = true; - num_pipes++; - } - } -} - -static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, - const struct dml_display_cfg_st *display_cfg, - struct dml_mode_support_info_st *evaluation_info, - enum dc_validate_mode validate_mode) -{ - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - - s->mode_support_params.mode_lib = &dml2->v20.dml_core_ctx; - s->mode_support_params.in_display_cfg = display_cfg; - if (validate_mode == DC_VALIDATE_MODE_ONLY) - s->mode_support_params.in_start_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; - else - s->mode_support_params.in_start_state_idx = 0; - s->mode_support_params.out_evaluation_info = evaluation_info; - - memset(evaluation_info, 0, sizeof(struct dml_mode_support_info_st)); - s->mode_support_params.out_lowest_state_idx = 0; - - return dml_mode_support_ex(&s->mode_support_params); -} - -static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrapper_optimize_configuration_params *p) -{ - int unused_dpps = p->ip_params->max_num_dpp; - int i; - int odms_needed; - int largest_blend_and_timing = 0; - bool optimization_done = false; - - for (i = 0; i < (int) p->cur_display_config->num_timings; i++) { - if (p->cur_display_config->plane.BlendingAndTiming[i] > largest_blend_and_timing) - largest_blend_and_timing = p->cur_display_config->plane.BlendingAndTiming[i]; - } - - if (p->new_policy != p->cur_policy) - *p->new_policy = *p->cur_policy; - - if (p->new_display_config != p->cur_display_config) - *p->new_display_config = *p->cur_display_config; - - - // Optimize Clocks - if (!optimization_done) { - if (largest_blend_and_timing == 0 && p->cur_policy->ODMUse[0] == dml_odm_use_policy_combine_as_needed && dml2->config.minimize_dispclk_using_odm) { - odms_needed = dml2_util_get_maximum_odm_combine_for_output(dml2->config.optimize_odm_4to1, - p->cur_display_config->output.OutputEncoder[0], p->cur_mode_support_info->DSCEnabled[0]) - 1; - - if (odms_needed <= unused_dpps) { - if (odms_needed == 1) { - p->new_policy->ODMUse[0] = dml_odm_use_policy_combine_2to1; - optimization_done = true; - } else if (odms_needed == 3) { - p->new_policy->ODMUse[0] = dml_odm_use_policy_combine_4to1; - optimization_done = true; - } else - optimization_done = false; - } - } - } - - return optimization_done; -} - -static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state, - enum dc_validate_mode validate_mode) -{ - struct dml2_calculate_lowest_supported_state_for_temp_read_scratch *s = &dml2->v20.scratch.dml2_calculate_lowest_supported_state_for_temp_read_scratch; - struct dml2_wrapper_scratch *s_global = &dml2->v20.scratch; - - unsigned int dml_result = 0; - int result = -1, i, j; - - build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); - - /* Zero out before each call before proceeding */ - memset(s, 0, sizeof(struct dml2_calculate_lowest_supported_state_for_temp_read_scratch)); - memset(&s_global->mode_support_params, 0, sizeof(struct dml_mode_support_ex_params_st)); - memset(&s_global->dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); - - for (i = 0; i < dml2->config.dcn_pipe_count; i++) { - /* Calling resource_build_scaling_params will populate the pipe params - * with the necessary information needed for correct DML calculations - * This is also done in DML1 driver code path and hence display_state - * cannot be const. - */ - struct pipe_ctx *pipe = &display_state->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state) { - if (!dml2->config.callbacks.build_scaling_params(pipe)) { - ASSERT(false); - return false; - } - } - } - - map_dc_state_into_dml_display_cfg(dml2, display_state, &s->cur_display_config); - - for (i = 0; i < dml2->v20.dml_core_ctx.states.num_states; i++) { - s->uclk_change_latencies[i] = dml2->v20.dml_core_ctx.states.state_array[i].dram_clock_change_latency_us; - } - - for (i = 0; i < 4; i++) { - for (j = 0; j < dml2->v20.dml_core_ctx.states.num_states; j++) { - dml2->v20.dml_core_ctx.states.state_array[j].dram_clock_change_latency_us = s_global->dummy_pstate_table[i].dummy_pstate_latency_us; - } - - dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info, - validate_mode); - - if (dml_result && s->evaluation_info.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive) { - map_hw_resources(dml2, &s->cur_display_config, &s->evaluation_info); - dml_result = dml_mode_programming(&dml2->v20.dml_core_ctx, s_global->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); - - ASSERT(dml_result); - - dml2_extract_watermark_set(&dml2->v20.g6_temp_read_watermark_set, &dml2->v20.dml_core_ctx); - dml2->v20.g6_temp_read_watermark_set.cstate_pstate.fclk_pstate_change_ns = dml2->v20.g6_temp_read_watermark_set.cstate_pstate.pstate_change_ns; - - result = s_global->mode_support_params.out_lowest_state_idx; - - while (dml2->v20.dml_core_ctx.states.state_array[result].dram_speed_mts < s_global->dummy_pstate_table[i].dram_speed_mts) - result++; - - break; - } - } - - for (i = 0; i < dml2->v20.dml_core_ctx.states.num_states; i++) { - dml2->v20.dml_core_ctx.states.state_array[i].dram_clock_change_latency_us = s->uclk_change_latencies[i]; - } - - return result; -} - -static void copy_dummy_pstate_table(struct dummy_pstate_entry *dest, struct dummy_pstate_entry *src, unsigned int num_entries) -{ - for (int i = 0; i < num_entries; i++) { - dest[i] = src[i]; - } -} - -static bool are_timings_requiring_odm_doing_blending(const struct dml_display_cfg_st *display_cfg, - const struct dml_mode_support_info_st *evaluation_info) -{ - unsigned int planes_per_timing[__DML_NUM_PLANES__] = {0}; - int i; - - for (i = 0; i < display_cfg->num_surfaces; i++) - planes_per_timing[display_cfg->plane.BlendingAndTiming[i]]++; - - for (i = 0; i < __DML_NUM_PLANES__; i++) { - if (planes_per_timing[i] > 1 && evaluation_info->ODMMode[i] != dml_odm_mode_bypass) - return true; - } - - return false; -} - -static bool does_configuration_meet_sw_policies(struct dml2_context *ctx, const struct dml_display_cfg_st *display_cfg, - const struct dml_mode_support_info_st *evaluation_info) -{ - bool pass = true; - - if (!ctx->config.enable_windowed_mpo_odm) { - if (are_timings_requiring_odm_doing_blending(display_cfg, evaluation_info)) - pass = false; - } - - return pass; -} - -static bool dml_mode_support_wrapper(struct dml2_context *dml2, - struct dc_state *display_state, - enum dc_validate_mode validate_mode) -{ - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - unsigned int result = 0, i; - unsigned int optimized_result = true; - - build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); - - /* Zero out before each call before proceeding */ - memset(&s->cur_display_config, 0, sizeof(struct dml_display_cfg_st)); - memset(&s->mode_support_params, 0, sizeof(struct dml_mode_support_ex_params_st)); - memset(&s->dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); - memset(&s->optimize_configuration_params, 0, sizeof(struct dml2_wrapper_optimize_configuration_params)); - - for (i = 0; i < dml2->config.dcn_pipe_count; i++) { - /* Calling resource_build_scaling_params will populate the pipe params - * with the necessary information needed for correct DML calculations - * This is also done in DML1 driver code path and hence display_state - * cannot be const. - */ - struct pipe_ctx *pipe = &display_state->res_ctx.pipe_ctx[i]; - - if (pipe->plane_state) { - if (!dml2->config.callbacks.build_scaling_params(pipe)) { - ASSERT(false); - return false; - } - } - } - - map_dc_state_into_dml_display_cfg(dml2, display_state, &s->cur_display_config); - if (!dml2->config.skip_hw_state_mapping) - dml2_apply_det_buffer_allocation_policy(dml2, &s->cur_display_config); - - result = pack_and_call_dml_mode_support_ex(dml2, - &s->cur_display_config, - &s->mode_support_info, - validate_mode); - - if (result) - result = does_configuration_meet_sw_policies(dml2, &s->cur_display_config, &s->mode_support_info); - - // Try to optimize - if (result) { - s->cur_policy = dml2->v20.dml_core_ctx.policy; - s->optimize_configuration_params.dml_core_ctx = &dml2->v20.dml_core_ctx; - s->optimize_configuration_params.config = &dml2->config; - s->optimize_configuration_params.ip_params = &dml2->v20.dml_core_ctx.ip; - s->optimize_configuration_params.cur_display_config = &s->cur_display_config; - s->optimize_configuration_params.cur_mode_support_info = &s->mode_support_info; - s->optimize_configuration_params.cur_policy = &s->cur_policy; - s->optimize_configuration_params.new_display_config = &s->new_display_config; - s->optimize_configuration_params.new_policy = &s->new_policy; - - while (optimized_result && optimize_configuration(dml2, &s->optimize_configuration_params)) { - dml2->v20.dml_core_ctx.policy = s->new_policy; - optimized_result = pack_and_call_dml_mode_support_ex(dml2, - &s->new_display_config, - &s->mode_support_info, - validate_mode); - - if (optimized_result) - optimized_result = does_configuration_meet_sw_policies(dml2, &s->new_display_config, &s->mode_support_info); - - // If the new optimized state is supposed, then set current = new - if (optimized_result) { - s->cur_display_config = s->new_display_config; - s->cur_policy = s->new_policy; - } else { - // Else, restore policy to current - dml2->v20.dml_core_ctx.policy = s->cur_policy; - } - } - - // Optimize ended with a failed config, so we need to restore DML state to last passing - if (!optimized_result) { - result = pack_and_call_dml_mode_support_ex(dml2, - &s->cur_display_config, - &s->mode_support_info, - validate_mode); - } - } - - if (result) - map_hw_resources(dml2, &s->cur_display_config, &s->mode_support_info); - - return result; -} - -static bool call_dml_mode_support_and_programming(struct dc_state *context, enum dc_validate_mode validate_mode) -{ - unsigned int result = 0; - unsigned int min_state = 0; - int min_state_for_g6_temp_read = 0; - - - if (!context) - return false; - - struct dml2_context *dml2 = context->bw_ctx.dml2; - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - - if (!context->streams[0]->sink->link->dc->caps.is_apu) { - min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context, - validate_mode); - - ASSERT(min_state_for_g6_temp_read >= 0); - } - - result = dml_mode_support_wrapper(dml2, context, validate_mode); - - /* Upon trying to sett certain frequencies in FRL, min_state_for_g6_temp_read is reported as -1. This leads to an invalid value of min_state causing crashes later on. - * Use the default logic for min_state only when min_state_for_g6_temp_read is a valid value. In other cases, use the value calculated by the DML directly. - */ - if (!context->streams[0]->sink->link->dc->caps.is_apu) { - if (min_state_for_g6_temp_read >= 0) - min_state = min_state_for_g6_temp_read > s->mode_support_params.out_lowest_state_idx ? min_state_for_g6_temp_read : s->mode_support_params.out_lowest_state_idx; - else - min_state = s->mode_support_params.out_lowest_state_idx; - } - - if (result) { - if (!context->streams[0]->sink->link->dc->caps.is_apu) { - result = dml_mode_programming(&dml2->v20.dml_core_ctx, min_state, &s->cur_display_config, true); - } else { - result = dml_mode_programming(&dml2->v20.dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); - } - } - return result; -} - -static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context, - enum dc_validate_mode validate_mode) -{ - struct dml2_context *dml2 = context->bw_ctx.dml2; - struct dml2_wrapper_scratch *s = &dml2->v20.scratch; - struct dml2_dcn_clocks out_clks; - unsigned int result = 0; - bool need_recalculation = false; - uint32_t cstate_enter_plus_exit_z8_ns; - - if (context->stream_count == 0) { - unsigned int lowest_state_idx = 0; - - out_clks.p_state_supported = true; - out_clks.dispclk_khz = 0; /* No requirement, and lowest index will generally be maximum dispclk. */ - out_clks.dcfclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dcfclk_mhz * 1000; - out_clks.fclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].fabricclk_mhz * 1000; - out_clks.uclk_mts = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dram_speed_mts; - out_clks.phyclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].phyclk_mhz * 1000; - out_clks.socclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].socclk_mhz * 1000; - out_clks.ref_dtbclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dtbclk_mhz * 1000; - context->bw_ctx.bw.dcn.clk.dtbclk_en = false; - dml2_copy_clocks_to_dc_state(&out_clks, context); - return true; - } - - /* Zero out before each call before proceeding */ - memset(&dml2->v20.scratch, 0, sizeof(struct dml2_wrapper_scratch)); - memset(&dml2->v20.dml_core_ctx.policy, 0, sizeof(struct dml_mode_eval_policy_st)); - memset(&dml2->v20.dml_core_ctx.ms, 0, sizeof(struct mode_support_st)); - memset(&dml2->v20.dml_core_ctx.mp, 0, sizeof(struct mode_program_st)); - - /* Initialize DET scratch */ - dml2_initialize_det_scratch(dml2); - - copy_dummy_pstate_table(s->dummy_pstate_table, in_dc->clk_mgr->bw_params->dummy_pstate_table, 4); - - result = call_dml_mode_support_and_programming(context, validate_mode); - /* Call map dc pipes to map the pipes based on the DML output. For correctly determining if recalculation - * is required or not, the resource context needs to correctly reflect the number of active pipes. We would - * only know the correct number if active pipes after dml2_map_dc_pipes is called. - */ - if (result && !dml2->config.skip_hw_state_mapping) - dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); - - /* Verify and update DET Buffer configuration if needed. dml2_verify_det_buffer_configuration will check if DET Buffer - * size needs to be updated. If yes it will update the DETOverride variable and set need_recalculation flag to true. - * Based on that flag, run mode support again. Verification needs to be run after dml_mode_programming because the getters - * return correct det buffer values only after dml_mode_programming is called. - */ - if (result && !dml2->config.skip_hw_state_mapping) { - need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); - if (need_recalculation) { - /* Engage the DML again if recalculation is required. */ - call_dml_mode_support_and_programming(context, validate_mode); - if (!dml2->config.skip_hw_state_mapping) { - dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); - } - need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); - ASSERT(need_recalculation == false); - } - } - - if (result) { - unsigned int lowest_state_idx = s->mode_support_params.out_lowest_state_idx; - out_clks.dispclk_khz = (unsigned int)dml2->v20.dml_core_ctx.mp.Dispclk_calculated * 1000; - out_clks.p_state_supported = s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported; - if (in_dc->config.use_default_clock_table && - (lowest_state_idx < dml2->v20.dml_core_ctx.states.num_states - 1)) { - lowest_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; - out_clks.dispclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dispclk_mhz * 1000; - } - - out_clks.dcfclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dcfclk_mhz * 1000; - out_clks.fclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].fabricclk_mhz * 1000; - out_clks.uclk_mts = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dram_speed_mts; - out_clks.phyclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].phyclk_mhz * 1000; - out_clks.socclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].socclk_mhz * 1000; - out_clks.ref_dtbclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dtbclk_mhz * 1000; - context->bw_ctx.bw.dcn.clk.dtbclk_en = is_dtbclk_required(in_dc, context); - - if (!dml2->config.skip_hw_state_mapping) { - /* Call dml2_calculate_rq_and_dlg_params */ - dml2_calculate_rq_and_dlg_params(in_dc, context, &context->res_ctx, dml2, in_dc->res_pool->pipe_count); - } - - dml2_copy_clocks_to_dc_state(&out_clks, context); - dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.a, &dml2->v20.dml_core_ctx); - dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.b, &dml2->v20.dml_core_ctx); - if (context->streams[0]->sink->link->dc->caps.is_apu) - dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.dml_core_ctx); - else - memcpy(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.g6_temp_read_watermark_set, sizeof(context->bw_ctx.bw.dcn.watermarks.c)); - dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.d, &dml2->v20.dml_core_ctx); - dml2_extract_writeback_wm(context, &dml2->v20.dml_core_ctx); - //copy for deciding zstate use - context->bw_ctx.dml.vba.StutterPeriod = context->bw_ctx.dml2->v20.dml_core_ctx.mp.StutterPeriod; - - cstate_enter_plus_exit_z8_ns = context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_z8_ns; - - if (context->bw_ctx.dml.vba.StutterPeriod < in_dc->debug.minimum_z8_residency_time && - cstate_enter_plus_exit_z8_ns < in_dc->debug.minimum_z8_residency_time * 1000) - cstate_enter_plus_exit_z8_ns = in_dc->debug.minimum_z8_residency_time * 1000; - - context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_z8_ns = cstate_enter_plus_exit_z8_ns; - } - - return result; -} - -static bool dml2_validate_only(struct dc_state *context, enum dc_validate_mode validate_mode) -{ - struct dml2_context *dml2; - unsigned int result = 0; - - if (!context || context->stream_count == 0) - return true; - - dml2 = context->bw_ctx.dml2; - - /* Zero out before each call before proceeding */ - memset(&dml2->v20.scratch, 0, sizeof(struct dml2_wrapper_scratch)); - memset(&dml2->v20.dml_core_ctx.policy, 0, sizeof(struct dml_mode_eval_policy_st)); - memset(&dml2->v20.dml_core_ctx.ms, 0, sizeof(struct mode_support_st)); - memset(&dml2->v20.dml_core_ctx.mp, 0, sizeof(struct mode_program_st)); - - build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); - - map_dc_state_into_dml_display_cfg(dml2, context, &dml2->v20.scratch.cur_display_config); - if (!dml2->config.skip_hw_state_mapping) - dml2_apply_det_buffer_allocation_policy(dml2, &dml2->v20.scratch.cur_display_config); - - result = pack_and_call_dml_mode_support_ex(dml2, - &dml2->v20.scratch.cur_display_config, - &dml2->v20.scratch.mode_support_info, - validate_mode); - - if (result) - result = does_configuration_meet_sw_policies(dml2, &dml2->v20.scratch.cur_display_config, &dml2->v20.scratch.mode_support_info); - - return result == 1; -} - -static void dml2_apply_debug_options(const struct dc *dc, struct dml2_context *dml2) -{ - if (dc->debug.override_odm_optimization) { - dml2->config.minimize_dispclk_using_odm = dc->debug.minimize_dispclk_using_odm; - } -} - -bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, - enum dc_validate_mode validate_mode) -{ - bool out = false; - - if (!dml2) - return false; - dml2_apply_debug_options(in_dc, dml2); - - /* DML2.1 validation path */ - if (dml2->architecture == dml2_architecture_21) { - out = dml21_validate(in_dc, context, dml2, validate_mode); - return out; - } - - DC_FP_START(); - - /* Use dml_validate_only for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ - if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) - out = dml2_validate_only(context, validate_mode); - else - out = dml2_validate_and_build_resource(in_dc, context, validate_mode); - - DC_FP_END(); - - return out; -} - -static inline struct dml2_context *dml2_allocate_memory(void) -{ - return (struct dml2_context *) vzalloc(sizeof(struct dml2_context)); -} - -static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) -{ - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { - dml21_reinit(in_dc, *dml2, config); - return; - } - - // Store config options - (*dml2)->config = *config; - - switch (in_dc->ctx->dce_version) { - case DCN_VERSION_3_5: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn35; - break; - case DCN_VERSION_3_51: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn351; - break; - case DCN_VERSION_3_6: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn36; - break; - case DCN_VERSION_3_2: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn32; - break; - case DCN_VERSION_3_21: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn321; - break; - case DCN_VERSION_4_01: - (*dml2)->v20.dml_core_ctx.project = dml_project_dcn401; - break; - default: - (*dml2)->v20.dml_core_ctx.project = dml_project_default; - break; - } - - DC_FP_START(); - - initialize_dml2_ip_params(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.ip); - - initialize_dml2_soc_bbox(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.soc); - - initialize_dml2_soc_states(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.soc, &(*dml2)->v20.dml_core_ctx.states); - - DC_FP_END(); -} - -bool dml2_create(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) -{ - // TODO : Temporarily add DCN_VERSION_3_2 for N-1 validation. Remove DCN_VERSION_3_2 after N-1 validation phase is complete. - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) - return dml21_create(in_dc, dml2, config); - - // Allocate Mode Lib Ctx - *dml2 = dml2_allocate_memory(); - - if (!(*dml2)) - return false; - - dml2_init(in_dc, config, dml2); - - return true; -} - -void dml2_destroy(struct dml2_context *dml2) -{ - if (!dml2) - return; - - if (dml2->architecture == dml2_architecture_21) - dml21_destroy(dml2); - vfree(dml2); -} - -void dml2_extract_dram_and_fclk_change_support(struct dml2_context *dml2, - unsigned int *fclk_change_support, unsigned int *dram_clk_change_support) -{ - *fclk_change_support = (unsigned int) dml2->v20.dml_core_ctx.ms.support.FCLKChangeSupport[0]; - *dram_clk_change_support = (unsigned int) dml2->v20.dml_core_ctx.ms.support.DRAMClockChangeSupport[0]; -} - -void dml2_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2) -{ - if (dml2->architecture == dml2_architecture_21) - dml21_prepare_mcache_programming(in_dc, context, dml2); -} - -void dml2_copy(struct dml2_context *dst_dml2, - struct dml2_context *src_dml2) -{ - if (src_dml2->architecture == dml2_architecture_21) { - dml21_copy(dst_dml2, src_dml2); - return; - } - /* copy Mode Lib Ctx */ - memcpy(dst_dml2, src_dml2, sizeof(struct dml2_context)); -} - -bool dml2_create_copy(struct dml2_context **dst_dml2, - struct dml2_context *src_dml2) -{ - if (src_dml2->architecture == dml2_architecture_21) - return dml21_create_copy(dst_dml2, src_dml2); - /* Allocate Mode Lib Ctx */ - *dst_dml2 = dml2_allocate_memory(); - - if (!(*dst_dml2)) - return false; - - /* copy Mode Lib Ctx */ - dml2_copy(*dst_dml2, src_dml2); - - return true; -} - -void dml2_reinit(const struct dc *in_dc, - const struct dml2_configuration_options *config, - struct dml2_context **dml2) -{ - if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { - dml21_reinit(in_dc, *dml2, config); - return; - } - - dml2_init(in_dc, config, dml2); -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h deleted file mode 100644 index c384e141cebc..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h +++ /dev/null @@ -1,309 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef _DML2_WRAPPER_H_ -#define _DML2_WRAPPER_H_ - -#include "os_types.h" - -#define DML2_MAX_NUM_DPM_LVL 30 - -struct dml2_context; -struct display_mode_lib_st; -struct dc; -struct pipe_ctx; -struct dc_plane_state; -struct dc_sink; -struct dc_stream_state; -struct resource_context; -struct display_stream_compressor; -struct dc_mcache_params; - -// Configuration of the MALL on the SoC -struct dml2_soc_mall_info { - // Cache line size of 0 means MALL is not enabled/present - unsigned int cache_line_size_bytes; - unsigned int cache_num_ways; - unsigned int max_cab_allocation_bytes; - - unsigned int mblk_width_pixels; - unsigned int mblk_size_bytes; - unsigned int mblk_height_4bpe_pixels; - unsigned int mblk_height_8bpe_pixels; -}; - -// Output of DML2 for clock requirements -struct dml2_dcn_clocks { - unsigned int dispclk_khz; - unsigned int dcfclk_khz; - unsigned int fclk_khz; - unsigned int uclk_mts; - unsigned int phyclk_khz; - unsigned int socclk_khz; - unsigned int ref_dtbclk_khz; - bool p_state_supported; - unsigned int cab_num_ways_required; - unsigned int dcfclk_khz_ds; -}; - -struct dml2_dc_callbacks { - struct dc *dc; - bool (*build_scaling_params)(struct pipe_ctx *pipe_ctx); - void (*build_test_pattern_params)(struct resource_context *res_ctx, struct pipe_ctx *otg_master); - bool (*can_support_mclk_switch_using_fw_based_vblank_stretch)(struct dc *dc, struct dc_state *context); - bool (*acquire_secondary_pipe_for_mpc_odm)(const struct dc *dc, struct dc_state *state, struct pipe_ctx *pri_pipe, struct pipe_ctx *sec_pipe, bool odm); - bool (*update_pipes_for_stream_with_slice_count)( - struct dc_state *new_ctx, - const struct dc_state *cur_ctx, - const struct resource_pool *pool, - const struct dc_stream_state *stream, - int new_slice_count); - bool (*update_pipes_for_plane_with_slice_count)( - struct dc_state *new_ctx, - const struct dc_state *cur_ctx, - const struct resource_pool *pool, - const struct dc_plane_state *plane, - int slice_count); - int (*get_odm_slice_index)(const struct pipe_ctx *opp_head); - int (*get_odm_slice_count)(const struct pipe_ctx *opp_head); - int (*get_mpc_slice_index)(const struct pipe_ctx *dpp_pipe); - int (*get_mpc_slice_count)(const struct pipe_ctx *dpp_pipe); - struct pipe_ctx *(*get_opp_head)(const struct pipe_ctx *pipe_ctx); - struct pipe_ctx *(*get_otg_master_for_stream)( - struct resource_context *res_ctx, - const struct dc_stream_state *stream); - int (*get_opp_heads_for_otg_master)(const struct pipe_ctx *otg_master, - struct resource_context *res_ctx, - struct pipe_ctx *opp_heads[MAX_PIPES]); - int (*get_dpp_pipes_for_plane)(const struct dc_plane_state *plane, - struct resource_context *res_ctx, - struct pipe_ctx *dpp_pipes[MAX_PIPES]); - struct dc_stream_status *(*get_stream_status)( - struct dc_state *state, - const struct dc_stream_state *stream); - struct dc_stream_state *(*get_stream_from_id)(const struct dc_state *state, unsigned int id); - unsigned int (*get_max_flickerless_instant_vtotal_increase)( - struct dc_stream_state *stream, - bool is_gaming); - bool (*allocate_mcache)(struct dc_state *context, const struct dc_mcache_params *mcache_params); -}; - -struct dml2_dc_svp_callbacks { - struct dc *dc; - bool (*build_scaling_params)(struct pipe_ctx *pipe_ctx); - struct dc_stream_state* (*create_phantom_stream)(const struct dc *dc, - struct dc_state *state, - struct dc_stream_state *main_stream); - struct dc_plane_state* (*create_phantom_plane)(const struct dc *dc, - struct dc_state *state, - struct dc_plane_state *main_plane); - enum dc_status (*add_phantom_stream)(const struct dc *dc, - struct dc_state *state, - struct dc_stream_state *phantom_stream, - struct dc_stream_state *main_stream); - bool (*add_phantom_plane)(const struct dc *dc, struct dc_stream_state *stream, struct dc_plane_state *plane_state, struct dc_state *context); - bool (*remove_phantom_plane)(const struct dc *dc, - struct dc_stream_state *stream, - struct dc_plane_state *plane_state, - struct dc_state *context); - enum dc_status (*remove_phantom_stream)(const struct dc *dc, - struct dc_state *state, - struct dc_stream_state *stream); - void (*release_phantom_plane)(const struct dc *dc, - struct dc_state *state, - struct dc_plane_state *plane); - void (*release_phantom_stream)(const struct dc *dc, - struct dc_state *state, - struct dc_stream_state *stream); - void (*release_dsc)(struct resource_context *res_ctx, const struct resource_pool *pool, struct display_stream_compressor **dsc); - enum mall_stream_type (*get_pipe_subvp_type)(const struct dc_state *state, const struct pipe_ctx *pipe_ctx); - enum mall_stream_type (*get_stream_subvp_type)(const struct dc_state *state, const struct dc_stream_state *stream); - struct dc_stream_state *(*get_paired_subvp_stream)(const struct dc_state *state, const struct dc_stream_state *stream); - bool (*remove_phantom_streams_and_planes)( - const struct dc *dc, - struct dc_state *state); - void (*release_phantom_streams_and_planes)( - const struct dc *dc, - struct dc_state *state); - unsigned int (*calculate_mall_ways_from_bytes)( - const struct dc *dc, - unsigned int total_size_in_mall_bytes); -}; - -struct dml2_clks_table_entry { - unsigned int dcfclk_mhz; - unsigned int fclk_mhz; - unsigned int memclk_mhz; - unsigned int socclk_mhz; - unsigned int dtbclk_mhz; - unsigned int dispclk_mhz; - unsigned int dppclk_mhz; - unsigned int dram_speed_mts; /*which is based on wck_ratio*/ -}; - -struct dml2_clks_num_entries { - unsigned int num_dcfclk_levels; - unsigned int num_fclk_levels; - unsigned int num_memclk_levels; - unsigned int num_socclk_levels; - unsigned int num_dtbclk_levels; - unsigned int num_dispclk_levels; - unsigned int num_dppclk_levels; -}; - -struct dml2_clks_limit_table { - struct dml2_clks_table_entry clk_entries[DML2_MAX_NUM_DPM_LVL]; - struct dml2_clks_num_entries num_entries_per_clk; - unsigned int num_states; -}; - -// Various overrides, per ASIC or per SKU specific, or for debugging purpose when/if available -struct dml2_soc_bbox_overrides { - double xtalclk_mhz; - double dchub_refclk_mhz; - double dprefclk_mhz; - double disp_pll_vco_speed_mhz; - double urgent_latency_us; - double sr_exit_latency_us; - double sr_enter_plus_exit_latency_us; - double sr_exit_z8_time_us; - double sr_enter_plus_exit_z8_time_us; - double dram_clock_change_latency_us; - double fclk_change_latency_us; - unsigned int dram_num_chan; - unsigned int dram_chanel_width_bytes; - struct dml2_clks_limit_table clks_table; -}; - -enum dml2_force_pstate_methods { - dml2_force_pstate_method_auto = 0, - dml2_force_pstate_method_vactive, - dml2_force_pstate_method_vblank, - dml2_force_pstate_method_drr, - dml2_force_pstate_method_subvp, -}; - -struct dml2_configuration_options { - int dcn_pipe_count; - bool use_native_pstate_optimization; - bool enable_windowed_mpo_odm; - bool use_native_soc_bb_construction; - bool skip_hw_state_mapping; - bool optimize_odm_4to1; - bool minimize_dispclk_using_odm; - bool override_det_buffer_size_kbytes; - struct dml2_dc_callbacks callbacks; - struct { - bool force_disable_subvp; - bool force_enable_subvp; - unsigned int subvp_fw_processing_delay_us; - unsigned int subvp_pstate_allow_width_us; - unsigned int subvp_prefetch_end_to_mall_start_us; - unsigned int subvp_swath_height_margin_lines; - struct dml2_dc_svp_callbacks callbacks; - } svp_pstate; - struct dml2_soc_mall_info mall_cfg; - struct dml2_soc_bbox_overrides bbox_overrides; - unsigned int max_segments_per_hubp; - unsigned int det_segment_size; - /* Only for debugging purposes when initializing SOCBB params via tool for DML21. */ - struct socbb_ip_params_external *external_socbb_ip_params; - struct { - bool force_pstate_method_enable; - enum dml2_force_pstate_methods force_pstate_method_values[MAX_PIPES]; - } pmo; - bool map_dc_pipes_with_callbacks; - - bool use_clock_dc_limits; - bool gpuvm_enable; - bool force_tdlut_enable; - void *bb_from_dmub; -}; - -/* - * dml2_create - Creates dml2_context. - * @in_dc: dc. - * @config: dml2 configuration options. - * @dml2: Created dml2 context. - * - * Create and destroy of DML2 is done as part of dc_state creation - * and dc_state_free. DML2 IP, SOC and STATES are initialized at - * creation time. - * - * Return: True if dml2 is successfully created, false otherwise. - */ -bool dml2_create(const struct dc *in_dc, - const struct dml2_configuration_options *config, - struct dml2_context **dml2); - -void dml2_destroy(struct dml2_context *dml2); -void dml2_copy(struct dml2_context *dst_dml2, - struct dml2_context *src_dml2); -bool dml2_create_copy(struct dml2_context **dst_dml2, - struct dml2_context *src_dml2); -void dml2_reinit(const struct dc *in_dc, - const struct dml2_configuration_options *config, - struct dml2_context **dml2); - -/* - * dml2_validate - Determines if a display configuration is supported or not. - * @in_dc: dc. - * @context: dc_state to be validated. - * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX will not populate context.res_ctx. - * - * DML1.0 compatible interface for validation. - * - * Based on fast_validate option internally would call: - * - * -dml2_validate_and_build_resource - for non fast_validate option - * Calculates if dc_state can be supported on the SOC, and attempts to - * optimize the power management feature supports versus minimum clocks. - * If supported, also builds out_new_hw_state to represent the hw programming - * for the new dc state. - * - * -dml2_validate_only - for fast_validate option - * Calculates if dc_state can be supported on the SOC (i.e. at maximum - * clocks) with all mandatory power features enabled. - - * Context: Two threads may not invoke this function concurrently unless they reference - * separate dc_states for validation. - * Return: True if mode is supported, false otherwise. - */ -bool dml2_validate(const struct dc *in_dc, - struct dc_state *context, - struct dml2_context *dml2, - enum dc_validate_mode validate_mode); - -/* - * dml2_extract_dram_and_fclk_change_support - Extracts the FCLK and UCLK change support info. - * @dml2: input dml2 context pointer. - * @fclk_change_support: output pointer holding the fclk change support info (vactive, vblank, unsupported). - * @dram_clk_change_support: output pointer holding the uclk change support info (vactive, vblank, unsupported). - */ -void dml2_extract_dram_and_fclk_change_support(struct dml2_context *dml2, - unsigned int *fclk_change_support, unsigned int *dram_clk_change_support); -void dml2_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2); -#endif //_DML2_WRAPPER_H_ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h b/drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h deleted file mode 100644 index 17f0972b1af7..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h +++ /dev/null @@ -1,32 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -#ifndef __DML_ASSERT_H__ -#define __DML_ASSERT_H__ - -#include "os_types.h" - -#endif //__DML_ASSERT_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h b/drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h deleted file mode 100644 index f7d30b47beff..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h +++ /dev/null @@ -1,33 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ - -/* This header intentinally does not include an #ifdef guard as it only contains includes for other headers*/ - -/* - * Standard Types - */ -#include "os_types.h" -#include "cmntypes.h" diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c b/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c deleted file mode 100644 index 00d22e542469..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c +++ /dev/null @@ -1,573 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#include "dml_display_rq_dlg_calc.h" -#include "display_mode_core.h" -#include "display_mode_util.h" - -static dml_bool_t is_dual_plane(enum dml_source_format_class source_format) -{ - dml_bool_t ret_val = 0; - - if ((source_format == dml_420_12) || (source_format == dml_420_8) || (source_format == dml_420_10) || (source_format == dml_rgbe_alpha)) - ret_val = 1; - - return ret_val; -} - -void dml_rq_dlg_get_rq_reg(dml_display_rq_regs_st *rq_regs, - struct display_mode_lib_st *mode_lib, - const dml_uint_t pipe_idx) -{ - dml_uint_t plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); - enum dml_source_format_class source_format = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[plane_idx]; - enum dml_swizzle_mode sw_mode = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[plane_idx]; - dml_bool_t dual_plane = is_dual_plane((enum dml_source_format_class)(source_format)); - - uint32 pixel_chunk_bytes = 0; - uint32 min_pixel_chunk_bytes = 0; - uint32 meta_chunk_bytes = 0; - uint32 min_meta_chunk_bytes = 0; - uint32 dpte_group_bytes = 0; - uint32 mpte_group_bytes = 0; - - uint32 p1_pixel_chunk_bytes = 0; - uint32 p1_min_pixel_chunk_bytes = 0; - uint32 p1_meta_chunk_bytes = 0; - uint32 p1_min_meta_chunk_bytes = 0; - uint32 p1_dpte_group_bytes = 0; - uint32 p1_mpte_group_bytes = 0; - - dml_uint_t detile_buf_size_in_bytes; - dml_uint_t detile_buf_plane1_addr = 0; - - dml_float_t stored_swath_l_bytes; - dml_float_t stored_swath_c_bytes; - dml_bool_t is_phantom_pipe; - - dml_uint_t pte_row_height_linear; - - dml_print("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); - - memset(rq_regs, 0, sizeof(*rq_regs)); - - pixel_chunk_bytes = (dml_uint_t)(dml_get_pixel_chunk_size_in_kbyte(mode_lib) * 1024); - min_pixel_chunk_bytes = (dml_uint_t)(dml_get_min_pixel_chunk_size_in_byte(mode_lib)); - - if (pixel_chunk_bytes == 64 * 1024) - min_pixel_chunk_bytes = 0; - - meta_chunk_bytes = (dml_uint_t)(dml_get_meta_chunk_size_in_kbyte(mode_lib) * 1024); - min_meta_chunk_bytes = (dml_uint_t)(dml_get_min_meta_chunk_size_in_byte(mode_lib)); - - dpte_group_bytes = (dml_uint_t)(dml_get_dpte_group_size_in_bytes(mode_lib, pipe_idx)); - mpte_group_bytes = (dml_uint_t)(dml_get_vm_group_size_in_bytes(mode_lib, pipe_idx)); - - p1_pixel_chunk_bytes = pixel_chunk_bytes; - p1_min_pixel_chunk_bytes = min_pixel_chunk_bytes; - p1_meta_chunk_bytes = meta_chunk_bytes; - p1_min_meta_chunk_bytes = min_meta_chunk_bytes; - p1_dpte_group_bytes = dpte_group_bytes; - p1_mpte_group_bytes = mpte_group_bytes; - - if (source_format == dml_rgbe_alpha) - p1_pixel_chunk_bytes = (dml_uint_t)(dml_get_alpha_pixel_chunk_size_in_kbyte(mode_lib) * 1024); - - rq_regs->rq_regs_l.chunk_size = (dml_uint_t)(dml_log2((dml_float_t) pixel_chunk_bytes) - 10); - rq_regs->rq_regs_c.chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_pixel_chunk_bytes) - 10); - - if (min_pixel_chunk_bytes == 0) - rq_regs->rq_regs_l.min_chunk_size = 0; - else - rq_regs->rq_regs_l.min_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) min_pixel_chunk_bytes) - 8 + 1); - - if (p1_min_pixel_chunk_bytes == 0) - rq_regs->rq_regs_c.min_chunk_size = 0; - else - rq_regs->rq_regs_c.min_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_min_pixel_chunk_bytes) - 8 + 1); - - rq_regs->rq_regs_l.meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) meta_chunk_bytes) - 10); - rq_regs->rq_regs_c.meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_meta_chunk_bytes) - 10); - - if (min_meta_chunk_bytes == 0) - rq_regs->rq_regs_l.min_meta_chunk_size = 0; - else - rq_regs->rq_regs_l.min_meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) min_meta_chunk_bytes) - 6 + 1); - - if (min_meta_chunk_bytes == 0) - rq_regs->rq_regs_c.min_meta_chunk_size = 0; - else - rq_regs->rq_regs_c.min_meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_min_meta_chunk_bytes) - 6 + 1); - - rq_regs->rq_regs_l.dpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) dpte_group_bytes) - 6); - rq_regs->rq_regs_l.mpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) mpte_group_bytes) - 6); - rq_regs->rq_regs_c.dpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) p1_dpte_group_bytes) - 6); - rq_regs->rq_regs_c.mpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) p1_mpte_group_bytes) - 6); - - detile_buf_size_in_bytes = (dml_uint_t)(dml_get_det_buffer_size_kbytes(mode_lib, pipe_idx) * 1024); - - pte_row_height_linear = (dml_uint_t)(dml_get_dpte_row_height_linear_l(mode_lib, pipe_idx)); - - if (sw_mode == dml_sw_linear) - ASSERT(pte_row_height_linear >= 8); - - rq_regs->rq_regs_l.pte_row_height_linear = (dml_uint_t)(dml_floor(dml_log2((dml_float_t) pte_row_height_linear), 1) - 3); - - if (dual_plane) { - dml_uint_t p1_pte_row_height_linear = (dml_uint_t)(dml_get_dpte_row_height_linear_c(mode_lib, pipe_idx)); - if (sw_mode == dml_sw_linear) - ASSERT(p1_pte_row_height_linear >= 8); - - rq_regs->rq_regs_c.pte_row_height_linear = (dml_uint_t)(dml_floor(dml_log2((dml_float_t) p1_pte_row_height_linear), 1) - 3); - } - - rq_regs->rq_regs_l.swath_height = (dml_uint_t)(dml_log2((dml_float_t) dml_get_swath_height_l(mode_lib, pipe_idx))); - rq_regs->rq_regs_c.swath_height = (dml_uint_t)(dml_log2((dml_float_t) dml_get_swath_height_c(mode_lib, pipe_idx))); - - if (pixel_chunk_bytes >= 32 * 1024 || (dual_plane && p1_pixel_chunk_bytes >= 32 * 1024)) { //32kb - rq_regs->drq_expansion_mode = 0; - } else { - rq_regs->drq_expansion_mode = 2; - } - rq_regs->prq_expansion_mode = 1; - rq_regs->mrq_expansion_mode = 1; - rq_regs->crq_expansion_mode = 1; - - stored_swath_l_bytes = dml_get_det_stored_buffer_size_l_bytes(mode_lib, pipe_idx); - stored_swath_c_bytes = dml_get_det_stored_buffer_size_c_bytes(mode_lib, pipe_idx); - is_phantom_pipe = dml_get_is_phantom_pipe(mode_lib, pipe_idx); - - // Note: detile_buf_plane1_addr is in unit of 1KB - if (dual_plane) { - if (is_phantom_pipe) { - detile_buf_plane1_addr = (dml_uint_t)((1024.0*1024.0) / 2.0 / 1024.0); // half to chroma - } else { - if (stored_swath_l_bytes / stored_swath_c_bytes <= 1.5) { - detile_buf_plane1_addr = (dml_uint_t)(detile_buf_size_in_bytes / 2.0 / 1024.0); // half to chroma -#ifdef __DML_VBA_DEBUG__ - dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); -#endif - } else { - detile_buf_plane1_addr = (dml_uint_t)(dml_round_to_multiple((dml_uint_t)((2.0 * detile_buf_size_in_bytes) / 3.0), 1024, 0) / 1024.0); // 2/3 to luma -#ifdef __DML_VBA_DEBUG__ - dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); -#endif - } - } - } - rq_regs->plane1_base_address = detile_buf_plane1_addr; - -#ifdef __DML_VBA_DEBUG__ - dml_print("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); - dml_print("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); - dml_print("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); - dml_print("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); - dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); - dml_print("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); -#endif - dml_print_rq_regs_st(rq_regs); - dml_print("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); -} - -// Note: currently taken in as is. -// Nice to decouple code from hw register implement and extract code that are repeated for luma and chroma. - - -void dml_rq_dlg_get_dlg_reg(dml_display_dlg_regs_st *disp_dlg_regs, - dml_display_ttu_regs_st *disp_ttu_regs, - struct display_mode_lib_st *mode_lib, - const dml_uint_t pipe_idx) -{ - dml_uint_t plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); - enum dml_source_format_class source_format = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[plane_idx]; - struct dml_timing_cfg_st *timing = &mode_lib->ms.cache_display_cfg.timing; - struct dml_plane_cfg_st *plane = &mode_lib->ms.cache_display_cfg.plane; - struct dml_hw_resource_st *hw = &mode_lib->ms.cache_display_cfg.hw; - dml_bool_t dual_plane = is_dual_plane(source_format); - dml_uint_t num_cursors = plane->NumberOfCursors[plane_idx]; - enum dml_odm_mode odm_mode = hw->ODMMode[plane_idx]; - - dml_uint_t htotal = timing->HTotal[plane_idx]; - dml_uint_t hactive = timing->HActive[plane_idx]; - dml_uint_t hblank_end = timing->HBlankEnd[plane_idx]; - dml_uint_t vblank_end = timing->VBlankEnd[plane_idx]; - dml_bool_t interlaced = timing->Interlace[plane_idx]; - dml_float_t pclk_freq_in_mhz = (dml_float_t) timing->PixelClock[plane_idx]; - dml_float_t refclk_freq_in_mhz = (hw->DLGRefClkFreqMHz > 0) ? (dml_float_t) hw->DLGRefClkFreqMHz : mode_lib->soc.refclk_mhz; - dml_float_t ref_freq_to_pix_freq = refclk_freq_in_mhz / pclk_freq_in_mhz; - - dml_uint_t vready_after_vcount0; - - dml_uint_t dst_x_after_scaler; - dml_uint_t dst_y_after_scaler; - - dml_float_t dst_y_prefetch; - dml_float_t dst_y_per_vm_vblank; - dml_float_t dst_y_per_row_vblank; - dml_float_t dst_y_per_vm_flip; - dml_float_t dst_y_per_row_flip; - - dml_float_t max_dst_y_per_vm_vblank = 32.0; //U5.2 - dml_float_t max_dst_y_per_row_vblank = 16.0; //U4.2 - - dml_float_t vratio_pre_l; - dml_float_t vratio_pre_c; - - dml_float_t refcyc_per_line_delivery_pre_l; - dml_float_t refcyc_per_line_delivery_l; - dml_float_t refcyc_per_line_delivery_pre_c = 0.; - dml_float_t refcyc_per_line_delivery_c = 0.; - dml_float_t refcyc_per_req_delivery_pre_l; - dml_float_t refcyc_per_req_delivery_l; - dml_float_t refcyc_per_req_delivery_pre_c = 0.; - dml_float_t refcyc_per_req_delivery_c = 0.; - dml_float_t refcyc_per_req_delivery_pre_cur0 = 0.; - dml_float_t refcyc_per_req_delivery_cur0 = 0.; - - dml_float_t dst_y_per_pte_row_nom_l; - dml_float_t dst_y_per_pte_row_nom_c; - dml_float_t dst_y_per_meta_row_nom_l; - dml_float_t dst_y_per_meta_row_nom_c; - dml_float_t refcyc_per_pte_group_nom_l; - dml_float_t refcyc_per_pte_group_nom_c; - dml_float_t refcyc_per_pte_group_vblank_l; - dml_float_t refcyc_per_pte_group_vblank_c; - dml_float_t refcyc_per_pte_group_flip_l; - dml_float_t refcyc_per_pte_group_flip_c; - dml_float_t refcyc_per_meta_chunk_nom_l; - dml_float_t refcyc_per_meta_chunk_nom_c; - dml_float_t refcyc_per_meta_chunk_vblank_l; - dml_float_t refcyc_per_meta_chunk_vblank_c; - dml_float_t refcyc_per_meta_chunk_flip_l; - dml_float_t refcyc_per_meta_chunk_flip_c; - - dml_float_t temp; - dml_float_t min_ttu_vblank; - dml_uint_t min_dst_y_next_start; - - dml_print("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); - dml_print("DML_DLG::%s: plane_idx = %d\n", __func__, plane_idx); - dml_print("DML_DLG: %s: htotal = %d\n", __func__, htotal); - dml_print("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, refclk_freq_in_mhz); - dml_print("DML_DLG: %s: hw->DLGRefClkFreqMHz = %3.2f\n", __func__, hw->DLGRefClkFreqMHz); - dml_print("DML_DLG: %s: soc.refclk_mhz = %3.2f\n", __func__, mode_lib->soc.refclk_mhz); - dml_print("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, pclk_freq_in_mhz); - dml_print("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, ref_freq_to_pix_freq); - dml_print("DML_DLG: %s: interlaced = %d\n", __func__, interlaced); - - memset(disp_dlg_regs, 0, sizeof(*disp_dlg_regs)); - memset(disp_ttu_regs, 0, sizeof(*disp_ttu_regs)); - - ASSERT(refclk_freq_in_mhz != 0); - ASSERT(pclk_freq_in_mhz != 0); - ASSERT(ref_freq_to_pix_freq < 4.0); - - // Need to figure out which side of odm combine we're in - // Assume the pipe instance under the same plane is in order - - if (odm_mode == dml_odm_mode_bypass) { - disp_dlg_regs->refcyc_h_blank_end = (dml_uint_t)((dml_float_t) hblank_end * ref_freq_to_pix_freq); - } else if (odm_mode == dml_odm_mode_combine_2to1 || odm_mode == dml_odm_mode_combine_4to1) { - // find out how many pipe are in this plane - dml_uint_t num_active_pipes = dml_get_num_active_pipes(&mode_lib->ms.cache_display_cfg); - dml_uint_t first_pipe_idx_in_plane = __DML_NUM_PLANES__; - dml_uint_t pipe_idx_in_combine = 0; // pipe index within the plane - dml_uint_t odm_combine_factor = (odm_mode == dml_odm_mode_combine_2to1 ? 2 : 4); - - for (dml_uint_t i = 0; i < num_active_pipes; i++) { - if (dml_get_plane_idx(mode_lib, i) == plane_idx) { - if (i < first_pipe_idx_in_plane) { - first_pipe_idx_in_plane = i; - } - } - } - pipe_idx_in_combine = pipe_idx - first_pipe_idx_in_plane; // DML assumes the pipes in the same plane will have continuous indexing (i.e. plane 0 use pipe 0, 1, and plane 1 uses pipe 2, 3, etc.) - - disp_dlg_regs->refcyc_h_blank_end = (dml_uint_t)(((dml_float_t) hblank_end + (dml_float_t) pipe_idx_in_combine * (dml_float_t) hactive / (dml_float_t) odm_combine_factor) * ref_freq_to_pix_freq); - dml_print("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); - dml_print("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, first_pipe_idx_in_plane); - dml_print("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, pipe_idx_in_combine); - dml_print("DML_DLG: %s: odm_combine_factor = %d\n", __func__, odm_combine_factor); - } - dml_print("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); - - ASSERT(disp_dlg_regs->refcyc_h_blank_end < (dml_uint_t)dml_pow(2, 13)); - - disp_dlg_regs->ref_freq_to_pix_freq = (dml_uint_t)(ref_freq_to_pix_freq * dml_pow(2, 19)); - temp = dml_pow(2, 8); - disp_dlg_regs->refcyc_per_htotal = (dml_uint_t)(ref_freq_to_pix_freq * (dml_float_t)htotal * temp); - disp_dlg_regs->dlg_vblank_end = interlaced ? (vblank_end / 2) : vblank_end; // 15 bits - - min_ttu_vblank = dml_get_min_ttu_vblank_in_us(mode_lib, pipe_idx); - min_dst_y_next_start = (dml_uint_t)(dml_get_min_dst_y_next_start(mode_lib, pipe_idx)); - - dml_print("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, min_ttu_vblank); - dml_print("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, min_dst_y_next_start); - dml_print("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, ref_freq_to_pix_freq); - - vready_after_vcount0 = (dml_uint_t)(dml_get_vready_at_or_after_vsync(mode_lib, pipe_idx)); - disp_dlg_regs->vready_after_vcount0 = vready_after_vcount0; - - dml_print("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); - - dst_x_after_scaler = (dml_uint_t)(dml_get_dst_x_after_scaler(mode_lib, pipe_idx)); - dst_y_after_scaler = (dml_uint_t)(dml_get_dst_y_after_scaler(mode_lib, pipe_idx)); - - dml_print("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, dst_x_after_scaler); - dml_print("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, dst_y_after_scaler); - - dst_y_prefetch = dml_get_dst_y_prefetch(mode_lib, pipe_idx); - dst_y_per_vm_vblank = dml_get_dst_y_per_vm_vblank(mode_lib, pipe_idx); - dst_y_per_row_vblank = dml_get_dst_y_per_row_vblank(mode_lib, pipe_idx); - dst_y_per_vm_flip = dml_get_dst_y_per_vm_flip(mode_lib, pipe_idx); - dst_y_per_row_flip = dml_get_dst_y_per_row_flip(mode_lib, pipe_idx); - - // magic! - if (htotal <= 75) { - max_dst_y_per_vm_vblank = 100.0; - max_dst_y_per_row_vblank = 100.0; - } - - dml_print("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, dst_y_prefetch); - dml_print("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, dst_y_per_vm_flip); - dml_print("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, dst_y_per_row_flip); - dml_print("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, dst_y_per_vm_vblank); - dml_print("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, dst_y_per_row_vblank); - - ASSERT(dst_y_per_vm_vblank < max_dst_y_per_vm_vblank); - ASSERT(dst_y_per_row_vblank < max_dst_y_per_row_vblank); - ASSERT(dst_y_prefetch > (dst_y_per_vm_vblank + dst_y_per_row_vblank)); - - vratio_pre_l = dml_get_vratio_prefetch_l(mode_lib, pipe_idx); - vratio_pre_c = dml_get_vratio_prefetch_c(mode_lib, pipe_idx); - - dml_print("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, vratio_pre_l); - dml_print("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, vratio_pre_c); - - // Active - refcyc_per_line_delivery_pre_l = dml_get_refcyc_per_line_delivery_pre_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_line_delivery_l = dml_get_refcyc_per_line_delivery_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, refcyc_per_line_delivery_pre_l); - dml_print("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, refcyc_per_line_delivery_l); - - if (dual_plane) { - refcyc_per_line_delivery_pre_c = dml_get_refcyc_per_line_delivery_pre_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_line_delivery_c = dml_get_refcyc_per_line_delivery_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, refcyc_per_line_delivery_pre_c); - dml_print("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, refcyc_per_line_delivery_c); - } - - disp_dlg_regs->refcyc_per_vm_dmdata = (dml_uint_t)(dml_get_refcyc_per_vm_dmdata_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); - disp_dlg_regs->dmdata_dl_delta = (dml_uint_t)(dml_get_dmdata_dl_delta_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); - - refcyc_per_req_delivery_pre_l = dml_get_refcyc_per_req_delivery_pre_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_req_delivery_l = dml_get_refcyc_per_req_delivery_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, refcyc_per_req_delivery_pre_l); - dml_print("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, refcyc_per_req_delivery_l); - - if (dual_plane) { - refcyc_per_req_delivery_pre_c = dml_get_refcyc_per_req_delivery_pre_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_req_delivery_c = dml_get_refcyc_per_req_delivery_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, refcyc_per_req_delivery_pre_c); - dml_print("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, refcyc_per_req_delivery_c); - } - - // TTU - Cursor - ASSERT(num_cursors <= 1); - if (num_cursors > 0) { - refcyc_per_req_delivery_pre_cur0 = dml_get_refcyc_per_cursor_req_delivery_pre_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_req_delivery_cur0 = dml_get_refcyc_per_cursor_req_delivery_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_cur0 = %3.2f\n", __func__, refcyc_per_req_delivery_pre_cur0); - dml_print("DML_DLG: %s: refcyc_per_req_delivery_cur0 = %3.2f\n", __func__, refcyc_per_req_delivery_cur0); - } - - // Assign to register structures - disp_dlg_regs->min_dst_y_next_start = (dml_uint_t)((dml_float_t) min_dst_y_next_start * dml_pow(2, 2)); - ASSERT(disp_dlg_regs->min_dst_y_next_start < (dml_uint_t)dml_pow(2, 18)); - - disp_dlg_regs->dst_y_after_scaler = dst_y_after_scaler; // in terms of line - disp_dlg_regs->refcyc_x_after_scaler = (dml_uint_t)((dml_float_t) dst_x_after_scaler * ref_freq_to_pix_freq); // in terms of refclk - disp_dlg_regs->dst_y_prefetch = (dml_uint_t)(dst_y_prefetch * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_vm_vblank = (dml_uint_t)(dst_y_per_vm_vblank * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_row_vblank = (dml_uint_t)(dst_y_per_row_vblank * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_vm_flip = (dml_uint_t)(dst_y_per_vm_flip * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_row_flip = (dml_uint_t)(dst_y_per_row_flip * dml_pow(2, 2)); - - disp_dlg_regs->vratio_prefetch = (dml_uint_t)(vratio_pre_l * dml_pow(2, 19)); - disp_dlg_regs->vratio_prefetch_c = (dml_uint_t)(vratio_pre_c * dml_pow(2, 19)); - - dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); - dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); - dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); - dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); - - disp_dlg_regs->refcyc_per_vm_group_vblank = (dml_uint_t)(dml_get_refcyc_per_vm_group_vblank_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); - disp_dlg_regs->refcyc_per_vm_group_flip = (dml_uint_t)(dml_get_refcyc_per_vm_group_flip_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); - disp_dlg_regs->refcyc_per_vm_req_vblank = (dml_uint_t)(dml_get_refcyc_per_vm_req_vblank_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10)); - disp_dlg_regs->refcyc_per_vm_req_flip = (dml_uint_t)(dml_get_refcyc_per_vm_req_flip_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10)); - - dst_y_per_pte_row_nom_l = dml_get_dst_y_per_pte_row_nom_l(mode_lib, pipe_idx); - dst_y_per_pte_row_nom_c = dml_get_dst_y_per_pte_row_nom_c(mode_lib, pipe_idx); - dst_y_per_meta_row_nom_l = dml_get_dst_y_per_meta_row_nom_l(mode_lib, pipe_idx); - dst_y_per_meta_row_nom_c = dml_get_dst_y_per_meta_row_nom_c(mode_lib, pipe_idx); - - refcyc_per_pte_group_nom_l = dml_get_refcyc_per_pte_group_nom_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_pte_group_nom_c = dml_get_refcyc_per_pte_group_nom_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_pte_group_vblank_l = dml_get_refcyc_per_pte_group_vblank_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_pte_group_vblank_c = dml_get_refcyc_per_pte_group_vblank_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_pte_group_flip_l = dml_get_refcyc_per_pte_group_flip_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_pte_group_flip_c = dml_get_refcyc_per_pte_group_flip_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - refcyc_per_meta_chunk_nom_l = dml_get_refcyc_per_meta_chunk_nom_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_meta_chunk_nom_c = dml_get_refcyc_per_meta_chunk_nom_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_meta_chunk_vblank_l = dml_get_refcyc_per_meta_chunk_vblank_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_meta_chunk_vblank_c = dml_get_refcyc_per_meta_chunk_vblank_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_meta_chunk_flip_l = dml_get_refcyc_per_meta_chunk_flip_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - refcyc_per_meta_chunk_flip_c = dml_get_refcyc_per_meta_chunk_flip_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; - - disp_dlg_regs->dst_y_per_pte_row_nom_l = (dml_uint_t)(dst_y_per_pte_row_nom_l * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_pte_row_nom_c = (dml_uint_t)(dst_y_per_pte_row_nom_c * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_meta_row_nom_l = (dml_uint_t)(dst_y_per_meta_row_nom_l * dml_pow(2, 2)); - disp_dlg_regs->dst_y_per_meta_row_nom_c = (dml_uint_t)(dst_y_per_meta_row_nom_c * dml_pow(2, 2)); - disp_dlg_regs->refcyc_per_pte_group_nom_l = (dml_uint_t)(refcyc_per_pte_group_nom_l); - disp_dlg_regs->refcyc_per_pte_group_nom_c = (dml_uint_t)(refcyc_per_pte_group_nom_c); - disp_dlg_regs->refcyc_per_pte_group_vblank_l = (dml_uint_t)(refcyc_per_pte_group_vblank_l); - disp_dlg_regs->refcyc_per_pte_group_vblank_c = (dml_uint_t)(refcyc_per_pte_group_vblank_c); - disp_dlg_regs->refcyc_per_pte_group_flip_l = (dml_uint_t)(refcyc_per_pte_group_flip_l); - disp_dlg_regs->refcyc_per_pte_group_flip_c = (dml_uint_t)(refcyc_per_pte_group_flip_c); - disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (dml_uint_t)(refcyc_per_meta_chunk_nom_l); - disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (dml_uint_t)(refcyc_per_meta_chunk_nom_c); - disp_dlg_regs->refcyc_per_meta_chunk_vblank_l = (dml_uint_t)(refcyc_per_meta_chunk_vblank_l); - disp_dlg_regs->refcyc_per_meta_chunk_vblank_c = (dml_uint_t)(refcyc_per_meta_chunk_vblank_c); - disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (dml_uint_t)(refcyc_per_meta_chunk_flip_l); - disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (dml_uint_t)(refcyc_per_meta_chunk_flip_c); - disp_dlg_regs->refcyc_per_line_delivery_pre_l = (dml_uint_t)dml_floor(refcyc_per_line_delivery_pre_l, 1); - disp_dlg_regs->refcyc_per_line_delivery_l = (dml_uint_t)dml_floor(refcyc_per_line_delivery_l, 1); - disp_dlg_regs->refcyc_per_line_delivery_pre_c = (dml_uint_t)dml_floor(refcyc_per_line_delivery_pre_c, 1); - disp_dlg_regs->refcyc_per_line_delivery_c = (dml_uint_t)dml_floor(refcyc_per_line_delivery_c, 1); - - disp_dlg_regs->chunk_hdl_adjust_cur0 = 3; - disp_dlg_regs->dst_y_offset_cur0 = 0; - disp_dlg_regs->chunk_hdl_adjust_cur1 = 3; - disp_dlg_regs->dst_y_offset_cur1 = 0; - - disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off - - disp_ttu_regs->refcyc_per_req_delivery_pre_l = (dml_uint_t)(refcyc_per_req_delivery_pre_l * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_l = (dml_uint_t)(refcyc_per_req_delivery_l * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_pre_c = (dml_uint_t)(refcyc_per_req_delivery_pre_c * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_c = (dml_uint_t)(refcyc_per_req_delivery_c * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_pre_cur0 = (dml_uint_t)(refcyc_per_req_delivery_pre_cur0 * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_cur0 = (dml_uint_t)(refcyc_per_req_delivery_cur0 * dml_pow(2, 10)); - disp_ttu_regs->refcyc_per_req_delivery_pre_cur1 = 0; - disp_ttu_regs->refcyc_per_req_delivery_cur1 = 0; - disp_ttu_regs->qos_level_low_wm = 0; - - disp_ttu_regs->qos_level_high_wm = (dml_uint_t)(4.0 * (dml_float_t)htotal * ref_freq_to_pix_freq); - - disp_ttu_regs->qos_level_flip = 14; - disp_ttu_regs->qos_level_fixed_l = 8; - disp_ttu_regs->qos_level_fixed_c = 8; - disp_ttu_regs->qos_level_fixed_cur0 = 8; - disp_ttu_regs->qos_ramp_disable_l = 0; - disp_ttu_regs->qos_ramp_disable_c = 0; - disp_ttu_regs->qos_ramp_disable_cur0 = 0; - disp_ttu_regs->min_ttu_vblank = (dml_uint_t)(min_ttu_vblank * refclk_freq_in_mhz); - - // CHECK for HW registers' range, assert or clamp - ASSERT(refcyc_per_req_delivery_pre_l < dml_pow(2, 13)); - ASSERT(refcyc_per_req_delivery_l < dml_pow(2, 13)); - ASSERT(refcyc_per_req_delivery_pre_c < dml_pow(2, 13)); - ASSERT(refcyc_per_req_delivery_c < dml_pow(2, 13)); - if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_group_vblank = (dml_uint_t)(dml_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_group_flip >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_group_flip = (dml_uint_t)(dml_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_req_vblank >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_req_vblank = (dml_uint_t)(dml_pow(2, 23) - 1); - - if (disp_dlg_regs->refcyc_per_vm_req_flip >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_vm_req_flip = (dml_uint_t)(dml_pow(2, 23) - 1); - - - ASSERT(disp_dlg_regs->dst_y_after_scaler < (dml_uint_t)8); - ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->dst_y_per_pte_row_nom_l < (dml_uint_t)dml_pow(2, 17)); - if (dual_plane) { - if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (dml_uint_t)dml_pow(2, 17)) { // FIXME what so special about chroma, can we just assert? - dml_print("DML_DLG: %s: Warning dst_y_per_pte_row_nom_c %u > register max U15.2 %u\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (dml_uint_t)dml_pow(2, 17) - 1); - } - } - ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_l < (dml_uint_t)dml_pow(2, 17)); - ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_c < (dml_uint_t)dml_pow(2, 17)); - - if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_pte_group_nom_l = (dml_uint_t)(dml_pow(2, 23) - 1); - if (dual_plane) { - if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_pte_group_nom_c = (dml_uint_t)(dml_pow(2, 23) - 1); - } - ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (dml_uint_t)dml_pow(2, 13)); - if (dual_plane) { - ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (dml_uint_t)dml_pow(2, 13)); - } - - if (disp_dlg_regs->refcyc_per_meta_chunk_nom_l >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (dml_uint_t)(dml_pow(2, 23) - 1); - if (dual_plane) { - if (disp_dlg_regs->refcyc_per_meta_chunk_nom_c >= (dml_uint_t)dml_pow(2, 23)) - disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (dml_uint_t)(dml_pow(2, 23) - 1); - } - ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_l < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_c < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (dml_uint_t)dml_pow(2, 13)); - ASSERT(disp_ttu_regs->qos_level_low_wm < (dml_uint_t) dml_pow(2, 14)); - ASSERT(disp_ttu_regs->qos_level_high_wm < (dml_uint_t) dml_pow(2, 14)); - ASSERT(disp_ttu_regs->min_ttu_vblank < (dml_uint_t) dml_pow(2, 24)); - - dml_print_ttu_regs_st(disp_ttu_regs); - dml_print_dlg_regs_st(disp_dlg_regs); - dml_print("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); -} - -void dml_rq_dlg_get_arb_params(struct display_mode_lib_st *mode_lib, dml_display_arb_params_st *arb_param) -{ - memset(arb_param, 0, sizeof(*arb_param)); - arb_param->max_req_outstanding = 256; - arb_param->min_req_outstanding = 256; // turn off the sat level feature if this set to max - arb_param->sat_level_us = 60; - arb_param->hvm_max_qos_commit_threshold = 0xf; - arb_param->hvm_min_req_outstand_commit_threshold = 0xa; - arb_param->compbuf_reserved_space_kbytes = 2 * 8; // assume max data chunk size of 8K -} diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h b/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h deleted file mode 100644 index bf491cf0582d..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h +++ /dev/null @@ -1,63 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef __DML_DISPLAY_RQ_DLG_CALC_H__ -#define __DML_DISPLAY_RQ_DLG_CALC_H__ - -#include "display_mode_core_structs.h" -#include "display_mode_lib_defines.h" - -struct display_mode_lib_st; - -// Function: dml_rq_dlg_get_rq_reg -// Main entry point for test to get the register values out of this DML class. -// This function calls and fucntions to calculate -// and then populate the rq_regs struct -// Input: -// Assume mode_program is already called -// Output: -// rq_regs - struct that holds all the RQ registers field value. -// See also: - -void dml_rq_dlg_get_rq_reg(dml_display_rq_regs_st *rq_regs, - struct display_mode_lib_st *mode_lib, - const dml_uint_t pipe_idx); - -// Function: dml_rq_dlg_get_dlg_reg -// Calculate and return DLG and TTU register struct given the system setting -// Output: -// dlg_regs - output DLG register struct -// ttu_regs - output DLG TTU register struct -// Input: -// Assume mode_program is already called -// pipe_idx - index that identifies the e2e_pipe_param that corresponding to this dlg -void dml_rq_dlg_get_dlg_reg(dml_display_dlg_regs_st *dlg_regs, - dml_display_ttu_regs_st *ttu_regs, - struct display_mode_lib_st *mode_lib, - const dml_uint_t pipe_idx); - -// Function: dml_rq_dlg_get_arb_params -void dml_rq_dlg_get_arb_params(struct display_mode_lib_st *mode_lib, dml_display_arb_params_st *arb_param); - -#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h b/drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h deleted file mode 100644 index 2a2f84e07ca8..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h +++ /dev/null @@ -1,31 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright 2023 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: AMD - * - */ -#ifndef __DML_LOGGING_H__ -#define __DML_LOGGING_H__ - -#define dml_print(...) ((void)0) - -#endif //__DML_LOGGING_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile new file mode 100644 index 000000000000..97e068b6bf6b --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile @@ -0,0 +1,140 @@ +# SPDX-License-Identifier: MIT */ +# +# Copyright 2023 Advanced Micro Devices, Inc. +# +# Permission is hereby granted, free of charge, to any person obtaining a +# copy of this software and associated documentation files (the "Software"), +# to deal in the Software without restriction, including without limitation +# the rights to use, copy, modify, merge, publish, distribute, sublicense, +# and/or sell copies of the Software, and to permit persons to whom the +# Software is furnished to do so, subject to the following conditions: +# +# The above copyright notice and this permission notice shall be included in +# all copies or substantial portions of the Software. +# +# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR +# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, +# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR +# OTHER DEALINGS IN THE SOFTWARE. +# +# Authors: AMD +# +# Makefile for dml2. + +dml2_ccflags := $(CC_FLAGS_FPU) +dml2_rcflags := $(CC_FLAGS_NO_FPU) + +ifneq ($(CONFIG_FRAME_WARN),0) + ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) + ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) + frame_warn_limit := 4096 + else + frame_warn_limit := 3072 + endif + else + frame_warn_limit := 2056 + endif + + ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) + frame_warn_flag := -Wframe-larger-than=$(frame_warn_limit) + endif +endif + +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0 +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_core +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_mcg/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_dpmm/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_pmo/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/inc +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/inc +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/ + +CFLAGS_$(AMDDALPATH)/dc/dml2_0/display_mode_core.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/display_mode_util.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_wrapper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_utils.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_policy.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_translation_helper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_mall_phantom.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml_display_rq_dlg_calc.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_dc_resource_mgmt.o := $(dml2_ccflags) + +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/display_mode_core.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/display_mode_util.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_wrapper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_policy.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_translation_helper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_mall_phantom.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml_display_rq_dlg_calc.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_dc_resource_mgmt.o := $(dml2_rcflags) + +DML2 = display_mode_core.o display_mode_util.o dml2_wrapper.o \ + dml2_utils.o dml2_policy.o dml2_translation_helper.o dml2_dc_resource_mgmt.o dml2_mall_phantom.o \ + dml_display_rq_dlg_calc.o + +AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2_0/,$(DML2)) + +AMD_DISPLAY_FILES += $(AMD_DAL_DML2) + +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml21_wrapper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_translation_helper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_utils.o := $(dml2_ccflags) + +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml21_wrapper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_translation_helper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_utils.o := $(dml2_rcflags) + +DML21 := src/dml2_top/dml2_top_interfaces.o +DML21 += src/dml2_top/dml2_top_soc15.o +DML21 += src/dml2_core/dml2_core_dcn4.o +DML21 += src/dml2_core/dml2_core_utils.o +DML21 += src/dml2_core/dml2_core_factory.o +DML21 += src/dml2_core/dml2_core_dcn4_calcs.o +DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o +DML21 += src/dml2_dpmm/dml2_dpmm_factory.o +DML21 += src/dml2_mcg/dml2_mcg_dcn4.o +DML21 += src/dml2_mcg/dml2_mcg_factory.o +DML21 += src/dml2_pmo/dml2_pmo_dcn3.o +DML21 += src/dml2_pmo/dml2_pmo_factory.o +DML21 += src/dml2_pmo/dml2_pmo_dcn4_fams2.o +DML21 += src/dml2_standalone_libraries/lib_float_math.o +DML21 += dml21_translation_helper.o +DML21 += dml21_wrapper.o +DML21 += dml21_utils.o + +AMD_DAL_DML21 = $(addprefix $(AMDDALPATH)/dc/dml2_0/dml21/,$(DML21)) + +AMD_DISPLAY_FILES += $(AMD_DAL_DML21) diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h b/drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h new file mode 100644 index 000000000000..b954c9648fbe --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h @@ -0,0 +1,94 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __CMNTYPES_H__ +#define __CMNTYPES_H__ + +#ifdef __GNUC__ +#if __GNUC__ == 4 && __GNUC_MINOR__ > 7 +typedef unsigned int uint; +#endif +#endif + +typedef signed char int8, *pint8; +typedef signed short int16, *pint16; +typedef signed int int32, *pint32; +typedef signed int64, *pint64; + +typedef unsigned char uint8, *puint8; +typedef unsigned short uint16, *puint16; +typedef unsigned int uint32, *puint32; +typedef unsigned uint64, *puint64; + +typedef unsigned long int ulong; +typedef unsigned char uchar; +typedef unsigned int uint; + +typedef void *pvoid; +typedef char *pchar; +typedef const void *const_pvoid; +typedef const char *const_pchar; + +typedef struct rgba_struct { + uint8 a; + uint8 r; + uint8 g; + uint8 b; +} rgba_t; + +typedef struct { + uint8 blue; + uint8 green; + uint8 red; + uint8 alpha; +} gen_color_t; + +typedef union { + uint32 val; + gen_color_t f; +} gen_color_u; + +// +// Types to make it easy to get or set the bits of a float/double. +// Avoids automatic casting from int to float and back. +// +#if 0 +typedef union { + uint32 i; + float f; +} uintfloat32; + +typedef union { + uint64 i; + double f; +} uintfloat64; + +#ifndef UNREFERENCED_PARAMETER +#define UNREFERENCED_PARAMETER(x) (x = x) +#endif +#endif + +#endif //__CMNTYPES_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c new file mode 100644 index 000000000000..c468f492b876 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c @@ -0,0 +1,10337 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "display_mode_core.h" +#include "display_mode_util.h" +#include "display_mode_lib_defines.h" + +#include "dml_assert.h" + +#define DML2_MAX_FMT_420_BUFFER_WIDTH 4096 +#define TB_BORROWED_MAX 400 +#define DML_MAX_VSTARTUP_START 1023 + +// --------------------------- +// Declaration Begins +// --------------------------- +static void CalculateBytePerPixelAndBlockSizes( + enum dml_source_format_class SourcePixelFormat, + enum dml_swizzle_mode SurfaceTiling, + // Output + dml_uint_t *BytePerPixelY, + dml_uint_t *BytePerPixelC, + dml_float_t *BytePerPixelDETY, + dml_float_t *BytePerPixelDETC, + dml_uint_t *BlockHeight256BytesY, + dml_uint_t *BlockHeight256BytesC, + dml_uint_t *BlockWidth256BytesY, + dml_uint_t *BlockWidth256BytesC, + dml_uint_t *MacroTileHeightY, + dml_uint_t *MacroTileHeightC, + dml_uint_t *MacroTileWidthY, + dml_uint_t *MacroTileWidthC); + +static dml_float_t CalculateWriteBackDISPCLK( + enum dml_source_format_class WritebackPixelFormat, + dml_float_t PixelClock, + dml_float_t WritebackHRatio, + dml_float_t WritebackVRatio, + dml_uint_t WritebackHTaps, + dml_uint_t WritebackVTaps, + dml_uint_t WritebackSourceWidth, + dml_uint_t WritebackDestinationWidth, + dml_uint_t HTotal, + dml_uint_t WritebackLineBufferSize, + dml_float_t DISPCLKDPPCLKVCOSpeed); + +static void CalculateVMRowAndSwath( + struct display_mode_lib_scratch_st *s, + struct CalculateVMRowAndSwath_params_st *p); + +static void CalculateOutputLink( + dml_float_t PHYCLKPerState, + dml_float_t PHYCLKD18PerState, + dml_float_t PHYCLKD32PerState, + dml_float_t Downspreading, + dml_bool_t IsMainSurfaceUsingTheIndicatedTiming, + enum dml_output_encoder_class Output, + enum dml_output_format_class OutputFormat, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_float_t PixelClockBackEnd, + dml_float_t ForcedOutputLinkBPP, + dml_uint_t DSCInputBitPerComponent, + dml_uint_t NumberOfDSCSlices, + dml_float_t AudioSampleRate, + dml_uint_t AudioSampleLayout, + enum dml_odm_mode ODMModeNoDSC, + enum dml_odm_mode ODMModeDSC, + enum dml_dsc_enable DSCEnable, + dml_uint_t OutputLinkDPLanes, + enum dml_output_link_dp_rate OutputLinkDPRate, + + // Output + dml_bool_t *RequiresDSC, + dml_bool_t *RequiresFEC, + dml_float_t *OutBpp, + enum dml_output_type_and_rate__type *OutputType, + enum dml_output_type_and_rate__rate *OutputRate, + dml_uint_t *RequiredSlots); + +static void CalculateODMMode( + dml_uint_t MaximumPixelsPerLinePerDSCUnit, + dml_uint_t HActive, + enum dml_output_encoder_class Output, + enum dml_output_format_class OutputFormat, + enum dml_odm_use_policy ODMUse, + dml_float_t StateDispclk, + dml_float_t MaxDispclk, + dml_bool_t DSCEnable, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t MaxNumDPP, + dml_float_t PixelClock, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKRampingMargin, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_uint_t NumberOfDSCSlices, + + // Output + dml_bool_t *TotalAvailablePipesSupport, + dml_uint_t *NumberOfDPP, + enum dml_odm_mode *ODMMode, + dml_float_t *RequiredDISPCLKPerSurface); + +static dml_float_t CalculateRequiredDispclk( + enum dml_odm_mode ODMMode, + dml_float_t PixelClock, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKRampingMargin, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_float_t MaxDispclkSingle); + +static void CalculateSinglePipeDPPCLKAndSCLThroughput( + dml_float_t HRatio, + dml_float_t HRatioChroma, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_float_t MaxDCHUBToPSCLThroughput, + dml_float_t MaxPSCLToLBThroughput, + dml_float_t PixelClock, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t HTaps, + dml_uint_t HTapsChroma, + dml_uint_t VTaps, + dml_uint_t VTapsChroma, + + // Output + dml_float_t *PSCL_THROUGHPUT, + dml_float_t *PSCL_THROUGHPUT_CHROMA, + dml_float_t *DPPCLKUsingSingleDPP); + +static void CalculateDPPCLK( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_float_t DPPCLKUsingSingleDPP[], + dml_uint_t DPPPerSurface[], + + // Output + dml_float_t *GlobalDPPCLK, + dml_float_t Dppclk[]); + +static void CalculateMALLUseForStaticScreen( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t MALLAllocatedForDCNFinal, + enum dml_use_mall_for_static_screen_mode *UseMALLForStaticScreen, + dml_uint_t SurfaceSizeInMALL[], + dml_bool_t one_row_per_frame_fits_in_buffer[], + + // Output + dml_bool_t UsesMALLForStaticScreen[]); + +static dml_uint_t dscceComputeDelay( + dml_uint_t bpc, + dml_float_t BPP, + dml_uint_t sliceWidth, + dml_uint_t numSlices, + enum dml_output_format_class pixelFormat, + enum dml_output_encoder_class Output); + +static dml_uint_t dscComputeDelay(enum dml_output_format_class pixelFormat, + enum dml_output_encoder_class Output); + +static dml_bool_t CalculatePrefetchSchedule(struct display_mode_lib_scratch_st *scratch, + struct CalculatePrefetchSchedule_params_st *p); + +static dml_float_t RoundToDFSGranularity(dml_float_t Clock, dml_bool_t round_up, dml_float_t VCOSpeed); + +static void CalculateDCCConfiguration( + dml_bool_t DCCEnabled, + dml_bool_t DCCProgrammingAssumesScanDirectionUnknown, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t SurfaceWidthLuma, + dml_uint_t SurfaceWidthChroma, + dml_uint_t SurfaceHeightLuma, + dml_uint_t SurfaceHeightChroma, + dml_uint_t nomDETInKByte, + dml_uint_t RequestHeight256ByteLuma, + dml_uint_t RequestHeight256ByteChroma, + enum dml_swizzle_mode TilingFormat, + dml_uint_t BytePerPixelY, + dml_uint_t BytePerPixelC, + dml_float_t BytePerPixelDETY, + dml_float_t BytePerPixelDETC, + enum dml_rotation_angle SourceScan, + // Output + dml_uint_t *MaxUncompressedBlockLuma, + dml_uint_t *MaxUncompressedBlockChroma, + dml_uint_t *MaxCompressedBlockLuma, + dml_uint_t *MaxCompressedBlockChroma, + dml_uint_t *IndependentBlockLuma, + dml_uint_t *IndependentBlockChroma); + +static dml_uint_t CalculatePrefetchSourceLines( + dml_float_t VRatio, + dml_uint_t VTaps, + dml_bool_t Interlace, + dml_bool_t ProgressiveToInterlaceUnitInOPP, + dml_uint_t SwathHeight, + enum dml_rotation_angle SourceScan, + dml_bool_t ViewportStationary, + dml_uint_t SwathWidth, + dml_uint_t ViewportHeight, + dml_uint_t ViewportXStart, + dml_uint_t ViewportYStart, + + // Output + dml_uint_t *VInitPreFill, + dml_uint_t *MaxNumSwath); + +static dml_uint_t CalculateVMAndRowBytes( + dml_bool_t ViewportStationary, + dml_bool_t DCCEnable, + dml_uint_t NumberOfDPPs, + dml_uint_t BlockHeight256Bytes, + dml_uint_t BlockWidth256Bytes, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t SurfaceTiling, + dml_uint_t BytePerPixel, + enum dml_rotation_angle SourceScan, + dml_uint_t SwathWidth, + dml_uint_t ViewportHeight, + dml_uint_t ViewportXStart, + dml_uint_t ViewportYStart, + dml_bool_t GPUVMEnable, + dml_uint_t GPUVMMaxPageTableLevels, + dml_uint_t GPUVMMinPageSizeKBytes, + dml_uint_t PTEBufferSizeInRequests, + dml_uint_t Pitch, + dml_uint_t DCCMetaPitch, + dml_uint_t MacroTileWidth, + dml_uint_t MacroTileHeight, + + // Output + dml_uint_t *MetaRowByte, + dml_uint_t *PixelPTEBytesPerRow, + dml_uint_t *PixelPTEBytesPerRowStorage, // for PTE buffer size check + dml_uint_t *dpte_row_width_ub, + dml_uint_t *dpte_row_height, + dml_uint_t *dpte_row_height_linear, + dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, + dml_uint_t *dpte_row_width_ub_one_row_per_frame, + dml_uint_t *dpte_row_height_one_row_per_frame, + dml_uint_t *MetaRequestWidth, + dml_uint_t *MetaRequestHeight, + dml_uint_t *meta_row_width, + dml_uint_t *meta_row_height, + dml_uint_t *PixelPTEReqWidth, + dml_uint_t *PixelPTEReqHeight, + dml_uint_t *PTERequestSize, + dml_uint_t *DPDE0BytesFrame, + dml_uint_t *MetaPTEBytesFrame); + +static dml_float_t CalculateTWait( + dml_uint_t PrefetchMode, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, + dml_bool_t SynchronizeDRRDisplaysForUCLKPStateChangeFinal, + dml_bool_t DRRDisplay, + dml_float_t DRAMClockChangeLatency, + dml_float_t FCLKChangeLatency, + dml_float_t UrgentLatency, + dml_float_t SREnterPlusExitTime); + +static void CalculatePrefetchMode( + enum dml_prefetch_modes AllowForPStateChangeOrStutterInVBlank, + dml_uint_t *MinPrefetchMode, + dml_uint_t *MaxPrefetchMode); + +static void CalculateRowBandwidth( + dml_bool_t GPUVMEnable, + enum dml_source_format_class SourcePixelFormat, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_bool_t DCCEnable, + dml_float_t LineTime, + dml_uint_t MetaRowByteLuma, + dml_uint_t MetaRowByteChroma, + dml_uint_t meta_row_height_luma, + dml_uint_t meta_row_height_chroma, + dml_uint_t PixelPTEBytesPerRowLuma, + dml_uint_t PixelPTEBytesPerRowChroma, + dml_uint_t dpte_row_height_luma, + dml_uint_t dpte_row_height_chroma, + // Output + dml_float_t *meta_row_bw, + dml_float_t *dpte_row_bw); + +static void CalculateFlipSchedule( + dml_float_t HostVMInefficiencyFactor, + dml_float_t UrgentExtraLatency, + dml_float_t UrgentLatency, + dml_uint_t GPUVMMaxPageTableLevels, + dml_bool_t HostVMEnable, + dml_uint_t HostVMMaxNonCachedPageTableLevels, + dml_bool_t GPUVMEnable, + dml_uint_t HostVMMinPageSize, + dml_float_t PDEAndMetaPTEBytesPerFrame, + dml_float_t MetaRowBytes, + dml_float_t DPTEBytesPerRow, + dml_float_t BandwidthAvailableForImmediateFlip, + dml_uint_t TotImmediateFlipBytes, + enum dml_source_format_class SourcePixelFormat, + dml_float_t LineTime, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_float_t Tno_bw, + dml_bool_t DCCEnable, + dml_uint_t dpte_row_height, + dml_uint_t meta_row_height, + dml_uint_t dpte_row_height_chroma, + dml_uint_t meta_row_height_chroma, + dml_bool_t use_one_row_for_frame_flip, + + // Output + dml_float_t *DestinationLinesToRequestVMInImmediateFlip, + dml_float_t *DestinationLinesToRequestRowInImmediateFlip, + dml_float_t *final_flip_bw, + dml_bool_t *ImmediateFlipSupportedForPipe); + +static dml_float_t CalculateWriteBackDelay( + enum dml_source_format_class WritebackPixelFormat, + dml_float_t WritebackHRatio, + dml_float_t WritebackVRatio, + dml_uint_t WritebackVTaps, + dml_uint_t WritebackDestinationWidth, + dml_uint_t WritebackDestinationHeight, + dml_uint_t WritebackSourceHeight, + dml_uint_t HTotal); + +static void CalculateVUpdateAndDynamicMetadataParameters( + dml_uint_t MaxInterDCNTileRepeaters, + dml_float_t Dppclk, + dml_float_t DISPCLK, + dml_float_t DCFClkDeepSleep, + dml_float_t PixelClock, + dml_uint_t HTotal, + dml_uint_t VBlank, + dml_uint_t DynamicMetadataTransmittedBytes, + dml_uint_t DynamicMetadataLinesBeforeActiveRequired, + dml_uint_t InterlaceEnable, + dml_bool_t ProgressiveToInterlaceUnitInOPP, + dml_float_t *TSetup, + dml_float_t *Tdmbf, + dml_float_t *Tdmec, + dml_float_t *Tdmsks, + dml_uint_t *VUpdateOffsetPix, + dml_uint_t *VUpdateWidthPix, + dml_uint_t *VReadyOffsetPix); + +static void PixelClockAdjustmentForProgressiveToInterlaceUnit(struct dml_display_cfg_st *display_cfg, dml_bool_t ptoi_supported); + +static dml_float_t TruncToValidBPP( + dml_float_t LinkBitRate, + dml_uint_t Lanes, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_float_t PixelClock, + dml_float_t DesiredBPP, + dml_bool_t DSCEnable, + enum dml_output_encoder_class Output, + enum dml_output_format_class Format, + dml_uint_t DSCInputBitPerComponent, + dml_uint_t DSCSlices, + dml_uint_t AudioRate, + dml_uint_t AudioLayout, + enum dml_odm_mode ODMModeNoDSC, + enum dml_odm_mode ODMModeDSC, + // Output + dml_uint_t *RequiredSlotsSingle); + +static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( + struct display_mode_lib_scratch_st *s, + struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *p); + +static void CalculateDCFCLKDeepSleep( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t BytePerPixelY[], + dml_uint_t BytePerPixelC[], + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_uint_t SwathWidthY[], + dml_uint_t SwathWidthC[], + dml_uint_t DPPPerSurface[], + dml_float_t HRatio[], + dml_float_t HRatioChroma[], + dml_float_t PixelClock[], + dml_float_t PSCL_THROUGHPUT[], + dml_float_t PSCL_THROUGHPUT_CHROMA[], + dml_float_t Dppclk[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_uint_t ReturnBusWidth, + + // Output + dml_float_t *DCFCLKDeepSleep); + +static void CalculateUrgentBurstFactor( + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, + dml_uint_t swath_width_luma_ub, + dml_uint_t swath_width_chroma_ub, + dml_uint_t SwathHeightY, + dml_uint_t SwathHeightC, + dml_float_t LineTime, + dml_float_t UrgentLatency, + dml_float_t CursorBufferSize, + dml_uint_t CursorWidth, + dml_uint_t CursorBPP, + dml_float_t VRatio, + dml_float_t VRatioC, + dml_float_t BytePerPixelInDETY, + dml_float_t BytePerPixelInDETC, + dml_uint_t DETBufferSizeY, + dml_uint_t DETBufferSizeC, + // Output + dml_float_t *UrgentBurstFactorCursor, + dml_float_t *UrgentBurstFactorLuma, + dml_float_t *UrgentBurstFactorChroma, + dml_bool_t *NotEnoughUrgentLatencyHiding); + +static dml_float_t RequiredDTBCLK( + dml_bool_t DSCEnable, + dml_float_t PixelClock, + enum dml_output_format_class OutputFormat, + dml_float_t OutputBpp, + dml_uint_t DSCSlices, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_uint_t AudioRate, + dml_uint_t AudioLayoutSingle); + +static void UseMinimumDCFCLK( + struct display_mode_lib_scratch_st *scratch, + struct UseMinimumDCFCLK_params_st *p); + +static void CalculatePixelDeliveryTimes( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_float_t VRatioPrefetchY[], + dml_float_t VRatioPrefetchC[], + dml_uint_t swath_width_luma_ub[], + dml_uint_t swath_width_chroma_ub[], + dml_uint_t DPPPerSurface[], + dml_float_t HRatio[], + dml_float_t HRatioChroma[], + dml_float_t PixelClock[], + dml_float_t PSCL_THROUGHPUT[], + dml_float_t PSCL_THROUGHPUT_CHROMA[], + dml_float_t Dppclk[], + dml_uint_t BytePerPixelC[], + enum dml_rotation_angle SourceScan[], + dml_uint_t NumberOfCursors[], + dml_uint_t CursorWidth[], + dml_uint_t CursorBPP[], + dml_uint_t BlockWidth256BytesY[], + dml_uint_t BlockHeight256BytesY[], + dml_uint_t BlockWidth256BytesC[], + dml_uint_t BlockHeight256BytesC[], + + // Output + dml_float_t DisplayPipeLineDeliveryTimeLuma[], + dml_float_t DisplayPipeLineDeliveryTimeChroma[], + dml_float_t DisplayPipeLineDeliveryTimeLumaPrefetch[], + dml_float_t DisplayPipeLineDeliveryTimeChromaPrefetch[], + dml_float_t DisplayPipeRequestDeliveryTimeLuma[], + dml_float_t DisplayPipeRequestDeliveryTimeChroma[], + dml_float_t DisplayPipeRequestDeliveryTimeLumaPrefetch[], + dml_float_t DisplayPipeRequestDeliveryTimeChromaPrefetch[], + dml_float_t CursorRequestDeliveryTime[], + dml_float_t CursorRequestDeliveryTimePrefetch[]); + +static void CalculateMetaAndPTETimes( + dml_bool_t use_one_row_for_frame[], + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t GPUVMEnable, + dml_uint_t MetaChunkSize, + dml_uint_t MinMetaChunkSizeBytes, + dml_uint_t HTotal[], + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_float_t DestinationLinesToRequestRowInVBlank[], + dml_float_t DestinationLinesToRequestRowInImmediateFlip[], + dml_bool_t DCCEnable[], + dml_float_t PixelClock[], + dml_uint_t BytePerPixelY[], + dml_uint_t BytePerPixelC[], + enum dml_rotation_angle SourceScan[], + dml_uint_t dpte_row_height[], + dml_uint_t dpte_row_height_chroma[], + dml_uint_t meta_row_width[], + dml_uint_t meta_row_width_chroma[], + dml_uint_t meta_row_height[], + dml_uint_t meta_row_height_chroma[], + dml_uint_t meta_req_width[], + dml_uint_t meta_req_width_chroma[], + dml_uint_t meta_req_height[], + dml_uint_t meta_req_height_chroma[], + dml_uint_t dpte_group_bytes[], + dml_uint_t PTERequestSizeY[], + dml_uint_t PTERequestSizeC[], + dml_uint_t PixelPTEReqWidthY[], + dml_uint_t PixelPTEReqHeightY[], + dml_uint_t PixelPTEReqWidthC[], + dml_uint_t PixelPTEReqHeightC[], + dml_uint_t dpte_row_width_luma_ub[], + dml_uint_t dpte_row_width_chroma_ub[], + + // Output + dml_float_t DST_Y_PER_PTE_ROW_NOM_L[], + dml_float_t DST_Y_PER_PTE_ROW_NOM_C[], + dml_float_t DST_Y_PER_META_ROW_NOM_L[], + dml_float_t DST_Y_PER_META_ROW_NOM_C[], + dml_float_t TimePerMetaChunkNominal[], + dml_float_t TimePerChromaMetaChunkNominal[], + dml_float_t TimePerMetaChunkVBlank[], + dml_float_t TimePerChromaMetaChunkVBlank[], + dml_float_t TimePerMetaChunkFlip[], + dml_float_t TimePerChromaMetaChunkFlip[], + dml_float_t time_per_pte_group_nom_luma[], + dml_float_t time_per_pte_group_vblank_luma[], + dml_float_t time_per_pte_group_flip_luma[], + dml_float_t time_per_pte_group_nom_chroma[], + dml_float_t time_per_pte_group_vblank_chroma[], + dml_float_t time_per_pte_group_flip_chroma[]); + +static void CalculateVMGroupAndRequestTimes( + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t GPUVMEnable, + dml_uint_t GPUVMMaxPageTableLevels, + dml_uint_t HTotal[], + dml_uint_t BytePerPixelC[], + dml_float_t DestinationLinesToRequestVMInVBlank[], + dml_float_t DestinationLinesToRequestVMInImmediateFlip[], + dml_bool_t DCCEnable[], + dml_float_t PixelClock[], + dml_uint_t dpte_row_width_luma_ub[], + dml_uint_t dpte_row_width_chroma_ub[], + dml_uint_t vm_group_bytes[], + dml_uint_t dpde0_bytes_per_frame_ub_l[], + dml_uint_t dpde0_bytes_per_frame_ub_c[], + dml_uint_t meta_pte_bytes_per_frame_ub_l[], + dml_uint_t meta_pte_bytes_per_frame_ub_c[], + + // Output + dml_float_t TimePerVMGroupVBlank[], + dml_float_t TimePerVMGroupFlip[], + dml_float_t TimePerVMRequestVBlank[], + dml_float_t TimePerVMRequestFlip[]); + +static void CalculateStutterEfficiency( + struct display_mode_lib_scratch_st *scratch, + struct CalculateStutterEfficiency_params_st *p); + +static void CalculateSwathAndDETConfiguration( + struct display_mode_lib_scratch_st *scratch, + struct CalculateSwathAndDETConfiguration_params_st *p); + +static void CalculateSwathWidth( + dml_bool_t ForceSingleDPP, + dml_uint_t NumberOfActiveSurfaces, + enum dml_source_format_class SourcePixelFormat[], + enum dml_rotation_angle SourceScan[], + dml_bool_t ViewportStationary[], + dml_uint_t ViewportWidth[], + dml_uint_t ViewportHeight[], + dml_uint_t ViewportXStart[], + dml_uint_t ViewportYStart[], + dml_uint_t ViewportXStartC[], + dml_uint_t ViewportYStartC[], + dml_uint_t SurfaceWidthY[], + dml_uint_t SurfaceWidthC[], + dml_uint_t SurfaceHeightY[], + dml_uint_t SurfaceHeightC[], + enum dml_odm_mode ODMMode[], + dml_uint_t BytePerPixY[], + dml_uint_t BytePerPixC[], + dml_uint_t Read256BytesBlockHeightY[], + dml_uint_t Read256BytesBlockHeightC[], + dml_uint_t Read256BytesBlockWidthY[], + dml_uint_t Read256BytesBlockWidthC[], + dml_uint_t BlendingAndTiming[], + dml_uint_t HActive[], + dml_float_t HRatio[], + dml_uint_t DPPPerSurface[], + + // Output + dml_uint_t SwathWidthSingleDPPY[], + dml_uint_t SwathWidthSingleDPPC[], + dml_uint_t SwathWidthY[], + dml_uint_t SwathWidthC[], + dml_uint_t MaximumSwathHeightY[], + dml_uint_t MaximumSwathHeightC[], + dml_uint_t swath_width_luma_ub[], + dml_uint_t swath_width_chroma_ub[]); + +static dml_float_t CalculateExtraLatency( + dml_uint_t RoundTripPingLatencyCycles, + dml_uint_t ReorderingBytes, + dml_float_t DCFCLK, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t PixelChunkSizeInKByte, + dml_uint_t TotalNumberOfDCCActiveDPP, + dml_uint_t MetaChunkSize, + dml_float_t ReturnBW, + dml_bool_t GPUVMEnable, + dml_bool_t HostVMEnable, + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t NumberOfDPP[], + dml_uint_t dpte_group_bytes[], + dml_float_t HostVMInefficiencyFactor, + dml_uint_t HostVMMinPageSize, + dml_uint_t HostVMMaxNonCachedPageTableLevels); + +static dml_uint_t CalculateExtraLatencyBytes( + dml_uint_t ReorderingBytes, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t PixelChunkSizeInKByte, + dml_uint_t TotalNumberOfDCCActiveDPP, + dml_uint_t MetaChunkSize, + dml_bool_t GPUVMEnable, + dml_bool_t HostVMEnable, + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t NumberOfDPP[], + dml_uint_t dpte_group_bytes[], + dml_float_t HostVMInefficiencyFactor, + dml_uint_t HostVMMinPageSize, + dml_uint_t HostVMMaxNonCachedPageTableLevels); + +static dml_float_t CalculateUrgentLatency( + dml_float_t UrgentLatencyPixelDataOnly, + dml_float_t UrgentLatencyPixelMixedWithVMData, + dml_float_t UrgentLatencyVMDataOnly, + dml_bool_t DoUrgentLatencyAdjustment, + dml_float_t UrgentLatencyAdjustmentFabricClockComponent, + dml_float_t UrgentLatencyAdjustmentFabricClockReference, + dml_float_t FabricClockSingle); + +static dml_bool_t UnboundedRequest( + enum dml_unbounded_requesting_policy UseUnboundedRequestingFinal, + dml_uint_t TotalNumberOfActiveDPP, + dml_bool_t NoChromaOrLinear, + enum dml_output_encoder_class Output); + +static void CalculateSurfaceSizeInMall( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t MALLAllocatedForDCN, + enum dml_use_mall_for_static_screen_mode UseMALLForStaticScreen[], + dml_bool_t DCCEnable[], + dml_bool_t ViewportStationary[], + dml_uint_t ViewportXStartY[], + dml_uint_t ViewportYStartY[], + dml_uint_t ViewportXStartC[], + dml_uint_t ViewportYStartC[], + dml_uint_t ViewportWidthY[], + dml_uint_t ViewportHeightY[], + dml_uint_t BytesPerPixelY[], + dml_uint_t ViewportWidthC[], + dml_uint_t ViewportHeightC[], + dml_uint_t BytesPerPixelC[], + dml_uint_t SurfaceWidthY[], + dml_uint_t SurfaceWidthC[], + dml_uint_t SurfaceHeightY[], + dml_uint_t SurfaceHeightC[], + dml_uint_t Read256BytesBlockWidthY[], + dml_uint_t Read256BytesBlockWidthC[], + dml_uint_t Read256BytesBlockHeightY[], + dml_uint_t Read256BytesBlockHeightC[], + dml_uint_t ReadBlockWidthY[], + dml_uint_t ReadBlockWidthC[], + dml_uint_t ReadBlockHeightY[], + dml_uint_t ReadBlockHeightC[], + + // Output + dml_uint_t SurfaceSizeInMALL[], + dml_bool_t *ExceededMALLSize); + +static void CalculateDETBufferSize( + dml_uint_t DETSizeOverride[], + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + dml_bool_t ForceSingleDPP, + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t UnboundedRequestEnabled, + dml_uint_t nomDETInKByte, + dml_uint_t MaxTotalDETInKByte, + dml_uint_t ConfigReturnBufferSizeInKByte, + dml_uint_t MinCompressedBufferSizeInKByte, + dml_uint_t ConfigReturnBufferSegmentSizeInkByte, + dml_uint_t CompressedBufferSegmentSizeInkByteFinal, + enum dml_source_format_class SourcePixelFormat[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_uint_t RotesY[], + dml_uint_t RoundedUpMaxSwathSizeBytesC[], + dml_uint_t DPPPerSurface[], + // Output + dml_uint_t DETBufferSizeInKByte[], + dml_uint_t *CompressedBufferSizeInkByte); + +static void CalculateMaxDETAndMinCompressedBufferSize( + dml_uint_t ConfigReturnBufferSizeInKByte, + dml_uint_t ConfigReturnBufferSegmentSizeInKByte, + dml_uint_t ROBBufferSizeInKByte, + dml_uint_t MaxNumDPP, + dml_bool_t nomDETInKByteOverrideEnable, + dml_uint_t nomDETInKByteOverrideValue, + + // Output + dml_uint_t *MaxTotalDETInKByte, + dml_uint_t *nomDETInKByte, + dml_uint_t *MinCompressedBufferSizeInKByte); + +static dml_uint_t DSCDelayRequirement( + dml_bool_t DSCEnabled, + enum dml_odm_mode ODMMode, + dml_uint_t DSCInputBitPerComponent, + dml_float_t OutputBpp, + dml_uint_t HActive, + dml_uint_t HTotal, + dml_uint_t NumberOfDSCSlices, + enum dml_output_format_class OutputFormat, + enum dml_output_encoder_class Output, + dml_float_t PixelClock, + dml_float_t PixelClockBackEnd); + +static dml_bool_t CalculateVActiveBandwithSupport( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + dml_bool_t NotUrgentLatencyHiding[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[]); + +static void CalculatePrefetchBandwithSupport( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + dml_bool_t NotUrgentLatencyHiding[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_float_t cursor_bw_pre[], + dml_float_t prefetch_vmrow_bw[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[], + + // Output + dml_float_t *PrefetchBandwidth, + dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch, + dml_float_t *FractionOfUrgentBandwidth, + dml_bool_t *PrefetchBandwidthSupport); + +static dml_float_t CalculateBandwidthAvailableForImmediateFlip( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t cursor_bw_pre[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[]); + +static void CalculateImmediateFlipBandwithSupport( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + enum dml_immediate_flip_requirement ImmediateFlipRequirement[], + dml_float_t final_flip_bw[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_float_t cursor_bw_pre[], + dml_float_t prefetch_vmrow_bw[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[], + + // Output + dml_float_t *TotalBandwidth, + dml_float_t *TotalBandwidthNotIncludingMALLPrefetch, + dml_float_t *FractionOfUrgentBandwidth, + dml_bool_t *ImmediateFlipBandwidthSupport); + +// --------------------------- +// Declaration Ends +// --------------------------- + +static dml_uint_t dscceComputeDelay( + dml_uint_t bpc, + dml_float_t BPP, + dml_uint_t sliceWidth, + dml_uint_t numSlices, + enum dml_output_format_class pixelFormat, + enum dml_output_encoder_class Output) +{ + // valid bpc = source bits per component in the set of {8, 10, 12} + // valid bpp = increments of 1/16 of a bit + // min = 6/7/8 in N420/N422/444, respectively + // max = such that compression is 1:1 + //valid sliceWidth = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode) + //valid numSlices = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4} + //valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420} + + // fixed value + dml_uint_t rcModelSize = 8192; + + // N422/N420 operate at 2 pixels per clock + dml_uint_t pixelsPerClock, lstall, D, initalXmitDelay, w, s, ix, wx, p, l0, a, ax, L, + Delay, pixels; + + if (pixelFormat == dml_420) + pixelsPerClock = 2; + // #all other modes operate at 1 pixel per clock + else if (pixelFormat == dml_444) + pixelsPerClock = 1; + else if (pixelFormat == dml_n422) + pixelsPerClock = 2; + else + pixelsPerClock = 1; + + //initial transmit delay as per PPS + initalXmitDelay = (dml_uint_t)(dml_round(rcModelSize / 2.0 / BPP / pixelsPerClock, 1)); + + //compute ssm delay + if (bpc == 8) + D = 81; + else if (bpc == 10) + D = 89; + else + D = 113; + + //divide by pixel per cycle to compute slice width as seen by DSC + w = sliceWidth / pixelsPerClock; + + //422 mode has an additional cycle of delay + if (pixelFormat == dml_420 || pixelFormat == dml_444 || pixelFormat == dml_n422) + s = 0; + else + s = 1; + + //main calculation for the dscce + ix = initalXmitDelay + 45; + wx = (w + 2) / 3; + p = 3 * wx - w; + l0 = ix / w; + a = ix + p * l0; + ax = (a + 2) / 3 + D + 6 + 1; + L = (ax + wx - 1) / wx; + if ((ix % w) == 0 && p != 0) + lstall = 1; + else + lstall = 0; + Delay = L * wx * (numSlices - 1) + ax + s + lstall + 22; + + //dsc processes 3 pixel containers per cycle and a container can contain 1 or 2 pixels + pixels = Delay * 3 * pixelsPerClock; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: bpc: %u\n", __func__, bpc); + dml_print("DML::%s: BPP: %f\n", __func__, BPP); + dml_print("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); + dml_print("DML::%s: numSlices: %u\n", __func__, numSlices); + dml_print("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); + dml_print("DML::%s: Output: %u\n", __func__, Output); + dml_print("DML::%s: pixels: %u\n", __func__, pixels); +#endif + return pixels; +} + +static dml_uint_t dscComputeDelay(enum dml_output_format_class pixelFormat, enum dml_output_encoder_class Output) +{ + dml_uint_t Delay = 0; + + if (pixelFormat == dml_420) { + // sfr + Delay = Delay + 2; + // dsccif + Delay = Delay + 0; + // dscc - input deserializer + Delay = Delay + 3; + // dscc gets pixels every other cycle + Delay = Delay + 2; + // dscc - input cdc fifo + Delay = Delay + 12; + // dscc gets pixels every other cycle + Delay = Delay + 13; + // dscc - cdc uncertainty + Delay = Delay + 2; + // dscc - output cdc fifo + Delay = Delay + 7; + // dscc gets pixels every other cycle + Delay = Delay + 3; + // dscc - cdc uncertainty + Delay = Delay + 2; + // dscc - output serializer + Delay = Delay + 1; + // sft + Delay = Delay + 1; + } else if (pixelFormat == dml_n422) { + // sfr + Delay = Delay + 2; + // dsccif + Delay = Delay + 1; + // dscc - input deserializer + Delay = Delay + 5; + // dscc - input cdc fifo + Delay = Delay + 25; + // dscc - cdc uncertainty + Delay = Delay + 2; + // dscc - output cdc fifo + Delay = Delay + 10; + // dscc - cdc uncertainty + Delay = Delay + 2; + // dscc - output serializer + Delay = Delay + 1; + // sft + Delay = Delay + 1; + } else { + // sfr + Delay = Delay + 2; + // dsccif + Delay = Delay + 0; + // dscc - input deserializer + Delay = Delay + 3; + // dscc - input cdc fifo + Delay = Delay + 12; + // dscc - cdc uncertainty + Delay = Delay + 2; + // dscc - output cdc fifo + Delay = Delay + 7; + // dscc - output serializer + Delay = Delay + 1; + // dscc - cdc uncertainty + Delay = Delay + 2; + // sft + Delay = Delay + 1; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); + dml_print("DML::%s: Delay = %u\n", __func__, Delay); +#endif + + return Delay; +} + +static dml_bool_t CalculatePrefetchSchedule(struct display_mode_lib_scratch_st *scratch, + struct CalculatePrefetchSchedule_params_st *p) +{ + struct CalculatePrefetchSchedule_locals_st *s = &scratch->CalculatePrefetchSchedule_locals; + + s->MyError = false; + s->DPPCycles = 0; + s->DISPCLKCycles = 0; + s->DSTTotalPixelsAfterScaler = 0.0; + s->LineTime = 0.0; + s->dst_y_prefetch_equ = 0.0; + s->prefetch_bw_oto = 0.0; + s->Tvm_oto = 0.0; + s->Tr0_oto = 0.0; + s->Tvm_oto_lines = 0.0; + s->Tr0_oto_lines = 0.0; + s->dst_y_prefetch_oto = 0.0; + s->TimeForFetchingMetaPTE = 0.0; + s->TimeForFetchingRowInVBlank = 0.0; + s->LinesToRequestPrefetchPixelData = 0.0; + s->HostVMDynamicLevelsTrips = 0; + s->trip_to_mem = 0.0; + s->Tvm_trips = 0.0; + s->Tr0_trips = 0.0; + s->Tvm_trips_rounded = 0.0; + s->Tr0_trips_rounded = 0.0; + s->max_Tsw = 0.0; + s->Lsw_oto = 0.0; + s->Tpre_rounded = 0.0; + s->prefetch_bw_equ = 0.0; + s->Tvm_equ = 0.0; + s->Tr0_equ = 0.0; + s->Tdmbf = 0.0; + s->Tdmec = 0.0; + s->Tdmsks = 0.0; + s->prefetch_sw_bytes = 0.0; + s->prefetch_bw_pr = 0.0; + s->bytes_pp = 0.0; + s->dep_bytes = 0.0; + s->min_Lsw_oto = 0.0; + s->Tsw_est1 = 0.0; + s->Tsw_est3 = 0.0; + + if (p->GPUVMEnable == true && p->HostVMEnable == true) { + s->HostVMDynamicLevelsTrips = p->HostVMMaxNonCachedPageTableLevels; + } else { + s->HostVMDynamicLevelsTrips = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); + dml_print("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->GPUVMPageTableLevels); + dml_print("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); + dml_print("DML::%s: VStartup = %u\n", __func__, p->VStartup); + dml_print("DML::%s: MaxVStartup = %u\n", __func__, p->MaxVStartup); + dml_print("DML::%s: HostVMEnable = %u\n", __func__, p->HostVMEnable); + dml_print("DML::%s: HostVMInefficiencyFactor= %f\n", __func__, p->HostVMInefficiencyFactor); + dml_print("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); +#endif + CalculateVUpdateAndDynamicMetadataParameters( + p->MaxInterDCNTileRepeaters, + p->myPipe->Dppclk, + p->myPipe->Dispclk, + p->myPipe->DCFClkDeepSleep, + p->myPipe->PixelClock, + p->myPipe->HTotal, + p->myPipe->VBlank, + p->DynamicMetadataTransmittedBytes, + p->DynamicMetadataLinesBeforeActiveRequired, + p->myPipe->InterlaceEnable, + p->myPipe->ProgressiveToInterlaceUnitInOPP, + p->TSetup, + + // Output + &s->Tdmbf, + &s->Tdmec, + &s->Tdmsks, + p->VUpdateOffsetPix, + p->VUpdateWidthPix, + p->VReadyOffsetPix); + + s->LineTime = p->myPipe->HTotal / p->myPipe->PixelClock; + s->trip_to_mem = p->UrgentLatency; + s->Tvm_trips = p->UrgentExtraLatency + s->trip_to_mem * (p->GPUVMPageTableLevels * (s->HostVMDynamicLevelsTrips + 1) - 1); + + if (p->DynamicMetadataVMEnabled == true) { + *p->Tdmdl = p->TWait + s->Tvm_trips + s->trip_to_mem; + } else { + *p->Tdmdl = p->TWait + p->UrgentExtraLatency; + } + +#ifdef __DML_VBA_ALLOW_DELTA__ + if (DynamicMetadataEnable == false) { + *Tdmdl = 0.0; + } +#endif + + if (p->DynamicMetadataEnable == true) { + if (p->VStartup * s->LineTime < *p->TSetup + *p->Tdmdl + s->Tdmbf + s->Tdmec + s->Tdmsks) { + *p->NotEnoughTimeForDynamicMetadata = true; + dml_print("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); + dml_print("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + dml_print("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + dml_print("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + dml_print("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + } else { + *p->NotEnoughTimeForDynamicMetadata = false; + } + } else { + *p->NotEnoughTimeForDynamicMetadata = false; + } + + *p->Tdmdl_vm = (p->DynamicMetadataEnable == true && p->DynamicMetadataVMEnabled == true && p->GPUVMEnable == true ? p->TWait + s->Tvm_trips : 0); + + if (p->myPipe->ScalerEnabled) + s->DPPCycles = (dml_uint_t)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCL); + else + s->DPPCycles = (dml_uint_t)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCLLBOnly); + + s->DPPCycles = (dml_uint_t)(s->DPPCycles + p->myPipe->NumberOfCursors * p->DPPCLKDelayCNVCCursor); + + s->DISPCLKCycles = (dml_uint_t)p->DISPCLKDelaySubtotal; + + if (p->myPipe->Dppclk == 0.0 || p->myPipe->Dispclk == 0.0) + return true; + + *p->DSTXAfterScaler = (dml_uint_t) dml_round(s->DPPCycles * p->myPipe->PixelClock / p->myPipe->Dppclk + s->DISPCLKCycles * p->myPipe->PixelClock / p->myPipe->Dispclk + p->DSCDelay, 1.0); + *p->DSTXAfterScaler = (dml_uint_t) dml_round(*p->DSTXAfterScaler + (p->myPipe->ODMMode != dml_odm_mode_bypass ? 18 : 0) + (p->myPipe->DPPPerSurface - 1) * p->DPP_RECOUT_WIDTH + + ((p->myPipe->ODMMode == dml_odm_mode_split_1to2 || p->myPipe->ODMMode == dml_odm_mode_mso_1to2) ? (dml_float_t)p->myPipe->HActive / 2.0 : 0) + + ((p->myPipe->ODMMode == dml_odm_mode_mso_1to4) ? (dml_float_t)p->myPipe->HActive * 3.0 / 4.0 : 0), 1.0); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); + dml_print("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); + dml_print("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); + dml_print("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); + dml_print("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); + dml_print("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); + dml_print("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); + dml_print("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); + dml_print("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); +#endif + + if (p->OutputFormat == dml_420 || (p->myPipe->InterlaceEnable && p->myPipe->ProgressiveToInterlaceUnitInOPP)) + *p->DSTYAfterScaler = 1; + else + *p->DSTYAfterScaler = 0; + + s->DSTTotalPixelsAfterScaler = *p->DSTYAfterScaler * p->myPipe->HTotal + *p->DSTXAfterScaler; + *p->DSTYAfterScaler = (dml_uint_t)(dml_floor(s->DSTTotalPixelsAfterScaler / p->myPipe->HTotal, 1)); + *p->DSTXAfterScaler = (dml_uint_t)(s->DSTTotalPixelsAfterScaler - ((dml_float_t) (*p->DSTYAfterScaler * p->myPipe->HTotal))); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); + dml_print("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); +#endif + + s->MyError = false; + + s->Tr0_trips = s->trip_to_mem * (s->HostVMDynamicLevelsTrips + 1); + + if (p->GPUVMEnable == true) { + s->Tvm_trips_rounded = dml_ceil(4.0 * s->Tvm_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; + s->Tr0_trips_rounded = dml_ceil(4.0 * s->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; + if (p->GPUVMPageTableLevels >= 3) { + *p->Tno_bw = p->UrgentExtraLatency + s->trip_to_mem * (dml_float_t) ((p->GPUVMPageTableLevels - 2) * (s->HostVMDynamicLevelsTrips + 1) - 1); + } else if (p->GPUVMPageTableLevels == 1 && p->myPipe->DCCEnable != true) { + s->Tr0_trips_rounded = dml_ceil(4.0 * p->UrgentExtraLatency / s->LineTime, 1.0) / 4.0 * s->LineTime; + *p->Tno_bw = p->UrgentExtraLatency; + } else { + *p->Tno_bw = 0; + } + } else if (p->myPipe->DCCEnable == true) { + s->Tvm_trips_rounded = s->LineTime / 4.0; + s->Tr0_trips_rounded = dml_ceil(4.0 * s->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; + *p->Tno_bw = 0; + } else { + s->Tvm_trips_rounded = s->LineTime / 4.0; + s->Tr0_trips_rounded = s->LineTime / 2.0; + *p->Tno_bw = 0; + } + s->Tvm_trips_rounded = dml_max(s->Tvm_trips_rounded, s->LineTime / 4.0); + s->Tr0_trips_rounded = dml_max(s->Tr0_trips_rounded, s->LineTime / 4.0); + + if (p->myPipe->SourcePixelFormat == dml_420_8 || p->myPipe->SourcePixelFormat == dml_420_10 || p->myPipe->SourcePixelFormat == dml_420_12) { + s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC / 4; + } else { + s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC; + } + + s->prefetch_bw_pr = s->bytes_pp * p->myPipe->PixelClock / (dml_float_t)p->myPipe->DPPPerSurface; + if (p->myPipe->VRatio < 1.0) + s->prefetch_bw_pr = p->myPipe->VRatio * s->prefetch_bw_pr; + + s->max_Tsw = (dml_max(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) * s->LineTime); + + s->prefetch_sw_bytes = p->PrefetchSourceLinesY * p->swath_width_luma_ub * p->myPipe->BytePerPixelY + p->PrefetchSourceLinesC * p->swath_width_chroma_ub * p->myPipe->BytePerPixelC; + s->prefetch_bw_oto = dml_max(s->prefetch_bw_pr, s->prefetch_sw_bytes / s->max_Tsw); + + s->min_Lsw_oto = dml_max(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML_MAX_VRATIO_PRE_OTO__; + s->min_Lsw_oto = dml_max(s->min_Lsw_oto, 1.0); + s->Lsw_oto = dml_ceil(4.0 * dml_max(s->prefetch_sw_bytes / s->prefetch_bw_oto / s->LineTime, s->min_Lsw_oto), 1.0) / 4.0; + + if (p->GPUVMEnable == true) { + s->Tvm_oto = dml_max3( + s->Tvm_trips, + *p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->prefetch_bw_oto, + s->LineTime / 4.0); + } else + s->Tvm_oto = s->LineTime / 4.0; + + if ((p->GPUVMEnable == true || p->myPipe->DCCEnable == true)) { + s->Tr0_oto = dml_max4( + s->Tr0_trips, + (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_oto, + (s->LineTime - s->Tvm_oto)/2.0, + s->LineTime / 4.0); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Tr0_oto max0 = %f\n", __func__, (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_oto); + dml_print("DML::%s: Tr0_oto max1 = %f\n", __func__, s->Tr0_trips); + dml_print("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime - s->Tvm_oto); + dml_print("DML::%s: Tr0_oto max3 = %f\n", __func__, s->LineTime / 4); +#endif + } else + s->Tr0_oto = (s->LineTime - s->Tvm_oto) / 2.0; + + s->Tvm_oto_lines = dml_ceil(4.0 * s->Tvm_oto / s->LineTime, 1) / 4.0; + s->Tr0_oto_lines = dml_ceil(4.0 * s->Tr0_oto / s->LineTime, 1) / 4.0; + s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto; + + s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + dml_max(p->TWait + p->TCalc, *p->Tdmdl)) / s->LineTime - (*p->DSTYAfterScaler + (dml_float_t) *p->DSTXAfterScaler / (dml_float_t)p->myPipe->HTotal); + s->dst_y_prefetch_equ = dml_min(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); + dml_print("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); + dml_print("DML::%s: *Tno_bw = %f\n", __func__, *p->Tno_bw); + dml_print("DML::%s: UrgentExtraLatency = %f\n", __func__, p->UrgentExtraLatency); + dml_print("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); + dml_print("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + dml_print("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + dml_print("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); + dml_print("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + dml_print("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); + dml_print("DML::%s: prefetch_sw_bytes = %f\n", __func__, s->prefetch_sw_bytes); + dml_print("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); + dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, p->PDEAndMetaPTEBytesFrame); + dml_print("DML::%s: MetaRowByte = %u\n", __func__, p->MetaRowByte); + dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + dml_print("DML::%s: Tvm_trips = %f\n", __func__, s->Tvm_trips); + dml_print("DML::%s: Tr0_trips = %f\n", __func__, s->Tr0_trips); + dml_print("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); + dml_print("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); + dml_print("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); + dml_print("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); + dml_print("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); + dml_print("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); + dml_print("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); + dml_print("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); +#endif + + s->dst_y_prefetch_equ = dml_floor(4.0 * (s->dst_y_prefetch_equ + 0.125), 1) / 4.0; + s->Tpre_rounded = s->dst_y_prefetch_equ * s->LineTime; + + dml_print("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); + + dml_print("DML::%s: LineTime: %f\n", __func__, s->LineTime); + dml_print("DML::%s: VStartup: %u\n", __func__, p->VStartup); + dml_print("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); + dml_print("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); + dml_print("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); + dml_print("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); + dml_print("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + dml_print("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + dml_print("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + dml_print("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); + dml_print("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + dml_print("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); + dml_print("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); + + s->dep_bytes = dml_max(p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor, p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor); + + if (s->prefetch_sw_bytes < s->dep_bytes) { + s->prefetch_sw_bytes = 2 * s->dep_bytes; + } + + *p->DestinationLinesToRequestVMInVBlank = 0; + *p->DestinationLinesToRequestRowInVBlank = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixDataBWLuma = 0; + if (s->dst_y_prefetch_equ > 1) { + + if (s->Tpre_rounded - *p->Tno_bw > 0) { + s->PrefetchBandwidth1 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + 2 * p->MetaRowByte + + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + + s->prefetch_sw_bytes) + / (s->Tpre_rounded - *p->Tno_bw); + s->Tsw_est1 = s->prefetch_sw_bytes / s->PrefetchBandwidth1; + } else + s->PrefetchBandwidth1 = 0; + + if (p->VStartup == p->MaxVStartup && (s->Tsw_est1 / s->LineTime < s->min_Lsw_oto) && s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw > 0) { + s->PrefetchBandwidth1 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + 2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / + (s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw); + } + + if (s->Tpre_rounded - *p->Tno_bw - 2 * s->Tr0_trips_rounded > 0) + s->PrefetchBandwidth2 = (p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor + s->prefetch_sw_bytes) / + (s->Tpre_rounded - *p->Tno_bw - 2 * s->Tr0_trips_rounded); + else + s->PrefetchBandwidth2 = 0; + + if (s->Tpre_rounded - s->Tvm_trips_rounded > 0) { + s->PrefetchBandwidth3 = (2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + s->prefetch_sw_bytes) / + (s->Tpre_rounded - s->Tvm_trips_rounded); + s->Tsw_est3 = s->prefetch_sw_bytes / s->PrefetchBandwidth3; + } + else + s->PrefetchBandwidth3 = 0; + + + if (p->VStartup == p->MaxVStartup && (s->Tsw_est3 / s->LineTime < s->min_Lsw_oto) && s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded > 0) { + s->PrefetchBandwidth3 = (2 * p->MetaRowByte + 2 * p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / (s->Tpre_rounded - s->min_Lsw_oto * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded); + } + + if (s->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded > 0) + s->PrefetchBandwidth4 = s->prefetch_sw_bytes / (s->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded); + else + s->PrefetchBandwidth4 = 0; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Tpre_rounded: %f\n", __func__, s->Tpre_rounded); + dml_print("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); + dml_print("DML::%s: Tvm_trips_rounded: %f\n", __func__, s->Tvm_trips_rounded); + dml_print("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); + dml_print("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); + dml_print("DML::%s: PrefetchBandwidth1: %f\n", __func__, s->PrefetchBandwidth1); + dml_print("DML::%s: PrefetchBandwidth2: %f\n", __func__, s->PrefetchBandwidth2); + dml_print("DML::%s: PrefetchBandwidth3: %f\n", __func__, s->PrefetchBandwidth3); + dml_print("DML::%s: PrefetchBandwidth4: %f\n", __func__, s->PrefetchBandwidth4); +#endif + { + dml_bool_t Case1OK; + dml_bool_t Case2OK; + dml_bool_t Case3OK; + + if (s->PrefetchBandwidth1 > 0) { + if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth1 >= s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth1 >= s->Tr0_trips_rounded) { + Case1OK = true; + } else { + Case1OK = false; + } + } else { + Case1OK = false; + } + + if (s->PrefetchBandwidth2 > 0) { + if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth2 >= s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth2 < s->Tr0_trips_rounded) { + Case2OK = true; + } else { + Case2OK = false; + } + } else { + Case2OK = false; + } + + if (s->PrefetchBandwidth3 > 0) { + if (*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->PrefetchBandwidth3 < s->Tvm_trips_rounded && (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->PrefetchBandwidth3 >= s->Tr0_trips_rounded) { + Case3OK = true; + } else { + Case3OK = false; + } + } else { + Case3OK = false; + } + + if (Case1OK) { + s->prefetch_bw_equ = s->PrefetchBandwidth1; + } else if (Case2OK) { + s->prefetch_bw_equ = s->PrefetchBandwidth2; + } else if (Case3OK) { + s->prefetch_bw_equ = s->PrefetchBandwidth3; + } else { + s->prefetch_bw_equ = s->PrefetchBandwidth4; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Case1OK: %u\n", __func__, Case1OK); + dml_print("DML::%s: Case2OK: %u\n", __func__, Case2OK); + dml_print("DML::%s: Case3OK: %u\n", __func__, Case3OK); + dml_print("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); +#endif + + if (s->prefetch_bw_equ > 0) { + if (p->GPUVMEnable == true) { + s->Tvm_equ = dml_max3(*p->Tno_bw + p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / s->prefetch_bw_equ, s->Tvm_trips, s->LineTime / 4); + } else { + s->Tvm_equ = s->LineTime / 4; + } + + if ((p->GPUVMEnable == true || p->myPipe->DCCEnable == true)) { + s->Tr0_equ = dml_max4((p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / s->prefetch_bw_equ, s->Tr0_trips, (s->LineTime - s->Tvm_equ) / 2, s->LineTime / 4); + } else { + s->Tr0_equ = (s->LineTime - s->Tvm_equ) / 2; + } + } else { + s->Tvm_equ = 0; + s->Tr0_equ = 0; + dml_print("DML::%s: prefetch_bw_equ equals 0!\n", __func__); + } + } + + + if (s->dst_y_prefetch_oto < s->dst_y_prefetch_equ) { + *p->DestinationLinesForPrefetch = s->dst_y_prefetch_oto; + s->TimeForFetchingMetaPTE = s->Tvm_oto; + s->TimeForFetchingRowInVBlank = s->Tr0_oto; + + *p->DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; + *p->DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + } else { + *p->DestinationLinesForPrefetch = s->dst_y_prefetch_equ; + s->TimeForFetchingMetaPTE = s->Tvm_equ; + s->TimeForFetchingRowInVBlank = s->Tr0_equ; + + if (p->VStartup == p->MaxVStartup && p->EnhancedPrefetchScheduleAccelerationFinal != 0) { + *p->DestinationLinesToRequestVMInVBlank = dml_floor(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; + *p->DestinationLinesToRequestRowInVBlank = dml_floor(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + } else { + *p->DestinationLinesToRequestVMInVBlank = dml_ceil(4.0 * s->TimeForFetchingMetaPTE / s->LineTime, 1.0) / 4.0; + *p->DestinationLinesToRequestRowInVBlank = dml_ceil(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + } + } + + s->LinesToRequestPrefetchPixelData = *p->DestinationLinesForPrefetch - *p->DestinationLinesToRequestVMInVBlank - 2 * *p->DestinationLinesToRequestRowInVBlank; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DestinationLinesForPrefetch = %f\n", __func__, *p->DestinationLinesForPrefetch); + dml_print("DML::%s: DestinationLinesToRequestVMInVBlank = %f\n", __func__, *p->DestinationLinesToRequestVMInVBlank); + dml_print("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); + dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); + dml_print("DML::%s: DestinationLinesToRequestRowInVBlank = %f\n", __func__, *p->DestinationLinesToRequestRowInVBlank); + dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + dml_print("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); +#endif + + if (s->LinesToRequestPrefetchPixelData >= 1 && s->prefetch_bw_equ > 0) { + *p->VRatioPrefetchY = (dml_float_t)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData; + *p->VRatioPrefetchY = dml_max(*p->VRatioPrefetchY, 1.0); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + dml_print("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); + dml_print("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); +#endif + if ((p->SwathHeightY > 4) && (p->VInitPreFillY > 3)) { + if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillY - 3.0) / 2.0) { + *p->VRatioPrefetchY = dml_max(*p->VRatioPrefetchY, + (dml_float_t)p->MaxNumSwathY * p->SwathHeightY / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillY - 3.0) / 2.0)); + } else { + s->MyError = true; + dml_print("DML::%s: MyErr set. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); + *p->VRatioPrefetchY = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + dml_print("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + dml_print("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); +#endif + } + + *p->VRatioPrefetchC = (dml_float_t)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData; + *p->VRatioPrefetchC = dml_max(*p->VRatioPrefetchC, 1.0); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + dml_print("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); + dml_print("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); +#endif + if ((p->SwathHeightC > 4) && (p->VInitPreFillC > 3)) { + if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillC - 3.0) / 2.0) { + *p->VRatioPrefetchC = dml_max(*p->VRatioPrefetchC, (dml_float_t)p->MaxNumSwathC * p->SwathHeightC / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillC - 3.0) / 2.0)); + } else { + s->MyError = true; + dml_print("DML::%s: MyErr set. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); + *p->VRatioPrefetchC = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + dml_print("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + dml_print("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); +#endif + } + + *p->RequiredPrefetchPixDataBWLuma = (dml_float_t)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData + * p->myPipe->BytePerPixelY + * p->swath_width_luma_ub / s->LineTime; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + dml_print("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); + dml_print("DML::%s: RequiredPrefetchPixDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixDataBWLuma); +#endif + *p->RequiredPrefetchPixDataBWChroma = (dml_float_t)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData + *p->myPipe->BytePerPixelC + *p->swath_width_chroma_ub / s->LineTime; + } else { + s->MyError = true; + dml_print("DML:%s: MyErr set. LinesToRequestPrefetchPixelData: %f, should be > 0\n", __func__, s->LinesToRequestPrefetchPixelData); + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixDataBWLuma = 0; + *p->RequiredPrefetchPixDataBWChroma = 0; + } + + dml_print("DML: Tpre: %fus - sum of time to request meta pte, 2 x data pte + meta data, swaths\n", (dml_float_t)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingMetaPTE); + dml_print("DML: Tvm: %fus - time to fetch page tables for meta surface\n", s->TimeForFetchingMetaPTE); + dml_print("DML: Tr0: %fus - time to fetch first row of data pagetables and first row of meta data (done in parallel)\n", s->TimeForFetchingRowInVBlank); + dml_print("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (dml_float_t)s->LinesToRequestPrefetchPixelData * s->LineTime); + dml_print("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((dml_float_t) (*p->DSTXAfterScaler) / (dml_float_t)p->myPipe->HTotal)) * s->LineTime); + dml_print("DML: Tvstartup - TSetup - Tcalc - Twait - Tpre - To > 0\n"); + dml_print("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingMetaPTE - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((dml_float_t) (*p->DSTXAfterScaler) / (dml_float_t)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); + dml_print("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); + + } else { + s->MyError = true; + dml_print("DML::%s: MyErr set, dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); + s->TimeForFetchingMetaPTE = 0; + s->TimeForFetchingRowInVBlank = 0; + *p->DestinationLinesToRequestVMInVBlank = 0; + *p->DestinationLinesToRequestRowInVBlank = 0; + s->LinesToRequestPrefetchPixelData = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixDataBWLuma = 0; + *p->RequiredPrefetchPixDataBWChroma = 0; + } + + { + dml_float_t prefetch_vm_bw; + dml_float_t prefetch_row_bw; + + if (p->PDEAndMetaPTEBytesFrame == 0) { + prefetch_vm_bw = 0; + } else if (*p->DestinationLinesToRequestVMInVBlank > 0) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, p->PDEAndMetaPTEBytesFrame); + dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + dml_print("DML::%s: DestinationLinesToRequestVMInVBlank = %f\n", __func__, *p->DestinationLinesToRequestVMInVBlank); + dml_print("DML::%s: LineTime = %f\n", __func__, s->LineTime); +#endif + prefetch_vm_bw = p->PDEAndMetaPTEBytesFrame * p->HostVMInefficiencyFactor / (*p->DestinationLinesToRequestVMInVBlank * s->LineTime); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); +#endif + } else { + prefetch_vm_bw = 0; + s->MyError = true; + dml_print("DML::%s: MyErr set. DestinationLinesToRequestVMInVBlank=%f (should be > 0)\n", __func__, *p->DestinationLinesToRequestVMInVBlank); + } + + if (p->MetaRowByte + p->PixelPTEBytesPerRow == 0) { + prefetch_row_bw = 0; + } else if (*p->DestinationLinesToRequestRowInVBlank > 0) { + prefetch_row_bw = (p->MetaRowByte + p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor) / (*p->DestinationLinesToRequestRowInVBlank * s->LineTime); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: MetaRowByte = %u\n", __func__, p->MetaRowByte); + dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + dml_print("DML::%s: DestinationLinesToRequestRowInVBlank = %f\n", __func__, *p->DestinationLinesToRequestRowInVBlank); + dml_print("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); +#endif + } else { + prefetch_row_bw = 0; + s->MyError = true; + dml_print("DML::%s: MyErr set. DestinationLinesToRequestRowInVBlank=%f (should be > 0)\n", __func__, *p->DestinationLinesToRequestRowInVBlank); + } + + *p->prefetch_vmrow_bw = dml_max(prefetch_vm_bw, prefetch_row_bw); + } + + if (s->MyError) { + s->TimeForFetchingMetaPTE = 0; + s->TimeForFetchingRowInVBlank = 0; + *p->DestinationLinesToRequestVMInVBlank = 0; + *p->DestinationLinesToRequestRowInVBlank = 0; + *p->DestinationLinesForPrefetch = 0; + s->LinesToRequestPrefetchPixelData = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixDataBWLuma = 0; + *p->RequiredPrefetchPixDataBWChroma = 0; + } + + return s->MyError; +} // CalculatePrefetchSchedule + +static void CalculateBytePerPixelAndBlockSizes( + enum dml_source_format_class SourcePixelFormat, + enum dml_swizzle_mode SurfaceTiling, + + // Output + dml_uint_t *BytePerPixelY, + dml_uint_t *BytePerPixelC, + dml_float_t *BytePerPixelDETY, + dml_float_t *BytePerPixelDETC, + dml_uint_t *BlockHeight256BytesY, + dml_uint_t *BlockHeight256BytesC, + dml_uint_t *BlockWidth256BytesY, + dml_uint_t *BlockWidth256BytesC, + dml_uint_t *MacroTileHeightY, + dml_uint_t *MacroTileHeightC, + dml_uint_t *MacroTileWidthY, + dml_uint_t *MacroTileWidthC) +{ + if (SourcePixelFormat == dml_444_64) { + *BytePerPixelDETY = 8; + *BytePerPixelDETC = 0; + *BytePerPixelY = 8; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml_444_32 || SourcePixelFormat == dml_rgbe) { + *BytePerPixelDETY = 4; + *BytePerPixelDETC = 0; + *BytePerPixelY = 4; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml_444_16 || SourcePixelFormat == dml_mono_16) { + *BytePerPixelDETY = 2; + *BytePerPixelDETC = 0; + *BytePerPixelY = 2; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml_444_8 || SourcePixelFormat == dml_mono_8) { + *BytePerPixelDETY = 1; + *BytePerPixelDETC = 0; + *BytePerPixelY = 1; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml_rgbe_alpha) { + *BytePerPixelDETY = 4; + *BytePerPixelDETC = 1; + *BytePerPixelY = 4; + *BytePerPixelC = 1; + } else if (SourcePixelFormat == dml_420_8) { + *BytePerPixelDETY = 1; + *BytePerPixelDETC = 2; + *BytePerPixelY = 1; + *BytePerPixelC = 2; + } else if (SourcePixelFormat == dml_420_12) { + *BytePerPixelDETY = 2; + *BytePerPixelDETC = 4; + *BytePerPixelY = 2; + *BytePerPixelC = 4; + } else { + *BytePerPixelDETY = (dml_float_t) (4.0 / 3); + *BytePerPixelDETC = (dml_float_t) (8.0 / 3); + *BytePerPixelY = 2; + *BytePerPixelC = 4; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); + dml_print("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); + dml_print("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); + dml_print("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); + dml_print("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); +#endif + if ((SourcePixelFormat == dml_444_64 || SourcePixelFormat == dml_444_32 + || SourcePixelFormat == dml_444_16 + || SourcePixelFormat == dml_444_8 + || SourcePixelFormat == dml_mono_16 + || SourcePixelFormat == dml_mono_8 + || SourcePixelFormat == dml_rgbe)) { + if (SurfaceTiling == dml_sw_linear) { + *BlockHeight256BytesY = 1; + } else if (SourcePixelFormat == dml_444_64) { + *BlockHeight256BytesY = 4; + } else if (SourcePixelFormat == dml_444_8) { + *BlockHeight256BytesY = 16; + } else { + *BlockHeight256BytesY = 8; + } + *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; + *BlockHeight256BytesC = 0; + *BlockWidth256BytesC = 0; + } else { + if (SurfaceTiling == dml_sw_linear) { + *BlockHeight256BytesY = 1; + *BlockHeight256BytesC = 1; + } else if (SourcePixelFormat == dml_rgbe_alpha) { + *BlockHeight256BytesY = 8; + *BlockHeight256BytesC = 16; + } else if (SourcePixelFormat == dml_420_8) { + *BlockHeight256BytesY = 16; + *BlockHeight256BytesC = 8; + } else { + *BlockHeight256BytesY = 8; + *BlockHeight256BytesC = 8; + } + *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; + *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); + dml_print("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); + dml_print("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); + dml_print("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); +#endif + + if (SurfaceTiling == dml_sw_linear) { + *MacroTileHeightY = *BlockHeight256BytesY; + *MacroTileWidthY = 256 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 256 / *BytePerPixelC / *MacroTileHeightC; + } + } else if (SurfaceTiling == dml_sw_64kb_d || SurfaceTiling == dml_sw_64kb_d_t || SurfaceTiling == dml_sw_64kb_d_x || SurfaceTiling == dml_sw_64kb_r_x) { + *MacroTileHeightY = 16 * *BlockHeight256BytesY; + *MacroTileWidthY = 65536 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = 16 * *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 65536 / *BytePerPixelC / *MacroTileHeightC; + } + } else { + *MacroTileHeightY = 32 * *BlockHeight256BytesY; + *MacroTileWidthY = 65536 * 4 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = 32 * *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 65536 * 4 / *BytePerPixelC / *MacroTileHeightC; + } + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); + dml_print("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); + dml_print("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); + dml_print("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); +#endif +} // CalculateBytePerPixelAndBlockSizes + +static noinline_for_stack dml_float_t CalculateTWait( + dml_uint_t PrefetchMode, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, + dml_bool_t SynchronizeDRRDisplaysForUCLKPStateChangeFinal, + dml_bool_t DRRDisplay, + dml_float_t DRAMClockChangeLatency, + dml_float_t FCLKChangeLatency, + dml_float_t UrgentLatency, + dml_float_t SREnterPlusExitTime) +{ + dml_float_t TWait = 0.0; + + if (PrefetchMode == 0 && + !(UseMALLForPStateChange == dml_use_mall_pstate_change_full_frame) && !(UseMALLForPStateChange == dml_use_mall_pstate_change_sub_viewport) && + !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe) && !(SynchronizeDRRDisplaysForUCLKPStateChangeFinal && DRRDisplay)) { + TWait = dml_max3(DRAMClockChangeLatency + UrgentLatency, SREnterPlusExitTime, UrgentLatency); + } else if (PrefetchMode <= 1 && !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe)) { + TWait = dml_max3(FCLKChangeLatency + UrgentLatency, SREnterPlusExitTime, UrgentLatency); + } else if (PrefetchMode <= 2 && !(UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe)) { + TWait = dml_max(SREnterPlusExitTime, UrgentLatency); + } else { + TWait = UrgentLatency; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: PrefetchMode = %u\n", __func__, PrefetchMode); + dml_print("DML::%s: TWait = %f\n", __func__, TWait); +#endif + return TWait; +} // CalculateTWait + + +/// @brief Calculate the "starting point" for prefetch calculation +/// if AllowForPStateChangeOrStutterInVBlank is set as a particular requirement, then the mode evalulation +/// will only be done at the given mode. If no specific requirement (i.e. *_if_possible), then will just go from +/// try all the prefetch mode in decreasing order of "difficulty" (start from 0 which means all power saving +/// features). +static void CalculatePrefetchMode( + enum dml_prefetch_modes AllowForPStateChangeOrStutterInVBlank, + dml_uint_t *MinPrefetchMode, + dml_uint_t *MaxPrefetchMode) +{ + if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_uclk_fclk_and_stutter_if_possible) { + *MinPrefetchMode = 0; // consider all pwr saving features + *MaxPrefetchMode = 3; // consider just urgent latency + } else { + if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_none) { + *MinPrefetchMode = 3; + } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_stutter) { + *MinPrefetchMode = 2; + } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_fclk_and_stutter) { + *MinPrefetchMode = 1; + } else if (AllowForPStateChangeOrStutterInVBlank == dml_prefetch_support_uclk_fclk_and_stutter) { + *MinPrefetchMode = 0; + } else { + dml_print("ERROR: Invalid AllowForPStateChangeOrStutterInVBlank setting! val=%u\n", AllowForPStateChangeOrStutterInVBlank); + ASSERT(0); + } + *MaxPrefetchMode = *MinPrefetchMode; + } +} // CalculatePrefetchMode + +static dml_float_t CalculateWriteBackDISPCLK( + enum dml_source_format_class WritebackPixelFormat, + dml_float_t PixelClock, + dml_float_t WritebackHRatio, + dml_float_t WritebackVRatio, + dml_uint_t WritebackHTaps, + dml_uint_t WritebackVTaps, + dml_uint_t WritebackSourceWidth, + dml_uint_t WritebackDestinationWidth, + dml_uint_t HTotal, + dml_uint_t WritebackLineBufferSize, + dml_float_t DISPCLKDPPCLKVCOSpeed) +{ + dml_float_t DISPCLK_H, DISPCLK_V, DISPCLK_HB; + + DISPCLK_H = PixelClock * dml_ceil(WritebackHTaps / 8.0, 1) / WritebackHRatio; + DISPCLK_V = PixelClock * (WritebackVTaps * dml_ceil(WritebackDestinationWidth / 6.0, 1) + 8.0) / (dml_float_t) HTotal; + DISPCLK_HB = PixelClock * WritebackVTaps * (WritebackDestinationWidth * WritebackVTaps - WritebackLineBufferSize / 57.0) / 6.0 / (dml_float_t) WritebackSourceWidth; + return RoundToDFSGranularity(dml_max3(DISPCLK_H, DISPCLK_V, DISPCLK_HB), 1, DISPCLKDPPCLKVCOSpeed); +} + +static dml_float_t CalculateWriteBackDelay( + enum dml_source_format_class WritebackPixelFormat, + dml_float_t WritebackHRatio, + dml_float_t WritebackVRatio, + dml_uint_t WritebackVTaps, + dml_uint_t WritebackDestinationWidth, + dml_uint_t WritebackDestinationHeight, + dml_uint_t WritebackSourceHeight, + dml_uint_t HTotal) +{ + dml_float_t CalculateWriteBackDelay; + dml_float_t Line_length; + dml_float_t Output_lines_last_notclamped; + dml_float_t WritebackVInit; + + WritebackVInit = (WritebackVRatio + WritebackVTaps + 1) / 2; + Line_length = dml_max((dml_float_t) WritebackDestinationWidth, dml_ceil((dml_float_t)WritebackDestinationWidth / 6.0, 1.0) * WritebackVTaps); + Output_lines_last_notclamped = WritebackDestinationHeight - 1 - dml_ceil(((dml_float_t)WritebackSourceHeight - (dml_float_t) WritebackVInit) / (dml_float_t)WritebackVRatio, 1.0); + if (Output_lines_last_notclamped < 0) { + CalculateWriteBackDelay = 0; + } else { + CalculateWriteBackDelay = Output_lines_last_notclamped * Line_length + (HTotal - WritebackDestinationWidth) + 80; + } + return CalculateWriteBackDelay; +} + +static void CalculateVUpdateAndDynamicMetadataParameters( + dml_uint_t MaxInterDCNTileRepeaters, + dml_float_t Dppclk, + dml_float_t Dispclk, + dml_float_t DCFClkDeepSleep, + dml_float_t PixelClock, + dml_uint_t HTotal, + dml_uint_t VBlank, + dml_uint_t DynamicMetadataTransmittedBytes, + dml_uint_t DynamicMetadataLinesBeforeActiveRequired, + dml_uint_t InterlaceEnable, + dml_bool_t ProgressiveToInterlaceUnitInOPP, + + // Output + dml_float_t *TSetup, + dml_float_t *Tdmbf, + dml_float_t *Tdmec, + dml_float_t *Tdmsks, + dml_uint_t *VUpdateOffsetPix, + dml_uint_t *VUpdateWidthPix, + dml_uint_t *VReadyOffsetPix) +{ + dml_float_t TotalRepeaterDelayTime; + TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2 / Dppclk + 3 / Dispclk); + *VUpdateWidthPix = (dml_uint_t)(dml_ceil((14.0 / DCFClkDeepSleep + 12.0 / Dppclk + TotalRepeaterDelayTime) * PixelClock, 1.0)); + *VReadyOffsetPix = (dml_uint_t)(dml_ceil(dml_max(150.0 / Dppclk, TotalRepeaterDelayTime + 20.0 / DCFClkDeepSleep + 10.0 / Dppclk) * PixelClock, 1.0)); + *VUpdateOffsetPix = (dml_uint_t)(dml_ceil(HTotal / 4.0, 1.0)); + *TSetup = (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock; + *Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / Dispclk; + *Tdmec = HTotal / PixelClock; + + if (DynamicMetadataLinesBeforeActiveRequired == 0) { + *Tdmsks = VBlank * HTotal / PixelClock / 2.0; + } else { + *Tdmsks = DynamicMetadataLinesBeforeActiveRequired * HTotal / PixelClock; + } + if (InterlaceEnable == 1 && ProgressiveToInterlaceUnitInOPP == false) { + *Tdmsks = *Tdmsks / 2; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); + dml_print("DML::%s: VBlank = %u\n", __func__, VBlank); + dml_print("DML::%s: HTotal = %u\n", __func__, HTotal); + dml_print("DML::%s: PixelClock = %f\n", __func__, PixelClock); + dml_print("DML::%s: Dppclk = %f\n", __func__, Dppclk); + dml_print("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); + dml_print("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); + dml_print("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); + + dml_print("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); + dml_print("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); + dml_print("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); + + dml_print("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); +#endif +} + +static void CalculateRowBandwidth( + dml_bool_t GPUVMEnable, + enum dml_source_format_class SourcePixelFormat, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_bool_t DCCEnable, + dml_float_t LineTime, + dml_uint_t MetaRowByteLuma, + dml_uint_t MetaRowByteChroma, + dml_uint_t meta_row_height_luma, + dml_uint_t meta_row_height_chroma, + dml_uint_t PixelPTEBytesPerRowLuma, + dml_uint_t PixelPTEBytesPerRowChroma, + dml_uint_t dpte_row_height_luma, + dml_uint_t dpte_row_height_chroma, + // Output + dml_float_t *meta_row_bw, + dml_float_t *dpte_row_bw) +{ + if (DCCEnable != true) { + *meta_row_bw = 0; + } else if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12 || SourcePixelFormat == dml_rgbe_alpha) { + *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime) + + VRatioChroma * MetaRowByteChroma + / (meta_row_height_chroma * LineTime); + } else { + *meta_row_bw = VRatio * MetaRowByteLuma / (meta_row_height_luma * LineTime); + } + + if (GPUVMEnable != true) { + *dpte_row_bw = 0; + } else if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12 || SourcePixelFormat == dml_rgbe_alpha) { + *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime) + + VRatioChroma * PixelPTEBytesPerRowChroma + / (dpte_row_height_chroma * LineTime); + } else { + *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime); + } +} + +/// @brief Determine immediate flip schedule given bw remaining after considering the prefetch schedule +/// @param BandwidthAvailableForImmediateFlip Bandwidth available for iflip for all planes +static void CalculateFlipSchedule( + dml_float_t HostVMInefficiencyFactor, + dml_float_t UrgentExtraLatency, + dml_float_t UrgentLatency, + dml_uint_t GPUVMMaxPageTableLevels, + dml_bool_t HostVMEnable, + dml_uint_t HostVMMaxNonCachedPageTableLevels, + dml_bool_t GPUVMEnable, + dml_uint_t HostVMMinPageSize, + dml_float_t PDEAndMetaPTEBytesPerFrame, + dml_float_t MetaRowBytes, + dml_float_t DPTEBytesPerRow, + dml_float_t BandwidthAvailableForImmediateFlip, + dml_uint_t TotImmediateFlipBytes, + enum dml_source_format_class SourcePixelFormat, + dml_float_t LineTime, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_float_t Tno_bw, + dml_bool_t DCCEnable, + dml_uint_t dpte_row_height, + dml_uint_t meta_row_height, + dml_uint_t dpte_row_height_chroma, + dml_uint_t meta_row_height_chroma, + dml_bool_t use_one_row_for_frame_flip, + + // Output + dml_float_t *DestinationLinesToRequestVMInImmediateFlip, + dml_float_t *DestinationLinesToRequestRowInImmediateFlip, + dml_float_t *final_flip_bw, + dml_bool_t *ImmediateFlipSupportedForPipe) +{ + dml_float_t min_row_time = 0.0; + dml_uint_t HostVMDynamicLevelsTrips = 0; + dml_float_t TimeForFetchingMetaPTEImmediateFlip = 0; + dml_float_t TimeForFetchingRowInVBlankImmediateFlip = 0; + dml_float_t ImmediateFlipBW = 0; // @brief The immediate flip bandwidth for this pipe + + if (GPUVMEnable == true && HostVMEnable == true) { + HostVMDynamicLevelsTrips = HostVMMaxNonCachedPageTableLevels; + } else { + HostVMDynamicLevelsTrips = 0; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); + dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); + dml_print("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); + dml_print("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); +#endif + + if (TotImmediateFlipBytes > 0) { + if (use_one_row_for_frame_flip) { + ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + 2.0 * DPTEBytesPerRow) * BandwidthAvailableForImmediateFlip / (dml_float_t) TotImmediateFlipBytes; + } else { + ImmediateFlipBW = (PDEAndMetaPTEBytesPerFrame + MetaRowBytes + DPTEBytesPerRow) * BandwidthAvailableForImmediateFlip / (dml_float_t) TotImmediateFlipBytes; + } + if (GPUVMEnable == true) { + TimeForFetchingMetaPTEImmediateFlip = dml_max3(Tno_bw + PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / ImmediateFlipBW, + UrgentExtraLatency + UrgentLatency * (GPUVMMaxPageTableLevels * (HostVMDynamicLevelsTrips + 1) - 1), + LineTime / 4.0); + } else { + TimeForFetchingMetaPTEImmediateFlip = 0; + } + if ((GPUVMEnable == true || DCCEnable == true)) { + TimeForFetchingRowInVBlankImmediateFlip = dml_max3((MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / ImmediateFlipBW, UrgentLatency * (HostVMDynamicLevelsTrips + 1), LineTime / 4.0); + } else { + TimeForFetchingRowInVBlankImmediateFlip = 0; + } + + *DestinationLinesToRequestVMInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingMetaPTEImmediateFlip / LineTime), 1.0) / 4.0; + *DestinationLinesToRequestRowInImmediateFlip = dml_ceil(4.0 * (TimeForFetchingRowInVBlankImmediateFlip / LineTime), 1.0) / 4.0; + + if (GPUVMEnable == true) { + *final_flip_bw = dml_max(PDEAndMetaPTEBytesPerFrame * HostVMInefficiencyFactor / (*DestinationLinesToRequestVMInImmediateFlip * LineTime), + (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime)); + } else if ((GPUVMEnable == true || DCCEnable == true)) { + *final_flip_bw = (MetaRowBytes + DPTEBytesPerRow * HostVMInefficiencyFactor) / (*DestinationLinesToRequestRowInImmediateFlip * LineTime); + } else { + *final_flip_bw = 0; + } + } else { + TimeForFetchingMetaPTEImmediateFlip = 0; + TimeForFetchingRowInVBlankImmediateFlip = 0; + *DestinationLinesToRequestVMInImmediateFlip = 0; + *DestinationLinesToRequestRowInImmediateFlip = 0; + *final_flip_bw = 0; + } + + if (SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_rgbe_alpha) { + if (GPUVMEnable == true && DCCEnable != true) { + min_row_time = dml_min(dpte_row_height * LineTime / VRatio, dpte_row_height_chroma * LineTime / VRatioChroma); + } else if (GPUVMEnable != true && DCCEnable == true) { + min_row_time = dml_min(meta_row_height * LineTime / VRatio, meta_row_height_chroma * LineTime / VRatioChroma); + } else { + min_row_time = dml_min4(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio, dpte_row_height_chroma * LineTime / VRatioChroma, meta_row_height_chroma * LineTime / VRatioChroma); + } + } else { + if (GPUVMEnable == true && DCCEnable != true) { + min_row_time = dpte_row_height * LineTime / VRatio; + } else if (GPUVMEnable != true && DCCEnable == true) { + min_row_time = meta_row_height * LineTime / VRatio; + } else { + min_row_time = dml_min(dpte_row_height * LineTime / VRatio, meta_row_height * LineTime / VRatio); + } + } + + if (*DestinationLinesToRequestVMInImmediateFlip >= 32 || *DestinationLinesToRequestRowInImmediateFlip >= 16 || TimeForFetchingMetaPTEImmediateFlip + 2 * TimeForFetchingRowInVBlankImmediateFlip > min_row_time) { + *ImmediateFlipSupportedForPipe = false; + } else { + *ImmediateFlipSupportedForPipe = true; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); + dml_print("DML::%s: DCCEnable = %u\n", __func__, DCCEnable); + + dml_print("DML::%s: MetaRowBytes = %f\n", __func__, MetaRowBytes); + dml_print("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); + dml_print("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); + dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); + dml_print("DML::%s: ImmediateFlipBW = %f\n", __func__, ImmediateFlipBW); + dml_print("DML::%s: PDEAndMetaPTEBytesPerFrame = %f\n", __func__, PDEAndMetaPTEBytesPerFrame); + dml_print("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); + dml_print("DML::%s: LineTime = %f\n", __func__, LineTime); + dml_print("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); + + dml_print("DML::%s: DestinationLinesToRequestVMInImmediateFlip = %f\n", __func__, *DestinationLinesToRequestVMInImmediateFlip); + dml_print("DML::%s: DestinationLinesToRequestRowInImmediateFlip = %f\n", __func__, *DestinationLinesToRequestRowInImmediateFlip); + dml_print("DML::%s: TimeForFetchingMetaPTEImmediateFlip = %f\n", __func__, TimeForFetchingMetaPTEImmediateFlip); + dml_print("DML::%s: TimeForFetchingRowInVBlankImmediateFlip = %f\n", __func__, TimeForFetchingRowInVBlankImmediateFlip); + dml_print("DML::%s: min_row_time = %f\n", __func__, min_row_time); + dml_print("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); +#endif +} // CalculateFlipSchedule + +static dml_float_t RoundToDFSGranularity(dml_float_t Clock, dml_bool_t round_up, dml_float_t VCOSpeed) +{ + if (Clock <= 0.0) + return 0.0; + else { + if (round_up) + return VCOSpeed * 4.0 / dml_floor(VCOSpeed * 4.0 / Clock, 1.0); + else + return VCOSpeed * 4.0 / dml_ceil(VCOSpeed * 4.0 / Clock, 1.0); + } +} + +static void CalculateDCCConfiguration( + dml_bool_t DCCEnabled, + dml_bool_t DCCProgrammingAssumesScanDirectionUnknown, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t SurfaceWidthLuma, + dml_uint_t SurfaceWidthChroma, + dml_uint_t SurfaceHeightLuma, + dml_uint_t SurfaceHeightChroma, + dml_uint_t nomDETInKByte, + dml_uint_t RequestHeight256ByteLuma, + dml_uint_t RequestHeight256ByteChroma, + enum dml_swizzle_mode TilingFormat, + dml_uint_t BytePerPixelY, + dml_uint_t BytePerPixelC, + dml_float_t BytePerPixelDETY, + dml_float_t BytePerPixelDETC, + enum dml_rotation_angle SourceScan, + // Output + dml_uint_t *MaxUncompressedBlockLuma, + dml_uint_t *MaxUncompressedBlockChroma, + dml_uint_t *MaxCompressedBlockLuma, + dml_uint_t *MaxCompressedBlockChroma, + dml_uint_t *IndependentBlockLuma, + dml_uint_t *IndependentBlockChroma) +{ + dml_uint_t DETBufferSizeForDCC = nomDETInKByte * 1024; + + dml_uint_t yuv420; + dml_uint_t horz_div_l; + dml_uint_t horz_div_c; + dml_uint_t vert_div_l; + dml_uint_t vert_div_c; + + dml_uint_t swath_buf_size; + dml_float_t detile_buf_vp_horz_limit; + dml_float_t detile_buf_vp_vert_limit; + + dml_uint_t MAS_vp_horz_limit; + dml_uint_t MAS_vp_vert_limit; + dml_uint_t max_vp_horz_width; + dml_uint_t max_vp_vert_height; + dml_uint_t eff_surf_width_l; + dml_uint_t eff_surf_width_c; + dml_uint_t eff_surf_height_l; + dml_uint_t eff_surf_height_c; + + dml_uint_t full_swath_bytes_horz_wc_l; + dml_uint_t full_swath_bytes_horz_wc_c; + dml_uint_t full_swath_bytes_vert_wc_l; + dml_uint_t full_swath_bytes_vert_wc_c; + + dml_uint_t req128_horz_wc_l; + dml_uint_t req128_horz_wc_c; + dml_uint_t req128_vert_wc_l; + dml_uint_t req128_vert_wc_c; + + dml_uint_t segment_order_horz_contiguous_luma; + dml_uint_t segment_order_horz_contiguous_chroma; + dml_uint_t segment_order_vert_contiguous_luma; + dml_uint_t segment_order_vert_contiguous_chroma; + + typedef enum{ + REQ_256Bytes, + REQ_128BytesNonContiguous, + REQ_128BytesContiguous, + REQ_NA + } RequestType; + + RequestType RequestLuma; + RequestType RequestChroma; + + yuv420 = ((SourcePixelFormat == dml_420_8 || SourcePixelFormat == dml_420_10 || SourcePixelFormat == dml_420_12) ? 1 : 0); + horz_div_l = 1; + horz_div_c = 1; + vert_div_l = 1; + vert_div_c = 1; + + if (BytePerPixelY == 1) + vert_div_l = 0; + if (BytePerPixelC == 1) + vert_div_c = 0; + + if (BytePerPixelC == 0) { + swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 256; + detile_buf_vp_horz_limit = (dml_float_t) swath_buf_size / ((dml_float_t) RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l)); + detile_buf_vp_vert_limit = (dml_float_t) swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l)); + } else { + swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 2 * 256; + detile_buf_vp_horz_limit = (dml_float_t) swath_buf_size / ((dml_float_t) RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l) + (dml_float_t) RequestHeight256ByteChroma * BytePerPixelC / (1 + horz_div_c) / (1 + yuv420)); + detile_buf_vp_vert_limit = (dml_float_t) swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l) + 256.0 / RequestHeight256ByteChroma / (1 + vert_div_c) / (1 + yuv420)); + } + + if (SourcePixelFormat == dml_420_10) { + detile_buf_vp_horz_limit = 1.5 * detile_buf_vp_horz_limit; + detile_buf_vp_vert_limit = 1.5 * detile_buf_vp_vert_limit; + } + + detile_buf_vp_horz_limit = dml_floor(detile_buf_vp_horz_limit - 1, 16); + detile_buf_vp_vert_limit = dml_floor(detile_buf_vp_vert_limit - 1, 16); + + MAS_vp_horz_limit = SourcePixelFormat == dml_rgbe_alpha ? 3840 : 6144; + MAS_vp_vert_limit = SourcePixelFormat == dml_rgbe_alpha ? 3840 : (BytePerPixelY == 8 ? 3072 : 6144); + max_vp_horz_width = (dml_uint_t)(dml_min((dml_float_t) MAS_vp_horz_limit, detile_buf_vp_horz_limit)); + max_vp_vert_height = (dml_uint_t)(dml_min((dml_float_t) MAS_vp_vert_limit, detile_buf_vp_vert_limit)); + eff_surf_width_l = (SurfaceWidthLuma > max_vp_horz_width ? max_vp_horz_width : SurfaceWidthLuma); + eff_surf_width_c = eff_surf_width_l / (1 + yuv420); + eff_surf_height_l = (SurfaceHeightLuma > max_vp_vert_height ? max_vp_vert_height : SurfaceHeightLuma); + eff_surf_height_c = eff_surf_height_l / (1 + yuv420); + + full_swath_bytes_horz_wc_l = eff_surf_width_l * RequestHeight256ByteLuma * BytePerPixelY; + full_swath_bytes_vert_wc_l = eff_surf_height_l * 256 / RequestHeight256ByteLuma; + if (BytePerPixelC > 0) { + full_swath_bytes_horz_wc_c = eff_surf_width_c * RequestHeight256ByteChroma * BytePerPixelC; + full_swath_bytes_vert_wc_c = eff_surf_height_c * 256 / RequestHeight256ByteChroma; + } else { + full_swath_bytes_horz_wc_c = 0; + full_swath_bytes_vert_wc_c = 0; + } + + if (SourcePixelFormat == dml_420_10) { + full_swath_bytes_horz_wc_l = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_horz_wc_l * 2.0 / 3.0, 256.0)); + full_swath_bytes_horz_wc_c = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_horz_wc_c * 2.0 / 3.0, 256.0)); + full_swath_bytes_vert_wc_l = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_vert_wc_l * 2.0 / 3.0, 256.0)); + full_swath_bytes_vert_wc_c = (dml_uint_t)(dml_ceil((dml_float_t) full_swath_bytes_vert_wc_c * 2.0 / 3.0, 256.0)); + } + + if (2 * full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 0; + req128_horz_wc_c = 0; + } else if (full_swath_bytes_horz_wc_l < 1.5 * full_swath_bytes_horz_wc_c && 2 * full_swath_bytes_horz_wc_l + full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 0; + req128_horz_wc_c = 1; + } else if (full_swath_bytes_horz_wc_l >= 1.5 * full_swath_bytes_horz_wc_c && full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 1; + req128_horz_wc_c = 0; + } else { + req128_horz_wc_l = 1; + req128_horz_wc_c = 1; + } + + if (2 * full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 0; + req128_vert_wc_c = 0; + } else if (full_swath_bytes_vert_wc_l < 1.5 * full_swath_bytes_vert_wc_c && 2 * full_swath_bytes_vert_wc_l + full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 0; + req128_vert_wc_c = 1; + } else if (full_swath_bytes_vert_wc_l >= 1.5 * full_swath_bytes_vert_wc_c && full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 1; + req128_vert_wc_c = 0; + } else { + req128_vert_wc_l = 1; + req128_vert_wc_c = 1; + } + + if (BytePerPixelY == 2) { + segment_order_horz_contiguous_luma = 0; + segment_order_vert_contiguous_luma = 1; + } else { + segment_order_horz_contiguous_luma = 1; + segment_order_vert_contiguous_luma = 0; + } + + if (BytePerPixelC == 2) { + segment_order_horz_contiguous_chroma = 0; + segment_order_vert_contiguous_chroma = 1; + } else { + segment_order_horz_contiguous_chroma = 1; + segment_order_vert_contiguous_chroma = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); + dml_print("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + dml_print("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); + dml_print("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); + dml_print("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); + dml_print("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); + dml_print("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); + dml_print("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); + dml_print("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); +#endif + + if (DCCProgrammingAssumesScanDirectionUnknown == true) { + if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { + RequestLuma = REQ_256Bytes; + } else if ((req128_horz_wc_l == 1 && segment_order_horz_contiguous_luma == 0) || (req128_vert_wc_l == 1 && segment_order_vert_contiguous_luma == 0)) { + RequestLuma = REQ_128BytesNonContiguous; + } else { + RequestLuma = REQ_128BytesContiguous; + } + if (req128_horz_wc_c == 0 && req128_vert_wc_c == 0) { + RequestChroma = REQ_256Bytes; + } else if ((req128_horz_wc_c == 1 && segment_order_horz_contiguous_chroma == 0) || (req128_vert_wc_c == 1 && segment_order_vert_contiguous_chroma == 0)) { + RequestChroma = REQ_128BytesNonContiguous; + } else { + RequestChroma = REQ_128BytesContiguous; + } + } else if (!dml_is_vertical_rotation(SourceScan)) { + if (req128_horz_wc_l == 0) { + RequestLuma = REQ_256Bytes; + } else if (segment_order_horz_contiguous_luma == 0) { + RequestLuma = REQ_128BytesNonContiguous; + } else { + RequestLuma = REQ_128BytesContiguous; + } + if (req128_horz_wc_c == 0) { + RequestChroma = REQ_256Bytes; + } else if (segment_order_horz_contiguous_chroma == 0) { + RequestChroma = REQ_128BytesNonContiguous; + } else { + RequestChroma = REQ_128BytesContiguous; + } + } else { + if (req128_vert_wc_l == 0) { + RequestLuma = REQ_256Bytes; + } else if (segment_order_vert_contiguous_luma == 0) { + RequestLuma = REQ_128BytesNonContiguous; + } else { + RequestLuma = REQ_128BytesContiguous; + } + if (req128_vert_wc_c == 0) { + RequestChroma = REQ_256Bytes; + } else if (segment_order_vert_contiguous_chroma == 0) { + RequestChroma = REQ_128BytesNonContiguous; + } else { + RequestChroma = REQ_128BytesContiguous; + } + } + + if (RequestLuma == REQ_256Bytes) { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 256; + *IndependentBlockLuma = 0; + } else if (RequestLuma == REQ_128BytesContiguous) { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 128; + *IndependentBlockLuma = 128; + } else { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 64; + *IndependentBlockLuma = 64; + } + + if (RequestChroma == REQ_256Bytes) { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 256; + *IndependentBlockChroma = 0; + } else if (RequestChroma == REQ_128BytesContiguous) { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 128; + *IndependentBlockChroma = 128; + } else { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 64; + *IndependentBlockChroma = 64; + } + + if (DCCEnabled != true || BytePerPixelC == 0) { + *MaxUncompressedBlockChroma = 0; + *MaxCompressedBlockChroma = 0; + *IndependentBlockChroma = 0; + } + + if (DCCEnabled != true) { + *MaxUncompressedBlockLuma = 0; + *MaxCompressedBlockLuma = 0; + *IndependentBlockLuma = 0; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); + dml_print("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); + dml_print("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); + dml_print("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); + dml_print("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); + dml_print("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); +#endif + +} // CalculateDCCConfiguration + +static dml_uint_t CalculatePrefetchSourceLines( + dml_float_t VRatio, + dml_uint_t VTaps, + dml_bool_t Interlace, + dml_bool_t ProgressiveToInterlaceUnitInOPP, + dml_uint_t SwathHeight, + enum dml_rotation_angle SourceScan, + dml_bool_t ViewportStationary, + dml_uint_t SwathWidth, + dml_uint_t ViewportHeight, + dml_uint_t ViewportXStart, + dml_uint_t ViewportYStart, + + // Output + dml_uint_t *VInitPreFill, + dml_uint_t *MaxNumSwath) +{ + + dml_uint_t vp_start_rot = 0; + dml_uint_t sw0_tmp = 0; + dml_uint_t MaxPartialSwath = 0; + dml_float_t numLines = 0; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VRatio = %f\n", __func__, VRatio); + dml_print("DML::%s: VTaps = %u\n", __func__, VTaps); + dml_print("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); + dml_print("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); + dml_print("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); + dml_print("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); +#endif + if (ProgressiveToInterlaceUnitInOPP) + *VInitPreFill = (dml_uint_t)(dml_floor((VRatio + (dml_float_t) VTaps + 1) / 2.0, 1)); + else + *VInitPreFill = (dml_uint_t)(dml_floor((VRatio + (dml_float_t) VTaps + 1 + Interlace * 0.5 * VRatio) / 2.0, 1)); + + if (ViewportStationary) { + if (SourceScan == dml_rotation_180 || SourceScan == dml_rotation_180m) { + vp_start_rot = SwathHeight - (((dml_uint_t) (ViewportYStart + ViewportHeight - 1) % SwathHeight) + 1); + } else if (SourceScan == dml_rotation_270 || SourceScan == dml_rotation_90m) { + vp_start_rot = ViewportXStart; + } else if (SourceScan == dml_rotation_90 || SourceScan == dml_rotation_270m) { + vp_start_rot = SwathHeight - (((dml_uint_t)(ViewportYStart + SwathWidth - 1) % SwathHeight) + 1); + } else { + vp_start_rot = ViewportYStart; + } + sw0_tmp = SwathHeight - (vp_start_rot % SwathHeight); + if (sw0_tmp < *VInitPreFill) { + *MaxNumSwath = (dml_uint_t)(dml_ceil((*VInitPreFill - sw0_tmp) / (dml_float_t) SwathHeight, 1) + 1); + } else { + *MaxNumSwath = 1; + } + MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (vp_start_rot + *VInitPreFill - 1) % SwathHeight)); + } else { + *MaxNumSwath = (dml_uint_t)(dml_ceil((*VInitPreFill - 1.0) / (dml_float_t) SwathHeight, 1) + 1); + if (*VInitPreFill > 1) { + MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (*VInitPreFill - 2) % SwathHeight)); + } else { + MaxPartialSwath = (dml_uint_t)(dml_max(1, (dml_uint_t) (*VInitPreFill + SwathHeight - 2) % SwathHeight)); + } + } + numLines = *MaxNumSwath * SwathHeight + MaxPartialSwath; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); + dml_print("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); + dml_print("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); + dml_print("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); + dml_print("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); +#endif + return (dml_uint_t)(numLines); + +} // CalculatePrefetchSourceLines + +static dml_uint_t CalculateVMAndRowBytes( + dml_bool_t ViewportStationary, + dml_bool_t DCCEnable, + dml_uint_t NumberOfDPPs, + dml_uint_t BlockHeight256Bytes, + dml_uint_t BlockWidth256Bytes, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t SurfaceTiling, + dml_uint_t BytePerPixel, + enum dml_rotation_angle SourceScan, + dml_uint_t SwathWidth, + dml_uint_t ViewportHeight, + dml_uint_t ViewportXStart, + dml_uint_t ViewportYStart, + dml_bool_t GPUVMEnable, + dml_uint_t GPUVMMaxPageTableLevels, + dml_uint_t GPUVMMinPageSizeKBytes, + dml_uint_t PTEBufferSizeInRequests, + dml_uint_t Pitch, + dml_uint_t DCCMetaPitch, + dml_uint_t MacroTileWidth, + dml_uint_t MacroTileHeight, + + // Output + dml_uint_t *MetaRowByte, + dml_uint_t *PixelPTEBytesPerRow, // for bandwidth calculation + dml_uint_t *PixelPTEBytesPerRowStorage, // for PTE buffer size check + dml_uint_t *dpte_row_width_ub, + dml_uint_t *dpte_row_height, + dml_uint_t *dpte_row_height_linear, + dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, + dml_uint_t *dpte_row_width_ub_one_row_per_frame, + dml_uint_t *dpte_row_height_one_row_per_frame, + dml_uint_t *MetaRequestWidth, + dml_uint_t *MetaRequestHeight, + dml_uint_t *meta_row_width, + dml_uint_t *meta_row_height, + dml_uint_t *PixelPTEReqWidth, + dml_uint_t *PixelPTEReqHeight, + dml_uint_t *PTERequestSize, + dml_uint_t *DPDE0BytesFrame, + dml_uint_t *MetaPTEBytesFrame) +{ + dml_uint_t MPDEBytesFrame; + dml_uint_t DCCMetaSurfaceBytes; + dml_uint_t ExtraDPDEBytesFrame; + dml_uint_t PDEAndMetaPTEBytesFrame; + dml_uint_t MacroTileSizeBytes; + dml_uint_t vp_height_meta_ub; + dml_uint_t vp_height_dpte_ub; + + dml_uint_t PixelPTEReqWidth_linear = 0; // VBA_DELTA. VBA doesn't calculate this + + *MetaRequestHeight = 8 * BlockHeight256Bytes; + *MetaRequestWidth = 8 * BlockWidth256Bytes; + if (SurfaceTiling == dml_sw_linear) { + *meta_row_height = 32; + *meta_row_width = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *MetaRequestWidth - 1, *MetaRequestWidth) - dml_floor(ViewportXStart, *MetaRequestWidth)); + } else if (!dml_is_vertical_rotation(SourceScan)) { + *meta_row_height = *MetaRequestHeight; + if (ViewportStationary && NumberOfDPPs == 1) { + *meta_row_width = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *MetaRequestWidth - 1, *MetaRequestWidth) - dml_floor(ViewportXStart, *MetaRequestWidth)); + } else { + *meta_row_width = (dml_uint_t)(dml_ceil(SwathWidth - 1, *MetaRequestWidth) + *MetaRequestWidth); + } + *MetaRowByte = (dml_uint_t)(*meta_row_width * *MetaRequestHeight * BytePerPixel / 256.0); + } else { + *meta_row_height = *MetaRequestWidth; + if (ViewportStationary && NumberOfDPPs == 1) { + *meta_row_width = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + *MetaRequestHeight - 1, *MetaRequestHeight) - dml_floor(ViewportYStart, *MetaRequestHeight)); + } else { + *meta_row_width = (dml_uint_t)(dml_ceil(SwathWidth - 1, *MetaRequestHeight) + *MetaRequestHeight); + } + *MetaRowByte = (dml_uint_t)(*meta_row_width * *MetaRequestWidth * BytePerPixel / 256.0); + } + + if (ViewportStationary && (NumberOfDPPs == 1 || !dml_is_vertical_rotation(SourceScan))) { + vp_height_meta_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + 64 * BlockHeight256Bytes - 1, 64 * BlockHeight256Bytes) - dml_floor(ViewportYStart, 64 * BlockHeight256Bytes)); + } else if (!dml_is_vertical_rotation(SourceScan)) { + vp_height_meta_ub = (dml_uint_t)(dml_ceil(ViewportHeight - 1, 64 * BlockHeight256Bytes) + 64 * BlockHeight256Bytes); + } else { + vp_height_meta_ub = (dml_uint_t)(dml_ceil(SwathWidth - 1, 64 * BlockHeight256Bytes) + 64 * BlockHeight256Bytes); + } + + DCCMetaSurfaceBytes = (dml_uint_t)(DCCMetaPitch * vp_height_meta_ub * BytePerPixel / 256.0); + + if (GPUVMEnable == true) { + *MetaPTEBytesFrame = (dml_uint_t)((dml_ceil((dml_float_t) (DCCMetaSurfaceBytes - 4.0 * 1024.0) / (8 * 4.0 * 1024), 1) + 1) * 64); + MPDEBytesFrame = 128 * (GPUVMMaxPageTableLevels - 1); + } else { + *MetaPTEBytesFrame = 0; + MPDEBytesFrame = 0; + } + + if (DCCEnable != true) { + *MetaPTEBytesFrame = 0; + MPDEBytesFrame = 0; + *MetaRowByte = 0; + } + + MacroTileSizeBytes = MacroTileWidth * BytePerPixel * MacroTileHeight; + + if (ViewportStationary && (NumberOfDPPs == 1 || !dml_is_vertical_rotation(SourceScan))) { + vp_height_dpte_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + MacroTileHeight - 1, MacroTileHeight) - dml_floor(ViewportYStart, MacroTileHeight)); + } else if (!dml_is_vertical_rotation(SourceScan)) { + vp_height_dpte_ub = (dml_uint_t)(dml_ceil(ViewportHeight - 1, MacroTileHeight) + MacroTileHeight); + } else { + vp_height_dpte_ub = (dml_uint_t)(dml_ceil(SwathWidth - 1, MacroTileHeight) + MacroTileHeight); + } + + if (GPUVMEnable == true && GPUVMMaxPageTableLevels > 1) { + *DPDE0BytesFrame = (dml_uint_t)(64 * (dml_ceil((dml_float_t) (Pitch * vp_height_dpte_ub * BytePerPixel - MacroTileSizeBytes) / (dml_float_t) (8 * 2097152), 1) + 1)); + ExtraDPDEBytesFrame = 128 * (GPUVMMaxPageTableLevels - 2); + } else { + *DPDE0BytesFrame = 0; + ExtraDPDEBytesFrame = 0; + } + + PDEAndMetaPTEBytesFrame = *MetaPTEBytesFrame + MPDEBytesFrame + *DPDE0BytesFrame + ExtraDPDEBytesFrame; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DCCEnable = %u\n", __func__, DCCEnable); + dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); + dml_print("DML::%s: SwModeLinear = %u\n", __func__, SurfaceTiling == dml_sw_linear); + dml_print("DML::%s: BytePerPixel = %u\n", __func__, BytePerPixel); + dml_print("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, GPUVMMaxPageTableLevels); + dml_print("DML::%s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); + dml_print("DML::%s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); + dml_print("DML::%s: MacroTileHeight = %u\n", __func__, MacroTileHeight); + dml_print("DML::%s: MacroTileWidth = %u\n", __func__, MacroTileWidth); + dml_print("DML::%s: MetaPTEBytesFrame = %u\n", __func__, *MetaPTEBytesFrame); + dml_print("DML::%s: MPDEBytesFrame = %u\n", __func__, MPDEBytesFrame); + dml_print("DML::%s: DPDE0BytesFrame = %u\n", __func__, *DPDE0BytesFrame); + dml_print("DML::%s: ExtraDPDEBytesFrame= %u\n", __func__, ExtraDPDEBytesFrame); + dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, PDEAndMetaPTEBytesFrame); + dml_print("DML::%s: ViewportHeight = %u\n", __func__, ViewportHeight); + dml_print("DML::%s: SwathWidth = %u\n", __func__, SwathWidth); + dml_print("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); +#endif + + if (SurfaceTiling == dml_sw_linear) { + *PixelPTEReqHeight = 1; + *PixelPTEReqWidth = GPUVMMinPageSizeKBytes * 1024 * 8 / BytePerPixel; + PixelPTEReqWidth_linear = GPUVMMinPageSizeKBytes * 1024 * 8 / BytePerPixel; + *PTERequestSize = 64; + } else if (GPUVMMinPageSizeKBytes == 4) { + *PixelPTEReqHeight = 16 * BlockHeight256Bytes; + *PixelPTEReqWidth = 16 * BlockWidth256Bytes; + *PTERequestSize = 128; + } else { + *PixelPTEReqHeight = MacroTileHeight; + *PixelPTEReqWidth = 8 * 1024 * GPUVMMinPageSizeKBytes / (MacroTileHeight * BytePerPixel); + *PTERequestSize = 64; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); + dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u (after HostVM factor)\n", __func__, PDEAndMetaPTEBytesFrame); + dml_print("DML::%s: PixelPTEReqHeight = %u\n", __func__, *PixelPTEReqHeight); + dml_print("DML::%s: PixelPTEReqWidth = %u\n", __func__, *PixelPTEReqWidth); + dml_print("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); + dml_print("DML::%s: PTERequestSize = %u\n", __func__, *PTERequestSize); + dml_print("DML::%s: Pitch = %u\n", __func__, Pitch); +#endif + + *dpte_row_height_one_row_per_frame = vp_height_dpte_ub; + *dpte_row_width_ub_one_row_per_frame = (dml_uint_t)((dml_ceil(((dml_float_t)Pitch * (dml_float_t) *dpte_row_height_one_row_per_frame / (dml_float_t) *PixelPTEReqHeight - 1) / (dml_float_t) *PixelPTEReqWidth, 1) + 1) * (dml_float_t) *PixelPTEReqWidth); + *PixelPTEBytesPerRow_one_row_per_frame = (dml_uint_t)((dml_float_t) *dpte_row_width_ub_one_row_per_frame / (dml_float_t) *PixelPTEReqWidth * *PTERequestSize); + + if (SurfaceTiling == dml_sw_linear) { + *dpte_row_height = (dml_uint_t)(dml_min(128, 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1))); + dml_print("DML::%s: dpte_row_height term 1 = %u\n", __func__, PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch); + dml_print("DML::%s: dpte_row_height term 2 = %f\n", __func__, dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch)); + dml_print("DML::%s: dpte_row_height term 3 = %f\n", __func__, dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1)); + dml_print("DML::%s: dpte_row_height term 4 = %u\n", __func__, 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * *PixelPTEReqWidth / Pitch), 1)); + dml_print("DML::%s: dpte_row_height = %u\n", __func__, *dpte_row_height); + + *dpte_row_width_ub = (dml_uint_t)(dml_ceil(((dml_float_t) Pitch * (dml_float_t) *dpte_row_height - 1), (dml_float_t) *PixelPTEReqWidth) + *PixelPTEReqWidth); + *PixelPTEBytesPerRow = (dml_uint_t)((dml_float_t) *dpte_row_width_ub / (dml_float_t) *PixelPTEReqWidth * *PTERequestSize); + + // VBA_DELTA, VBA doesn't have programming value for pte row height linear. + *dpte_row_height_linear = 1 << (dml_uint_t) dml_floor(dml_log2(PTEBufferSizeInRequests * PixelPTEReqWidth_linear / Pitch), 1); + if (*dpte_row_height_linear > 128) + *dpte_row_height_linear = 128; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *dpte_row_width_ub); +#endif + + } else if (!dml_is_vertical_rotation(SourceScan)) { + *dpte_row_height = *PixelPTEReqHeight; + + if (GPUVMMinPageSizeKBytes > 64) { + *dpte_row_width_ub = (dml_uint_t)((dml_ceil(((dml_float_t) Pitch * (dml_float_t) *dpte_row_height / (dml_float_t) *PixelPTEReqHeight - 1) / (dml_float_t) *PixelPTEReqWidth, 1) + 1) * *PixelPTEReqWidth); + } else if (ViewportStationary && (NumberOfDPPs == 1)) { + *dpte_row_width_ub = (dml_uint_t)(dml_floor(ViewportXStart + SwathWidth + *PixelPTEReqWidth - 1, *PixelPTEReqWidth) - dml_floor(ViewportXStart, *PixelPTEReqWidth)); + } else { + *dpte_row_width_ub = (dml_uint_t)((dml_ceil((dml_float_t) (SwathWidth - 1) / (dml_float_t)*PixelPTEReqWidth, 1) + 1.0) * *PixelPTEReqWidth); + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *dpte_row_width_ub); +#endif + + ASSERT(*PixelPTEReqWidth); + if (*PixelPTEReqWidth != 0) + *PixelPTEBytesPerRow = *dpte_row_width_ub / *PixelPTEReqWidth * *PTERequestSize; + } else { + *dpte_row_height = (dml_uint_t)(dml_min(*PixelPTEReqWidth, MacroTileWidth)); + + if (ViewportStationary && (NumberOfDPPs == 1)) { + *dpte_row_width_ub = (dml_uint_t)(dml_floor(ViewportYStart + ViewportHeight + *PixelPTEReqHeight - 1, *PixelPTEReqHeight) - dml_floor(ViewportYStart, *PixelPTEReqHeight)); + } else { + *dpte_row_width_ub = (dml_uint_t)((dml_ceil((dml_float_t) (SwathWidth - 1) / (dml_float_t) *PixelPTEReqHeight, 1) + 1) * *PixelPTEReqHeight); + } + + *PixelPTEBytesPerRow = (dml_uint_t)((dml_float_t) *dpte_row_width_ub / (dml_float_t) *PixelPTEReqHeight * *PTERequestSize); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *dpte_row_width_ub); +#endif + } + + if (GPUVMEnable != true) + *PixelPTEBytesPerRow = 0; + + *PixelPTEBytesPerRowStorage = *PixelPTEBytesPerRow; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); + dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); + dml_print("DML::%s: dpte_row_height = %u\n", __func__, *dpte_row_height); + dml_print("DML::%s: dpte_row_height_linear = %u\n", __func__, *dpte_row_height_linear); + dml_print("DML::%s: dpte_row_width_ub = %u\n", __func__, *dpte_row_width_ub); + dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *PixelPTEBytesPerRow); + dml_print("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *PixelPTEBytesPerRowStorage); + dml_print("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); + dml_print("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *dpte_row_height_one_row_per_frame); + dml_print("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *dpte_row_width_ub_one_row_per_frame); + dml_print("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *PixelPTEBytesPerRow_one_row_per_frame); +#endif + + dml_print("DML: vm_bytes = meta_pte_bytes_per_frame (per_pipe) = MetaPTEBytesFrame = : %i\n", *MetaPTEBytesFrame); + + return PDEAndMetaPTEBytesFrame; +} // CalculateVMAndRowBytes + +static void PixelClockAdjustmentForProgressiveToInterlaceUnit(struct dml_display_cfg_st *display_cfg, dml_bool_t ptoi_supported) +{ + dml_uint_t num_active_planes = dml_get_num_active_planes(display_cfg); + + //Progressive To Interlace Unit Effect + for (dml_uint_t k = 0; k < num_active_planes; ++k) { + display_cfg->output.PixelClockBackEnd[k] = display_cfg->timing.PixelClock[k]; + if (display_cfg->timing.Interlace[k] == 1 && ptoi_supported == true) { + display_cfg->timing.PixelClock[k] = 2 * display_cfg->timing.PixelClock[k]; + } + } +} + +static dml_float_t TruncToValidBPP( + dml_float_t LinkBitRate, + dml_uint_t Lanes, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_float_t PixelClock, + dml_float_t DesiredBPP, + dml_bool_t DSCEnable, + enum dml_output_encoder_class Output, + enum dml_output_format_class Format, + dml_uint_t DSCInputBitPerComponent, + dml_uint_t DSCSlices, + dml_uint_t AudioRate, + dml_uint_t AudioLayout, + enum dml_odm_mode ODMModeNoDSC, + enum dml_odm_mode ODMModeDSC, + + // Output + dml_uint_t *RequiredSlots) +{ + dml_float_t MaxLinkBPP; + dml_uint_t MinDSCBPP; + dml_float_t MaxDSCBPP; + dml_uint_t NonDSCBPP0; + dml_uint_t NonDSCBPP1; + dml_uint_t NonDSCBPP2; + + if (Format == dml_420) { + NonDSCBPP0 = 12; + NonDSCBPP1 = 15; + NonDSCBPP2 = 18; + MinDSCBPP = 6; + MaxDSCBPP = 1.5 * DSCInputBitPerComponent - 1.0 / 16; + } else if (Format == dml_444) { + NonDSCBPP0 = 24; + NonDSCBPP1 = 30; + NonDSCBPP2 = 36; + MinDSCBPP = 8; + MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16; + } else { + if (Output == dml_hdmi) { + NonDSCBPP0 = 24; + NonDSCBPP1 = 24; + NonDSCBPP2 = 24; + } else { + NonDSCBPP0 = 16; + NonDSCBPP1 = 20; + NonDSCBPP2 = 24; + } + if (Format == dml_n422) { + MinDSCBPP = 7; + MaxDSCBPP = 2 * DSCInputBitPerComponent - 1.0 / 16.0; + } else { + MinDSCBPP = 8; + MaxDSCBPP = 3 * DSCInputBitPerComponent - 1.0 / 16.0; + } + } + + if (Output == dml_dp2p0) { + MaxLinkBPP = LinkBitRate * Lanes / PixelClock * 128.0 / 132.0 * 383.0 / 384.0 * 65536.0 / 65540.0; + } else if (DSCEnable && Output == dml_dp) { + MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock * (1 - 2.4 / 100); + } else { + MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock; + } + + if (DSCEnable) { + if (ODMModeDSC == dml_odm_mode_combine_4to1) { + MaxLinkBPP = dml_min(MaxLinkBPP, 16); + } else if (ODMModeDSC == dml_odm_mode_combine_2to1) { + MaxLinkBPP = dml_min(MaxLinkBPP, 32); + } else if (ODMModeDSC == dml_odm_mode_split_1to2) { + MaxLinkBPP = 2 * MaxLinkBPP; + } + } else { + if (ODMModeNoDSC == dml_odm_mode_combine_4to1) { + MaxLinkBPP = dml_min(MaxLinkBPP, 16); + } else if (ODMModeNoDSC == dml_odm_mode_combine_2to1) { + MaxLinkBPP = dml_min(MaxLinkBPP, 32); + } else if (ODMModeNoDSC == dml_odm_mode_split_1to2) { + MaxLinkBPP = 2 * MaxLinkBPP; + } + } + + *RequiredSlots = (dml_uint_t)(dml_ceil(DesiredBPP / MaxLinkBPP * 64, 1)); + + if (DesiredBPP == 0) { + if (DSCEnable) { + if (MaxLinkBPP < MinDSCBPP) { + return __DML_DPP_INVALID__; + } else if (MaxLinkBPP >= MaxDSCBPP) { + return MaxDSCBPP; + } else { + return dml_floor(16.0 * MaxLinkBPP, 1.0) / 16.0; + } + } else { + if (MaxLinkBPP >= NonDSCBPP2) { + return NonDSCBPP2; + } else if (MaxLinkBPP >= NonDSCBPP1) { + return NonDSCBPP1; + } else if (MaxLinkBPP >= NonDSCBPP0) { + return NonDSCBPP0; + } else { + return __DML_DPP_INVALID__; + } + } + } else { + if (!((DSCEnable == false && (DesiredBPP == NonDSCBPP2 || DesiredBPP == NonDSCBPP1 || DesiredBPP == NonDSCBPP0)) || + (DSCEnable && DesiredBPP >= MinDSCBPP && DesiredBPP <= MaxDSCBPP))) { + return __DML_DPP_INVALID__; + } else { + return DesiredBPP; + } + } +} // TruncToValidBPP + +static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( + struct display_mode_lib_scratch_st *scratch, + struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *p) +{ + struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals_st *s = &scratch->CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals; + + s->TotalActiveWriteback = 0; + p->Watermark->UrgentWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency; + p->Watermark->USRRetrainingWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency + p->mmSOCParameters.USRRetrainingLatency + p->mmSOCParameters.SMNLatency; + p->Watermark->DRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->Watermark->UrgentWatermark; + p->Watermark->FCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->Watermark->UrgentWatermark; + p->Watermark->StutterExitWatermark = p->mmSOCParameters.SRExitTime + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; + p->Watermark->StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitTime + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; + p->Watermark->Z8StutterExitWatermark = p->mmSOCParameters.SRExitZ8Time + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; + p->Watermark->Z8StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitZ8Time + p->mmSOCParameters.ExtraLatency + 10 / p->DCFClkDeepSleep; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); + dml_print("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); + dml_print("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); + dml_print("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); + dml_print("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); + dml_print("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); + dml_print("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); + dml_print("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); + dml_print("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); + dml_print("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); + dml_print("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); +#endif + + s->TotalActiveWriteback = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->WritebackEnable[k] == true) { + s->TotalActiveWriteback = s->TotalActiveWriteback + 1; + } + } + + if (s->TotalActiveWriteback <= 1) { + p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency; + } else { + p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; + } + if (p->USRRetrainingRequiredFinal) + p->Watermark->WritebackUrgentWatermark = p->Watermark->WritebackUrgentWatermark + p->mmSOCParameters.USRRetrainingLatency; + + if (s->TotalActiveWriteback <= 1) { + p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency; + p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency; + } else { + p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; + p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024 / 32 / p->SOCCLK; + } + + if (p->USRRetrainingRequiredFinal) + p->Watermark->WritebackDRAMClockChangeWatermark = p->Watermark->WritebackDRAMClockChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; + + if (p->USRRetrainingRequiredFinal) + p->Watermark->WritebackFCLKChangeWatermark = p->Watermark->WritebackFCLKChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); + dml_print("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); + dml_print("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); + dml_print("DML::%s: USRRetrainingRequiredFinal = %u\n", __func__, p->USRRetrainingRequiredFinal); + dml_print("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); +#endif + + s->TotalPixelBW = 0.0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + s->TotalPixelBW = s->TotalPixelBW + p->DPPPerSurface[k] + * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * p->VRatio[k] + p->SwathWidthC[k] * p->BytePerPixelDETC[k] * p->VRatioChroma[k]) / (p->HTotal[k] / p->PixelClock[k]); + } + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + + s->LBLatencyHidingSourceLinesY[k] = (dml_uint_t)(dml_min((dml_float_t)p->MaxLineBufferLines, dml_floor((dml_float_t)p->LineBufferSize / (dml_float_t)p->LBBitPerPixel[k] / ((dml_float_t)p->SwathWidthY[k] / dml_max(p->HRatio[k], 1.0)), 1)) - (p->VTaps[k] - 1)); + s->LBLatencyHidingSourceLinesC[k] = (dml_uint_t)(dml_min((dml_float_t)p->MaxLineBufferLines, dml_floor((dml_float_t)p->LineBufferSize / (dml_float_t)p->LBBitPerPixel[k] / ((dml_float_t)p->SwathWidthC[k] / dml_max(p->HRatioChroma[k], 1.0)), 1)) - (p->VTapsChroma[k] - 1)); + + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); + dml_print("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); + dml_print("DML::%s: k=%u, LBBitPerPixel = %u\n", __func__, k, p->LBBitPerPixel[k]); + dml_print("DML::%s: k=%u, HRatio = %f\n", __func__, k, p->HRatio[k]); + dml_print("DML::%s: k=%u, VTaps = %u\n", __func__, k, p->VTaps[k]); +#endif + + s->EffectiveLBLatencyHidingY = s->LBLatencyHidingSourceLinesY[k] / p->VRatio[k] * (p->HTotal[k] / p->PixelClock[k]); + s->EffectiveLBLatencyHidingC = s->LBLatencyHidingSourceLinesC[k] / p->VRatioChroma[k] * (p->HTotal[k] / p->PixelClock[k]); + + s->EffectiveDETBufferSizeY = p->DETBufferSizeY[k]; + if (p->UnboundedRequestEnabled) { + s->EffectiveDETBufferSizeY = s->EffectiveDETBufferSizeY + p->CompressedBufferSizeInkByte * 1024 * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * p->VRatio[k]) / (p->HTotal[k] / p->PixelClock[k]) / s->TotalPixelBW; + } + + s->LinesInDETY[k] = (dml_float_t)s->EffectiveDETBufferSizeY / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; + s->LinesInDETYRoundedDownToSwath[k] = (dml_uint_t)(dml_floor(s->LinesInDETY[k], p->SwathHeightY[k])); + s->FullDETBufferingTimeY = s->LinesInDETYRoundedDownToSwath[k] * (p->HTotal[k] / p->PixelClock[k]) / p->VRatio[k]; + + s->ActiveClockChangeLatencyHidingY = s->EffectiveLBLatencyHidingY + s->FullDETBufferingTimeY - ((dml_float_t)p->DSTXAfterScaler[k] / (dml_float_t)p->HTotal[k] + (dml_float_t)p->DSTYAfterScaler[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; + + if (p->NumberOfActiveSurfaces > 1) { + s->ActiveClockChangeLatencyHidingY = s->ActiveClockChangeLatencyHidingY - (1.0 - 1.0 / (dml_float_t)p->NumberOfActiveSurfaces) * (dml_float_t)p->SwathHeightY[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k] / p->VRatio[k]; + } + + if (p->BytePerPixelDETC[k] > 0) { + s->LinesInDETC[k] = p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k]; + s->LinesInDETCRoundedDownToSwath[k] = (dml_uint_t)(dml_floor(s->LinesInDETC[k], p->SwathHeightC[k])); + s->FullDETBufferingTimeC = s->LinesInDETCRoundedDownToSwath[k] * (p->HTotal[k] / p->PixelClock[k]) / p->VRatioChroma[k]; + s->ActiveClockChangeLatencyHidingC = s->EffectiveLBLatencyHidingC + s->FullDETBufferingTimeC - ((dml_float_t)p->DSTXAfterScaler[k] / (dml_float_t)p->HTotal[k] + (dml_float_t)p->DSTYAfterScaler[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; + if (p->NumberOfActiveSurfaces > 1) { + s->ActiveClockChangeLatencyHidingC = s->ActiveClockChangeLatencyHidingC - (1.0 - 1.0 / (dml_float_t)p->NumberOfActiveSurfaces) * (dml_float_t)p->SwathHeightC[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k] / p->VRatioChroma[k]; + } + s->ActiveClockChangeLatencyHiding = dml_min(s->ActiveClockChangeLatencyHidingY, s->ActiveClockChangeLatencyHidingC); + } else { + s->ActiveClockChangeLatencyHiding = s->ActiveClockChangeLatencyHidingY; + } + + s->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->UrgentWatermark - p->Watermark->DRAMClockChangeWatermark; + s->ActiveFCLKChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->UrgentWatermark - p->Watermark->FCLKChangeWatermark; + s->USRRetrainingLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->USRRetrainingWatermark; + + if (p->WritebackEnable[k]) { + s->WritebackLatencyHiding = (dml_float_t)p->WritebackInterfaceBufferSize * 1024.0 / ((dml_float_t)p->WritebackDestinationWidth[k] * (dml_float_t)p->WritebackDestinationHeight[k] / ((dml_float_t)p->WritebackSourceHeight[k] * (dml_float_t)p->HTotal[k] / p->PixelClock[k]) * 4.0); + if (p->WritebackPixelFormat[k] == dml_444_64) { + s->WritebackLatencyHiding = s->WritebackLatencyHiding / 2; + } + s->WritebackDRAMClockChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackDRAMClockChangeWatermark; + + s->WritebackFCLKChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackFCLKChangeWatermark; + + s->ActiveDRAMClockChangeLatencyMargin[k] = dml_min(s->ActiveDRAMClockChangeLatencyMargin[k], s->WritebackFCLKChangeLatencyMargin); + s->ActiveFCLKChangeLatencyMargin[k] = dml_min(s->ActiveFCLKChangeLatencyMargin[k], s->WritebackDRAMClockChangeLatencyMargin); + } + p->MaxActiveDRAMClockChangeLatencySupported[k] = (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) ? 0 : (s->ActiveDRAMClockChangeLatencyMargin[k] + p->mmSOCParameters.DRAMClockChangeLatency); + p->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveDRAMClockChangeLatencyMargin[k]; + } + + *p->USRRetrainingSupport = true; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && (s->USRRetrainingLatencyMargin[k] < 0)) { + *p->USRRetrainingSupport = false; + } + } + + s->FoundCriticalSurface = false; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && ((!s->FoundCriticalSurface) + || ((s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency) < *p->MaxActiveFCLKChangeLatencySupported))) { + s->FoundCriticalSurface = true; + *p->MaxActiveFCLKChangeLatencySupported = s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency; + } + } + + for (dml_uint_t i = 0; i < p->NumberOfActiveSurfaces; ++i) { + for (dml_uint_t j = 0; j < p->NumberOfActiveSurfaces; ++j) { + if (i == j || + (p->BlendingAndTiming[i] == i && p->BlendingAndTiming[j] == i) || + (p->BlendingAndTiming[j] == j && p->BlendingAndTiming[i] == j) || + (p->BlendingAndTiming[i] == p->BlendingAndTiming[j] && p->BlendingAndTiming[i] != i) || + (p->SynchronizeTimingsFinal && p->PixelClock[i] == p->PixelClock[j] && p->HTotal[i] == p->HTotal[j] && p->VTotal[i] == p->VTotal[j] && p->VActive[i] == p->VActive[j]) || + (p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && (p->DRRDisplay[i] || p->DRRDisplay[j]))) { + s->SynchronizedSurfaces[i][j] = true; + } else { + s->SynchronizedSurfaces[i][j] = false; + } + } + } + + s->FCLKChangeSupportNumber = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if ((p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) && (s->ActiveFCLKChangeLatencyMargin[k] < 0)) { + if (!(p->PrefetchMode[k] <= 1)) { + s->FCLKChangeSupportNumber = 3; + } else if (s->FCLKChangeSupportNumber == 0) { + s->FCLKChangeSupportNumber = ((p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && p->DRRDisplay[k]) ? 2 : 1); + s->LastSurfaceWithoutMargin = k; + } else if (((s->FCLKChangeSupportNumber == 1) && (p->DRRDisplay[k] || (!s->SynchronizedSurfaces[s->LastSurfaceWithoutMargin][k]))) || (s->FCLKChangeSupportNumber == 2)) + s->FCLKChangeSupportNumber = 3; + } + } + + if (s->FCLKChangeSupportNumber == 0) { + *p->FCLKChangeSupport = dml_fclock_change_vactive; + } else if ((s->FCLKChangeSupportNumber == 1) || (s->FCLKChangeSupportNumber == 2)) { + *p->FCLKChangeSupport = dml_fclock_change_vblank; + } else { + *p->FCLKChangeSupport = dml_fclock_change_unsupported; + } + + s->DRAMClockChangeMethod = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame) + s->DRAMClockChangeMethod = 1; + else if (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) + s->DRAMClockChangeMethod = 2; + } + + s->DRAMClockChangeSupportNumber = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (((s->DRAMClockChangeMethod == 0) && (s->ActiveDRAMClockChangeLatencyMargin[k] < 0)) || + ((s->DRAMClockChangeMethod == 1) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_full_frame)) || + ((s->DRAMClockChangeMethod == 2) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_sub_viewport) && (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe))) { + if (p->PrefetchMode[k] != 0) { // Don't need to support DRAM clock change, PrefetchMode 0 means needs DRAM clock change support + s->DRAMClockChangeSupportNumber = 3; + } else if (s->DRAMClockChangeSupportNumber == 0) { + s->DRAMClockChangeSupportNumber = (p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal && p->DRRDisplay[k]) ? 2 : 1; + s->LastSurfaceWithoutMargin = k; + } else if (((s->DRAMClockChangeSupportNumber == 1) && (p->DRRDisplay[k] || !s->SynchronizedSurfaces[s->LastSurfaceWithoutMargin][k])) || (s->DRAMClockChangeSupportNumber == 2)) { + s->DRAMClockChangeSupportNumber = 3; + } + } + } + + if (s->DRAMClockChangeMethod == 0) { // No MALL usage + if (s->DRAMClockChangeSupportNumber == 0) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive; + } else if (s->DRAMClockChangeSupportNumber == 1) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank; + } else if (s->DRAMClockChangeSupportNumber == 2) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr; + } else { + *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; + } + } else if (s->DRAMClockChangeMethod == 1) { // Any pipe using MALL full frame + if (s->DRAMClockChangeSupportNumber == 0) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive_w_mall_full_frame; + } else if (s->DRAMClockChangeSupportNumber == 1) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_w_mall_full_frame; + } else if (s->DRAMClockChangeSupportNumber == 2) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr_w_mall_full_frame; + } else { + *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; + } + } else { // Any pipe using MALL subviewport + if (s->DRAMClockChangeSupportNumber == 0) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vactive_w_mall_sub_vp; + } else if (s->DRAMClockChangeSupportNumber == 1) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_w_mall_sub_vp; + } else if (s->DRAMClockChangeSupportNumber == 2) { + *p->DRAMClockChangeSupport = dml_dram_clock_change_vblank_drr_w_mall_sub_vp; + } else { + *p->DRAMClockChangeSupport = dml_dram_clock_change_unsupported; + } + } + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + s->dst_y_pstate = (dml_uint_t)(dml_ceil((p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.UrgentLatency) / (p->HTotal[k] / p->PixelClock[k]), 1)); + s->src_y_pstate_l = (dml_uint_t)(dml_ceil(s->dst_y_pstate * p->VRatio[k], p->SwathHeightY[k])); + s->src_y_ahead_l = (dml_uint_t)(dml_floor(p->DETBufferSizeY[k] / p->BytePerPixelDETY[k] / p->SwathWidthY[k], p->SwathHeightY[k]) + s->LBLatencyHidingSourceLinesY[k]); + s->sub_vp_lines_l = s->src_y_pstate_l + s->src_y_ahead_l + p->meta_row_height[k]; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + dml_print("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + dml_print("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + dml_print("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + dml_print("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); + dml_print("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); + dml_print("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); + dml_print("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); + dml_print("DML::%s: k=%u, meta_row_height = %u\n", __func__, k, p->meta_row_height[k]); + dml_print("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); +#endif + p->SubViewportLinesNeededInMALL[k] = s->sub_vp_lines_l; + + if (p->BytePerPixelDETC[k] > 0) { + s->src_y_pstate_c = (dml_uint_t)(dml_ceil(s->dst_y_pstate * p->VRatioChroma[k], p->SwathHeightC[k])); + s->src_y_ahead_c = (dml_uint_t)(dml_floor(p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k], p->SwathHeightC[k]) + s->LBLatencyHidingSourceLinesC[k]); + s->sub_vp_lines_c = s->src_y_pstate_c + s->src_y_ahead_c + p->meta_row_height_chroma[k]; + p->SubViewportLinesNeededInMALL[k] = (dml_uint_t)(dml_max(s->sub_vp_lines_l, s->sub_vp_lines_c)); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); + dml_print("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); + dml_print("DML::%s: k=%u, meta_row_height_chroma = %u\n", __func__, k, p->meta_row_height_chroma[k]); + dml_print("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); +#endif + } + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->DRAMClockChangeSupport); + dml_print("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->FCLKChangeSupport); + dml_print("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); + dml_print("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); +#endif +} // CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport + +static void CalculateDCFCLKDeepSleep( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t BytePerPixelY[], + dml_uint_t BytePerPixelC[], + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_uint_t SwathWidthY[], + dml_uint_t SwathWidthC[], + dml_uint_t DPPPerSurface[], + dml_float_t HRatio[], + dml_float_t HRatioChroma[], + dml_float_t PixelClock[], + dml_float_t PSCL_THROUGHPUT[], + dml_float_t PSCL_THROUGHPUT_CHROMA[], + dml_float_t Dppclk[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_uint_t ReturnBusWidth, + + // Output + dml_float_t *DCFClkDeepSleep) +{ + dml_float_t DisplayPipeLineDeliveryTimeLuma; + dml_float_t DisplayPipeLineDeliveryTimeChroma; + dml_float_t DCFClkDeepSleepPerSurface[__DML_NUM_PLANES__]; + dml_float_t ReadBandwidth = 0.0; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + + if (VRatio[k] <= 1) { + DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChroma = 0; + } else { + if (VRatioChroma[k] <= 1) { + DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } + + if (BytePerPixelC[k] > 0) { + DCFClkDeepSleepPerSurface[k] = dml_max(__DML_MIN_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 32.0 / DisplayPipeLineDeliveryTimeLuma, + __DML_MIN_DCFCLK_FACTOR__ * SwathWidthC[k] * BytePerPixelC[k] / 32.0 / DisplayPipeLineDeliveryTimeChroma); + } else { + DCFClkDeepSleepPerSurface[k] = __DML_MIN_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 64.0 / DisplayPipeLineDeliveryTimeLuma; + } + DCFClkDeepSleepPerSurface[k] = dml_max(DCFClkDeepSleepPerSurface[k], PixelClock[k] / 16); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, PixelClock = %f\n", __func__, k, PixelClock[k]); + dml_print("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); +#endif + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + ReadBandwidth = ReadBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; + } + + *DCFClkDeepSleep = dml_max(8.0, __DML_MIN_DCFCLK_FACTOR__ * ReadBandwidth / (dml_float_t) ReturnBusWidth); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: __DML_MIN_DCFCLK_FACTOR__ = %f\n", __func__, __DML_MIN_DCFCLK_FACTOR__); + dml_print("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); + dml_print("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); + dml_print("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); +#endif + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + *DCFClkDeepSleep = dml_max(*DCFClkDeepSleep, DCFClkDeepSleepPerSurface[k]); + } + dml_print("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); +} // CalculateDCFCLKDeepSleep + +static void CalculateUrgentBurstFactor( + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange, + dml_uint_t swath_width_luma_ub, + dml_uint_t swath_width_chroma_ub, + dml_uint_t SwathHeightY, + dml_uint_t SwathHeightC, + dml_float_t LineTime, + dml_float_t UrgentLatency, + dml_float_t CursorBufferSize, + dml_uint_t CursorWidth, + dml_uint_t CursorBPP, + dml_float_t VRatio, + dml_float_t VRatioC, + dml_float_t BytePerPixelInDETY, + dml_float_t BytePerPixelInDETC, + dml_uint_t DETBufferSizeY, + dml_uint_t DETBufferSizeC, + // Output + dml_float_t *UrgentBurstFactorCursor, + dml_float_t *UrgentBurstFactorLuma, + dml_float_t *UrgentBurstFactorChroma, + dml_bool_t *NotEnoughUrgentLatencyHiding) +{ + dml_float_t LinesInDETLuma; + dml_float_t LinesInDETChroma; + dml_uint_t LinesInCursorBuffer; + dml_float_t CursorBufferSizeInTime; + dml_float_t DETBufferSizeInTimeLuma; + dml_float_t DETBufferSizeInTimeChroma; + + *NotEnoughUrgentLatencyHiding = 0; + + if (CursorWidth > 0) { + LinesInCursorBuffer = 1 << (dml_uint_t) dml_floor(dml_log2(CursorBufferSize * 1024.0 / (CursorWidth * CursorBPP / 8.0)), 1.0); + if (VRatio > 0) { + CursorBufferSizeInTime = LinesInCursorBuffer * LineTime / VRatio; + if (CursorBufferSizeInTime - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorCursor = 0; + } else { + *UrgentBurstFactorCursor = CursorBufferSizeInTime / (CursorBufferSizeInTime - UrgentLatency); + } + } else { + *UrgentBurstFactorCursor = 1; + } + } + + LinesInDETLuma = (UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe ? 1024*1024 : DETBufferSizeY) / BytePerPixelInDETY / swath_width_luma_ub; + + if (VRatio > 0) { + DETBufferSizeInTimeLuma = dml_floor(LinesInDETLuma, SwathHeightY) * LineTime / VRatio; + if (DETBufferSizeInTimeLuma - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorLuma = 0; + } else { + *UrgentBurstFactorLuma = DETBufferSizeInTimeLuma / (DETBufferSizeInTimeLuma - UrgentLatency); + } + } else { + *UrgentBurstFactorLuma = 1; + } + + if (BytePerPixelInDETC > 0) { + LinesInDETChroma = (UseMALLForPStateChange == dml_use_mall_pstate_change_phantom_pipe ? 1024*1024 : DETBufferSizeC) / BytePerPixelInDETC / swath_width_chroma_ub; + + if (VRatioC > 0) { + DETBufferSizeInTimeChroma = dml_floor(LinesInDETChroma, SwathHeightC) * LineTime / VRatioC; + if (DETBufferSizeInTimeChroma - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorChroma = 0; + } else { + *UrgentBurstFactorChroma = DETBufferSizeInTimeChroma / (DETBufferSizeInTimeChroma - UrgentLatency); + } + } else { + *UrgentBurstFactorChroma = 1; + } + } +} // CalculateUrgentBurstFactor + +static void CalculatePixelDeliveryTimes( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_float_t VRatioPrefetchY[], + dml_float_t VRatioPrefetchC[], + dml_uint_t swath_width_luma_ub[], + dml_uint_t swath_width_chroma_ub[], + dml_uint_t DPPPerSurface[], + dml_float_t HRatio[], + dml_float_t HRatioChroma[], + dml_float_t PixelClock[], + dml_float_t PSCL_THROUGHPUT[], + dml_float_t PSCL_THROUGHPUT_CHROMA[], + dml_float_t Dppclk[], + dml_uint_t BytePerPixelC[], + enum dml_rotation_angle SourceScan[], + dml_uint_t NumberOfCursors[], + dml_uint_t CursorWidth[], + dml_uint_t CursorBPP[], + dml_uint_t BlockWidth256BytesY[], + dml_uint_t BlockHeight256BytesY[], + dml_uint_t BlockWidth256BytesC[], + dml_uint_t BlockHeight256BytesC[], + + // Output + dml_float_t DisplayPipeLineDeliveryTimeLuma[], + dml_float_t DisplayPipeLineDeliveryTimeChroma[], + dml_float_t DisplayPipeLineDeliveryTimeLumaPrefetch[], + dml_float_t DisplayPipeLineDeliveryTimeChromaPrefetch[], + dml_float_t DisplayPipeRequestDeliveryTimeLuma[], + dml_float_t DisplayPipeRequestDeliveryTimeChroma[], + dml_float_t DisplayPipeRequestDeliveryTimeLumaPrefetch[], + dml_float_t DisplayPipeRequestDeliveryTimeChromaPrefetch[], + dml_float_t CursorRequestDeliveryTime[], + dml_float_t CursorRequestDeliveryTimePrefetch[]) +{ + dml_float_t req_per_swath_ub; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : HRatio = %f\n", __func__, k, HRatio[k]); + dml_print("DML::%s: k=%u : VRatio = %f\n", __func__, k, VRatio[k]); + dml_print("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, HRatioChroma[k]); + dml_print("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, VRatioChroma[k]); + dml_print("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); + dml_print("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); + dml_print("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); + dml_print("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); + dml_print("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); + dml_print("DML::%s: k=%u : PixelClock = %f\n", __func__, k, PixelClock[k]); + dml_print("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); +#endif + + if (VRatio[k] <= 1) { + DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChroma[k] = 0; + } else { + if (VRatioChroma[k] <= 1) { + DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } + + if (VRatioPrefetchY[k] <= 1) { + DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] * DPPPerSurface[k] / HRatio[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0; + } else { + if (VRatioPrefetchC[k] <= 1) { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] * DPPPerSurface[k] / HRatioChroma[k] / PixelClock[k]; + } else { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); + dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); + dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); + dml_print("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); +#endif + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_vertical_rotation(SourceScan[k])) { + req_per_swath_ub = swath_width_luma_ub[k] / BlockWidth256BytesY[k]; + } else { + req_per_swath_ub = swath_width_luma_ub[k] / BlockHeight256BytesY[k]; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : req_per_swath_ub = %f (Luma)\n", __func__, k, req_per_swath_ub); +#endif + + DisplayPipeRequestDeliveryTimeLuma[k] = DisplayPipeLineDeliveryTimeLuma[k] / req_per_swath_ub; + DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = DisplayPipeLineDeliveryTimeLumaPrefetch[k] / req_per_swath_ub; + if (BytePerPixelC[k] == 0) { + DisplayPipeRequestDeliveryTimeChroma[k] = 0; + DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = 0; + } else { + if (!dml_is_vertical_rotation(SourceScan[k])) { + req_per_swath_ub = swath_width_chroma_ub[k] / BlockWidth256BytesC[k]; + } else { + req_per_swath_ub = swath_width_chroma_ub[k] / BlockHeight256BytesC[k]; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : req_per_swath_ub = %f (Chroma)\n", __func__, k, req_per_swath_ub); +#endif + DisplayPipeRequestDeliveryTimeChroma[k] = DisplayPipeLineDeliveryTimeChroma[k] / req_per_swath_ub; + DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); + dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); + dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); + dml_print("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); +#endif + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + dml_uint_t cursor_req_per_width; + cursor_req_per_width = (dml_uint_t)(dml_ceil((dml_float_t) CursorWidth[k] * (dml_float_t) CursorBPP[k] / 256.0 / 8.0, 1.0)); + if (NumberOfCursors[k] > 0) { + if (VRatio[k] <= 1) { + CursorRequestDeliveryTime[k] = (dml_float_t) CursorWidth[k] / HRatio[k] / PixelClock[k] / cursor_req_per_width; + } else { + CursorRequestDeliveryTime[k] = (dml_float_t) CursorWidth[k] / PSCL_THROUGHPUT[k] / Dppclk[k] / cursor_req_per_width; + } + if (VRatioPrefetchY[k] <= 1) { + CursorRequestDeliveryTimePrefetch[k] = (dml_float_t) CursorWidth[k] / HRatio[k] / PixelClock[k] / cursor_req_per_width; + } else { + CursorRequestDeliveryTimePrefetch[k] = (dml_float_t) CursorWidth[k] / PSCL_THROUGHPUT[k] / Dppclk[k] / cursor_req_per_width; + } + } else { + CursorRequestDeliveryTime[k] = 0; + CursorRequestDeliveryTimePrefetch[k] = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u : NumberOfCursors = %u\n", __func__, k, NumberOfCursors[k]); + dml_print("DML::%s: k=%u : CursorRequestDeliveryTime = %f\n", __func__, k, CursorRequestDeliveryTime[k]); + dml_print("DML::%s: k=%u : CursorRequestDeliveryTimePrefetch = %f\n", __func__, k, CursorRequestDeliveryTimePrefetch[k]); +#endif + } +} // CalculatePixelDeliveryTimes + +static void CalculateMetaAndPTETimes( + dml_bool_t use_one_row_for_frame[], + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t GPUVMEnable, + dml_uint_t MetaChunkSize, + dml_uint_t MinMetaChunkSizeBytes, + dml_uint_t HTotal[], + dml_float_t VRatio[], + dml_float_t VRatioChroma[], + dml_float_t DestinationLinesToRequestRowInVBlank[], + dml_float_t DestinationLinesToRequestRowInImmediateFlip[], + dml_bool_t DCCEnable[], + dml_float_t PixelClock[], + dml_uint_t BytePerPixelY[], + dml_uint_t BytePerPixelC[], + enum dml_rotation_angle SourceScan[], + dml_uint_t dpte_row_height[], + dml_uint_t dpte_row_height_chroma[], + dml_uint_t meta_row_width[], + dml_uint_t meta_row_width_chroma[], + dml_uint_t meta_row_height[], + dml_uint_t meta_row_height_chroma[], + dml_uint_t meta_req_width[], + dml_uint_t meta_req_width_chroma[], + dml_uint_t meta_req_height[], + dml_uint_t meta_req_height_chroma[], + dml_uint_t dpte_group_bytes[], + dml_uint_t PTERequestSizeY[], + dml_uint_t PTERequestSizeC[], + dml_uint_t PixelPTEReqWidthY[], + dml_uint_t PixelPTEReqHeightY[], + dml_uint_t PixelPTEReqWidthC[], + dml_uint_t PixelPTEReqHeightC[], + dml_uint_t dpte_row_width_luma_ub[], + dml_uint_t dpte_row_width_chroma_ub[], + + // Output + dml_float_t DST_Y_PER_PTE_ROW_NOM_L[], + dml_float_t DST_Y_PER_PTE_ROW_NOM_C[], + dml_float_t DST_Y_PER_META_ROW_NOM_L[], + dml_float_t DST_Y_PER_META_ROW_NOM_C[], + dml_float_t TimePerMetaChunkNominal[], + dml_float_t TimePerChromaMetaChunkNominal[], + dml_float_t TimePerMetaChunkVBlank[], + dml_float_t TimePerChromaMetaChunkVBlank[], + dml_float_t TimePerMetaChunkFlip[], + dml_float_t TimePerChromaMetaChunkFlip[], + dml_float_t time_per_pte_group_nom_luma[], + dml_float_t time_per_pte_group_vblank_luma[], + dml_float_t time_per_pte_group_flip_luma[], + dml_float_t time_per_pte_group_nom_chroma[], + dml_float_t time_per_pte_group_vblank_chroma[], + dml_float_t time_per_pte_group_flip_chroma[]) +{ + dml_uint_t meta_chunk_width; + dml_uint_t min_meta_chunk_width; + dml_uint_t meta_chunk_per_row_int; + dml_uint_t meta_row_remainder; + dml_uint_t meta_chunk_threshold; + dml_uint_t meta_chunks_per_row_ub; + dml_uint_t meta_chunk_width_chroma; + dml_uint_t min_meta_chunk_width_chroma; + dml_uint_t meta_chunk_per_row_int_chroma; + dml_uint_t meta_row_remainder_chroma; + dml_uint_t meta_chunk_threshold_chroma; + dml_uint_t meta_chunks_per_row_ub_chroma; + dml_uint_t dpte_group_width_luma; + dml_uint_t dpte_groups_per_row_luma_ub; + dml_uint_t dpte_group_width_chroma; + dml_uint_t dpte_groups_per_row_chroma_ub; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + DST_Y_PER_PTE_ROW_NOM_L[k] = dpte_row_height[k] / VRatio[k]; + if (BytePerPixelC[k] == 0) { + DST_Y_PER_PTE_ROW_NOM_C[k] = 0; + } else { + DST_Y_PER_PTE_ROW_NOM_C[k] = dpte_row_height_chroma[k] / VRatioChroma[k]; + } + DST_Y_PER_META_ROW_NOM_L[k] = meta_row_height[k] / VRatio[k]; + if (BytePerPixelC[k] == 0) { + DST_Y_PER_META_ROW_NOM_C[k] = 0; + } else { + DST_Y_PER_META_ROW_NOM_C[k] = meta_row_height_chroma[k] / VRatioChroma[k]; + } + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (DCCEnable[k] == true) { + meta_chunk_width = MetaChunkSize * 1024 * 256 / BytePerPixelY[k] / meta_row_height[k]; + min_meta_chunk_width = MinMetaChunkSizeBytes * 256 / BytePerPixelY[k] / meta_row_height[k]; + meta_chunk_per_row_int = meta_row_width[k] / meta_chunk_width; + meta_row_remainder = meta_row_width[k] % meta_chunk_width; + if (!dml_is_vertical_rotation(SourceScan[k])) { + meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_width[k]; + } else { + meta_chunk_threshold = 2 * min_meta_chunk_width - meta_req_height[k]; + } + if (meta_row_remainder <= meta_chunk_threshold) { + meta_chunks_per_row_ub = meta_chunk_per_row_int + 1; + } else { + meta_chunks_per_row_ub = meta_chunk_per_row_int + 2; + } + TimePerMetaChunkNominal[k] = meta_row_height[k] / VRatio[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; + TimePerMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; + TimePerMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub; + if (BytePerPixelC[k] == 0) { + TimePerChromaMetaChunkNominal[k] = 0; + TimePerChromaMetaChunkVBlank[k] = 0; + TimePerChromaMetaChunkFlip[k] = 0; + } else { + meta_chunk_width_chroma = MetaChunkSize * 1024 * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; + min_meta_chunk_width_chroma = MinMetaChunkSizeBytes * 256 / BytePerPixelC[k] / meta_row_height_chroma[k]; + meta_chunk_per_row_int_chroma = (dml_uint_t)((dml_float_t) meta_row_width_chroma[k] / meta_chunk_width_chroma); + meta_row_remainder_chroma = meta_row_width_chroma[k] % meta_chunk_width_chroma; + if (!dml_is_vertical_rotation(SourceScan[k])) { + meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_width_chroma[k]; + } else { + meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - meta_req_height_chroma[k]; + } + if (meta_row_remainder_chroma <= meta_chunk_threshold_chroma) { + meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 1; + } else { + meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 2; + } + TimePerChromaMetaChunkNominal[k] = meta_row_height_chroma[k] / VRatioChroma[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; + TimePerChromaMetaChunkVBlank[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; + TimePerChromaMetaChunkFlip[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / meta_chunks_per_row_ub_chroma; + } + } else { + TimePerMetaChunkNominal[k] = 0; + TimePerMetaChunkVBlank[k] = 0; + TimePerMetaChunkFlip[k] = 0; + TimePerChromaMetaChunkNominal[k] = 0; + TimePerChromaMetaChunkVBlank[k] = 0; + TimePerChromaMetaChunkFlip[k] = 0; + } + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (GPUVMEnable == true) { + if (!dml_is_vertical_rotation(SourceScan[k])) { + dpte_group_width_luma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeY[k] * PixelPTEReqWidthY[k]); + } else { + dpte_group_width_luma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeY[k] * PixelPTEReqHeightY[k]); + } + + if (use_one_row_for_frame[k]) { + dpte_groups_per_row_luma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_luma_ub[k] / (dml_float_t) dpte_group_width_luma / 2.0, 1.0)); + } else { + dpte_groups_per_row_luma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_luma_ub[k] / (dml_float_t) dpte_group_width_luma, 1.0)); + } + + dml_print("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, use_one_row_for_frame[k]); + dml_print("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, dpte_group_bytes[k]); + dml_print("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, PTERequestSizeY[k]); + dml_print("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, PixelPTEReqWidthY[k]); + dml_print("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, PixelPTEReqHeightY[k]); + dml_print("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, dpte_row_width_luma_ub[k]); + dml_print("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); + dml_print("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); + + time_per_pte_group_nom_luma[k] = DST_Y_PER_PTE_ROW_NOM_L[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; + time_per_pte_group_vblank_luma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; + time_per_pte_group_flip_luma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_luma_ub; + if (BytePerPixelC[k] == 0) { + time_per_pte_group_nom_chroma[k] = 0; + time_per_pte_group_vblank_chroma[k] = 0; + time_per_pte_group_flip_chroma[k] = 0; + } else { + if (!dml_is_vertical_rotation(SourceScan[k])) { + dpte_group_width_chroma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeC[k] * PixelPTEReqWidthC[k]); + } else { + dpte_group_width_chroma = (dml_uint_t)((dml_float_t) dpte_group_bytes[k] / (dml_float_t) PTERequestSizeC[k] * PixelPTEReqHeightC[k]); + } + + if (use_one_row_for_frame[k]) { + dpte_groups_per_row_chroma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_chroma_ub[k] / (dml_float_t) dpte_group_width_chroma / 2.0, 1.0)); + } else { + dpte_groups_per_row_chroma_ub = (dml_uint_t)(dml_ceil((dml_float_t) dpte_row_width_chroma_ub[k] / (dml_float_t) dpte_group_width_chroma, 1.0)); + } + dml_print("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, dpte_row_width_chroma_ub[k]); + dml_print("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); + dml_print("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); + + time_per_pte_group_nom_chroma[k] = DST_Y_PER_PTE_ROW_NOM_C[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; + time_per_pte_group_vblank_chroma[k] = DestinationLinesToRequestRowInVBlank[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; + time_per_pte_group_flip_chroma[k] = DestinationLinesToRequestRowInImmediateFlip[k] * HTotal[k] / PixelClock[k] / dpte_groups_per_row_chroma_ub; + } + } else { + time_per_pte_group_nom_luma[k] = 0; + time_per_pte_group_vblank_luma[k] = 0; + time_per_pte_group_flip_luma[k] = 0; + time_per_pte_group_nom_chroma[k] = 0; + time_per_pte_group_vblank_chroma[k] = 0; + time_per_pte_group_flip_chroma[k] = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, DestinationLinesToRequestRowInVBlank = %f\n", __func__, k, DestinationLinesToRequestRowInVBlank[k]); + dml_print("DML::%s: k=%u, DestinationLinesToRequestRowInImmediateFlip = %f\n", __func__, k, DestinationLinesToRequestRowInImmediateFlip[k]); + + dml_print("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, DST_Y_PER_PTE_ROW_NOM_L[k]); + dml_print("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, DST_Y_PER_PTE_ROW_NOM_C[k]); + dml_print("DML::%s: k=%u, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, DST_Y_PER_META_ROW_NOM_L[k]); + dml_print("DML::%s: k=%u, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, DST_Y_PER_META_ROW_NOM_C[k]); + dml_print("DML::%s: k=%u, TimePerMetaChunkNominal = %f\n", __func__, k, TimePerMetaChunkNominal[k]); + dml_print("DML::%s: k=%u, TimePerMetaChunkVBlank = %f\n", __func__, k, TimePerMetaChunkVBlank[k]); + dml_print("DML::%s: k=%u, TimePerMetaChunkFlip = %f\n", __func__, k, TimePerMetaChunkFlip[k]); + dml_print("DML::%s: k=%u, TimePerChromaMetaChunkNominal = %f\n", __func__, k, TimePerChromaMetaChunkNominal[k]); + dml_print("DML::%s: k=%u, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, TimePerChromaMetaChunkVBlank[k]); + dml_print("DML::%s: k=%u, TimePerChromaMetaChunkFlip = %f\n", __func__, k, TimePerChromaMetaChunkFlip[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, time_per_pte_group_nom_luma[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, time_per_pte_group_vblank_luma[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, time_per_pte_group_flip_luma[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, time_per_pte_group_nom_chroma[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, time_per_pte_group_vblank_chroma[k]); + dml_print("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, time_per_pte_group_flip_chroma[k]); +#endif + } +} // CalculateMetaAndPTETimes + +static void CalculateVMGroupAndRequestTimes( + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t GPUVMEnable, + dml_uint_t GPUVMMaxPageTableLevels, + dml_uint_t HTotal[], + dml_uint_t BytePerPixelC[], + dml_float_t DestinationLinesToRequestVMInVBlank[], + dml_float_t DestinationLinesToRequestVMInImmediateFlip[], + dml_bool_t DCCEnable[], + dml_float_t PixelClock[], + dml_uint_t dpte_row_width_luma_ub[], + dml_uint_t dpte_row_width_chroma_ub[], + dml_uint_t vm_group_bytes[], + dml_uint_t dpde0_bytes_per_frame_ub_l[], + dml_uint_t dpde0_bytes_per_frame_ub_c[], + dml_uint_t meta_pte_bytes_per_frame_ub_l[], + dml_uint_t meta_pte_bytes_per_frame_ub_c[], + + // Output + dml_float_t TimePerVMGroupVBlank[], + dml_float_t TimePerVMGroupFlip[], + dml_float_t TimePerVMRequestVBlank[], + dml_float_t TimePerVMRequestFlip[]) +{ + dml_uint_t num_group_per_lower_vm_stage; + dml_uint_t num_req_per_lower_vm_stage; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + dml_print("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); +#endif + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, DCCEnable = %u\n", __func__, k, DCCEnable[k]); + dml_print("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); + dml_print("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); + dml_print("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); + dml_print("DML::%s: k=%u, meta_pte_bytes_per_frame_ub_l = %u\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); + dml_print("DML::%s: k=%u, meta_pte_bytes_per_frame_ub_c = %u\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); +#endif + + if (GPUVMEnable == true && (DCCEnable[k] == true || GPUVMMaxPageTableLevels > 1)) { + if (DCCEnable[k] == false) { + if (BytePerPixelC[k] > 0) { + num_group_per_lower_vm_stage = (dml_uint_t) (dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_l[k] / (dml_float_t) vm_group_bytes[k], 1.0) + + dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_c[k] / (dml_float_t) vm_group_bytes[k], 1.0)); + } else { + num_group_per_lower_vm_stage = (dml_uint_t) (dml_ceil((dml_float_t) dpde0_bytes_per_frame_ub_l[k] / (dml_float_t) vm_group_bytes[k], 1.0)); + } + } else { + if (GPUVMMaxPageTableLevels == 1) { + if (BytePerPixelC[k] > 0) { + num_group_per_lower_vm_stage = (dml_uint_t)(dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0) + + dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0)); + } else { + num_group_per_lower_vm_stage = (dml_uint_t)(dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1.0)); + } + } else { + if (BytePerPixelC[k] > 0) { + num_group_per_lower_vm_stage = (dml_uint_t)(2.0 + dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + + dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + + dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + + dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_c[k]) / (dml_float_t) (vm_group_bytes[k]), 1)); + } else { + num_group_per_lower_vm_stage = (dml_uint_t)(1.0 + dml_ceil((dml_float_t) (dpde0_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1) + + dml_ceil((dml_float_t) (meta_pte_bytes_per_frame_ub_l[k]) / (dml_float_t) (vm_group_bytes[k]), 1)); + } + } + } + + if (DCCEnable[k] == false) { + if (BytePerPixelC[k] > 0) { + num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + dpde0_bytes_per_frame_ub_c[k] / 64; + } else { + num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64; + } + } else { + if (GPUVMMaxPageTableLevels == 1) { + if (BytePerPixelC[k] > 0) { + num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_c[k] / 64; + } else { + num_req_per_lower_vm_stage = meta_pte_bytes_per_frame_ub_l[k] / 64; + } + } else { + if (BytePerPixelC[k] > 0) { + num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + dpde0_bytes_per_frame_ub_c[k] / 64 + meta_pte_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_c[k] / 64; + } else { + num_req_per_lower_vm_stage = dpde0_bytes_per_frame_ub_l[k] / 64 + meta_pte_bytes_per_frame_ub_l[k] / 64; + } + } + } + + TimePerVMGroupVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] / num_group_per_lower_vm_stage; + TimePerVMGroupFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] / num_group_per_lower_vm_stage; + TimePerVMRequestVBlank[k] = DestinationLinesToRequestVMInVBlank[k] * HTotal[k] / PixelClock[k] / num_req_per_lower_vm_stage; + TimePerVMRequestFlip[k] = DestinationLinesToRequestVMInImmediateFlip[k] * HTotal[k] / PixelClock[k] / num_req_per_lower_vm_stage; + + if (GPUVMMaxPageTableLevels > 2) { + TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; + TimePerVMGroupFlip[k] = TimePerVMGroupFlip[k] / 2; + TimePerVMRequestVBlank[k] = TimePerVMRequestVBlank[k] / 2; + TimePerVMRequestFlip[k] = TimePerVMRequestFlip[k] / 2; + } + + } else { + TimePerVMGroupVBlank[k] = 0; + TimePerVMGroupFlip[k] = 0; + TimePerVMRequestVBlank[k] = 0; + TimePerVMRequestFlip[k] = 0; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); + dml_print("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); + dml_print("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); + dml_print("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); +#endif + } +} // CalculateVMGroupAndRequestTimes + +static void CalculateStutterEfficiency(struct display_mode_lib_scratch_st *scratch, + struct CalculateStutterEfficiency_params_st *p) +{ + dml_float_t DETBufferingTimeY = 0; + dml_float_t SwathWidthYCriticalSurface = 0; + dml_float_t SwathHeightYCriticalSurface = 0; + dml_float_t VActiveTimeCriticalSurface = 0; + dml_float_t FrameTimeCriticalSurface = 0; + dml_uint_t BytePerPixelYCriticalSurface = 0; + dml_float_t LinesToFinishSwathTransferStutterCriticalSurface = 0; + dml_uint_t DETBufferSizeYCriticalSurface = 0; + dml_float_t MinTTUVBlankCriticalSurface = 0; + dml_uint_t BlockWidth256BytesYCriticalSurface = 0; + dml_bool_t SinglePlaneCriticalSurface = 0; + dml_bool_t SinglePipeCriticalSurface = 0; + dml_float_t TotalCompressedReadBandwidth = 0; + dml_float_t TotalRowReadBandwidth = 0; + dml_float_t AverageDCCCompressionRate = 0; + dml_float_t EffectiveCompressedBufferSize = 0; + dml_float_t PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = 0; + dml_float_t StutterBurstTime = 0; + dml_uint_t TotalActiveWriteback = 0; + dml_float_t LinesInDETY = 0; + dml_float_t LinesInDETYRoundedDownToSwath = 0; + dml_float_t MaximumEffectiveCompressionLuma = 0; + dml_float_t MaximumEffectiveCompressionChroma = 0; + dml_float_t TotalZeroSizeRequestReadBandwidth = 0; + dml_float_t TotalZeroSizeCompressedReadBandwidth = 0; + dml_float_t AverageDCCZeroSizeFraction = 0; + dml_float_t AverageZeroSizeCompressionRate = 0; + + dml_bool_t FoundCriticalSurface = false; + + dml_uint_t TotalNumberOfActiveOTG = 0; + dml_float_t SinglePixelClock = 0; + dml_uint_t SingleHTotal = 0; + dml_uint_t SingleVTotal = 0; + dml_bool_t SameTiming = true; + + dml_float_t LastStutterPeriod = 0.0; + dml_float_t LastZ8StutterPeriod = 0.0; + + dml_uint_t SwathSizeCriticalSurface; + dml_uint_t LastChunkOfSwathSize; + dml_uint_t MissingPartOfLastSwathOfDETSize; + + TotalZeroSizeRequestReadBandwidth = 0; + TotalZeroSizeCompressedReadBandwidth = 0; + TotalRowReadBandwidth = 0; + TotalCompressedReadBandwidth = 0; + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { + if (p->DCCEnable[k] == true) { + if ((dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockWidth256BytesY[k] > p->SwathHeightY[k]) || (!dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockHeight256BytesY[k] > p->SwathHeightY[k]) || p->DCCYMaxUncompressedBlock[k] < 256) { + MaximumEffectiveCompressionLuma = 2; + } else { + MaximumEffectiveCompressionLuma = 4; + } + TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] / dml_min(p->NetDCCRateLuma[k], MaximumEffectiveCompressionLuma); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + dml_print("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->NetDCCRateLuma[k]); + dml_print("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, MaximumEffectiveCompressionLuma); +#endif + TotalZeroSizeRequestReadBandwidth = TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->DCCFractionOfZeroSizeRequestsLuma[k]; + TotalZeroSizeCompressedReadBandwidth = TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->DCCFractionOfZeroSizeRequestsLuma[k] / MaximumEffectiveCompressionLuma; + + if (p->ReadBandwidthSurfaceChroma[k] > 0) { + if ((dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockWidth256BytesC[k] > p->SwathHeightC[k]) || (!dml_is_vertical_rotation(p->SourceScan[k]) && p->BlockHeight256BytesC[k] > p->SwathHeightC[k]) || p->DCCCMaxUncompressedBlock[k] < 256) { + MaximumEffectiveCompressionChroma = 2; + } else { + MaximumEffectiveCompressionChroma = 4; + } + TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] / dml_min(p->NetDCCRateChroma[k], MaximumEffectiveCompressionChroma); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); + dml_print("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->NetDCCRateChroma[k]); + dml_print("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, MaximumEffectiveCompressionChroma); +#endif + TotalZeroSizeRequestReadBandwidth = TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->DCCFractionOfZeroSizeRequestsChroma[k]; + TotalZeroSizeCompressedReadBandwidth = TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->DCCFractionOfZeroSizeRequestsChroma[k] / MaximumEffectiveCompressionChroma; + } + } else { + TotalCompressedReadBandwidth = TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] + p->ReadBandwidthSurfaceChroma[k]; + } + TotalRowReadBandwidth = TotalRowReadBandwidth + p->DPPPerSurface[k] * (p->meta_row_bw[k] + p->dpte_row_bw[k]); + } + } + + AverageDCCCompressionRate = p->TotalDataReadBandwidth / TotalCompressedReadBandwidth; + AverageDCCZeroSizeFraction = TotalZeroSizeRequestReadBandwidth / p->TotalDataReadBandwidth; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); + dml_print("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, TotalCompressedReadBandwidth); + dml_print("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, TotalZeroSizeRequestReadBandwidth); + dml_print("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, TotalZeroSizeCompressedReadBandwidth); + dml_print("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, MaximumEffectiveCompressionLuma); + dml_print("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, MaximumEffectiveCompressionChroma); + dml_print("DML::%s: AverageDCCCompressionRate = %f\n", __func__, AverageDCCCompressionRate); + dml_print("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, AverageDCCZeroSizeFraction); + dml_print("DML::%s: CompbufReservedSpace64B = %u\n", __func__, p->CompbufReservedSpace64B); + dml_print("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); + dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, p->CompressedBufferSizeInkByte); +#endif + if (AverageDCCZeroSizeFraction == 1) { + AverageZeroSizeCompressionRate = TotalZeroSizeRequestReadBandwidth / TotalZeroSizeCompressedReadBandwidth; + EffectiveCompressedBufferSize = (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageZeroSizeCompressionRate + ((dml_float_t)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 * AverageZeroSizeCompressionRate; + } else if (AverageDCCZeroSizeFraction > 0) { + AverageZeroSizeCompressionRate = TotalZeroSizeRequestReadBandwidth / TotalZeroSizeCompressedReadBandwidth; + EffectiveCompressedBufferSize = dml_min((dml_float_t)p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate, + (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate + 1 / AverageDCCCompressionRate)) + + dml_min(((dml_float_t)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate, + ((dml_float_t)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate)); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate); + dml_print("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate + 1 / AverageDCCCompressionRate)); + dml_print("DML::%s: min 3 = %f\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate); + dml_print("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (AverageDCCZeroSizeFraction / AverageZeroSizeCompressionRate)); +#endif + } else { + EffectiveCompressedBufferSize = dml_min((dml_float_t)p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate, + (dml_float_t)p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageDCCCompressionRate) + + ((dml_float_t)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * AverageDCCCompressionRate; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * AverageDCCCompressionRate); + dml_print("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * AverageDCCCompressionRate); +#endif + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); + dml_print("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, AverageZeroSizeCompressionRate); + dml_print("DML::%s: EffectiveCompressedBufferSize = %f\n", __func__, EffectiveCompressedBufferSize); +#endif + + *p->StutterPeriod = 0; + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { + LinesInDETY = ((dml_float_t)p->DETBufferSizeY[k] + (p->UnboundedRequestEnabled == true ? EffectiveCompressedBufferSize : 0) * p->ReadBandwidthSurfaceLuma[k] / p->TotalDataReadBandwidth) / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; + LinesInDETYRoundedDownToSwath = dml_floor(LinesInDETY, p->SwathHeightY[k]); + DETBufferingTimeY = LinesInDETYRoundedDownToSwath * ((dml_float_t)p->HTotal[k] / p->PixelClock[k]) / p->VRatio[k]; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + dml_print("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + dml_print("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + dml_print("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); + dml_print("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, LinesInDETY); + dml_print("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, LinesInDETYRoundedDownToSwath); + dml_print("DML::%s: k=%u, HTotal = %u\n", __func__, k, p->HTotal[k]); + dml_print("DML::%s: k=%u, PixelClock = %f\n", __func__, k, p->PixelClock[k]); + dml_print("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->VRatio[k]); + dml_print("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, DETBufferingTimeY); + dml_print("DML::%s: k=%u,PixelClock = %f\n", __func__, k, p->PixelClock[k]); +#endif + + if (!FoundCriticalSurface || DETBufferingTimeY < *p->StutterPeriod) { + dml_bool_t isInterlaceTiming = p->Interlace[k] && !p->ProgressiveToInterlaceUnitInOPP; + + FoundCriticalSurface = true; + *p->StutterPeriod = DETBufferingTimeY; + FrameTimeCriticalSurface = (isInterlaceTiming ? dml_floor((dml_float_t)p->VTotal[k]/2.0, 1.0) : p->VTotal[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; + VActiveTimeCriticalSurface = (isInterlaceTiming ? dml_floor((dml_float_t)p->VActive[k]/2.0, 1.0) : p->VActive[k]) * (dml_float_t)p->HTotal[k] / p->PixelClock[k]; + BytePerPixelYCriticalSurface = p->BytePerPixelY[k]; + SwathWidthYCriticalSurface = p->SwathWidthY[k]; + SwathHeightYCriticalSurface = p->SwathHeightY[k]; + BlockWidth256BytesYCriticalSurface = p->BlockWidth256BytesY[k]; + LinesToFinishSwathTransferStutterCriticalSurface = p->SwathHeightY[k] - (LinesInDETY - LinesInDETYRoundedDownToSwath); + DETBufferSizeYCriticalSurface = p->DETBufferSizeY[k]; + MinTTUVBlankCriticalSurface = p->MinTTUVBlank[k]; + SinglePlaneCriticalSurface = (p->ReadBandwidthSurfaceChroma[k] == 0); + SinglePipeCriticalSurface = (p->DPPPerSurface[k] == 1); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); + dml_print("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); + dml_print("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, MinTTUVBlankCriticalSurface); + dml_print("DML::%s: k=%u, FrameTimeCriticalSurface = %f\n", __func__, k, FrameTimeCriticalSurface); + dml_print("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, VActiveTimeCriticalSurface); + dml_print("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, BytePerPixelYCriticalSurface); + dml_print("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, SwathWidthYCriticalSurface); + dml_print("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, SwathHeightYCriticalSurface); + dml_print("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, BlockWidth256BytesYCriticalSurface); + dml_print("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, SinglePlaneCriticalSurface); + dml_print("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, SinglePipeCriticalSurface); + dml_print("DML::%s: k=%u, LinesToFinishSwathTransferStutterCriticalSurface = %f\n", __func__, k, LinesToFinishSwathTransferStutterCriticalSurface); +#endif + } + } + } + + PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = dml_min(*p->StutterPeriod * p->TotalDataReadBandwidth, EffectiveCompressedBufferSize); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, p->ROBBufferSizeInKByte); + dml_print("DML::%s: AverageDCCCompressionRate = %f\n", __func__, AverageDCCCompressionRate); + dml_print("DML::%s: StutterPeriod * TotalDataReadBandwidth = %f\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth); + dml_print("DML::%s: ROBBufferSizeInKByte * 1024 * AverageDCCCompressionRate + EffectiveCompressedBufferSize = %f\n", __func__, p->ROBBufferSizeInKByte * 1024 * AverageDCCCompressionRate + EffectiveCompressedBufferSize); + dml_print("DML::%s: EffectiveCompressedBufferSize = %f\n", __func__, EffectiveCompressedBufferSize); + dml_print("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f\n", __func__, PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer); + dml_print("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); + dml_print("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); + dml_print("DML::%s: TotalRowReadBandwidth = %f\n", __func__, TotalRowReadBandwidth); + dml_print("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); +#endif + + StutterBurstTime = PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / AverageDCCCompressionRate / p->ReturnBW + (*p->StutterPeriod * p->TotalDataReadBandwidth - PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64) + *p->StutterPeriod * TotalRowReadBandwidth / p->ReturnBW; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Part 1 = %f\n", __func__, PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / AverageDCCCompressionRate / p->ReturnBW); + dml_print("DML::%s: StutterPeriod * TotalDataReadBandwidth = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth)); + dml_print("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); + dml_print("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * TotalRowReadBandwidth / p->ReturnBW); + dml_print("DML::%s: StutterBurstTime = %f\n", __func__, StutterBurstTime); +#endif + StutterBurstTime = dml_max(StutterBurstTime, LinesToFinishSwathTransferStutterCriticalSurface * BytePerPixelYCriticalSurface * SwathWidthYCriticalSurface / p->ReturnBW); + + dml_print("DML::%s: Time to finish residue swath=%f\n", __func__, LinesToFinishSwathTransferStutterCriticalSurface * BytePerPixelYCriticalSurface * SwathWidthYCriticalSurface / p->ReturnBW); + + TotalActiveWriteback = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->WritebackEnable[k]) { + TotalActiveWriteback = TotalActiveWriteback + 1; + } + } + + if (TotalActiveWriteback == 0) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); + dml_print("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); + dml_print("DML::%s: StutterBurstTime = %f (final)\n", __func__, StutterBurstTime); + dml_print("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); +#endif + *p->StutterEfficiencyNotIncludingVBlank = dml_max(0., 1 - (p->SRExitTime + StutterBurstTime) / *p->StutterPeriod) * 100; + *p->Z8StutterEfficiencyNotIncludingVBlank = dml_max(0., 1 - (p->SRExitZ8Time + StutterBurstTime) / *p->StutterPeriod) * 100; + *p->NumberOfStutterBurstsPerFrame = (*p->StutterEfficiencyNotIncludingVBlank > 0 ? (dml_uint_t)(dml_ceil(VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); + *p->Z8NumberOfStutterBurstsPerFrame = (*p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? (dml_uint_t)(dml_ceil(VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); + } else { + *p->StutterEfficiencyNotIncludingVBlank = 0.; + *p->Z8StutterEfficiencyNotIncludingVBlank = 0.; + *p->NumberOfStutterBurstsPerFrame = 0; + *p->Z8NumberOfStutterBurstsPerFrame = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, VActiveTimeCriticalSurface); + dml_print("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + dml_print("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); + dml_print("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); + dml_print("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); +#endif + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { + if (p->BlendingAndTiming[k] == k) { + if (TotalNumberOfActiveOTG == 0) { + SinglePixelClock = p->PixelClock[k]; + SingleHTotal = p->HTotal[k]; + SingleVTotal = p->VTotal[k]; + } else if (SinglePixelClock != p->PixelClock[k] || SingleHTotal != p->HTotal[k] || SingleVTotal != p->VTotal[k]) { + SameTiming = false; + } + TotalNumberOfActiveOTG = TotalNumberOfActiveOTG + 1; + } + } + } + + if (*p->StutterEfficiencyNotIncludingVBlank > 0) { + LastStutterPeriod = VActiveTimeCriticalSurface - (*p->NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; + + if ((p->SynchronizeTimingsFinal || TotalNumberOfActiveOTG == 1) && SameTiming && + LastStutterPeriod + MinTTUVBlankCriticalSurface > p->StutterEnterPlusExitWatermark) { + *p->StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitTime + StutterBurstTime * VActiveTimeCriticalSurface / *p->StutterPeriod) / FrameTimeCriticalSurface) * 100; + } else { + *p->StutterEfficiency = *p->StutterEfficiencyNotIncludingVBlank; + } + } else { + *p->StutterEfficiency = 0; + } + + if (*p->Z8StutterEfficiencyNotIncludingVBlank > 0) { + LastZ8StutterPeriod = VActiveTimeCriticalSurface - (*p->NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; + if ((p->SynchronizeTimingsFinal || TotalNumberOfActiveOTG == 1) && SameTiming && LastZ8StutterPeriod + MinTTUVBlankCriticalSurface > p->Z8StutterEnterPlusExitWatermark) { + *p->Z8StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitZ8Time + StutterBurstTime * VActiveTimeCriticalSurface / *p->StutterPeriod) / FrameTimeCriticalSurface) * 100; + } else { + *p->Z8StutterEfficiency = *p->Z8StutterEfficiencyNotIncludingVBlank; + } + } else { + *p->Z8StutterEfficiency = 0.; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: LastZ8StutterPeriod = %f\n", __func__, LastZ8StutterPeriod); + dml_print("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); + dml_print("DML::%s: StutterBurstTime = %f\n", __func__, StutterBurstTime); + dml_print("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); + dml_print("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); + dml_print("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); + dml_print("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + dml_print("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); +#endif + + SwathSizeCriticalSurface = (dml_uint_t)(BytePerPixelYCriticalSurface * SwathHeightYCriticalSurface * dml_ceil(SwathWidthYCriticalSurface, BlockWidth256BytesYCriticalSurface)); + LastChunkOfSwathSize = SwathSizeCriticalSurface % (p->PixelChunkSizeInKByte * 1024); + MissingPartOfLastSwathOfDETSize = (dml_uint_t)(dml_ceil(DETBufferSizeYCriticalSurface, SwathSizeCriticalSurface) - DETBufferSizeYCriticalSurface); + + *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = !(!p->UnboundedRequestEnabled && (p->NumberOfActiveSurfaces == 1) && SinglePlaneCriticalSurface && SinglePipeCriticalSurface && (LastChunkOfSwathSize > 0) && + (LastChunkOfSwathSize <= 4096) && (MissingPartOfLastSwathOfDETSize > 0) && (MissingPartOfLastSwathOfDETSize <= LastChunkOfSwathSize)); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: SwathSizeCriticalSurface = %u\n", __func__, SwathSizeCriticalSurface); + dml_print("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, DETBufferSizeYCriticalSurface); + dml_print("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); + dml_print("DML::%s: LastChunkOfSwathSize = %u\n", __func__, LastChunkOfSwathSize); + dml_print("DML::%s: MissingPartOfLastSwathOfDETSize = %u\n", __func__, MissingPartOfLastSwathOfDETSize); + dml_print("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); +#endif +} // CalculateStutterEfficiency + +/// \CalculateSwathAndDETConfiguration +/// @brief Calculates swath width and different return buffers sizing (DET, CDB, etc.) +static void CalculateSwathAndDETConfiguration(struct display_mode_lib_scratch_st *scratch, + struct CalculateSwathAndDETConfiguration_params_st *p) +{ + dml_uint_t MaximumSwathHeightY[__DML_NUM_PLANES__]; + dml_uint_t MaximumSwathHeightC[__DML_NUM_PLANES__]; + dml_uint_t RoundedUpMaxSwathSizeBytesY[__DML_NUM_PLANES__]; + dml_uint_t RoundedUpMaxSwathSizeBytesC[__DML_NUM_PLANES__]; + dml_uint_t RoundedUpSwathSizeBytesY[__DML_NUM_PLANES__]; + dml_uint_t RoundedUpSwathSizeBytesC[__DML_NUM_PLANES__]; + dml_uint_t SwathWidthSingleDPP[__DML_NUM_PLANES__]; + dml_uint_t SwathWidthSingleDPPChroma[__DML_NUM_PLANES__]; + + dml_uint_t TotalActiveDPP = 0; + dml_bool_t NoChromaOrLinearSurfaces = true; + dml_uint_t SurfaceDoingUnboundedRequest = 0; + + dml_uint_t DETBufferSizeInKByteForSwathCalculation; + + const long TTUFIFODEPTH = 8; + const long MAXIMUMCOMPRESSION = 4; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + dml_print("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); + } +#endif + CalculateSwathWidth(p->ForceSingleDPP, + p->NumberOfActiveSurfaces, + p->SourcePixelFormat, + p->SourceScan, + p->ViewportStationary, + p->ViewportWidth, + p->ViewportHeight, + p->ViewportXStart, + p->ViewportYStart, + p->ViewportXStartC, + p->ViewportYStartC, + p->SurfaceWidthY, + p->SurfaceWidthC, + p->SurfaceHeightY, + p->SurfaceHeightC, + p->ODMMode, + p->BytePerPixY, + p->BytePerPixC, + p->Read256BytesBlockHeightY, + p->Read256BytesBlockHeightC, + p->Read256BytesBlockWidthY, + p->Read256BytesBlockWidthC, + p->BlendingAndTiming, + p->HActive, + p->HRatio, + p->DPPPerSurface, + + // Output + SwathWidthSingleDPP, + SwathWidthSingleDPPChroma, + p->SwathWidth, + p->SwathWidthChroma, + MaximumSwathHeightY, + MaximumSwathHeightC, + p->swath_width_luma_ub, + p->swath_width_chroma_ub); + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + RoundedUpMaxSwathSizeBytesY[k] = (dml_uint_t)(p->swath_width_luma_ub[k] * p->BytePerPixDETY[k] * MaximumSwathHeightY[k]); + RoundedUpMaxSwathSizeBytesC[k] = (dml_uint_t)(p->swath_width_chroma_ub[k] * p->BytePerPixDETC[k] * MaximumSwathHeightC[k]); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); + dml_print("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); + dml_print("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); + dml_print("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); + dml_print("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); + dml_print("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); + dml_print("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); +#endif + if (p->SourcePixelFormat[k] == dml_420_10) { + RoundedUpMaxSwathSizeBytesY[k] = (dml_uint_t)(dml_ceil((dml_float_t) RoundedUpMaxSwathSizeBytesY[k], 256)); + RoundedUpMaxSwathSizeBytesC[k] = (dml_uint_t)(dml_ceil((dml_float_t) RoundedUpMaxSwathSizeBytesC[k], 256)); + } + } + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + TotalActiveDPP = TotalActiveDPP + (p->ForceSingleDPP ? 1 : p->DPPPerSurface[k]); + if (p->DPPPerSurface[k] > 0) + SurfaceDoingUnboundedRequest = k; + if (p->SourcePixelFormat[k] == dml_420_8 || p->SourcePixelFormat[k] == dml_420_10 || + p->SourcePixelFormat[k] == dml_420_12 || p->SourcePixelFormat[k] == dml_rgbe_alpha + || p->SurfaceTiling[k] == dml_sw_linear) { + NoChromaOrLinearSurfaces = false; + } + } + + *p->UnboundedRequestEnabled = UnboundedRequest(p->UseUnboundedRequestingFinal, TotalActiveDPP, + NoChromaOrLinearSurfaces, p->Output[0]); + + CalculateDETBufferSize(p->DETSizeOverride, + p->UseMALLForPStateChange, + p->ForceSingleDPP, + p->NumberOfActiveSurfaces, + *p->UnboundedRequestEnabled, + p->nomDETInKByte, + p->MaxTotalDETInKByte, + p->ConfigReturnBufferSizeInKByte, + p->MinCompressedBufferSizeInKByte, + p->ConfigReturnBufferSegmentSizeInkByte, + p->CompressedBufferSegmentSizeInkByteFinal, + p->SourcePixelFormat, + p->ReadBandwidthLuma, + p->ReadBandwidthChroma, + RoundedUpMaxSwathSizeBytesY, + RoundedUpMaxSwathSizeBytesC, + p->DPPPerSurface, + + // Output + p->DETBufferSizeInKByte, // per hubp pipe + p->CompressedBufferSizeInkByte); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); + dml_print("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); + dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); + dml_print("DML::%s: UseUnboundedRequestingFinal = %u\n", __func__, p->UseUnboundedRequestingFinal); + dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); + dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); +#endif + + *p->ViewportSizeSupport = true; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + + DETBufferSizeInKByteForSwathCalculation = (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe ? 1024 : p->DETBufferSizeInKByte[k]); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); +#endif + + if (RoundedUpMaxSwathSizeBytesY[k] + RoundedUpMaxSwathSizeBytesC[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k]; + p->SwathHeightC[k] = MaximumSwathHeightC[k]; + RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k]; + RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k]; + } else if (RoundedUpMaxSwathSizeBytesY[k] >= 1.5 * RoundedUpMaxSwathSizeBytesC[k] && RoundedUpMaxSwathSizeBytesY[k] / 2 + RoundedUpMaxSwathSizeBytesC[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; + p->SwathHeightC[k] = MaximumSwathHeightC[k]; + RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k] / 2; + RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k]; + } else if (RoundedUpMaxSwathSizeBytesY[k] < 1.5 * RoundedUpMaxSwathSizeBytesC[k] && RoundedUpMaxSwathSizeBytesY[k] + RoundedUpMaxSwathSizeBytesC[k] / 2 <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k]; + p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; + RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k]; + RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k] / 2; + } else { + p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; + p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; + RoundedUpSwathSizeBytesY[k] = RoundedUpMaxSwathSizeBytesY[k] / 2; + RoundedUpSwathSizeBytesC[k] = RoundedUpMaxSwathSizeBytesC[k] / 2; + } + + if ((RoundedUpMaxSwathSizeBytesY[k] / 2 + RoundedUpMaxSwathSizeBytesC[k] / 2 > DETBufferSizeInKByteForSwathCalculation * 1024 / 2) || + p->SwathWidth[k] > p->MaximumSwathWidthLuma[k] || (p->SwathHeightC[k] > 0 && p->SwathWidthChroma[k] > p->MaximumSwathWidthChroma[k])) { + *p->ViewportSizeSupport = false; + p->ViewportSizeSupportPerSurface[k] = false; + } else { + p->ViewportSizeSupportPerSurface[k] = true; + } + + if (p->SwathHeightC[k] == 0) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u All DET for plane0\n", __func__, k); +#endif + p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024; + p->DETBufferSizeC[k] = 0; + } else if (RoundedUpSwathSizeBytesY[k] <= 1.5 * RoundedUpSwathSizeBytesC[k]) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u Half DET for plane0, half for plane1\n", __func__, k); +#endif + p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; + p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; + } else { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u 2/3 DET for plane0, 1/3 for plane1\n", __func__, k); +#endif + p->DETBufferSizeY[k] = (dml_uint_t)(dml_floor(p->DETBufferSizeInKByte[k] * 1024 * 2 / 3, 1024)); + p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 - p->DETBufferSizeY[k]; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + dml_print("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); + dml_print("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); + dml_print("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); + dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); + dml_print("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + dml_print("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); + dml_print("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); +#endif + + } + + *p->compbuf_reserved_space_64b = 2 * p->PixelChunkSizeInKByte * 1024 / 64; + if (*p->UnboundedRequestEnabled) { + *p->compbuf_reserved_space_64b = dml_max(*p->compbuf_reserved_space_64b, + (dml_float_t)(p->ROBBufferSizeInKByte * 1024/64) + - (dml_float_t)(RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest] * TTUFIFODEPTH / MAXIMUMCOMPRESSION/64)); + } + *p->compbuf_reserved_space_zs = 2 * p->PixelChunkSizeInKByte * 1024 / 256; +} // CalculateSwathAndDETConfiguration + +static void CalculateSwathWidth( + dml_bool_t ForceSingleDPP, + dml_uint_t NumberOfActiveSurfaces, + enum dml_source_format_class SourcePixelFormat[], + enum dml_rotation_angle SourceScan[], + dml_bool_t ViewportStationary[], + dml_uint_t ViewportWidth[], + dml_uint_t ViewportHeight[], + dml_uint_t ViewportXStart[], + dml_uint_t ViewportYStart[], + dml_uint_t ViewportXStartC[], + dml_uint_t ViewportYStartC[], + dml_uint_t SurfaceWidthY[], + dml_uint_t SurfaceWidthC[], + dml_uint_t SurfaceHeightY[], + dml_uint_t SurfaceHeightC[], + enum dml_odm_mode ODMMode[], + dml_uint_t BytePerPixY[], + dml_uint_t BytePerPixC[], + dml_uint_t Read256BytesBlockHeightY[], + dml_uint_t Read256BytesBlockHeightC[], + dml_uint_t Read256BytesBlockWidthY[], + dml_uint_t Read256BytesBlockWidthC[], + dml_uint_t BlendingAndTiming[], + dml_uint_t HActive[], + dml_float_t HRatio[], + dml_uint_t DPPPerSurface[], + + // Output + dml_uint_t SwathWidthSingleDPPY[], + dml_uint_t SwathWidthSingleDPPC[], + dml_uint_t SwathWidthY[], // per-pipe + dml_uint_t SwathWidthC[], // per-pipe + dml_uint_t MaximumSwathHeightY[], + dml_uint_t MaximumSwathHeightC[], + dml_uint_t swath_width_luma_ub[], // per-pipe + dml_uint_t swath_width_chroma_ub[]) // per-pipe +{ + enum dml_odm_mode MainSurfaceODMMode; + dml_uint_t surface_width_ub_l; + dml_uint_t surface_height_ub_l; + dml_uint_t surface_width_ub_c = 0; + dml_uint_t surface_height_ub_c = 0; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); +#endif + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_vertical_rotation(SourceScan[k])) { + SwathWidthSingleDPPY[k] = ViewportWidth[k]; + } else { + SwathWidthSingleDPPY[k] = ViewportHeight[k]; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u ViewportWidth=%u\n", __func__, k, ViewportWidth[k]); + dml_print("DML::%s: k=%u ViewportHeight=%u\n", __func__, k, ViewportHeight[k]); + dml_print("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); +#endif + + MainSurfaceODMMode = ODMMode[k]; + for (dml_uint_t j = 0; j < NumberOfActiveSurfaces; ++j) { + if (BlendingAndTiming[k] == j) { + MainSurfaceODMMode = ODMMode[j]; + } + } + + if (ForceSingleDPP) { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + } else { + if (MainSurfaceODMMode == dml_odm_mode_combine_4to1) { + SwathWidthY[k] = (dml_uint_t)(dml_min(SwathWidthSingleDPPY[k], dml_round(HActive[k] / 4.0 * HRatio[k], true))); + } else if (MainSurfaceODMMode == dml_odm_mode_combine_2to1) { + SwathWidthY[k] = (dml_uint_t)(dml_min(SwathWidthSingleDPPY[k], dml_round(HActive[k] / 2.0 * HRatio[k], true))); + } else if (DPPPerSurface[k] == 2) { + SwathWidthY[k] = SwathWidthSingleDPPY[k] / 2; + } else { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + } + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u HActive=%u\n", __func__, k, HActive[k]); + dml_print("DML::%s: k=%u HRatio=%f\n", __func__, k, HRatio[k]); + dml_print("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); + dml_print("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); + dml_print("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); +#endif + + if (SourcePixelFormat[k] == dml_420_8 || SourcePixelFormat[k] == dml_420_10 || SourcePixelFormat[k] == dml_420_12) { + SwathWidthC[k] = SwathWidthY[k] / 2; + SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k] / 2; + } else { + SwathWidthC[k] = SwathWidthY[k]; + SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k]; + } + + if (ForceSingleDPP == true) { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + SwathWidthC[k] = SwathWidthSingleDPPC[k]; + } + + surface_width_ub_l = (dml_uint_t)dml_ceil(SurfaceWidthY[k], Read256BytesBlockWidthY[k]); + surface_height_ub_l = (dml_uint_t)dml_ceil(SurfaceHeightY[k], Read256BytesBlockHeightY[k]); + + if (!dml_is_vertical_rotation(SourceScan[k])) { + MaximumSwathHeightY[k] = Read256BytesBlockHeightY[k]; + MaximumSwathHeightC[k] = Read256BytesBlockHeightC[k]; + if (ViewportStationary[k] && DPPPerSurface[k] == 1) { + swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_l, dml_floor(ViewportXStart[k] + SwathWidthY[k] + Read256BytesBlockWidthY[k] - 1, Read256BytesBlockWidthY[k]) - dml_floor(ViewportXStart[k], Read256BytesBlockWidthY[k]))); + } else { + swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_l, dml_ceil(SwathWidthY[k] - 1, Read256BytesBlockWidthY[k]) + Read256BytesBlockWidthY[k])); + } + if (BytePerPixC[k] > 0) { + surface_width_ub_c = (dml_uint_t)dml_ceil(SurfaceWidthC[k], Read256BytesBlockWidthC[k]); + if (ViewportStationary[k] && DPPPerSurface[k] == 1) { + swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_c, dml_floor(ViewportXStartC[k] + SwathWidthC[k] + Read256BytesBlockWidthC[k] - 1, Read256BytesBlockWidthC[k]) - dml_floor(ViewportXStartC[k], Read256BytesBlockWidthC[k]))); + } else { + swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_width_ub_c, dml_ceil(SwathWidthC[k] - 1, Read256BytesBlockWidthC[k]) + Read256BytesBlockWidthC[k])); + } + } else { + swath_width_chroma_ub[k] = 0; + } + } else { + MaximumSwathHeightY[k] = Read256BytesBlockWidthY[k]; + MaximumSwathHeightC[k] = Read256BytesBlockWidthC[k]; + + if (ViewportStationary[k] && DPPPerSurface[k] == 1) { + swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_l, dml_floor(ViewportYStart[k] + SwathWidthY[k] + Read256BytesBlockHeightY[k] - 1, Read256BytesBlockHeightY[k]) - dml_floor(ViewportYStart[k], Read256BytesBlockHeightY[k]))); + } else { + swath_width_luma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_l, dml_ceil(SwathWidthY[k] - 1, Read256BytesBlockHeightY[k]) + Read256BytesBlockHeightY[k])); + } + if (BytePerPixC[k] > 0) { + surface_height_ub_c = (dml_uint_t)dml_ceil(SurfaceHeightC[k], Read256BytesBlockHeightC[k]); + if (ViewportStationary[k] && DPPPerSurface[k] == 1) { + swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_c, dml_floor(ViewportYStartC[k] + SwathWidthC[k] + Read256BytesBlockHeightC[k] - 1, Read256BytesBlockHeightC[k]) - dml_floor(ViewportYStartC[k], Read256BytesBlockHeightC[k]))); + } else { + swath_width_chroma_ub[k] = (dml_uint_t)(dml_min(surface_height_ub_c, dml_ceil(SwathWidthC[k] - 1, Read256BytesBlockHeightC[k]) + Read256BytesBlockHeightC[k])); + } + } else { + swath_width_chroma_ub[k] = 0; + } + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); + dml_print("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); + dml_print("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); + dml_print("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); + dml_print("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); + dml_print("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); + dml_print("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); + dml_print("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); + dml_print("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, ViewportStationary[k]); + dml_print("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); + dml_print("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); + dml_print("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); + dml_print("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); + dml_print("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); +#endif + + } +} // CalculateSwathWidth + +static noinline_for_stack dml_float_t CalculateExtraLatency( + dml_uint_t RoundTripPingLatencyCycles, + dml_uint_t ReorderingBytes, + dml_float_t DCFCLK, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t PixelChunkSizeInKByte, + dml_uint_t TotalNumberOfDCCActiveDPP, + dml_uint_t MetaChunkSize, + dml_float_t ReturnBW, + dml_bool_t GPUVMEnable, + dml_bool_t HostVMEnable, + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t NumberOfDPP[], + dml_uint_t dpte_group_bytes[], + dml_float_t HostVMInefficiencyFactor, + dml_uint_t HostVMMinPageSize, + dml_uint_t HostVMMaxNonCachedPageTableLevels) +{ + dml_float_t ExtraLatencyBytes; + dml_float_t ExtraLatency; + + ExtraLatencyBytes = CalculateExtraLatencyBytes( + ReorderingBytes, + TotalNumberOfActiveDPP, + PixelChunkSizeInKByte, + TotalNumberOfDCCActiveDPP, + MetaChunkSize, + GPUVMEnable, + HostVMEnable, + NumberOfActiveSurfaces, + NumberOfDPP, + dpte_group_bytes, + HostVMInefficiencyFactor, + HostVMMinPageSize, + HostVMMaxNonCachedPageTableLevels); + + ExtraLatency = (RoundTripPingLatencyCycles + __DML_ARB_TO_RET_DELAY__) / DCFCLK + ExtraLatencyBytes / ReturnBW; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); + dml_print("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); + dml_print("DML::%s: ExtraLatencyBytes=%f\n", __func__, ExtraLatencyBytes); + dml_print("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); + dml_print("DML::%s: ExtraLatency=%f\n", __func__, ExtraLatency); +#endif + + return ExtraLatency; +} // CalculateExtraLatency + +static dml_uint_t CalculateHostVMDynamicLevels( + dml_bool_t GPUVMEnable, + dml_bool_t HostVMEnable, + dml_uint_t HostVMMinPageSize, + dml_uint_t HostVMMaxNonCachedPageTableLevels) +{ + dml_uint_t HostVMDynamicLevels = 0; + + if (GPUVMEnable && HostVMEnable) { + if (HostVMMinPageSize < 2048) + HostVMDynamicLevels = HostVMMaxNonCachedPageTableLevels; + else if (HostVMMinPageSize >= 2048 && HostVMMinPageSize < 1048576) + HostVMDynamicLevels = (dml_uint_t) dml_max(0, (dml_float_t) HostVMMaxNonCachedPageTableLevels - 1); + else + HostVMDynamicLevels = (dml_uint_t) dml_max(0, (dml_float_t) HostVMMaxNonCachedPageTableLevels - 2); + } else { + HostVMDynamicLevels = 0; + } + return HostVMDynamicLevels; +} + +static dml_uint_t CalculateExtraLatencyBytes(dml_uint_t ReorderingBytes, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t PixelChunkSizeInKByte, + dml_uint_t TotalNumberOfDCCActiveDPP, + dml_uint_t MetaChunkSize, + dml_bool_t GPUVMEnable, + dml_bool_t HostVMEnable, + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t NumberOfDPP[], + dml_uint_t dpte_group_bytes[], + dml_float_t HostVMInefficiencyFactor, + dml_uint_t HostVMMinPageSize, + dml_uint_t HostVMMaxNonCachedPageTableLevels) +{ + dml_uint_t HostVMDynamicLevels = CalculateHostVMDynamicLevels(GPUVMEnable, HostVMEnable, HostVMMinPageSize, HostVMMaxNonCachedPageTableLevels); + dml_float_t ret = ReorderingBytes + (TotalNumberOfActiveDPP * PixelChunkSizeInKByte + TotalNumberOfDCCActiveDPP * MetaChunkSize) * 1024.0; + + if (GPUVMEnable == true) { + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + ret = ret + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactor; + } + } + return (dml_uint_t)(ret); +} + +static dml_float_t CalculateUrgentLatency( + dml_float_t UrgentLatencyPixelDataOnly, + dml_float_t UrgentLatencyPixelMixedWithVMData, + dml_float_t UrgentLatencyVMDataOnly, + dml_bool_t DoUrgentLatencyAdjustment, + dml_float_t UrgentLatencyAdjustmentFabricClockComponent, + dml_float_t UrgentLatencyAdjustmentFabricClockReference, + dml_float_t FabricClock) +{ + dml_float_t ret; + + ret = dml_max3(UrgentLatencyPixelDataOnly, UrgentLatencyPixelMixedWithVMData, UrgentLatencyVMDataOnly); + if (DoUrgentLatencyAdjustment == true) { + ret = ret + UrgentLatencyAdjustmentFabricClockComponent * (UrgentLatencyAdjustmentFabricClockReference / FabricClock - 1); + } + return ret; +} + +static dml_float_t RequiredDTBCLK( + dml_bool_t DSCEnable, + dml_float_t PixelClock, + enum dml_output_format_class OutputFormat, + dml_float_t OutputBpp, + dml_uint_t DSCSlices, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_uint_t AudioRate, + dml_uint_t AudioLayout) +{ + if (DSCEnable != true) { + return dml_max(PixelClock / 4.0 * OutputBpp / 24.0, 25.0); + } else { + dml_float_t PixelWordRate = PixelClock / (OutputFormat == dml_444 ? 1 : 2); + dml_float_t HCActive = dml_ceil(DSCSlices * dml_ceil(OutputBpp * dml_ceil(HActive / DSCSlices, 1) / 8.0, 1) / 3.0, 1); + dml_float_t HCBlank = 64 + 32 * dml_ceil(AudioRate * (AudioLayout == 1 ? 1 : 0.25) * HTotal / (PixelClock * 1000), 1); + dml_float_t AverageTribyteRate = PixelWordRate * (HCActive + HCBlank) / HTotal; + dml_float_t HActiveTribyteRate = PixelWordRate * HCActive / HActive; + return dml_max4(PixelWordRate / 4.0, AverageTribyteRate / 4.0, HActiveTribyteRate / 4.0, 25.0) * 1.002; + } +} + +static void UseMinimumDCFCLK(struct display_mode_lib_scratch_st *scratch, struct UseMinimumDCFCLK_params_st *p) +{ + struct UseMinimumDCFCLK_locals_st *s = &scratch->UseMinimumDCFCLK_locals; + + s->NormalEfficiency = p->PercentOfIdealSDPPortBWReceivedAfterUrgLatency / 100.0; + for (dml_uint_t j = 0; j < 2; ++j) { + + + s->TotalMaxPrefetchFlipDPTERowBandwidth[j] = 0; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + s->TotalMaxPrefetchFlipDPTERowBandwidth[j] = s->TotalMaxPrefetchFlipDPTERowBandwidth[j] + p->NoOfDPP[j][k] * p->DPTEBytesPerRow[j][k] / (15.75 * p->HTotal[k] / p->PixelClock[k]); + } + + for (dml_uint_t k = 0; k <= p->NumberOfActiveSurfaces - 1; ++k) { + s->NoOfDPPState[k] = p->NoOfDPP[j][k]; + } + + s->DPTEBandwidth = s->TotalMaxPrefetchFlipDPTERowBandwidth[j]; + + s->DCFCLKRequiredForAverageBandwidth = dml_max(p->ProjectedDCFCLKDeepSleep[j], s->DPTEBandwidth / s->NormalEfficiency / p->ReturnBusWidth); + + s->ExtraLatencyBytes = CalculateExtraLatencyBytes(p->ReorderingBytes, p->TotalNumberOfActiveDPP[j], p->PixelChunkSizeInKByte, p->TotalNumberOfDCCActiveDPP[j], + p->MetaChunkSize, p->GPUVMEnable, p->HostVMEnable, p->NumberOfActiveSurfaces, s->NoOfDPPState, p->dpte_group_bytes, + 1, p->HostVMMinPageSize, p->HostVMMaxNonCachedPageTableLevels); + s->ExtraLatencyCycles = p->RoundTripPingLatencyCycles + __DML_ARB_TO_RET_DELAY__ + s->ExtraLatencyBytes / s->NormalEfficiency / p->ReturnBusWidth; + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + dml_float_t DCFCLKCyclesRequiredInPrefetch; + dml_float_t PrefetchTime; + + s->PixelDCFCLKCyclesRequiredInPrefetch[k] = (p->PrefetchLinesY[j][k] * p->swath_width_luma_ub_all_states[j][k] * p->BytePerPixelY[k] + p->PrefetchLinesC[j][k] * p->swath_width_chroma_ub_all_states[j][k] * p->BytePerPixelC[k]) / s->NormalEfficiency / p->ReturnBusWidth; + DCFCLKCyclesRequiredInPrefetch = 2 * s->ExtraLatencyCycles / s->NoOfDPPState[k] + p->PDEAndMetaPTEBytesPerFrame[j][k] / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth * (p->GPUVMMaxPageTableLevels > 2 ? 1 : 0) + 2 * p->DPTEBytesPerRow[j][k] / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth + 2 * p->MetaRowBytes[j][k] / s->NormalEfficiency / p->ReturnBusWidth + s->PixelDCFCLKCyclesRequiredInPrefetch[k]; + s->PrefetchPixelLinesTime[k] = dml_max(p->PrefetchLinesY[j][k], p->PrefetchLinesC[j][k]) * p->HTotal[k] / p->PixelClock[k]; + s->DynamicMetadataVMExtraLatency[k] = (p->GPUVMEnable == true && p->DynamicMetadataEnable[k] == true && p->DynamicMetadataVMEnabled == true) ? p->UrgLatency * p->GPUVMMaxPageTableLevels * (p->HostVMEnable == true ? p->HostVMMaxNonCachedPageTableLevels + 1 : 1) : 0; + + s->MinimumTWait = CalculateTWait(p->MaxPrefetchMode, + p->UseMALLForPStateChange[k], + p->SynchronizeDRRDisplaysForUCLKPStateChangeFinal, + p->DRRDisplay[k], + p->DRAMClockChangeLatencyFinal, + p->FCLKChangeLatency, + p->UrgLatency, + p->SREnterPlusExitTime); + + PrefetchTime = (p->MaximumVStartup[j][k] - 1) * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - p->UrgLatency * ((p->GPUVMMaxPageTableLevels <= 2 ? p->GPUVMMaxPageTableLevels : p->GPUVMMaxPageTableLevels - 2) * (p->HostVMEnable == true ? p->HostVMMaxNonCachedPageTableLevels + 1 : 1) - 1) - s->DynamicMetadataVMExtraLatency[k]; + + if (PrefetchTime > 0) { + dml_float_t ExpectedVRatioPrefetch; + ExpectedVRatioPrefetch = s->PrefetchPixelLinesTime[k] / (PrefetchTime * s->PixelDCFCLKCyclesRequiredInPrefetch[k] / DCFCLKCyclesRequiredInPrefetch); + s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = s->NoOfDPPState[k] * s->PixelDCFCLKCyclesRequiredInPrefetch[k] / s->PrefetchPixelLinesTime[k] * dml_max(1.0, ExpectedVRatioPrefetch) * dml_max(1.0, ExpectedVRatioPrefetch / 4); + if (p->HostVMEnable == true || p->ImmediateFlipRequirement == true) { + s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = s->DCFCLKRequiredForPeakBandwidthPerSurface[k] + s->NoOfDPPState[k] * s->DPTEBandwidth / s->NormalEfficiency / s->NormalEfficiency / p->ReturnBusWidth; + } + } else { + s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = p->DCFCLKPerState; + } + if (p->DynamicMetadataEnable[k] == true) { + dml_float_t TSetupPipe; + dml_float_t TdmbfPipe; + dml_float_t TdmsksPipe; + dml_float_t TdmecPipe; + dml_float_t AllowedTimeForUrgentExtraLatency; + + CalculateVUpdateAndDynamicMetadataParameters( + p->MaxInterDCNTileRepeaters, + p->RequiredDPPCLKPerSurface[j][k], + p->RequiredDISPCLK[j], + p->ProjectedDCFCLKDeepSleep[j], + p->PixelClock[k], + p->HTotal[k], + p->VTotal[k] - p->VActive[k], + p->DynamicMetadataTransmittedBytes[k], + p->DynamicMetadataLinesBeforeActiveRequired[k], + p->Interlace[k], + p->ProgressiveToInterlaceUnitInOPP, + + // Output + &TSetupPipe, + &TdmbfPipe, + &TdmecPipe, + &TdmsksPipe, + &s->dummy1, + &s->dummy2, + &s->dummy3); + + AllowedTimeForUrgentExtraLatency = p->MaximumVStartup[j][k] * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - TSetupPipe - TdmbfPipe - TdmecPipe - TdmsksPipe - s->DynamicMetadataVMExtraLatency[k]; + if (AllowedTimeForUrgentExtraLatency > 0) { + s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = dml_max(s->DCFCLKRequiredForPeakBandwidthPerSurface[k], s->ExtraLatencyCycles / AllowedTimeForUrgentExtraLatency); + } else { + s->DCFCLKRequiredForPeakBandwidthPerSurface[k] = p->DCFCLKPerState; + } + } + } + s->DCFCLKRequiredForPeakBandwidth = 0; + for (dml_uint_t k = 0; k <= p->NumberOfActiveSurfaces - 1; ++k) { + s->DCFCLKRequiredForPeakBandwidth = s->DCFCLKRequiredForPeakBandwidth + s->DCFCLKRequiredForPeakBandwidthPerSurface[k]; + } + s->MinimumTvmPlus2Tr0 = p->UrgLatency * (p->GPUVMEnable == true ? (p->HostVMEnable == true ? (p->GPUVMMaxPageTableLevels + 2) * (p->HostVMMaxNonCachedPageTableLevels + 1) - 1 : p->GPUVMMaxPageTableLevels + 1) : 0); + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + dml_float_t MaximumTvmPlus2Tr0PlusTsw; + MaximumTvmPlus2Tr0PlusTsw = (p->MaximumVStartup[j][k] - 2) * p->HTotal[k] / p->PixelClock[k] - s->MinimumTWait - s->DynamicMetadataVMExtraLatency[k]; + if (MaximumTvmPlus2Tr0PlusTsw <= s->MinimumTvmPlus2Tr0 + s->PrefetchPixelLinesTime[k] / 4) { + s->DCFCLKRequiredForPeakBandwidth = p->DCFCLKPerState; + } else { + s->DCFCLKRequiredForPeakBandwidth = dml_max3(s->DCFCLKRequiredForPeakBandwidth, + 2 * s->ExtraLatencyCycles / (MaximumTvmPlus2Tr0PlusTsw - s->MinimumTvmPlus2Tr0 - s->PrefetchPixelLinesTime[k] / 4), + (2 * s->ExtraLatencyCycles + s->PixelDCFCLKCyclesRequiredInPrefetch[k]) / (MaximumTvmPlus2Tr0PlusTsw - s->MinimumTvmPlus2Tr0)); + } + } + p->DCFCLKState[j] = dml_min(p->DCFCLKPerState, 1.05 * dml_max(s->DCFCLKRequiredForAverageBandwidth, s->DCFCLKRequiredForPeakBandwidth)); + } +} + + +static dml_bool_t UnboundedRequest(enum dml_unbounded_requesting_policy UseUnboundedRequestingFinal, + dml_uint_t TotalNumberOfActiveDPP, + dml_bool_t NoChromaOrLinear, + enum dml_output_encoder_class Output) +{ + dml_bool_t ret_val = false; + + ret_val = (UseUnboundedRequestingFinal != dml_unbounded_requesting_disable + && TotalNumberOfActiveDPP == 1 && NoChromaOrLinear); + if (UseUnboundedRequestingFinal == dml_unbounded_requesting_edp_only && Output != dml_edp) { + ret_val = false; + } + return (ret_val); +} + +static void CalculateSurfaceSizeInMall( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t MALLAllocatedForDCN, + enum dml_use_mall_for_static_screen_mode UseMALLForStaticScreen[], + dml_bool_t DCCEnable[], + dml_bool_t ViewportStationary[], + dml_uint_t ViewportXStartY[], + dml_uint_t ViewportYStartY[], + dml_uint_t ViewportXStartC[], + dml_uint_t ViewportYStartC[], + dml_uint_t ViewportWidthY[], + dml_uint_t ViewportHeightY[], + dml_uint_t BytesPerPixelY[], + dml_uint_t ViewportWidthC[], + dml_uint_t ViewportHeightC[], + dml_uint_t BytesPerPixelC[], + dml_uint_t SurfaceWidthY[], + dml_uint_t SurfaceWidthC[], + dml_uint_t SurfaceHeightY[], + dml_uint_t SurfaceHeightC[], + dml_uint_t Read256BytesBlockWidthY[], + dml_uint_t Read256BytesBlockWidthC[], + dml_uint_t Read256BytesBlockHeightY[], + dml_uint_t Read256BytesBlockHeightC[], + dml_uint_t ReadBlockWidthY[], + dml_uint_t ReadBlockWidthC[], + dml_uint_t ReadBlockHeightY[], + dml_uint_t ReadBlockHeightC[], + + // Output + dml_uint_t SurfaceSizeInMALL[], + dml_bool_t *ExceededMALLSize) +{ + dml_uint_t TotalSurfaceSizeInMALL = 0; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (ViewportStationary[k]) { + SurfaceSizeInMALL[k] = (dml_uint_t)(dml_min(dml_ceil(SurfaceWidthY[k], ReadBlockWidthY[k]), dml_floor(ViewportXStartY[k] + ViewportWidthY[k] + ReadBlockWidthY[k] - 1, ReadBlockWidthY[k]) - dml_floor(ViewportXStartY[k], ReadBlockWidthY[k])) * + dml_min(dml_ceil(SurfaceHeightY[k], ReadBlockHeightY[k]), dml_floor(ViewportYStartY[k] + ViewportHeightY[k] + ReadBlockHeightY[k] - 1, ReadBlockHeightY[k]) - dml_floor(ViewportYStartY[k], ReadBlockHeightY[k])) * + BytesPerPixelY[k]); + + if (ReadBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_min(dml_ceil(SurfaceWidthC[k], ReadBlockWidthC[k]), dml_floor(ViewportXStartC[k] + ViewportWidthC[k] + ReadBlockWidthC[k] - 1, ReadBlockWidthC[k]) - dml_floor(ViewportXStartC[k], ReadBlockWidthC[k])) * + dml_min(dml_ceil(SurfaceHeightC[k], ReadBlockHeightC[k]), dml_floor(ViewportYStartC[k] + ViewportHeightC[k] + ReadBlockHeightC[k] - 1, ReadBlockHeightC[k]) - dml_floor(ViewportYStartC[k], ReadBlockHeightC[k])) * BytesPerPixelC[k]); + } + if (DCCEnable[k] == true) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_min(dml_ceil(SurfaceWidthY[k], 8 * Read256BytesBlockWidthY[k]), dml_floor(ViewportXStartY[k] + ViewportWidthY[k] + 8 * Read256BytesBlockWidthY[k] - 1, 8 * Read256BytesBlockWidthY[k]) - dml_floor(ViewportXStartY[k], 8 * Read256BytesBlockWidthY[k])) * + dml_min(dml_ceil(SurfaceHeightY[k], 8 * Read256BytesBlockHeightY[k]), dml_floor(ViewportYStartY[k] + ViewportHeightY[k] + 8 * Read256BytesBlockHeightY[k] - 1, 8 * Read256BytesBlockHeightY[k]) - dml_floor(ViewportYStartY[k], 8 * Read256BytesBlockHeightY[k])) * BytesPerPixelY[k] / 256); + if (Read256BytesBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_min(dml_ceil(SurfaceWidthC[k], 8 * Read256BytesBlockWidthC[k]), dml_floor(ViewportXStartC[k] + ViewportWidthC[k] + 8 * Read256BytesBlockWidthC[k] - 1, 8 * Read256BytesBlockWidthC[k]) - dml_floor(ViewportXStartC[k], 8 * Read256BytesBlockWidthC[k])) * + dml_min(dml_ceil(SurfaceHeightC[k], 8 * Read256BytesBlockHeightC[k]), dml_floor(ViewportYStartC[k] + ViewportHeightC[k] + 8 * Read256BytesBlockHeightC[k] - 1, 8 * Read256BytesBlockHeightC[k]) - dml_floor(ViewportYStartC[k], 8 * Read256BytesBlockHeightC[k])) * BytesPerPixelC[k] / 256); + } + } + } else { + SurfaceSizeInMALL[k] = (dml_uint_t)(dml_ceil(dml_min(SurfaceWidthY[k], ViewportWidthY[k] + ReadBlockWidthY[k] - 1), ReadBlockWidthY[k]) * dml_ceil(dml_min(SurfaceHeightY[k], ViewportHeightY[k] + ReadBlockHeightY[k] - 1), ReadBlockHeightY[k]) * BytesPerPixelY[k]); + if (ReadBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_ceil(dml_min(SurfaceWidthC[k], ViewportWidthC[k] + ReadBlockWidthC[k] - 1), ReadBlockWidthC[k]) * + dml_ceil(dml_min(SurfaceHeightC[k], ViewportHeightC[k] + ReadBlockHeightC[k] - 1), ReadBlockHeightC[k]) * BytesPerPixelC[k]); + } + if (DCCEnable[k] == true) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_ceil(dml_min(SurfaceWidthY[k], ViewportWidthY[k] + 8 * Read256BytesBlockWidthY[k] - 1), 8 * Read256BytesBlockWidthY[k]) * + dml_ceil(dml_min(SurfaceHeightY[k], ViewportHeightY[k] + 8 * Read256BytesBlockHeightY[k] - 1), 8 * Read256BytesBlockHeightY[k]) * BytesPerPixelY[k] / 256); + + if (Read256BytesBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (dml_uint_t)(SurfaceSizeInMALL[k] + + dml_ceil(dml_min(SurfaceWidthC[k], ViewportWidthC[k] + 8 * Read256BytesBlockWidthC[k] - 1), 8 * Read256BytesBlockWidthC[k]) * + dml_ceil(dml_min(SurfaceHeightC[k], ViewportHeightC[k] + 8 * Read256BytesBlockHeightC[k] - 1), 8 * Read256BytesBlockHeightC[k]) * BytesPerPixelC[k] / 256); + } + } + } + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable) + TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; + } + *ExceededMALLSize = (TotalSurfaceSizeInMALL > MALLAllocatedForDCN * 1024 * 1024); +} // CalculateSurfaceSizeInMall + +static void CalculateDETBufferSize( + dml_uint_t DETSizeOverride[], + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + dml_bool_t ForceSingleDPP, + dml_uint_t NumberOfActiveSurfaces, + dml_bool_t UnboundedRequestEnabled, + dml_uint_t nomDETInKByte, + dml_uint_t MaxTotalDETInKByte, + dml_uint_t ConfigReturnBufferSizeInKByte, + dml_uint_t MinCompressedBufferSizeInKByte, + dml_uint_t ConfigReturnBufferSegmentSizeInkByte, + dml_uint_t CompressedBufferSegmentSizeInkByteFinal, + enum dml_source_format_class SourcePixelFormat[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_uint_t RoundedUpMaxSwathSizeBytesY[], + dml_uint_t RoundedUpMaxSwathSizeBytesC[], + dml_uint_t DPPPerSurface[], + // Output + dml_uint_t DETBufferSizeInKByte[], + dml_uint_t *CompressedBufferSizeInkByte) +{ + dml_uint_t DETBufferSizePoolInKByte; + dml_uint_t NextDETBufferPieceInKByte; + dml_bool_t DETPieceAssignedToThisSurfaceAlready[__DML_NUM_PLANES__]; + dml_bool_t NextPotentialSurfaceToAssignDETPieceFound; + dml_uint_t NextSurfaceToAssignDETPiece; + dml_float_t TotalBandwidth; + dml_float_t BandwidthOfSurfacesNotAssignedDETPiece; + dml_uint_t max_minDET; + dml_uint_t minDET; + dml_uint_t minDET_pipe; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + dml_print("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + dml_print("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); + dml_print("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); + dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + dml_print("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); + dml_print("DML::%s: CompressedBufferSegmentSizeInkByteFinal = %u\n", __func__, CompressedBufferSegmentSizeInkByteFinal); +#endif + + // Note: Will use default det size if that fits 2 swaths + if (UnboundedRequestEnabled) { + if (DETSizeOverride[0] > 0) { + DETBufferSizeInKByte[0] = DETSizeOverride[0]; + } else { + DETBufferSizeInKByte[0] = (dml_uint_t) dml_max(128.0, dml_ceil(2.0 * ((dml_float_t) RoundedUpMaxSwathSizeBytesY[0] + (dml_float_t) RoundedUpMaxSwathSizeBytesC[0]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte)); + } + *CompressedBufferSizeInkByte = ConfigReturnBufferSizeInKByte - DETBufferSizeInKByte[0]; + } else { + DETBufferSizePoolInKByte = MaxTotalDETInKByte; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + DETBufferSizeInKByte[k] = 0; + if (SourcePixelFormat[k] == dml_420_8 || SourcePixelFormat[k] == dml_420_10 || SourcePixelFormat[k] == dml_420_12) { + max_minDET = nomDETInKByte - ConfigReturnBufferSegmentSizeInkByte; + } else { + max_minDET = nomDETInKByte; + } + minDET = 128; + minDET_pipe = 0; + + // add DET resource until can hold 2 full swaths + while (minDET <= max_minDET && minDET_pipe == 0) { + if (2.0 * ((dml_float_t) RoundedUpMaxSwathSizeBytesY[k] + (dml_float_t) RoundedUpMaxSwathSizeBytesC[k]) / 1024.0 <= minDET) + minDET_pipe = minDET; + minDET = minDET + ConfigReturnBufferSegmentSizeInkByte; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u minDET = %u\n", __func__, k, minDET); + dml_print("DML::%s: k=%u max_minDET = %u\n", __func__, k, max_minDET); + dml_print("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, minDET_pipe); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesY = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesY[k]); + dml_print("DML::%s: k=%u RoundedUpMaxSwathSizeBytesC = %u\n", __func__, k, RoundedUpMaxSwathSizeBytesC[k]); +#endif + + if (minDET_pipe == 0) { + minDET_pipe = (dml_uint_t)(dml_max(128, dml_ceil(((dml_float_t)RoundedUpMaxSwathSizeBytesY[k] + (dml_float_t)RoundedUpMaxSwathSizeBytesC[k]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte))); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, minDET_pipe); +#endif + } + + if (UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) { + DETBufferSizeInKByte[k] = 0; + } else if (DETSizeOverride[k] > 0) { + DETBufferSizeInKByte[k] = DETSizeOverride[k]; + DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * DETSizeOverride[k]; + } else if ((ForceSingleDPP ? 1 : DPPPerSurface[k]) * minDET_pipe <= DETBufferSizePoolInKByte) { + DETBufferSizeInKByte[k] = minDET_pipe; + DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * minDET_pipe; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); + dml_print("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, DETSizeOverride[k]); + dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + dml_print("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, DETBufferSizePoolInKByte); +#endif + } + + TotalBandwidth = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) + TotalBandwidth = TotalBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: --- Before bandwidth adjustment ---\n", __func__); + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + } + dml_print("DML::%s: --- DET allocation with bandwidth ---\n", __func__); +#endif + dml_print("DML::%s: TotalBandwidth = %f\n", __func__, TotalBandwidth); + BandwidthOfSurfacesNotAssignedDETPiece = TotalBandwidth; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + + if (UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) { + DETPieceAssignedToThisSurfaceAlready[k] = true; + } else if (DETSizeOverride[k] > 0 || (((dml_float_t) (ForceSingleDPP ? 1 : DPPPerSurface[k]) * (dml_float_t) DETBufferSizeInKByte[k] / (dml_float_t) MaxTotalDETInKByte) >= ((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / TotalBandwidth))) { + DETPieceAssignedToThisSurfaceAlready[k] = true; + BandwidthOfSurfacesNotAssignedDETPiece = BandwidthOfSurfacesNotAssignedDETPiece - ReadBandwidthLuma[k] - ReadBandwidthChroma[k]; + } else { + DETPieceAssignedToThisSurfaceAlready[k] = false; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); + dml_print("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, BandwidthOfSurfacesNotAssignedDETPiece); +#endif + } + + for (dml_uint_t j = 0; j < NumberOfActiveSurfaces; ++j) { + NextPotentialSurfaceToAssignDETPieceFound = false; + NextSurfaceToAssignDETPiece = 0; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); + dml_print("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); + dml_print("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[NextSurfaceToAssignDETPiece]); + dml_print("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); + dml_print("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, NextSurfaceToAssignDETPiece); +#endif + if (!DETPieceAssignedToThisSurfaceAlready[k] && (!NextPotentialSurfaceToAssignDETPieceFound || + ReadBandwidthLuma[k] + ReadBandwidthChroma[k] < ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece])) { + NextSurfaceToAssignDETPiece = k; + NextPotentialSurfaceToAssignDETPieceFound = true; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); + dml_print("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); +#endif + } + + if (NextPotentialSurfaceToAssignDETPieceFound) { + // Note: To show the banker's rounding behavior in VBA and also the fact that the DET buffer size varies due to precision issue + // + //dml_float_t tmp1 = ((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / + // ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * 64.0)); + //dml_float_t tmp2 = dml_round((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / + // ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * 64.0)); + // + //dml_print("DML::%s: j=%u, tmp1 = %f\n", __func__, j, tmp1); + //dml_print("DML::%s: j=%u, tmp2 = %f\n", __func__, j, tmp2); + + NextDETBufferPieceInKByte = (dml_uint_t)(dml_min( + dml_round((dml_float_t) DETBufferSizePoolInKByte * (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]) / BandwidthOfSurfacesNotAssignedDETPiece / + ((ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte), true) + * (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte, + dml_floor((dml_float_t) DETBufferSizePoolInKByte, (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte))); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, DETBufferSizePoolInKByte); + dml_print("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, NextSurfaceToAssignDETPiece); + dml_print("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, NextSurfaceToAssignDETPiece, ReadBandwidthLuma[NextSurfaceToAssignDETPiece]); + dml_print("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, NextSurfaceToAssignDETPiece, ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); + dml_print("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, BandwidthOfSurfacesNotAssignedDETPiece); + dml_print("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, NextDETBufferPieceInKByte); + dml_print("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[NextSurfaceToAssignDETPiece]); +#endif + + DETBufferSizeInKByte[NextSurfaceToAssignDETPiece] = DETBufferSizeInKByte[NextSurfaceToAssignDETPiece] + NextDETBufferPieceInKByte / (ForceSingleDPP ? 1 : DPPPerSurface[NextSurfaceToAssignDETPiece]); +#ifdef __DML_VBA_DEBUG__ + dml_print("to %u\n", DETBufferSizeInKByte[NextSurfaceToAssignDETPiece]); +#endif + + DETBufferSizePoolInKByte = DETBufferSizePoolInKByte - NextDETBufferPieceInKByte; + DETPieceAssignedToThisSurfaceAlready[NextSurfaceToAssignDETPiece] = true; + BandwidthOfSurfacesNotAssignedDETPiece = BandwidthOfSurfacesNotAssignedDETPiece - (ReadBandwidthLuma[NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[NextSurfaceToAssignDETPiece]); + } + } + *CompressedBufferSizeInkByte = MinCompressedBufferSizeInKByte; + } + *CompressedBufferSizeInkByte = *CompressedBufferSizeInkByte * CompressedBufferSegmentSizeInkByteFinal / ConfigReturnBufferSegmentSizeInkByte; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: --- After bandwidth adjustment ---\n", __func__); + dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + dml_print("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); + } +#endif +} // CalculateDETBufferSize + + +/// @brief Calculate the bound for return buffer sizing +static void CalculateMaxDETAndMinCompressedBufferSize( + dml_uint_t ConfigReturnBufferSizeInKByte, + dml_uint_t ConfigReturnBufferSegmentSizeInKByte, + dml_uint_t ROBBufferSizeInKByte, + dml_uint_t MaxNumDPP, + dml_bool_t nomDETInKByteOverrideEnable, // VBA_DELTA, allow DV to override default DET size + dml_uint_t nomDETInKByteOverrideValue, // VBA_DELTA + + // Output + dml_uint_t *MaxTotalDETInKByte, + dml_uint_t *nomDETInKByte, + dml_uint_t *MinCompressedBufferSizeInKByte) +{ + *MaxTotalDETInKByte = ConfigReturnBufferSizeInKByte - ConfigReturnBufferSegmentSizeInKByte; + *nomDETInKByte = (dml_uint_t)(dml_floor((dml_float_t) *MaxTotalDETInKByte / (dml_float_t) MaxNumDPP, ConfigReturnBufferSegmentSizeInKByte)); + *MinCompressedBufferSizeInKByte = ConfigReturnBufferSizeInKByte - *MaxTotalDETInKByte; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + dml_print("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); + dml_print("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); + dml_print("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); + dml_print("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); + dml_print("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); +#endif + + if (nomDETInKByteOverrideEnable) { + *nomDETInKByte = nomDETInKByteOverrideValue; + dml_print("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); + } +} // CalculateMaxDETAndMinCompressedBufferSize + +/// @brief Calculate all the RQ request attributes, like row height and # swath +static void CalculateVMRowAndSwath(struct display_mode_lib_scratch_st *scratch, + struct CalculateVMRowAndSwath_params_st *p) +{ + struct CalculateVMRowAndSwath_locals_st *s = &scratch->CalculateVMRowAndSwath_locals; + + s->HostVMDynamicLevels = CalculateHostVMDynamicLevels(p->GPUVMEnable, p->HostVMEnable, p->HostVMMinPageSize, p->HostVMMaxNonCachedPageTableLevels); + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->HostVMEnable == true) { + p->vm_group_bytes[k] = 512; + p->dpte_group_bytes[k] = 512; + } else if (p->GPUVMEnable == true) { + p->vm_group_bytes[k] = 2048; + if (p->GPUVMMinPageSizeKBytes[k] >= 64 && dml_is_vertical_rotation(p->myPipe[k].SourceScan)) { + p->dpte_group_bytes[k] = 512; + } else { + p->dpte_group_bytes[k] = 2048; + } + } else { + p->vm_group_bytes[k] = 0; + p->dpte_group_bytes[k] = 0; + } + + if (p->myPipe[k].SourcePixelFormat == dml_420_8 || p->myPipe[k].SourcePixelFormat == dml_420_10 || + p->myPipe[k].SourcePixelFormat == dml_420_12 || p->myPipe[k].SourcePixelFormat == dml_rgbe_alpha) { + if ((p->myPipe[k].SourcePixelFormat == dml_420_10 || p->myPipe[k].SourcePixelFormat == dml_420_12) && !dml_is_vertical_rotation(p->myPipe[k].SourceScan)) { + s->PTEBufferSizeInRequestsForLuma[k] = (p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma) / 2; + s->PTEBufferSizeInRequestsForChroma[k] = s->PTEBufferSizeInRequestsForLuma[k]; + } else { + s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma; + s->PTEBufferSizeInRequestsForChroma[k] = p->PTEBufferSizeInRequestsChroma; + } + + s->PDEAndMetaPTEBytesFrameC = CalculateVMAndRowBytes( + p->myPipe[k].ViewportStationary, + p->myPipe[k].DCCEnable, + p->myPipe[k].DPPPerSurface, + p->myPipe[k].BlockHeight256BytesC, + p->myPipe[k].BlockWidth256BytesC, + p->myPipe[k].SourcePixelFormat, + p->myPipe[k].SurfaceTiling, + p->myPipe[k].BytePerPixelC, + p->myPipe[k].SourceScan, + p->SwathWidthC[k], + p->myPipe[k].ViewportHeightChroma, + p->myPipe[k].ViewportXStartC, + p->myPipe[k].ViewportYStartC, + p->GPUVMEnable, + p->GPUVMMaxPageTableLevels, + p->GPUVMMinPageSizeKBytes[k], + s->PTEBufferSizeInRequestsForChroma[k], + p->myPipe[k].PitchC, + p->myPipe[k].DCCMetaPitchC, + p->myPipe[k].BlockWidthC, + p->myPipe[k].BlockHeightC, + + // Output + &s->MetaRowByteC[k], + &s->PixelPTEBytesPerRowC[k], + &s->PixelPTEBytesPerRowStorageC[k], + &p->dpte_row_width_chroma_ub[k], + &p->dpte_row_height_chroma[k], + &p->dpte_row_height_linear_chroma[k], + &s->PixelPTEBytesPerRowC_one_row_per_frame[k], + &s->dpte_row_width_chroma_ub_one_row_per_frame[k], + &s->dpte_row_height_chroma_one_row_per_frame[k], + &p->meta_req_width_chroma[k], + &p->meta_req_height_chroma[k], + &p->meta_row_width_chroma[k], + &p->meta_row_height_chroma[k], + &p->PixelPTEReqWidthC[k], + &p->PixelPTEReqHeightC[k], + &p->PTERequestSizeC[k], + &p->dpde0_bytes_per_frame_ub_c[k], + &p->meta_pte_bytes_per_frame_ub_c[k]); + + p->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines ( + p->myPipe[k].VRatioChroma, + p->myPipe[k].VTapsChroma, + p->myPipe[k].InterlaceEnable, + p->myPipe[k].ProgressiveToInterlaceUnitInOPP, + p->myPipe[k].SwathHeightC, + p->myPipe[k].SourceScan, + p->myPipe[k].ViewportStationary, + p->SwathWidthC[k], + p->myPipe[k].ViewportHeightChroma, + p->myPipe[k].ViewportXStartC, + p->myPipe[k].ViewportYStartC, + + // Output + &p->VInitPreFillC[k], + &p->MaxNumSwathC[k]); + } else { + s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma; + s->PTEBufferSizeInRequestsForChroma[k] = 0; + s->PixelPTEBytesPerRowC[k] = 0; + s->PixelPTEBytesPerRowStorageC[k] = 0; + s->PDEAndMetaPTEBytesFrameC = 0; + s->MetaRowByteC[k] = 0; + p->MaxNumSwathC[k] = 0; + p->PrefetchSourceLinesC[k] = 0; + s->dpte_row_height_chroma_one_row_per_frame[k] = 0; + s->dpte_row_width_chroma_ub_one_row_per_frame[k] = 0; + s->PixelPTEBytesPerRowC_one_row_per_frame[k] = 0; + } + + s->PDEAndMetaPTEBytesFrameY = CalculateVMAndRowBytes( + p->myPipe[k].ViewportStationary, + p->myPipe[k].DCCEnable, + p->myPipe[k].DPPPerSurface, + p->myPipe[k].BlockHeight256BytesY, + p->myPipe[k].BlockWidth256BytesY, + p->myPipe[k].SourcePixelFormat, + p->myPipe[k].SurfaceTiling, + p->myPipe[k].BytePerPixelY, + p->myPipe[k].SourceScan, + p->SwathWidthY[k], + p->myPipe[k].ViewportHeight, + p->myPipe[k].ViewportXStart, + p->myPipe[k].ViewportYStart, + p->GPUVMEnable, + p->GPUVMMaxPageTableLevels, + p->GPUVMMinPageSizeKBytes[k], + s->PTEBufferSizeInRequestsForLuma[k], + p->myPipe[k].PitchY, + p->myPipe[k].DCCMetaPitchY, + p->myPipe[k].BlockWidthY, + p->myPipe[k].BlockHeightY, + + // Output + &s->MetaRowByteY[k], + &s->PixelPTEBytesPerRowY[k], + &s->PixelPTEBytesPerRowStorageY[k], + &p->dpte_row_width_luma_ub[k], + &p->dpte_row_height_luma[k], + &p->dpte_row_height_linear_luma[k], + &s->PixelPTEBytesPerRowY_one_row_per_frame[k], + &s->dpte_row_width_luma_ub_one_row_per_frame[k], + &s->dpte_row_height_luma_one_row_per_frame[k], + &p->meta_req_width[k], + &p->meta_req_height[k], + &p->meta_row_width[k], + &p->meta_row_height[k], + &p->PixelPTEReqWidthY[k], + &p->PixelPTEReqHeightY[k], + &p->PTERequestSizeY[k], + &p->dpde0_bytes_per_frame_ub_l[k], + &p->meta_pte_bytes_per_frame_ub_l[k]); + + p->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines( + p->myPipe[k].VRatio, + p->myPipe[k].VTaps, + p->myPipe[k].InterlaceEnable, + p->myPipe[k].ProgressiveToInterlaceUnitInOPP, + p->myPipe[k].SwathHeightY, + p->myPipe[k].SourceScan, + p->myPipe[k].ViewportStationary, + p->SwathWidthY[k], + p->myPipe[k].ViewportHeight, + p->myPipe[k].ViewportXStart, + p->myPipe[k].ViewportYStart, + + // Output + &p->VInitPreFillY[k], + &p->MaxNumSwathY[k]); + + p->PDEAndMetaPTEBytesFrame[k] = (s->PDEAndMetaPTEBytesFrameY + s->PDEAndMetaPTEBytesFrameC) * (1 + 8 * s->HostVMDynamicLevels); + p->MetaRowByte[k] = s->MetaRowByteY[k] + s->MetaRowByteC[k]; + + if (s->PixelPTEBytesPerRowStorageY[k] <= 64 * s->PTEBufferSizeInRequestsForLuma[k] && s->PixelPTEBytesPerRowStorageC[k] <= 64 * s->PTEBufferSizeInRequestsForChroma[k]) { + p->PTEBufferSizeNotExceeded[k] = true; + } else { + p->PTEBufferSizeNotExceeded[k] = false; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY = %u\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC = %u\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); + dml_print("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); + dml_print("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); + dml_print("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); +#endif + } + s->one_row_per_frame_fits_in_buffer[k] = (s->PixelPTEBytesPerRowY_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForLuma[k] && + s->PixelPTEBytesPerRowC_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForChroma[k]); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrame = %u\n", __func__, k, p->PDEAndMetaPTEBytesFrame[k]); + dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrameY = %u\n", __func__, k, s->PDEAndMetaPTEBytesFrameY); + dml_print("DML::%s: k=%u, PDEAndMetaPTEBytesFrameC = %u\n", __func__, k, s->PDEAndMetaPTEBytesFrameC); + dml_print("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); + dml_print("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); +#endif + } + + CalculateMALLUseForStaticScreen( + p->NumberOfActiveSurfaces, + p->MALLAllocatedForDCN, + p->UseMALLForStaticScreen, // mode + p->SurfaceSizeInMALL, + s->one_row_per_frame_fits_in_buffer, + // Output + p->UsesMALLForStaticScreen); // boolen + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->PTEBufferModeOverrideEn[k] == 1) { + p->PTE_BUFFER_MODE[k] = p->PTEBufferModeOverrideVal[k]; + } + p->PTE_BUFFER_MODE[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->UsesMALLForStaticScreen[k] || (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) || + (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) || (p->GPUVMMinPageSizeKBytes[k] > 64); + p->BIGK_FRAGMENT_SIZE[k] = (dml_uint_t)(dml_log2(p->GPUVMMinPageSizeKBytes[k] * 1024) - 12); + } + + for (dml_uint_t k = 0; k < p->NumberOfActiveSurfaces; ++k) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); + dml_print("DML::%s: k=%u, UsesMALLForStaticScreen = %u\n", __func__, k, p->UsesMALLForStaticScreen[k]); +#endif + p->use_one_row_for_frame[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->UsesMALLForStaticScreen[k] || (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) || + (p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) || (p->GPUVMMinPageSizeKBytes[k] > 64 && dml_is_vertical_rotation(p->myPipe[k].SourceScan)); + + p->use_one_row_for_frame_flip[k] = p->use_one_row_for_frame[k] && !(p->UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame); + + if (p->use_one_row_for_frame[k]) { + p->dpte_row_height_luma[k] = s->dpte_row_height_luma_one_row_per_frame[k]; + p->dpte_row_width_luma_ub[k] = s->dpte_row_width_luma_ub_one_row_per_frame[k]; + s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY_one_row_per_frame[k]; + p->dpte_row_height_chroma[k] = s->dpte_row_height_chroma_one_row_per_frame[k]; + p->dpte_row_width_chroma_ub[k] = s->dpte_row_width_chroma_ub_one_row_per_frame[k]; + s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC_one_row_per_frame[k]; + p->PTEBufferSizeNotExceeded[k] = s->one_row_per_frame_fits_in_buffer[k]; + } + + if (p->MetaRowByte[k] <= p->DCCMetaBufferSizeBytes) { + p->DCCMetaBufferSizeNotExceeded[k] = true; + } else { + p->DCCMetaBufferSizeNotExceeded[k] = false; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, MetaRowByte = %u\n", __func__, k, p->MetaRowByte[k]); + dml_print("DML::%s: k=%u, DCCMetaBufferSizeBytes = %u\n", __func__, k, p->DCCMetaBufferSizeBytes); + dml_print("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); +#endif + } + s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY[k] * (1 + 8 * s->HostVMDynamicLevels); + s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC[k] * (1 + 8 * s->HostVMDynamicLevels); + p->PixelPTEBytesPerRow[k] = s->PixelPTEBytesPerRowY[k] + s->PixelPTEBytesPerRowC[k]; + if (p->use_one_row_for_frame[k]) + p->PixelPTEBytesPerRow[k] = p->PixelPTEBytesPerRow[k] / 2; + + CalculateRowBandwidth( + p->GPUVMEnable, + p->myPipe[k].SourcePixelFormat, + p->myPipe[k].VRatio, + p->myPipe[k].VRatioChroma, + p->myPipe[k].DCCEnable, + p->myPipe[k].HTotal / p->myPipe[k].PixelClock, + s->MetaRowByteY[k], + s->MetaRowByteC[k], + p->meta_row_height[k], + p->meta_row_height_chroma[k], + s->PixelPTEBytesPerRowY[k], + s->PixelPTEBytesPerRowC[k], + p->dpte_row_height_luma[k], + p->dpte_row_height_chroma[k], + + // Output + &p->meta_row_bw[k], + &p->dpte_row_bw[k]); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); + dml_print("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); + dml_print("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->UseMALLForPStateChange[k]); + dml_print("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); + dml_print("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowY = %u\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + dml_print("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); + dml_print("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRowC = %u\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + dml_print("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); + dml_print("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); + dml_print("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); + dml_print("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); +#endif + } +} + +static void CalculateOutputLink( + dml_float_t PHYCLKPerState, + dml_float_t PHYCLKD18PerState, + dml_float_t PHYCLKD32PerState, + dml_float_t Downspreading, + dml_bool_t IsMainSurfaceUsingTheIndicatedTiming, + enum dml_output_encoder_class Output, + enum dml_output_format_class OutputFormat, + dml_uint_t HTotal, + dml_uint_t HActive, + dml_float_t PixelClockBackEnd, + dml_float_t ForcedOutputLinkBPP, + dml_uint_t DSCInputBitPerComponent, + dml_uint_t NumberOfDSCSlices, + dml_float_t AudioSampleRate, + dml_uint_t AudioSampleLayout, + enum dml_odm_mode ODMModeNoDSC, + enum dml_odm_mode ODMModeDSC, + enum dml_dsc_enable DSCEnable, + dml_uint_t OutputLinkDPLanes, + enum dml_output_link_dp_rate OutputLinkDPRate, + + // Output + dml_bool_t *RequiresDSC, + dml_bool_t *RequiresFEC, + dml_float_t *OutBpp, + enum dml_output_type_and_rate__type *OutputType, + enum dml_output_type_and_rate__rate *OutputRate, + dml_uint_t *RequiredSlots) +{ + dml_bool_t LinkDSCEnable; + dml_uint_t dummy; + *RequiresDSC = false; + *RequiresFEC = false; + *OutBpp = 0; + + *OutputType = dml_output_type_unknown; + *OutputRate = dml_output_rate_unknown; + + if (IsMainSurfaceUsingTheIndicatedTiming) { + if (Output == dml_hdmi) { + *RequiresDSC = false; + *RequiresFEC = false; + *OutBpp = TruncToValidBPP(dml_min(600, PHYCLKPerState) * 10, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, false, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = "HDMI"; + *OutputType = dml_output_type_hdmi; + + } else if (Output == dml_dp || Output == dml_dp2p0 || Output == dml_edp) { + if (DSCEnable == dml_dsc_enable) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml_dp || Output == dml_dp2p0) { + *RequiresFEC = true; + } else { + *RequiresFEC = false; + } + } else { + *RequiresDSC = false; + LinkDSCEnable = false; + if (Output == dml_dp2p0) { + *RequiresFEC = true; + } else { + *RequiresFEC = false; + } + } + if (Output == dml_dp2p0) { + *OutBpp = 0; + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr10) && PHYCLKD32PerState >= 10000 / 32.0) { + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && PHYCLKD32PerState < 13500 / 32.0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR10"; + *OutputType = dml_output_type_dp2p0; + *OutputRate = dml_output_rate_dp_rate_uhbr10; + } + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr13p5) && *OutBpp == 0 && PHYCLKD32PerState >= 13500 / 32.0) { + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && PHYCLKD32PerState < 20000 / 32 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR13p5"; + *OutputType = dml_output_type_dp2p0; + *OutputRate = dml_output_rate_dp_rate_uhbr13p5; + } + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_uhbr20) && *OutBpp == 0 && PHYCLKD32PerState >= 20000 / 32) { + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR20"; + *OutputType = dml_output_type_dp2p0; + *OutputRate = dml_output_rate_dp_rate_uhbr20; + } + } else { // output is dp or edp + *OutBpp = 0; + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr) && PHYCLKPerState >= 270) { + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && PHYCLKPerState < 540 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR"; + *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; + *OutputRate = dml_output_rate_dp_rate_hbr; + } + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr2) && *OutBpp == 0 && PHYCLKPerState >= 540) { + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && PHYCLKPerState < 810 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR2"; + *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; + *OutputRate = dml_output_rate_dp_rate_hbr2; + } + if ((OutputLinkDPRate == dml_dp_rate_na || OutputLinkDPRate == dml_dp_rate_hbr3) && *OutBpp == 0 && PHYCLKPerState >= 810) { // VBA_ERROR, vba code doesn't have hbr3 check + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && DSCEnable == dml_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP((1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (dml_uint_t)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR3"; + *OutputType = (Output == dml_dp) ? dml_output_type_dp : dml_output_type_edp; + *OutputRate = dml_output_rate_dp_rate_hbr3; + } + } + } + } +} + +/// @brief Determine the ODM mode and number of DPP used per plane based on dispclk, dsc usage, odm usage policy +static void CalculateODMMode( + dml_uint_t MaximumPixelsPerLinePerDSCUnit, + dml_uint_t HActive, + enum dml_output_encoder_class Output, + enum dml_output_format_class OutputFormat, + enum dml_odm_use_policy ODMUse, + dml_float_t StateDispclk, + dml_float_t MaxDispclk, + dml_bool_t DSCEnable, + dml_uint_t TotalNumberOfActiveDPP, + dml_uint_t MaxNumDPP, + dml_float_t PixelClock, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKRampingMargin, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_uint_t NumberOfDSCSlices, + + // Output + dml_bool_t *TotalAvailablePipesSupport, + dml_uint_t *NumberOfDPP, + enum dml_odm_mode *ODMMode, + dml_float_t *RequiredDISPCLKPerSurface) +{ + + dml_float_t SurfaceRequiredDISPCLKWithoutODMCombine; + dml_float_t SurfaceRequiredDISPCLKWithODMCombineTwoToOne; + dml_float_t SurfaceRequiredDISPCLKWithODMCombineFourToOne; + + SurfaceRequiredDISPCLKWithoutODMCombine = CalculateRequiredDispclk(dml_odm_mode_bypass, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); + SurfaceRequiredDISPCLKWithODMCombineTwoToOne = CalculateRequiredDispclk(dml_odm_mode_combine_2to1, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); + SurfaceRequiredDISPCLKWithODMCombineFourToOne = CalculateRequiredDispclk(dml_odm_mode_combine_4to1, PixelClock, DISPCLKDPPCLKDSCCLKDownSpreading, DISPCLKRampingMargin, DISPCLKDPPCLKVCOSpeed, MaxDispclk); + *TotalAvailablePipesSupport = true; + + if (OutputFormat == dml_420) { + if (HActive > 4 * DML2_MAX_FMT_420_BUFFER_WIDTH) + *TotalAvailablePipesSupport = false; + else if (HActive > 2 * DML2_MAX_FMT_420_BUFFER_WIDTH) + ODMUse = dml_odm_use_policy_combine_4to1; + else if (HActive > DML2_MAX_FMT_420_BUFFER_WIDTH && ODMUse != dml_odm_use_policy_combine_4to1) + ODMUse = dml_odm_use_policy_combine_2to1; + if (Output == dml_hdmi && ODMUse == dml_odm_use_policy_combine_2to1) + *TotalAvailablePipesSupport = false; + if ((Output == dml_hdmi || Output == dml_dp || Output == dml_edp) && ODMUse == dml_odm_use_policy_combine_4to1) + *TotalAvailablePipesSupport = false; + } + + if (ODMUse == dml_odm_use_policy_bypass || ODMUse == dml_odm_use_policy_combine_as_needed) + *ODMMode = dml_odm_mode_bypass; + else if (ODMUse == dml_odm_use_policy_combine_2to1) + *ODMMode = dml_odm_mode_combine_2to1; + else if (ODMUse == dml_odm_use_policy_combine_4to1) + *ODMMode = dml_odm_mode_combine_4to1; + else if (ODMUse == dml_odm_use_policy_split_1to2) + *ODMMode = dml_odm_mode_split_1to2; + else if (ODMUse == dml_odm_use_policy_mso_1to2) + *ODMMode = dml_odm_mode_mso_1to2; + else if (ODMUse == dml_odm_use_policy_mso_1to4) + *ODMMode = dml_odm_mode_mso_1to4; + + *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithoutODMCombine; + *NumberOfDPP = 0; + + if (!(Output == dml_hdmi || Output == dml_dp || Output == dml_edp) && (ODMUse == dml_odm_use_policy_combine_4to1 || (ODMUse == dml_odm_use_policy_combine_as_needed && + (SurfaceRequiredDISPCLKWithODMCombineTwoToOne > StateDispclk || (DSCEnable && (HActive > 2 * MaximumPixelsPerLinePerDSCUnit)) || NumberOfDSCSlices > 8)))) { + if (TotalNumberOfActiveDPP + 4 <= MaxNumDPP) { + *ODMMode = dml_odm_mode_combine_4to1; + *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithODMCombineFourToOne; + *NumberOfDPP = 4; + } else { + *TotalAvailablePipesSupport = false; + } + } else if (Output != dml_hdmi && (ODMUse == dml_odm_use_policy_combine_2to1 || (ODMUse == dml_odm_use_policy_combine_as_needed && + ((SurfaceRequiredDISPCLKWithoutODMCombine > StateDispclk && SurfaceRequiredDISPCLKWithODMCombineTwoToOne <= StateDispclk) || + (DSCEnable && (HActive > MaximumPixelsPerLinePerDSCUnit)) || (NumberOfDSCSlices <= 8 && NumberOfDSCSlices > 4))))) { + if (TotalNumberOfActiveDPP + 2 <= MaxNumDPP) { + *ODMMode = dml_odm_mode_combine_2to1; + *RequiredDISPCLKPerSurface = SurfaceRequiredDISPCLKWithODMCombineTwoToOne; + *NumberOfDPP = 2; + } else { + *TotalAvailablePipesSupport = false; + } + } else { + if (TotalNumberOfActiveDPP + 1 <= MaxNumDPP) { + *NumberOfDPP = 1; + } else { + *TotalAvailablePipesSupport = false; + } + } +} + +/// @brief Calculate the required DISPCLK given the odm mode and pixclk +static dml_float_t CalculateRequiredDispclk( + enum dml_odm_mode ODMMode, + dml_float_t PixelClock, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKRampingMargin, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_float_t MaxDispclk) +{ + dml_float_t RequiredDispclk = 0.; + dml_float_t PixelClockAfterODM; + + dml_float_t DISPCLKWithRampingRoundedToDFSGranularity; + dml_float_t DISPCLKWithoutRampingRoundedToDFSGranularity; + dml_float_t MaxDispclkRoundedDownToDFSGranularity; + + if (ODMMode == dml_odm_mode_combine_4to1) { + PixelClockAfterODM = PixelClock / 4; + } else if (ODMMode == dml_odm_mode_combine_2to1) { + PixelClockAfterODM = PixelClock / 2; + } else { + PixelClockAfterODM = PixelClock; + } + + DISPCLKWithRampingRoundedToDFSGranularity = RoundToDFSGranularity(PixelClockAfterODM * (1.0 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0) * (1 + DISPCLKRampingMargin / 100.0), 1, DISPCLKDPPCLKVCOSpeed); + DISPCLKWithoutRampingRoundedToDFSGranularity = RoundToDFSGranularity(PixelClockAfterODM * (1.0 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0), 1, DISPCLKDPPCLKVCOSpeed); + MaxDispclkRoundedDownToDFSGranularity = RoundToDFSGranularity(MaxDispclk, 0, DISPCLKDPPCLKVCOSpeed); + + if (DISPCLKWithoutRampingRoundedToDFSGranularity > MaxDispclkRoundedDownToDFSGranularity) { + RequiredDispclk = DISPCLKWithoutRampingRoundedToDFSGranularity; + } else if (DISPCLKWithRampingRoundedToDFSGranularity > MaxDispclkRoundedDownToDFSGranularity) { + RequiredDispclk = MaxDispclkRoundedDownToDFSGranularity; + } else { + RequiredDispclk = DISPCLKWithRampingRoundedToDFSGranularity; + } + + return RequiredDispclk; +} + +/// @brief Determine DPPCLK if there only one DPP per plane, main factor is the pixel rate and DPP scaling parameter +static void CalculateSinglePipeDPPCLKAndSCLThroughput( + dml_float_t HRatio, + dml_float_t HRatioChroma, + dml_float_t VRatio, + dml_float_t VRatioChroma, + dml_float_t MaxDCHUBToPSCLThroughput, + dml_float_t MaxPSCLToLBThroughput, + dml_float_t PixelClock, + enum dml_source_format_class SourcePixelFormat, + dml_uint_t HTaps, + dml_uint_t HTapsChroma, + dml_uint_t VTaps, + dml_uint_t VTapsChroma, + + // Output + dml_float_t *PSCL_THROUGHPUT, + dml_float_t *PSCL_THROUGHPUT_CHROMA, + dml_float_t *DPPCLKUsingSingleDPP) +{ + dml_float_t DPPCLKUsingSingleDPPLuma; + dml_float_t DPPCLKUsingSingleDPPChroma; + + if (HRatio > 1) { + *PSCL_THROUGHPUT = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatio / dml_ceil((dml_float_t) HTaps / 6.0, 1.0)); + } else { + *PSCL_THROUGHPUT = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); + } + + DPPCLKUsingSingleDPPLuma = PixelClock * dml_max3(VTaps / 6 * dml_min(1, HRatio), HRatio * VRatio / *PSCL_THROUGHPUT, 1); + + if ((HTaps > 6 || VTaps > 6) && DPPCLKUsingSingleDPPLuma < 2 * PixelClock) + DPPCLKUsingSingleDPPLuma = 2 * PixelClock; + + if ((SourcePixelFormat != dml_420_8 && SourcePixelFormat != dml_420_10 && SourcePixelFormat != dml_420_12 && SourcePixelFormat != dml_rgbe_alpha)) { + *PSCL_THROUGHPUT_CHROMA = 0; + *DPPCLKUsingSingleDPP = DPPCLKUsingSingleDPPLuma; + } else { + if (HRatioChroma > 1) { + *PSCL_THROUGHPUT_CHROMA = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatioChroma / dml_ceil((dml_float_t) HTapsChroma / 6.0, 1.0)); + } else { + *PSCL_THROUGHPUT_CHROMA = dml_min(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); + } + DPPCLKUsingSingleDPPChroma = PixelClock * dml_max3(VTapsChroma / 6 * dml_min(1, HRatioChroma), + HRatioChroma * VRatioChroma / *PSCL_THROUGHPUT_CHROMA, 1); + if ((HTapsChroma > 6 || VTapsChroma > 6) && DPPCLKUsingSingleDPPChroma < 2 * PixelClock) + DPPCLKUsingSingleDPPChroma = 2 * PixelClock; + *DPPCLKUsingSingleDPP = dml_max(DPPCLKUsingSingleDPPLuma, DPPCLKUsingSingleDPPChroma); + } +} + +/// @brief Calculate the actual dppclk freq +/// @param DPPCLKUsingSingleDPP DppClk freq required if there is only 1 DPP per plane +/// @param DPPPerSurface Number of DPP for each plane +static void CalculateDPPCLK( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t DISPCLKDPPCLKDSCCLKDownSpreading, + dml_float_t DISPCLKDPPCLKVCOSpeed, + dml_float_t DPPCLKUsingSingleDPP[], + dml_uint_t DPPPerSurface[], + + // Output + dml_float_t *GlobalDPPCLK, + dml_float_t Dppclk[]) +{ + *GlobalDPPCLK = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + Dppclk[k] = DPPCLKUsingSingleDPP[k] / DPPPerSurface[k] * (1 + DISPCLKDPPCLKDSCCLKDownSpreading / 100.0); + *GlobalDPPCLK = dml_max(*GlobalDPPCLK, Dppclk[k]); + } + *GlobalDPPCLK = RoundToDFSGranularity(*GlobalDPPCLK, 1, DISPCLKDPPCLKVCOSpeed); + + dml_print("DML::%s: GlobalDPPCLK = %f\n", __func__, *GlobalDPPCLK); + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + Dppclk[k] = *GlobalDPPCLK / 255.0 * dml_ceil(Dppclk[k] * 255.0 / *GlobalDPPCLK, 1.0); + dml_print("DML::%s: Dppclk[%0d] = %f\n", __func__, k, Dppclk[k]); + } +} + +static void CalculateMALLUseForStaticScreen( + dml_uint_t NumberOfActiveSurfaces, + dml_uint_t MALLAllocatedForDCNFinal, + enum dml_use_mall_for_static_screen_mode *UseMALLForStaticScreen, + dml_uint_t SurfaceSizeInMALL[], + dml_bool_t one_row_per_frame_fits_in_buffer[], + + // Output + dml_bool_t UsesMALLForStaticScreen[]) +{ + + dml_uint_t SurfaceToAddToMALL; + dml_bool_t CanAddAnotherSurfaceToMALL; + dml_uint_t TotalSurfaceSizeInMALL; + + TotalSurfaceSizeInMALL = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + UsesMALLForStaticScreen[k] = (UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable); + if (UsesMALLForStaticScreen[k]) + TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, UsesMALLForStaticScreen = %u\n", __func__, k, UsesMALLForStaticScreen[k]); + dml_print("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); +#endif + } + + SurfaceToAddToMALL = 0; + CanAddAnotherSurfaceToMALL = true; + while (CanAddAnotherSurfaceToMALL) { + CanAddAnotherSurfaceToMALL = false; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k] <= MALLAllocatedForDCNFinal * 1024 * 1024 && + !UsesMALLForStaticScreen[k] && UseMALLForStaticScreen[k] != dml_use_mall_static_screen_disable && one_row_per_frame_fits_in_buffer[k] && + (!CanAddAnotherSurfaceToMALL || SurfaceSizeInMALL[k] < SurfaceSizeInMALL[SurfaceToAddToMALL])) { + CanAddAnotherSurfaceToMALL = true; + SurfaceToAddToMALL = k; + dml_print("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, UseMALLForStaticScreen[k]); + } + } + if (CanAddAnotherSurfaceToMALL) { + UsesMALLForStaticScreen[SurfaceToAddToMALL] = true; + TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[SurfaceToAddToMALL]; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); + dml_print("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); +#endif + } + } +} + +// @brief Calculate return bw for VM only traffic +dml_float_t dml_get_return_bw_mbps_vm_only( + const struct soc_bounding_box_st *soc, + dml_bool_t use_ideal_dram_bw_strobe, + dml_bool_t HostVMEnable, + dml_float_t DCFCLK, + dml_float_t FabricClock, + dml_float_t DRAMSpeed) +{ + dml_float_t VMDataOnlyReturnBW = + dml_min3(soc->return_bus_width_bytes * DCFCLK * soc->pct_ideal_sdp_bw_after_urgent / 100.0, + FabricClock * soc->fabric_datapath_to_dcn_data_return_bytes * soc->pct_ideal_sdp_bw_after_urgent / 100.0, + DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes * + ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : soc->pct_ideal_dram_bw_after_urgent_vm_only) / 100.0); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); + dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); + dml_print("DML::%s: DCFCLK = %f\n", __func__, DCFCLK); + dml_print("DML::%s: FabricClock = %f\n", __func__, FabricClock); + dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); + dml_print("DML::%s: VMDataOnlyReturnBW = %f\n", __func__, VMDataOnlyReturnBW); +#endif + return VMDataOnlyReturnBW; +} + +// Function: dml_get_return_bw_mbps +// Megabyte per second +dml_float_t dml_get_return_bw_mbps( + const struct soc_bounding_box_st *soc, + dml_bool_t use_ideal_dram_bw_strobe, + dml_bool_t HostVMEnable, + dml_float_t DCFCLK, + dml_float_t FabricClock, + dml_float_t DRAMSpeed) +{ + dml_float_t ReturnBW = 0.; + dml_float_t IdealSDPPortBandwidth = soc->return_bus_width_bytes * DCFCLK; + dml_float_t IdealFabricBandwidth = FabricClock * soc->fabric_datapath_to_dcn_data_return_bytes; + dml_float_t IdealDRAMBandwidth = DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes; + dml_float_t PixelDataOnlyReturnBW = dml_min3(IdealSDPPortBandwidth * soc->pct_ideal_sdp_bw_after_urgent / 100, + IdealFabricBandwidth * soc->pct_ideal_fabric_bw_after_urgent / 100, + IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : + soc->pct_ideal_dram_bw_after_urgent_pixel_only) / 100); + dml_float_t PixelMixedWithVMDataReturnBW = dml_min3(IdealSDPPortBandwidth * soc->pct_ideal_sdp_bw_after_urgent / 100, + IdealFabricBandwidth * soc->pct_ideal_fabric_bw_after_urgent / 100, + IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : + soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm) / 100); + + if (HostVMEnable != true) { + ReturnBW = PixelDataOnlyReturnBW; + } else { + ReturnBW = PixelMixedWithVMDataReturnBW; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); + dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); + dml_print("DML::%s: DCFCLK = %f\n", __func__, DCFCLK); + dml_print("DML::%s: FabricClock = %f\n", __func__, FabricClock); + dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); + dml_print("DML::%s: IdealSDPPortBandwidth = %f\n", __func__, IdealSDPPortBandwidth); + dml_print("DML::%s: IdealFabricBandwidth = %f\n", __func__, IdealFabricBandwidth); + dml_print("DML::%s: IdealDRAMBandwidth = %f\n", __func__, IdealDRAMBandwidth); + dml_print("DML::%s: PixelDataOnlyReturnBW = %f\n", __func__, PixelDataOnlyReturnBW); + dml_print("DML::%s: PixelMixedWithVMDataReturnBW = %f\n", __func__, PixelMixedWithVMDataReturnBW); + dml_print("DML::%s: ReturnBW = %f MBps\n", __func__, ReturnBW); +#endif + return ReturnBW; +} + +// Function: dml_get_return_dram_bw_mbps +// Megabyte per second +static dml_float_t dml_get_return_dram_bw_mbps( + const struct soc_bounding_box_st *soc, + dml_bool_t use_ideal_dram_bw_strobe, + dml_bool_t HostVMEnable, + dml_float_t DRAMSpeed) +{ + dml_float_t ReturnDRAMBW = 0.; + dml_float_t IdealDRAMBandwidth = DRAMSpeed * soc->num_chans * soc->dram_channel_width_bytes; + dml_float_t PixelDataOnlyReturnBW = IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : + soc->pct_ideal_dram_bw_after_urgent_pixel_only) / 100; + dml_float_t PixelMixedWithVMDataReturnBW = IdealDRAMBandwidth * ((use_ideal_dram_bw_strobe && !HostVMEnable) ? soc->pct_ideal_dram_bw_after_urgent_strobe : + soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm) / 100; + + if (HostVMEnable != true) { + ReturnDRAMBW = PixelDataOnlyReturnBW; + } else { + ReturnDRAMBW = PixelMixedWithVMDataReturnBW; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: use_ideal_dram_bw_strobe = %u\n", __func__, use_ideal_dram_bw_strobe); + dml_print("DML::%s: HostVMEnable = %u\n", __func__, HostVMEnable); + dml_print("DML::%s: DRAMSpeed = %f\n", __func__, DRAMSpeed); + dml_print("DML::%s: IdealDRAMBandwidth = %f\n", __func__, IdealDRAMBandwidth); + dml_print("DML::%s: PixelDataOnlyReturnBW = %f\n", __func__, PixelDataOnlyReturnBW); + dml_print("DML::%s: PixelMixedWithVMDataReturnBW = %f\n", __func__, PixelMixedWithVMDataReturnBW); + dml_print("DML::%s: ReturnDRAMBW = %f MBps\n", __func__, ReturnDRAMBW); +#endif + return ReturnDRAMBW; +} + +/// @brief BACKEND +static dml_uint_t DSCDelayRequirement( + dml_bool_t DSCEnabled, + enum dml_odm_mode ODMMode, + dml_uint_t DSCInputBitPerComponent, + dml_float_t OutputBpp, + dml_uint_t HActive, + dml_uint_t HTotal, + dml_uint_t NumberOfDSCSlices, + enum dml_output_format_class OutputFormat, + enum dml_output_encoder_class Output, + dml_float_t PixelClock, + dml_float_t PixelClockBackEnd) +{ + dml_uint_t DSCDelayRequirement_val = 0; + + if (DSCEnabled == true && OutputBpp != 0) { + if (ODMMode == dml_odm_mode_combine_4to1) { + DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)(dml_ceil((dml_float_t) HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), + (dml_uint_t) (NumberOfDSCSlices / 4.0), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); + } else if (ODMMode == dml_odm_mode_combine_2to1) { + DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)(dml_ceil((dml_float_t) HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), + (dml_uint_t) (NumberOfDSCSlices / 2.0), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); + } else { + DSCDelayRequirement_val = dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (dml_uint_t)((dml_float_t) dml_ceil(HActive / (dml_float_t) NumberOfDSCSlices, 1.0)), + NumberOfDSCSlices, OutputFormat, Output) + dscComputeDelay(OutputFormat, Output); + } + DSCDelayRequirement_val = (dml_uint_t)(DSCDelayRequirement_val + (HTotal - HActive) * dml_ceil((dml_float_t) DSCDelayRequirement_val / (dml_float_t) HActive, 1.0)); + DSCDelayRequirement_val = (dml_uint_t)(DSCDelayRequirement_val * PixelClock / PixelClockBackEnd); + + } else { + DSCDelayRequirement_val = 0; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DSCEnabled = %u\n", __func__, DSCEnabled); + dml_print("DML::%s: ODMMode = %u\n", __func__, ODMMode); + dml_print("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); + dml_print("DML::%s: HActive = %u\n", __func__, HActive); + dml_print("DML::%s: HTotal = %u\n", __func__, HTotal); + dml_print("DML::%s: PixelClock = %f\n", __func__, PixelClock); + dml_print("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); + dml_print("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); + dml_print("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); + dml_print("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); + dml_print("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); +#endif + + return DSCDelayRequirement_val; +} + +static noinline_for_stack dml_bool_t CalculateVActiveBandwithSupport(dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + dml_bool_t NotUrgentLatencyHiding[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[]) +{ + dml_bool_t NotEnoughUrgentLatencyHiding = false; + dml_bool_t CalculateVActiveBandwithSupport_val = false; + dml_float_t VActiveBandwith = 0; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (NotUrgentLatencyHiding[k]) { + NotEnoughUrgentLatencyHiding = true; + } + } + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + VActiveBandwith = VActiveBandwith + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * meta_row_bandwidth[k] + NumberOfDPP[k] * dpte_row_bandwidth[k]; + } + + CalculateVActiveBandwithSupport_val = (VActiveBandwith <= ReturnBW) && !NotEnoughUrgentLatencyHiding; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: NotEnoughUrgentLatencyHiding = %u\n", __func__, NotEnoughUrgentLatencyHiding); + dml_print("DML::%s: VActiveBandwith = %f\n", __func__, VActiveBandwith); + dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); + dml_print("DML::%s: CalculateVActiveBandwithSupport_val = %u\n", __func__, CalculateVActiveBandwithSupport_val); +#endif + return CalculateVActiveBandwithSupport_val; +} + +static void CalculatePrefetchBandwithSupport( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + dml_bool_t NotUrgentLatencyHiding[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_float_t cursor_bw_pre[], + dml_float_t prefetch_vmrow_bw[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[], + + // Output + dml_float_t *PrefetchBandwidth, + dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch, + dml_float_t *FractionOfUrgentBandwidth, + dml_bool_t *PrefetchBandwidthSupport) +{ + dml_bool_t NotEnoughUrgentLatencyHiding = false; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (NotUrgentLatencyHiding[k]) { + NotEnoughUrgentLatencyHiding = true; + } + } + + *PrefetchBandwidth = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + *PrefetchBandwidth = *PrefetchBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k] + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]), + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + } + + *PrefetchBandwidthNotIncludingMALLPrefetch = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) + *PrefetchBandwidthNotIncludingMALLPrefetch = *PrefetchBandwidthNotIncludingMALLPrefetch + + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + + cursor_bw[k] * UrgentBurstFactorCursor[k] + + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]), + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + } + + *PrefetchBandwidthSupport = (*PrefetchBandwidth <= ReturnBW) && !NotEnoughUrgentLatencyHiding; + *FractionOfUrgentBandwidth = *PrefetchBandwidth / ReturnBW; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); + dml_print("DML::%s: PrefetchBandwidth = %f\n", __func__, *PrefetchBandwidth); + dml_print("DML::%s: FractionOfUrgentBandwidth = %f\n", __func__, *FractionOfUrgentBandwidth); + dml_print("DML::%s: PrefetchBandwidthSupport = %u\n", __func__, *PrefetchBandwidthSupport); +#endif +} + +static noinline_for_stack dml_float_t CalculateBandwidthAvailableForImmediateFlip( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t cursor_bw_pre[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[]) +{ + dml_float_t ret_val = ReturnBW; + + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + ret_val = ret_val - dml_max(ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u\n", __func__, k); + dml_print("DML::%s: NumberOfDPP = %u\n", __func__, NumberOfDPP[k]); + dml_print("DML::%s: ReadBandwidthLuma = %f\n", __func__, ReadBandwidthLuma[k]); + dml_print("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, UrgentBurstFactorLuma[k]); + dml_print("DML::%s: ReadBandwidthChroma = %f\n", __func__, ReadBandwidthChroma[k]); + dml_print("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, UrgentBurstFactorChroma[k]); + dml_print("DML::%s: cursor_bw = %f\n", __func__, cursor_bw[k]); + dml_print("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, UrgentBurstFactorCursor[k]); + + dml_print("DML::%s: PrefetchBandwidthLuma = %f\n", __func__, PrefetchBandwidthLuma[k]); + dml_print("DML::%s: UrgentBurstFactorLumaPre = %f\n", __func__, UrgentBurstFactorLumaPre[k]); + dml_print("DML::%s: PrefetchBandwidthChroma = %f\n", __func__, PrefetchBandwidthChroma[k]); + dml_print("DML::%s: UrgentBurstFactorChromaPre = %f\n", __func__, UrgentBurstFactorChromaPre[k]); + dml_print("DML::%s: cursor_bw_pre = %f\n", __func__, cursor_bw_pre[k]); + dml_print("DML::%s: UrgentBurstFactorCursorPre = %f\n", __func__, UrgentBurstFactorCursorPre[k]); + dml_print("DML::%s: ret_val = %f\n", __func__, ret_val); +#endif + } + + return ret_val; +} + +static void CalculateImmediateFlipBandwithSupport( + dml_uint_t NumberOfActiveSurfaces, + dml_float_t ReturnBW, + enum dml_use_mall_for_pstate_change_mode UseMALLForPStateChange[], + enum dml_immediate_flip_requirement ImmediateFlipRequirement[], + dml_float_t final_flip_bw[], + dml_float_t ReadBandwidthLuma[], + dml_float_t ReadBandwidthChroma[], + dml_float_t PrefetchBandwidthLuma[], + dml_float_t PrefetchBandwidthChroma[], + dml_float_t cursor_bw[], + dml_float_t meta_row_bandwidth[], + dml_float_t dpte_row_bandwidth[], + dml_float_t cursor_bw_pre[], + dml_float_t prefetch_vmrow_bw[], + dml_uint_t NumberOfDPP[], + dml_float_t UrgentBurstFactorLuma[], + dml_float_t UrgentBurstFactorChroma[], + dml_float_t UrgentBurstFactorCursor[], + dml_float_t UrgentBurstFactorLumaPre[], + dml_float_t UrgentBurstFactorChromaPre[], + dml_float_t UrgentBurstFactorCursorPre[], + + // Output + dml_float_t *TotalBandwidth, + dml_float_t *TotalBandwidthNotIncludingMALLPrefetch, + dml_float_t *FractionOfUrgentBandwidth, + dml_bool_t *ImmediateFlipBandwidthSupport) +{ + *TotalBandwidth = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) { + + + + *TotalBandwidth = *TotalBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + NumberOfDPP[k] * final_flip_bw[k] + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], + NumberOfDPP[k] * (final_flip_bw[k] + PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + } else { + *TotalBandwidth = *TotalBandwidth + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]) + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k = %u\n", __func__, k); + dml_print("DML::%s: ImmediateFlipRequirement = %u\n", __func__, ImmediateFlipRequirement[k]); + dml_print("DML::%s: TotalBandwidth = %f\n", __func__, *TotalBandwidth); + dml_print("DML::%s: NumberOfDPP = %u\n", __func__, NumberOfDPP[k]); + dml_print("DML::%s: prefetch_vmrow_bw = %f\n", __func__, prefetch_vmrow_bw[k]); + dml_print("DML::%s: final_flip_bw = %f\n", __func__, final_flip_bw[k]); + dml_print("DML::%s: ReadBandwidthLuma = %f\n", __func__, ReadBandwidthLuma[k]); + dml_print("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, UrgentBurstFactorLuma[k]); + dml_print("DML::%s: ReadBandwidthChroma = %f\n", __func__, ReadBandwidthChroma[k]); + dml_print("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, UrgentBurstFactorChroma[k]); + dml_print("DML::%s: cursor_bw = %f\n", __func__, cursor_bw[k]); + dml_print("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, UrgentBurstFactorCursor[k]); + dml_print("DML::%s: PrefetchBandwidthLuma = %f\n", __func__, PrefetchBandwidthLuma[k]); + dml_print("DML::%s: UrgentBurstFactorLumaPre = %f\n", __func__, UrgentBurstFactorLumaPre[k]); + dml_print("DML::%s: PrefetchBandwidthChroma = %f\n", __func__, PrefetchBandwidthChroma[k]); + dml_print("DML::%s: UrgentBurstFactorChromaPre = %f\n", __func__, UrgentBurstFactorChromaPre[k]); + dml_print("DML::%s: cursor_bw_pre = %f\n", __func__, cursor_bw_pre[k]); + dml_print("DML::%s: UrgentBurstFactorCursorPre = %f\n", __func__, UrgentBurstFactorCursorPre[k]); + dml_print("DML::%s: meta_row_bandwidth = %f\n", __func__, meta_row_bandwidth[k]); + dml_print("DML::%s: dpte_row_bandwidth = %f\n", __func__, dpte_row_bandwidth[k]); +#endif + } + + *TotalBandwidthNotIncludingMALLPrefetch = 0; + for (dml_uint_t k = 0; k < NumberOfActiveSurfaces; ++k) { + if (UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { + if (ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) + *TotalBandwidthNotIncludingMALLPrefetch = *TotalBandwidthNotIncludingMALLPrefetch + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + NumberOfDPP[k] * final_flip_bw[k] + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], + NumberOfDPP[k] * (final_flip_bw[k] + PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + else + *TotalBandwidthNotIncludingMALLPrefetch = *TotalBandwidthNotIncludingMALLPrefetch + dml_max3(NumberOfDPP[k] * prefetch_vmrow_bw[k], + NumberOfDPP[k] * (meta_row_bandwidth[k] + dpte_row_bandwidth[k]) + + ReadBandwidthLuma[k] * UrgentBurstFactorLuma[k] + ReadBandwidthChroma[k] * UrgentBurstFactorChroma[k] + cursor_bw[k] * UrgentBurstFactorCursor[k], + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * UrgentBurstFactorLumaPre[k] + PrefetchBandwidthChroma[k] * UrgentBurstFactorChromaPre[k]) + + cursor_bw_pre[k] * UrgentBurstFactorCursorPre[k]); + } + } + + *ImmediateFlipBandwidthSupport = (*TotalBandwidth <= ReturnBW); + *FractionOfUrgentBandwidth = *TotalBandwidth / ReturnBW; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ReturnBW = %f\n", __func__, ReturnBW); + dml_print("DML::%s: TotalBandwidth = %f\n", __func__, *TotalBandwidth); + dml_print("DML::%s: ImmediateFlipBandwidthSupport = %u\n", __func__, *ImmediateFlipBandwidthSupport); +#endif +} + +static dml_uint_t MicroSecToVertLines(dml_uint_t num_us, dml_uint_t h_total, dml_float_t pixel_clock) +{ + dml_uint_t lines_time_in_ns = 1000.0 * (h_total * 1000.0) / (pixel_clock * 1000.0); + + return dml_ceil(1000.0 * num_us / lines_time_in_ns, 1.0); +} + +/// @brief Calculate the maximum vstartup for mode support and mode programming consideration +/// Bounded by min of actual vblank and input vblank_nom, dont want vstartup/ready to start too early if actual vbllank is huge +static dml_uint_t CalculateMaxVStartup( + dml_uint_t plane_idx, + dml_bool_t ptoi_supported, + dml_uint_t vblank_nom_default_us, + struct dml_timing_cfg_st *timing, + dml_float_t write_back_delay_us) +{ + dml_uint_t vblank_size = 0; + dml_uint_t max_vstartup_lines = 0; + const dml_uint_t max_allowed_vblank_nom = 1023; + + dml_float_t line_time_us = (dml_float_t) timing->HTotal[plane_idx] / timing->PixelClock[plane_idx]; + dml_uint_t vblank_actual = timing->VTotal[plane_idx] - timing->VActive[plane_idx]; + + dml_uint_t vblank_nom_default_in_line = MicroSecToVertLines(vblank_nom_default_us, timing->HTotal[plane_idx], + timing->PixelClock[plane_idx]); + dml_uint_t vblank_nom_input = (dml_uint_t)dml_min(vblank_actual, vblank_nom_default_in_line); + + // vblank_nom should not be smaller than (VSync (VTotal - VActive - VFrontPorch) + 2) + // + 2 is because + // 1 -> VStartup_start should be 1 line before VSync + // 1 -> always reserve 1 line between start of VBlank to VStartup signal + dml_uint_t vblank_nom_vsync_capped = dml_max(vblank_nom_input, + timing->VTotal[plane_idx] - timing->VActive[plane_idx] - timing->VFrontPorch[plane_idx] + 2); + dml_uint_t vblank_nom_max_allowed_capped = dml_min(vblank_nom_vsync_capped, max_allowed_vblank_nom); + dml_uint_t vblank_avail = (vblank_nom_max_allowed_capped == 0) ? + vblank_nom_default_in_line : vblank_nom_max_allowed_capped; + + vblank_size = (dml_uint_t) dml_min(vblank_actual, vblank_avail); + + if (timing->Interlace[plane_idx] && !ptoi_supported) + max_vstartup_lines = (dml_uint_t) (dml_floor(vblank_size/2.0, 1.0)); + else + max_vstartup_lines = vblank_size - (dml_uint_t) dml_max(1.0, dml_ceil(write_back_delay_us/line_time_us, 1.0)); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: plane_idx = %u\n", __func__, plane_idx); + dml_print("DML::%s: VBlankNom = %u\n", __func__, timing->VBlankNom[plane_idx]); + dml_print("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); + dml_print("DML::%s: line_time_us = %f\n", __func__, line_time_us); + dml_print("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); + dml_print("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); + dml_print("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); +#endif + max_vstartup_lines = (dml_uint_t) dml_min(max_vstartup_lines, DML_MAX_VSTARTUP_START); + return max_vstartup_lines; +} + +static noinline_for_stack void set_calculate_prefetch_schedule_params(struct display_mode_lib_st *mode_lib, + struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params, + dml_uint_t j, + dml_uint_t k) +{ + CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelayPerState[k]; + CalculatePrefetchSchedule_params->EnhancedPrefetchScheduleAccelerationFinal = mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal; + CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ms.ip.dppclk_delay_subtotal + mode_lib->ms.ip.dppclk_delay_cnvc_formatter; + CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ms.ip.dppclk_delay_scl; + CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ms.ip.dppclk_delay_scl_lb_only; + CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ms.ip.dppclk_delay_cnvc_cursor; + CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ms.ip.dispclk_delay_subtotal; + CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (dml_uint_t)(mode_lib->ms.SwathWidthYThisState[k] / mode_lib->ms.cache_display_cfg.plane.HRatio[k]); + CalculatePrefetchSchedule_params->OutputFormat = mode_lib->ms.cache_display_cfg.output.OutputFormat[k]; + CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; + CalculatePrefetchSchedule_params->GPUVMPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + CalculatePrefetchSchedule_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + CalculatePrefetchSchedule_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + CalculatePrefetchSchedule_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + CalculatePrefetchSchedule_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]; + CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; + CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired[k]; + CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes[k]; + CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; + CalculatePrefetchSchedule_params->UrgentExtraLatency = mode_lib->ms.ExtraLatency; + CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; + CalculatePrefetchSchedule_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k]; + CalculatePrefetchSchedule_params->MetaRowByte = mode_lib->ms.MetaRowBytes[j][k]; + CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[j][k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[j][k]; + CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; + CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[j][k]; + CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; + CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC[k]; + CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub_this_state[k]; + CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub_this_state[k]; + CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState[k]; + CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState[k]; + CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait; + CalculatePrefetchSchedule_params->DestinationLinesForPrefetch = &mode_lib->ms.LineTimesForPrefetch[k]; + CalculatePrefetchSchedule_params->DestinationLinesToRequestVMInVBlank = &mode_lib->ms.LinesForMetaPTE[k]; + CalculatePrefetchSchedule_params->DestinationLinesToRequestRowInVBlank = &mode_lib->ms.LinesForMetaAndDPTERow[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[j][k]; + CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[j][k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; + CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.support.NoTimeForDynamicMetadata[j][k]; + CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; +} + +static noinline_for_stack void dml_prefetch_check(struct display_mode_lib_st *mode_lib) +{ + struct dml_core_mode_support_locals_st *s = &mode_lib->scratch.dml_core_mode_support_locals; + struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; + struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; + struct DmlPipe *myPipe; + dml_uint_t j, k; + + for (j = 0; j < 2; ++j) { + mode_lib->ms.TimeCalc = 24 / mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; + mode_lib->ms.swath_width_luma_ub_this_state[k] = mode_lib->ms.swath_width_luma_ub_all_states[j][k]; + mode_lib->ms.swath_width_chroma_ub_this_state[k] = mode_lib->ms.swath_width_chroma_ub_all_states[j][k]; + mode_lib->ms.SwathWidthYThisState[k] = mode_lib->ms.SwathWidthYAllStates[j][k]; + mode_lib->ms.SwathWidthCThisState[k] = mode_lib->ms.SwathWidthCAllStates[j][k]; + mode_lib->ms.SwathHeightYThisState[k] = mode_lib->ms.SwathHeightYAllStates[j][k]; + mode_lib->ms.SwathHeightCThisState[k] = mode_lib->ms.SwathHeightCAllStates[j][k]; + mode_lib->ms.UnboundedRequestEnabledThisState = mode_lib->ms.UnboundedRequestEnabledAllStates[j]; + mode_lib->ms.CompressedBufferSizeInkByteThisState = mode_lib->ms.CompressedBufferSizeInkByteAllStates[j]; + mode_lib->ms.DETBufferSizeInKByteThisState[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[j][k]; + mode_lib->ms.DETBufferSizeYThisState[k] = mode_lib->ms.DETBufferSizeYAllStates[j][k]; + mode_lib->ms.DETBufferSizeCThisState[k] = mode_lib->ms.DETBufferSizeCAllStates[j][k]; + } + + mode_lib->ms.support.VActiveBandwithSupport[j] = CalculateVActiveBandwithSupport( + mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBWPerState[j], + mode_lib->ms.NotUrgentLatencyHiding, + mode_lib->ms.ReadBandwidthLuma, + mode_lib->ms.ReadBandwidthChroma, + mode_lib->ms.cursor_bw, + mode_lib->ms.meta_row_bandwidth_this_state, + mode_lib->ms.dpte_row_bandwidth_this_state, + mode_lib->ms.NoOfDPPThisState, + mode_lib->ms.UrgentBurstFactorLuma[j], + mode_lib->ms.UrgentBurstFactorChroma[j], + mode_lib->ms.UrgentBurstFactorCursor[j]); + + s->VMDataOnlyReturnBWPerState = dml_get_return_bw_mbps_vm_only( + &mode_lib->ms.soc, + mode_lib->ms.state.use_ideal_dram_bw_strobe, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.DCFCLKState[j], + mode_lib->ms.state.fabricclk_mhz, + mode_lib->ms.state.dram_speed_mts); + + s->HostVMInefficiencyFactor = 1; + if (mode_lib->ms.cache_display_cfg.plane.GPUVMEnable && mode_lib->ms.cache_display_cfg.plane.HostVMEnable) + s->HostVMInefficiencyFactor = mode_lib->ms.ReturnBWPerState[j] / s->VMDataOnlyReturnBWPerState; + + mode_lib->ms.ExtraLatency = CalculateExtraLatency( + mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles, + s->ReorderingBytes, + mode_lib->ms.DCFCLKState[j], + mode_lib->ms.TotalNumberOfActiveDPP[j], + mode_lib->ms.ip.pixel_chunk_size_kbytes, + mode_lib->ms.TotalNumberOfDCCActiveDPP[j], + mode_lib->ms.ip.meta_chunk_size_kbytes, + mode_lib->ms.ReturnBWPerState[j], + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.num_active_planes, + mode_lib->ms.NoOfDPPThisState, + mode_lib->ms.dpte_group_bytes, + s->HostVMInefficiencyFactor, + mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, + mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels); + + s->NextMaxVStartup = s->MaxVStartupAllPlanes[j]; + s->MaxVStartup = 0; + s->AllPrefetchModeTested = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculatePrefetchMode(mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k], &s->MinPrefetchMode[k], &s->MaxPrefetchMode[k]); + s->NextPrefetchMode[k] = s->MinPrefetchMode[k]; + } + + do { + s->MaxVStartup = s->NextMaxVStartup; + s->AllPrefetchModeTested = true; + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.PrefetchMode[k] = s->NextPrefetchMode[k]; + mode_lib->ms.TWait = CalculateTWait( + mode_lib->ms.PrefetchMode[k], + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal, + mode_lib->ms.cache_display_cfg.timing.DRRDisplay[k], + mode_lib->ms.state.dram_clock_change_latency_us, + mode_lib->ms.state.fclk_change_latency_us, + mode_lib->ms.UrgLatency, + mode_lib->ms.state.sr_enter_plus_exit_time_us); + + myPipe = &s->myPipe; + myPipe->Dppclk = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; + myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK[j]; + myPipe->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; + myPipe->DCFClkDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; + myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[j][k]; + myPipe->ScalerEnabled = mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k]; + myPipe->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; + myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; + myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; + myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; + myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; + myPipe->InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; + myPipe->NumberOfCursors = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k]; + myPipe->VBlank = mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]; + myPipe->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; + myPipe->HActive = mode_lib->ms.cache_display_cfg.timing.HActive[k]; + myPipe->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; + myPipe->ODMMode = mode_lib->ms.ODMModePerState[k]; + myPipe->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; + myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; + myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; + myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Calling CalculatePrefetchSchedule for j=%u, k=%u\n", __func__, j, k); + dml_print("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[j][k]); + dml_print("DML::%s: MaxVStartup = %u\n", __func__, s->MaxVStartup); + dml_print("DML::%s: NextPrefetchMode = %u\n", __func__, s->NextPrefetchMode[k]); + dml_print("DML::%s: AllowForPStateChangeOrStutterInVBlank = %u\n", __func__, mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k]); + dml_print("DML::%s: PrefetchMode = %u\n", __func__, mode_lib->ms.PrefetchMode[k]); +#endif + + CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactor; + CalculatePrefetchSchedule_params->myPipe = myPipe; + CalculatePrefetchSchedule_params->VStartup = (dml_uint_t)(dml_min(s->MaxVStartup, s->MaximumVStartup[j][k])); + CalculatePrefetchSchedule_params->MaxVStartup = s->MaximumVStartup[j][k]; + CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; + CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; + CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; + CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; + CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; + CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; + CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; + CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; + CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; + + set_calculate_prefetch_schedule_params(mode_lib, CalculatePrefetchSchedule_params, j, k); + + mode_lib->ms.support.NoTimeForPrefetch[j][k] = + CalculatePrefetchSchedule(&mode_lib->scratch, + CalculatePrefetchSchedule_params); + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateUrgentBurstFactor( + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + mode_lib->ms.swath_width_luma_ub_this_state[k], + mode_lib->ms.swath_width_chroma_ub_this_state[k], + mode_lib->ms.SwathHeightYThisState[k], + mode_lib->ms.SwathHeightCThisState[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.UrgLatency, + mode_lib->ms.ip.cursor_buffer_size, + mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], + mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], + mode_lib->ms.VRatioPreY[j][k], + mode_lib->ms.VRatioPreC[j][k], + mode_lib->ms.BytePerPixelInDETY[k], + mode_lib->ms.BytePerPixelInDETC[k], + mode_lib->ms.DETBufferSizeYThisState[k], + mode_lib->ms.DETBufferSizeCThisState[k], + /* Output */ + &mode_lib->ms.UrgentBurstFactorCursorPre[k], + &mode_lib->ms.UrgentBurstFactorLumaPre[k], + &mode_lib->ms.UrgentBurstFactorChromaPre[k], + &mode_lib->ms.NotUrgentLatencyHidingPre[k]); + + mode_lib->ms.cursor_bw_pre[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * + mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / + mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.VRatioPreY[j][k]; + } + + { + CalculatePrefetchBandwithSupport( + mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBWPerState[j], + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + mode_lib->ms.NotUrgentLatencyHidingPre, + mode_lib->ms.ReadBandwidthLuma, + mode_lib->ms.ReadBandwidthChroma, + mode_lib->ms.RequiredPrefetchPixelDataBWLuma, + mode_lib->ms.RequiredPrefetchPixelDataBWChroma, + mode_lib->ms.cursor_bw, + mode_lib->ms.meta_row_bandwidth_this_state, + mode_lib->ms.dpte_row_bandwidth_this_state, + mode_lib->ms.cursor_bw_pre, + mode_lib->ms.prefetch_vmrow_bw, + mode_lib->ms.NoOfDPPThisState, + mode_lib->ms.UrgentBurstFactorLuma[j], + mode_lib->ms.UrgentBurstFactorChroma[j], + mode_lib->ms.UrgentBurstFactorCursor[j], + mode_lib->ms.UrgentBurstFactorLumaPre, + mode_lib->ms.UrgentBurstFactorChromaPre, + mode_lib->ms.UrgentBurstFactorCursorPre, + + /* output */ + &s->dummy_single[0], // dml_float_t *PrefetchBandwidth + &s->dummy_single[1], // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch + &mode_lib->mp.FractionOfUrgentBandwidth, // dml_float_t *FractionOfUrgentBandwidth + &mode_lib->ms.support.PrefetchSupported[j]); + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.LineTimesForPrefetch[k] < 2.0 + || mode_lib->ms.LinesForMetaPTE[k] >= 32.0 + || mode_lib->ms.LinesForMetaAndDPTERow[k] >= 16.0 + || mode_lib->ms.support.NoTimeForPrefetch[j][k] == true) { + mode_lib->ms.support.PrefetchSupported[j] = false; + } + } + + mode_lib->ms.support.DynamicMetadataSupported[j] = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.support.NoTimeForDynamicMetadata[j][k] == true) { + mode_lib->ms.support.DynamicMetadataSupported[j] = false; + } + } + + mode_lib->ms.support.VRatioInPrefetchSupported[j] = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.support.NoTimeForPrefetch[j][k] == true || + mode_lib->ms.VRatioPreY[j][k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || + mode_lib->ms.VRatioPreC[j][k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || + ((s->MaxVStartup < s->MaximumVStartup[j][k] || mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal == 0) && + (mode_lib->ms.VRatioPreY[j][k] > __DML_MAX_VRATIO_PRE__ || mode_lib->ms.VRatioPreC[j][k] > __DML_MAX_VRATIO_PRE__))) { + mode_lib->ms.support.VRatioInPrefetchSupported[j] = false; + } + } + + s->AnyLinesForVMOrRowTooLarge = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.LinesForMetaAndDPTERow[k] >= 16 || mode_lib->ms.LinesForMetaPTE[k] >= 32) { + s->AnyLinesForVMOrRowTooLarge = true; + } + } + + if (mode_lib->ms.support.PrefetchSupported[j] == true && mode_lib->ms.support.VRatioInPrefetchSupported[j] == true) { + mode_lib->ms.BandwidthAvailableForImmediateFlip = CalculateBandwidthAvailableForImmediateFlip( + mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBWPerState[j], + mode_lib->ms.ReadBandwidthLuma, + mode_lib->ms.ReadBandwidthChroma, + mode_lib->ms.RequiredPrefetchPixelDataBWLuma, + mode_lib->ms.RequiredPrefetchPixelDataBWChroma, + mode_lib->ms.cursor_bw, + mode_lib->ms.cursor_bw_pre, + mode_lib->ms.NoOfDPPThisState, + mode_lib->ms.UrgentBurstFactorLuma[j], + mode_lib->ms.UrgentBurstFactorChroma[j], + mode_lib->ms.UrgentBurstFactorCursor[j], + mode_lib->ms.UrgentBurstFactorLumaPre, + mode_lib->ms.UrgentBurstFactorChromaPre, + mode_lib->ms.UrgentBurstFactorCursorPre); + + mode_lib->ms.TotImmediateFlipBytes = 0; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (!(mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_not_required)) { + mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * (mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k] + mode_lib->ms.MetaRowBytes[j][k]); + if (mode_lib->ms.use_one_row_for_frame_flip[j][k]) { + mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * (2 * mode_lib->ms.DPTEBytesPerRow[j][k]); + } else { + mode_lib->ms.TotImmediateFlipBytes = mode_lib->ms.TotImmediateFlipBytes + mode_lib->ms.NoOfDPP[j][k] * mode_lib->ms.DPTEBytesPerRow[j][k]; + } + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateFlipSchedule( + s->HostVMInefficiencyFactor, + mode_lib->ms.ExtraLatency, + mode_lib->ms.UrgLatency, + mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels, + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, + mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k], + mode_lib->ms.MetaRowBytes[j][k], + mode_lib->ms.DPTEBytesPerRow[j][k], + mode_lib->ms.BandwidthAvailableForImmediateFlip, + mode_lib->ms.TotImmediateFlipBytes, + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + mode_lib->ms.Tno_bw[k], + mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], + mode_lib->ms.dpte_row_height[k], + mode_lib->ms.meta_row_height[k], + mode_lib->ms.dpte_row_height_chroma[k], + mode_lib->ms.meta_row_height_chroma[k], + mode_lib->ms.use_one_row_for_frame_flip[j][k], // 24 + + /* Output */ + &mode_lib->ms.DestinationLinesToRequestVMInImmediateFlip[k], + &mode_lib->ms.DestinationLinesToRequestRowInImmediateFlip[k], + &mode_lib->ms.final_flip_bw[k], + &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); + } + + { + CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBWPerState[j], + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + mode_lib->ms.policy.ImmediateFlipRequirement, + mode_lib->ms.final_flip_bw, + mode_lib->ms.ReadBandwidthLuma, + mode_lib->ms.ReadBandwidthChroma, + mode_lib->ms.RequiredPrefetchPixelDataBWLuma, + mode_lib->ms.RequiredPrefetchPixelDataBWChroma, + mode_lib->ms.cursor_bw, + mode_lib->ms.meta_row_bandwidth_this_state, + mode_lib->ms.dpte_row_bandwidth_this_state, + mode_lib->ms.cursor_bw_pre, + mode_lib->ms.prefetch_vmrow_bw, + mode_lib->ms.NoOfDPP[j], // VBA_ERROR DPPPerSurface is not assigned at this point, should use NoOfDpp here + mode_lib->ms.UrgentBurstFactorLuma[j], + mode_lib->ms.UrgentBurstFactorChroma[j], + mode_lib->ms.UrgentBurstFactorCursor[j], + mode_lib->ms.UrgentBurstFactorLumaPre, + mode_lib->ms.UrgentBurstFactorChromaPre, + mode_lib->ms.UrgentBurstFactorCursorPre, + + /* output */ + &s->dummy_single[0], // dml_float_t *TotalBandwidth + &s->dummy_single[1], // dml_float_t *TotalBandwidthNotIncludingMALLPrefetch + &s->dummy_single[2], // dml_float_t *FractionOfUrgentBandwidth + &mode_lib->ms.support.ImmediateFlipSupportedForState[j]); // dml_bool_t *ImmediateFlipBandwidthSupport + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (!(mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_not_required) && (mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false)) + mode_lib->ms.support.ImmediateFlipSupportedForState[j] = false; + } + + } else { // if prefetch not support, assume iflip not supported + mode_lib->ms.support.ImmediateFlipSupportedForState[j] = false; + } + + if (s->MaxVStartup <= __DML_VBA_MIN_VSTARTUP__ || s->AnyLinesForVMOrRowTooLarge == false) { + s->NextMaxVStartup = s->MaxVStartupAllPlanes[j]; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + s->NextPrefetchMode[k] = s->NextPrefetchMode[k] + 1; + + if (s->NextPrefetchMode[k] <= s->MaxPrefetchMode[k]) + s->AllPrefetchModeTested = false; + } + } else { + s->NextMaxVStartup = s->NextMaxVStartup - 1; + } + } while (!((mode_lib->ms.support.PrefetchSupported[j] == true && mode_lib->ms.support.DynamicMetadataSupported[j] == true && + mode_lib->ms.support.VRatioInPrefetchSupported[j] == true && + // consider flip support is okay if when there is no hostvm and the user does't require a iflip OR the flip bw is ok + // If there is hostvm, DCN needs to support iflip for invalidation + ((s->ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j] == true)) || + (s->NextMaxVStartup == s->MaxVStartupAllPlanes[j] && s->AllPrefetchModeTested))); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.use_one_row_for_frame_this_state[k] = mode_lib->ms.use_one_row_for_frame[j][k]; + } + + s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; + s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; + s->mSOCParameters.WritebackLatency = mode_lib->ms.state.writeback_latency_us; + s->mSOCParameters.DRAMClockChangeLatency = mode_lib->ms.state.dram_clock_change_latency_us; + s->mSOCParameters.FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; + s->mSOCParameters.SRExitTime = mode_lib->ms.state.sr_exit_time_us; + s->mSOCParameters.SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; + s->mSOCParameters.SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; + s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->ms.state.sr_enter_plus_exit_z8_time_us; + s->mSOCParameters.USRRetrainingLatency = mode_lib->ms.state.usr_retraining_latency_us; + s->mSOCParameters.SMNLatency = mode_lib->ms.soc.smn_latency_us; + + CalculateWatermarks_params->USRRetrainingRequiredFinal = mode_lib->ms.policy.USRRetrainingRequiredFinal; + CalculateWatermarks_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateWatermarks_params->PrefetchMode = mode_lib->ms.PrefetchMode; + CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ms.ip.max_line_buffer_lines; + CalculateWatermarks_params->LineBufferSize = mode_lib->ms.ip.line_buffer_size_bits; + CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ms.ip.writeback_interface_buffer_size_kbytes; + CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLKState[j]; + CalculateWatermarks_params->ReturnBW = mode_lib->ms.ReturnBWPerState[j]; + CalculateWatermarks_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; + CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; + CalculateWatermarks_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; + CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateWatermarks_params->meta_row_height = mode_lib->ms.meta_row_height; + CalculateWatermarks_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; + CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; + CalculateWatermarks_params->WritebackChunkSize = mode_lib->ms.ip.writeback_chunk_size_kbytes; + CalculateWatermarks_params->SOCCLK = mode_lib->ms.state.socclk_mhz; + CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep[j]; + CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeYThisState; + CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeCThisState; + CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState; + CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState; + CalculateWatermarks_params->LBBitPerPixel = mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel; + CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; + CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; + CalculateWatermarks_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; + CalculateWatermarks_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; + CalculateWatermarks_params->VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps; + CalculateWatermarks_params->VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma; + CalculateWatermarks_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; + CalculateWatermarks_params->VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma; + CalculateWatermarks_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; + CalculateWatermarks_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; + CalculateWatermarks_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; + CalculateWatermarks_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; + CalculateWatermarks_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPPThisState; + CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; + CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; + CalculateWatermarks_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; + CalculateWatermarks_params->WritebackPixelFormat = mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat; + CalculateWatermarks_params->WritebackDestinationWidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth; + CalculateWatermarks_params->WritebackDestinationHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight; + CalculateWatermarks_params->WritebackSourceHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight; + CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabledThisState; + CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByteThisState; + + // Output + CalculateWatermarks_params->Watermark = &s->dummy_watermark; // Watermarks *Watermark + CalculateWatermarks_params->DRAMClockChangeSupport = &mode_lib->ms.support.DRAMClockChangeSupport[j]; + CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // dml_float_t *MaxActiveDRAMClockChangeLatencySupported[] + CalculateWatermarks_params->SubViewportLinesNeededInMALL = &mode_lib->ms.SubViewportLinesNeededInMALL[j]; // dml_uint_t SubViewportLinesNeededInMALL[] + CalculateWatermarks_params->FCLKChangeSupport = &mode_lib->ms.support.FCLKChangeSupport[j]; + CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // dml_float_t *MaxActiveFCLKChangeLatencySupported + CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport[j]; + CalculateWatermarks_params->ActiveDRAMClockChangeLatencyMargin = mode_lib->ms.support.ActiveDRAMClockChangeLatencyMargin; + + CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, + CalculateWatermarks_params); + + } // for j +} + +/// @brief The Mode Support function. +dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib) +{ + struct dml_core_mode_support_locals_st *s = &mode_lib->scratch.dml_core_mode_support_locals; + struct UseMinimumDCFCLK_params_st *UseMinimumDCFCLK_params = &mode_lib->scratch.UseMinimumDCFCLK_params; + struct CalculateSwathAndDETConfiguration_params_st *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; + struct CalculateVMRowAndSwath_params_st *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; + + dml_uint_t j, k, m; + + mode_lib->ms.num_active_planes = dml_get_num_active_planes(&mode_lib->ms.cache_display_cfg); + dml_print("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); + + CalculateMaxDETAndMinCompressedBufferSize( + mode_lib->ms.ip.config_return_buffer_size_in_kbytes, + mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes, + mode_lib->ms.ip.rob_buffer_size_kbytes, + mode_lib->ms.ip.max_num_dpp, + mode_lib->ms.policy.NomDETInKByteOverrideEnable, // VBA_DELTA + mode_lib->ms.policy.NomDETInKByteOverrideValue, // VBA_DELTA + + /* Output */ + &mode_lib->ms.MaxTotalDETInKByte, + &mode_lib->ms.NomDETInKByte, + &mode_lib->ms.MinCompressedBufferSizeInKByte); + + PixelClockAdjustmentForProgressiveToInterlaceUnit(&mode_lib->ms.cache_display_cfg, mode_lib->ms.ip.ptoi_supported); + + + /*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/ + + /*Scale Ratio, taps Support Check*/ + mode_lib->ms.support.ScaleRatioAndTapsSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k] == false + && ((mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) + || mode_lib->ms.cache_display_cfg.plane.HRatio[k] != 1.0 + || mode_lib->ms.cache_display_cfg.plane.HTaps[k] != 1.0 + || mode_lib->ms.cache_display_cfg.plane.VRatio[k] != 1.0 + || mode_lib->ms.cache_display_cfg.plane.VTaps[k] != 1.0)) { + mode_lib->ms.support.ScaleRatioAndTapsSupport = false; + } else if (mode_lib->ms.cache_display_cfg.plane.VTaps[k] < 1.0 || mode_lib->ms.cache_display_cfg.plane.VTaps[k] > 8.0 + || mode_lib->ms.cache_display_cfg.plane.HTaps[k] < 1.0 || mode_lib->ms.cache_display_cfg.plane.HTaps[k] > 8.0 + || (mode_lib->ms.cache_display_cfg.plane.HTaps[k] > 1.0 && (mode_lib->ms.cache_display_cfg.plane.HTaps[k] % 2) == 1) + || mode_lib->ms.cache_display_cfg.plane.HRatio[k] > mode_lib->ms.ip.max_hscl_ratio + || mode_lib->ms.cache_display_cfg.plane.VRatio[k] > mode_lib->ms.ip.max_vscl_ratio + || mode_lib->ms.cache_display_cfg.plane.HRatio[k] > mode_lib->ms.cache_display_cfg.plane.HTaps[k] + || mode_lib->ms.cache_display_cfg.plane.VRatio[k] > mode_lib->ms.cache_display_cfg.plane.VTaps[k] + || (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe + && (mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] < 1 || mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] > 8 || mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] < 1 || mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] > 8 || + (mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] > 1 && mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] % 2 == 1) || + mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k] > mode_lib->ms.ip.max_hscl_ratio || + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k] > mode_lib->ms.ip.max_vscl_ratio || + mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k] > mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k] || + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k] > mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]))) { + mode_lib->ms.support.ScaleRatioAndTapsSupport = false; + } + } + + /*Source Format, Pixel Format and Scan Support Check*/ + mode_lib->ms.support.SourceFormatPixelAndScanSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear && (!(!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k])) || mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true)) { + mode_lib->ms.support.SourceFormatPixelAndScanSupport = false; + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateBytePerPixelAndBlockSizes( + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], + + /* Output */ + &mode_lib->ms.BytePerPixelY[k], + &mode_lib->ms.BytePerPixelC[k], + &mode_lib->ms.BytePerPixelInDETY[k], + &mode_lib->ms.BytePerPixelInDETC[k], + &mode_lib->ms.Read256BlockHeightY[k], + &mode_lib->ms.Read256BlockHeightC[k], + &mode_lib->ms.Read256BlockWidthY[k], + &mode_lib->ms.Read256BlockWidthC[k], + &mode_lib->ms.MacroTileHeightY[k], + &mode_lib->ms.MacroTileHeightC[k], + &mode_lib->ms.MacroTileWidthY[k], + &mode_lib->ms.MacroTileWidthC[k]); + } + + /*Bandwidth Support Check*/ + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k])) { + mode_lib->ms.SwathWidthYSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportWidth[k]; + mode_lib->ms.SwathWidthCSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma[k]; + } else { + mode_lib->ms.SwathWidthYSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; + mode_lib->ms.SwathWidthCSingleDPP[k] = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; + } + } + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.ReadBandwidthLuma[k] = mode_lib->ms.SwathWidthYSingleDPP[k] * dml_ceil(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + mode_lib->ms.ReadBandwidthChroma[k] = mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * dml_ceil(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k] / 2.0; + } + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true + && mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k] == dml_444_64) { + mode_lib->ms.WriteBandwidth[k] = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] + * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] + / (mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] + * mode_lib->ms.cache_display_cfg.timing.HTotal[k] + / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 8.0; + } else if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + mode_lib->ms.WriteBandwidth[k] = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] + * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] + / (mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] + * mode_lib->ms.cache_display_cfg.timing.HTotal[k] + / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 4.0; + } else { + mode_lib->ms.WriteBandwidth[k] = 0.0; + } + } + + /*Writeback Latency support check*/ + mode_lib->ms.support.WritebackLatencySupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true && + (mode_lib->ms.WriteBandwidth[k] > mode_lib->ms.ip.writeback_interface_buffer_size_kbytes * 1024 / mode_lib->ms.state.writeback_latency_us)) { + mode_lib->ms.support.WritebackLatencySupport = false; + } + } + + /*Writeback Mode Support Check*/ + s->TotalNumberOfActiveWriteback = 0; + for (k = 0; k <= (dml_uint_t) mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + s->TotalNumberOfActiveWriteback = s->TotalNumberOfActiveWriteback + 1; + } + } + + mode_lib->ms.support.EnoughWritebackUnits = 1; + if (s->TotalNumberOfActiveWriteback > (dml_uint_t) mode_lib->ms.ip.max_num_wb) { + mode_lib->ms.support.EnoughWritebackUnits = false; + } + + /*Writeback Scale Ratio and Taps Support Check*/ + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] > mode_lib->ms.ip.writeback_max_hscl_ratio + || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] > mode_lib->ms.ip.writeback_max_vscl_ratio + || mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] < mode_lib->ms.ip.writeback_min_hscl_ratio + || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] < mode_lib->ms.ip.writeback_min_vscl_ratio + || mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] > (dml_uint_t) mode_lib->ms.ip.writeback_max_hscl_taps + || mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] > (dml_uint_t) mode_lib->ms.ip.writeback_max_vscl_taps + || mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k] > (dml_uint_t) mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] + || mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k] > (dml_uint_t) mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] + || (mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] > 2.0 && ((mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k] % 2) == 1))) { + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; + } + if (2.0 * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * (mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k] - 1) * 57 > mode_lib->ms.ip.writeback_line_buffer_buffer_size) { + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; + } + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateSinglePipeDPPCLKAndSCLThroughput( + mode_lib->ms.cache_display_cfg.plane.HRatio[k], + mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + mode_lib->ms.ip.max_dchub_pscl_bw_pix_per_clk, + mode_lib->ms.ip.max_pscl_lb_bw_pix_per_clk, + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.plane.HTaps[k], + mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k], + mode_lib->ms.cache_display_cfg.plane.VTaps[k], + mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k], + /* Output */ + &mode_lib->ms.PSCL_FACTOR[k], + &mode_lib->ms.PSCL_FACTOR_CHROMA[k], + &mode_lib->ms.MinDPPCLKUsingSingleDPP[k]); + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear) { + s->MaximumSwathWidthSupportLuma = 8192; + } else if (!dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelC[k] > 0 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) { + s->MaximumSwathWidthSupportLuma = 7680; + } else if (dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelC[k] > 0 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe_alpha) { + s->MaximumSwathWidthSupportLuma = 4320; + } else if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_rgbe_alpha) { + s->MaximumSwathWidthSupportLuma = 3840; + } else if (dml_is_vertical_rotation(mode_lib->ms.cache_display_cfg.plane.SourceScan[k]) && mode_lib->ms.BytePerPixelY[k] == 8 && mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { + s->MaximumSwathWidthSupportLuma = 3072; + } else { + s->MaximumSwathWidthSupportLuma = 6144; + } + + if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_8 || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_10 || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_12) { + s->MaximumSwathWidthSupportChroma = (dml_uint_t)(s->MaximumSwathWidthSupportLuma / 2.0); + } else { + s->MaximumSwathWidthSupportChroma = s->MaximumSwathWidthSupportLuma; + } + mode_lib->ms.MaximumSwathWidthInLineBufferLuma = mode_lib->ms.ip.line_buffer_size_bits * dml_max(mode_lib->ms.cache_display_cfg.plane.HRatio[k], 1.0) / mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel[k] / + (mode_lib->ms.cache_display_cfg.plane.VTaps[k] + dml_max(dml_ceil(mode_lib->ms.cache_display_cfg.plane.VRatio[k], 1.0) - 2, 0.0)); + if (mode_lib->ms.BytePerPixelC[k] == 0.0) { + mode_lib->ms.MaximumSwathWidthInLineBufferChroma = 0; + } else { + mode_lib->ms.MaximumSwathWidthInLineBufferChroma = + mode_lib->ms.ip.line_buffer_size_bits + * dml_max(mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], 1.0) + / mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel[k] + / (mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k] + + dml_max(dml_ceil(mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], 1.0) - 2, 0.0)); + } + mode_lib->ms.MaximumSwathWidthLuma[k] = dml_min(s->MaximumSwathWidthSupportLuma, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); + mode_lib->ms.MaximumSwathWidthChroma[k] = dml_min(s->MaximumSwathWidthSupportChroma, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); + } + + /*Number Of DSC Slices*/ + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && + mode_lib->ms.cache_display_cfg.output.DSCEnable[k] != dml_dsc_disable) { + mode_lib->ms.support.NumberOfDSCSlices[k] = mode_lib->ms.cache_display_cfg.output.DSCSlices[k]; + + if (mode_lib->ms.support.NumberOfDSCSlices[k] == 0) { + if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 4800) { + mode_lib->ms.support.NumberOfDSCSlices[k] = (dml_uint_t)(dml_ceil(mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 600, 4)); + } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 2400) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 8; + } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 1200) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 4; + } else if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] > 340) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 2; + } else { + mode_lib->ms.support.NumberOfDSCSlices[k] = 1; + } + } + } else { + mode_lib->ms.support.NumberOfDSCSlices[k] = 1; + } + } + + CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; + CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; + CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = 1; + CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; + CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->Output = mode_lib->ms.cache_display_cfg.output.OutputEncoder; + CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; + CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; + CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; + CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; + CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; + CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; + CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; + CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; + CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; + CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; + CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; + CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; + CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; + CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; + CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; + CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; + CalculateSwathAndDETConfiguration_params->ODMMode = s->dummy_odm_mode; + CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; + CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; + CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; + CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; + CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = s->dummy_integer_array[0]; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_integer_array[1]; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_integer_array[2]; + CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_integer_array[3]; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_integer_array[4]; + CalculateSwathAndDETConfiguration_params->SwathHeightY = s->dummy_integer_array[5]; + CalculateSwathAndDETConfiguration_params->SwathHeightC = s->dummy_integer_array[6]; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = s->dummy_integer_array[7]; + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &s->dummy_boolean[0]; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[2]; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &s->dummy_integer[1]; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &s->dummy_integer[0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = mode_lib->ms.SingleDPPViewportSizeSupportPerSurface; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[1]; + + CalculateSwathAndDETConfiguration(&mode_lib->scratch, + CalculateSwathAndDETConfiguration_params); /* dml_bool_t *ViewportSizeSupport */ + + s->MPCCombineMethodAsNeededForPStateChangeAndVoltage = false; + s->MPCCombineMethodAsPossible = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_as_needed_for_pstate_and_voltage) + s->MPCCombineMethodAsNeededForPStateChangeAndVoltage = true; + if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_as_possible) + s->MPCCombineMethodAsPossible = true; + } + mode_lib->ms.support.MPCCombineMethodIncompatible = s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && s->MPCCombineMethodAsPossible; + + for (j = 0; j < 2; j++) { + mode_lib->ms.TotalNumberOfActiveDPP[j] = 0; + mode_lib->ms.support.TotalAvailablePipesSupport[j] = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateODMMode( + mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit, + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.policy.ODMUse[k], + mode_lib->ms.state.dispclk_mhz, + mode_lib->ms.max_state.dispclk_mhz, + false, // DSCEnable + mode_lib->ms.TotalNumberOfActiveDPP[j], + mode_lib->ms.ip.max_num_dpp, + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.soc.dcn_downspread_percent, + mode_lib->ms.ip.dispclk_ramp_margin_percent, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, + mode_lib->ms.support.NumberOfDSCSlices[k], + + /* Output */ + &s->TotalAvailablePipesSupportNoDSC, + &s->NumberOfDPPNoDSC, + &s->ODMModeNoDSC, + &s->RequiredDISPCLKPerSurfaceNoDSC); + + CalculateODMMode( + mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit, + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.policy.ODMUse[k], + mode_lib->ms.state.dispclk_mhz, + mode_lib->ms.max_state.dispclk_mhz, + true, // DSCEnable + mode_lib->ms.TotalNumberOfActiveDPP[j], + mode_lib->ms.ip.max_num_dpp, + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.soc.dcn_downspread_percent, + mode_lib->ms.ip.dispclk_ramp_margin_percent, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, + mode_lib->ms.support.NumberOfDSCSlices[k], + + /* Output */ + &s->TotalAvailablePipesSupportDSC, + &s->NumberOfDPPDSC, + &s->ODMModeDSC, + &s->RequiredDISPCLKPerSurfaceDSC); + + CalculateOutputLink( + mode_lib->ms.state.phyclk_mhz, + mode_lib->ms.state.phyclk_d18_mhz, + mode_lib->ms.state.phyclk_d32_mhz, + mode_lib->ms.soc.phy_downspread_percent, + (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k), + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k], + mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k], + mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], + mode_lib->ms.support.NumberOfDSCSlices[k], + mode_lib->ms.cache_display_cfg.output.AudioSampleRate[k], + mode_lib->ms.cache_display_cfg.output.AudioSampleLayout[k], + s->ODMModeNoDSC, + s->ODMModeDSC, + mode_lib->ms.cache_display_cfg.output.DSCEnable[k], + mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k], + mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k], + + /* Output */ + &mode_lib->ms.RequiresDSC[k], + &mode_lib->ms.RequiresFEC[k], + &mode_lib->ms.OutputBppPerState[k], + &mode_lib->ms.OutputTypePerState[k], // VBA_DELTA, VBA uses a string to represent type and rate, but DML uses enum, don't want to rely on strng + &mode_lib->ms.OutputRatePerState[k], + &mode_lib->ms.RequiredSlots[k]); + + if (mode_lib->ms.RequiresDSC[k] == false) { + mode_lib->ms.ODMModePerState[k] = s->ODMModeNoDSC; + mode_lib->ms.RequiredDISPCLKPerSurface[j][k] = s->RequiredDISPCLKPerSurfaceNoDSC; + if (!s->TotalAvailablePipesSupportNoDSC) + mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; + mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + s->NumberOfDPPNoDSC; + } else { + mode_lib->ms.ODMModePerState[k] = s->ODMModeDSC; + mode_lib->ms.RequiredDISPCLKPerSurface[j][k] = s->RequiredDISPCLKPerSurfaceDSC; + if (!s->TotalAvailablePipesSupportDSC) + mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; + mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + s->NumberOfDPPDSC; + } + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { + mode_lib->ms.MPCCombine[j][k] = false; + mode_lib->ms.NoOfDPP[j][k] = 4; + } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { + mode_lib->ms.MPCCombine[j][k] = false; + mode_lib->ms.NoOfDPP[j][k] = 2; + } else if (mode_lib->ms.policy.MPCCombineUse[k] == dml_mpc_disabled) { + mode_lib->ms.MPCCombine[j][k] = false; + mode_lib->ms.NoOfDPP[j][k] = 1; + } else if (RoundToDFSGranularity(mode_lib->ms.MinDPPCLKUsingSingleDPP[k] * (1 + mode_lib->ms.soc.dcn_downspread_percent / 100), + 1, mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz) <= mode_lib->ms.state.dppclk_mhz && + mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k] == true) { + mode_lib->ms.MPCCombine[j][k] = false; + mode_lib->ms.NoOfDPP[j][k] = 1; + } else if (mode_lib->ms.TotalNumberOfActiveDPP[j] < (dml_uint_t) mode_lib->ms.ip.max_num_dpp) { + mode_lib->ms.MPCCombine[j][k] = true; + mode_lib->ms.NoOfDPP[j][k] = 2; + mode_lib->ms.TotalNumberOfActiveDPP[j] = (dml_uint_t) mode_lib->ms.TotalNumberOfActiveDPP[j] + 1; + } else { + mode_lib->ms.MPCCombine[j][k] = false; + mode_lib->ms.NoOfDPP[j][k] = 1; + mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; + } + } + + mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = 0; + s->NoChromaOrLinear = true; + for (k = 0; k < (dml_uint_t) mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.NoOfDPP[j][k] == 1) + mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] + 1; + if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_8 + || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_10 + || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_420_12 + || mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] == dml_rgbe_alpha + || mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k] == dml_sw_linear) { + s->NoChromaOrLinear = false; + } + } + + if (j == 1 && !UnboundedRequest(mode_lib->ms.policy.UseUnboundedRequesting, + mode_lib->ms.TotalNumberOfActiveDPP[j], s->NoChromaOrLinear, + mode_lib->ms.cache_display_cfg.output.OutputEncoder[0])) { + while (!(mode_lib->ms.TotalNumberOfActiveDPP[j] >= (dml_uint_t) mode_lib->ms.ip.max_num_dpp || mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] == 0)) { + s->BWOfNonCombinedSurfaceOfMaximumBandwidth = 0; + s->NumberOfNonCombinedSurfaceOfMaximumBandwidth = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.policy.MPCCombineUse[k] != dml_mpc_disabled && mode_lib->ms.policy.MPCCombineUse[k] != dml_mpc_as_needed_for_voltage && + mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k] > s->BWOfNonCombinedSurfaceOfMaximumBandwidth && + (mode_lib->ms.ODMModePerState[k] != dml_odm_mode_combine_2to1 && mode_lib->ms.ODMModePerState[k] != dml_odm_mode_combine_4to1) && + mode_lib->ms.MPCCombine[j][k] == false) { + s->BWOfNonCombinedSurfaceOfMaximumBandwidth = mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k]; + s->NumberOfNonCombinedSurfaceOfMaximumBandwidth = k; + } + } + mode_lib->ms.MPCCombine[j][s->NumberOfNonCombinedSurfaceOfMaximumBandwidth] = true; + mode_lib->ms.NoOfDPP[j][s->NumberOfNonCombinedSurfaceOfMaximumBandwidth] = 2; + mode_lib->ms.TotalNumberOfActiveDPP[j] = mode_lib->ms.TotalNumberOfActiveDPP[j] + 1; + mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] = mode_lib->ms.TotalNumberOfSingleDPPSurfaces[j] - 1; + } + } + + //DISPCLK/DPPCLK + mode_lib->ms.WritebackRequiredDISPCLK = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k]) { + mode_lib->ms.WritebackRequiredDISPCLK = dml_max(mode_lib->ms.WritebackRequiredDISPCLK, + CalculateWriteBackDISPCLK(mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceWidth[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.ip.writeback_line_buffer_buffer_size, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz)); + } + } + + mode_lib->ms.RequiredDISPCLK[j] = mode_lib->ms.WritebackRequiredDISPCLK; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.RequiredDISPCLK[j] = dml_max(mode_lib->ms.RequiredDISPCLK[j], mode_lib->ms.RequiredDISPCLKPerSurface[j][k]); + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; + } + + CalculateDPPCLK(mode_lib->ms.num_active_planes, + mode_lib->ms.soc.dcn_downspread_percent, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, + mode_lib->ms.MinDPPCLKUsingSingleDPP, + mode_lib->ms.NoOfDPPThisState, + /* Output */ + &mode_lib->ms.GlobalDPPCLK, + mode_lib->ms.RequiredDPPCLKThisState); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.RequiredDPPCLKPerSurface[j][k] = mode_lib->ms.RequiredDPPCLKThisState[k]; + } + + mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] = !((mode_lib->ms.RequiredDISPCLK[j] > mode_lib->ms.state.dispclk_mhz) || (mode_lib->ms.GlobalDPPCLK > mode_lib->ms.state.dppclk_mhz)); + + if (mode_lib->ms.TotalNumberOfActiveDPP[j] > (dml_uint_t) mode_lib->ms.ip.max_num_dpp) { + mode_lib->ms.support.TotalAvailablePipesSupport[j] = false; + } + } // j + + /* Total Available OTG, HDMIFRL, DP Support Check */ + s->TotalNumberOfActiveOTG = 0; + s->TotalNumberOfActiveHDMIFRL = 0; + s->TotalNumberOfActiveDP2p0 = 0; + s->TotalNumberOfActiveDP2p0Outputs = 0; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + s->TotalNumberOfActiveOTG = s->TotalNumberOfActiveOTG + 1; + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) + s->TotalNumberOfActiveHDMIFRL = s->TotalNumberOfActiveHDMIFRL + 1; + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0) { + s->TotalNumberOfActiveDP2p0 = s->TotalNumberOfActiveDP2p0 + 1; + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k || mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == false) { + s->TotalNumberOfActiveDP2p0Outputs = s->TotalNumberOfActiveDP2p0Outputs + 1; + } + } + } + } + + mode_lib->ms.support.NumberOfOTGSupport = (s->TotalNumberOfActiveOTG <= (dml_uint_t) mode_lib->ms.ip.max_num_otg); + mode_lib->ms.support.NumberOfHDMIFRLSupport = (s->TotalNumberOfActiveHDMIFRL <= (dml_uint_t) mode_lib->ms.ip.max_num_hdmi_frl_outputs); + mode_lib->ms.support.NumberOfDP2p0Support = (s->TotalNumberOfActiveDP2p0 <= (dml_uint_t) mode_lib->ms.ip.max_num_dp2p0_streams && s->TotalNumberOfActiveDP2p0Outputs <= (dml_uint_t) mode_lib->ms.ip.max_num_dp2p0_outputs); + + /* Display IO and DSC Support Check */ + mode_lib->ms.support.NonsupportedDSCInputBPC = false; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.output.OutputDisabled[k] == false && + !(mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 12.0 + || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 10.0 + || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] == 8.0 + || mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k] > (dml_uint_t) mode_lib->ms.ip.maximum_dsc_bits_per_component + )) { + mode_lib->ms.support.NonsupportedDSCInputBPC = true; + } + } + + mode_lib->ms.support.ExceededMultistreamSlots = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k) { + s->TotalSlots = mode_lib->ms.RequiredSlots[k]; + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[j] == k) + s->TotalSlots = s->TotalSlots + mode_lib->ms.RequiredSlots[j]; + } + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp && s->TotalSlots > 63) + mode_lib->ms.support.ExceededMultistreamSlots = true; + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 && s->TotalSlots > 64) + mode_lib->ms.support.ExceededMultistreamSlots = true; + } + } + mode_lib->ms.support.LinkCapacitySupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.output.OutputDisabled[k] == false && + mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) && mode_lib->ms.OutputBppPerState[k] == 0) { + mode_lib->ms.support.LinkCapacitySupport = false; + } + } + + mode_lib->ms.support.P2IWith420 = false; + mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = false; + mode_lib->ms.support.DSC422NativeNotSupported = false; + mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = false; + mode_lib->ms.support.LinkRateForMultistreamNotIndicated = false; + mode_lib->ms.support.BPPForMultistreamNotIndicated = false; + mode_lib->ms.support.MultistreamWithHDMIOreDP = false; + mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = false; + mode_lib->ms.support.NotEnoughLanesForMSO = false; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl)) { + if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420 && mode_lib->ms.cache_display_cfg.timing.Interlace[k] == 1 && mode_lib->ms.ip.ptoi_supported == true) + mode_lib->ms.support.P2IWith420 = true; + + if (mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable_if_necessary && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] != 0) + mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = true; + if ((mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable || mode_lib->ms.cache_display_cfg.output.DSCEnable[k] == dml_dsc_enable_if_necessary) && mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 && !mode_lib->ms.ip.dsc422_native_support) + mode_lib->ms.support.DSC422NativeNotSupported = true; + + if (((mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr2 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_hbr3) && + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp && mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_edp) || + ((mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr10 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr13p5 || mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_uhbr20) && + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp2p0)) + mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = true; + + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k && mode_lib->ms.cache_display_cfg.output.OutputLinkDPRate[k] == dml_dp_rate_na) + mode_lib->ms.support.LinkRateForMultistreamNotIndicated = true; + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] == 0) + mode_lib->ms.support.BPPForMultistreamNotIndicated = true; + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == j && mode_lib->ms.cache_display_cfg.output.ForcedOutputLinkBPP[k] == 0) + mode_lib->ms.support.BPPForMultistreamNotIndicated = true; + } + } + + if ((mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl)) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1 && mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == k) + mode_lib->ms.support.MultistreamWithHDMIOreDP = true; + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { + if (mode_lib->ms.cache_display_cfg.output.OutputMultistreamEn[k] == 1 && mode_lib->ms.cache_display_cfg.output.OutputMultistreamId[k] == j) + mode_lib->ms.support.MultistreamWithHDMIOreDP = true; + } + } + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] != dml_dp && (mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_split_1to2 || + mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to2 || mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to4)) + mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = true; + + if ((mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to2 && mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k] < 2) || + (mode_lib->ms.policy.ODMUse[k] == dml_odm_use_policy_mso_1to4 && mode_lib->ms.cache_display_cfg.output.OutputLinkDPLanes[k] < 4)) + mode_lib->ms.support.NotEnoughLanesForMSO = true; + } + } + + mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl && + RequiredDTBCLK( + mode_lib->ms.RequiresDSC[k], + mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.OutputBppPerState[k], + mode_lib->ms.support.NumberOfDSCSlices[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.output.AudioSampleRate[k], + mode_lib->ms.cache_display_cfg.output.AudioSampleLayout[k]) > mode_lib->ms.state.dtbclk_mhz) { + mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = true; + } + } + + mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK = true; + mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1 && mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi) { + mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK = false; + } + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k && mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1 && (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmi)) { + mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK = false; + } + } + + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = false; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + if (mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_dp2p0 || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_edp || + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) { + if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420) { + s->DSCFormatFactor = 2; + } else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_444) { + s->DSCFormatFactor = 1; + } else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) { + s->DSCFormatFactor = 2; + } else { + s->DSCFormatFactor = 1; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); +#endif + if (mode_lib->ms.RequiresDSC[k] == true) { + if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { + if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 12.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); + dml_print("DML::%s: k=%u, DSCCLKPerState = %f\n", __func__, k, mode_lib->ms.state.dscclk_mhz); + dml_print("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); +#endif + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; + } + } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { + if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 6.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; + } + } else { + if (mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 3.0 / (dml_float_t)s->DSCFormatFactor > (1.0 - mode_lib->ms.soc.dcn_downspread_percent / 100.0) * mode_lib->ms.state.dscclk_mhz) { + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; + } + } + } + } + } + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: DSCCLKRequiredMoreThanSupported = %u\n", __func__, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); +#endif + + /* Check DSC Unit and Slices Support */ + mode_lib->ms.support.NotEnoughDSCUnits = false; + mode_lib->ms.support.NotEnoughDSCSlices = false; + s->TotalDSCUnitsRequired = 0; + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.RequiresDSC[k] == true) { + if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_4to1) { + if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > 4 * (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; + s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 4; + if (mode_lib->ms.support.NumberOfDSCSlices[k] > 16) + mode_lib->ms.support.NotEnoughDSCSlices = true; + } else if (mode_lib->ms.ODMModePerState[k] == dml_odm_mode_combine_2to1) { + if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > 2 * (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; + s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 2; + if (mode_lib->ms.support.NumberOfDSCSlices[k] > 8) + mode_lib->ms.support.NotEnoughDSCSlices = true; + } else { + if (mode_lib->ms.cache_display_cfg.timing.HActive[k] > (dml_uint_t) mode_lib->ms.ip.maximum_pixels_per_line_per_dsc_unit) + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; + s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + 1; + if (mode_lib->ms.support.NumberOfDSCSlices[k] > 4) + mode_lib->ms.support.NotEnoughDSCSlices = true; + } + } + } + if (s->TotalDSCUnitsRequired > (dml_uint_t) mode_lib->ms.ip.num_dsc) { + mode_lib->ms.support.NotEnoughDSCUnits = true; + } + + /*DSC Delay per state*/ + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.DSCDelayPerState[k] = DSCDelayRequirement(mode_lib->ms.RequiresDSC[k], + mode_lib->ms.ODMModePerState[k], + mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], + mode_lib->ms.OutputBppPerState[k], + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.support.NumberOfDSCSlices[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { + for (j = 0; j <= mode_lib->ms.num_active_planes - 1; j++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == m && mode_lib->ms.RequiresDSC[m] == true) { + mode_lib->ms.DSCDelayPerState[k] = mode_lib->ms.DSCDelayPerState[m]; + } + } + } + } + + //Calculate Swath, DET Configuration, DCFCLKDeepSleep + // + for (j = 0; j < 2; ++j) { + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.RequiredDPPCLKThisState[k] = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; + mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; + mode_lib->ms.ODMModeThisState[k] = mode_lib->ms.ODMModePerState[k]; + } + + CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; + CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; + CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; + CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; + CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->Output = mode_lib->ms.cache_display_cfg.output.OutputEncoder; + CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; + CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; + CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; + CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; + CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; + CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; + CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; + CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; + CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; + CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; + CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; + CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; + CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; + CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; + CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; + CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; + CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.ODMModeThisState; + CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; + CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; + CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; + CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; + CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.NoOfDPPThisState; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub_this_state; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub_this_state; + CalculateSwathAndDETConfiguration_params->SwathWidth = mode_lib->ms.SwathWidthYThisState; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = mode_lib->ms.SwathWidthCThisState; + CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->ms.SwathHeightYThisState; + CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->ms.SwathHeightCThisState; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->ms.DETBufferSizeInKByteThisState; + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeYThisState; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeCThisState; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->ms.UnboundedRequestEnabledThisState; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[2]; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &s->dummy_integer[1]; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->ms.CompressedBufferSizeInkByteThisState; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = s->dummy_boolean_array[0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &mode_lib->ms.support.ViewportSizeSupport[j]; + + CalculateSwathAndDETConfiguration(&mode_lib->scratch, + CalculateSwathAndDETConfiguration_params); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.swath_width_luma_ub_all_states[j][k] = mode_lib->ms.swath_width_luma_ub_this_state[k]; + mode_lib->ms.swath_width_chroma_ub_all_states[j][k] = mode_lib->ms.swath_width_chroma_ub_this_state[k]; + mode_lib->ms.SwathWidthYAllStates[j][k] = mode_lib->ms.SwathWidthYThisState[k]; + mode_lib->ms.SwathWidthCAllStates[j][k] = mode_lib->ms.SwathWidthCThisState[k]; + mode_lib->ms.SwathHeightYAllStates[j][k] = mode_lib->ms.SwathHeightYThisState[k]; + mode_lib->ms.SwathHeightCAllStates[j][k] = mode_lib->ms.SwathHeightCThisState[k]; + mode_lib->ms.UnboundedRequestEnabledAllStates[j] = mode_lib->ms.UnboundedRequestEnabledThisState; + mode_lib->ms.CompressedBufferSizeInkByteAllStates[j] = mode_lib->ms.CompressedBufferSizeInkByteThisState; + mode_lib->ms.DETBufferSizeInKByteAllStates[j][k] = mode_lib->ms.DETBufferSizeInKByteThisState[k]; + mode_lib->ms.DETBufferSizeYAllStates[j][k] = mode_lib->ms.DETBufferSizeYThisState[k]; + mode_lib->ms.DETBufferSizeCAllStates[j][k] = mode_lib->ms.DETBufferSizeCThisState[k]; + } + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.cursor_bw[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + } + + CalculateSurfaceSizeInMall( + mode_lib->ms.num_active_planes, + mode_lib->ms.soc.mall_allocated_for_dcn_mbytes, + mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen, + mode_lib->ms.cache_display_cfg.surface.DCCEnable, + mode_lib->ms.cache_display_cfg.plane.ViewportStationary, + mode_lib->ms.cache_display_cfg.plane.ViewportXStart, + mode_lib->ms.cache_display_cfg.plane.ViewportYStart, + mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, + mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, + mode_lib->ms.cache_display_cfg.plane.ViewportWidth, + mode_lib->ms.cache_display_cfg.plane.ViewportHeight, + mode_lib->ms.BytePerPixelY, + mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma, + mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, + mode_lib->ms.Read256BlockWidthY, + mode_lib->ms.Read256BlockWidthC, + mode_lib->ms.Read256BlockHeightY, + mode_lib->ms.Read256BlockHeightC, + mode_lib->ms.MacroTileWidthY, + mode_lib->ms.MacroTileWidthC, + mode_lib->ms.MacroTileHeightY, + mode_lib->ms.MacroTileHeightC, + + /* Output */ + mode_lib->ms.SurfaceSizeInMALL, + &mode_lib->ms.support.ExceededMALLSize); + + for (j = 0; j < 2; j++) { + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.swath_width_luma_ub_this_state[k] = mode_lib->ms.swath_width_luma_ub_all_states[j][k]; + mode_lib->ms.swath_width_chroma_ub_this_state[k] = mode_lib->ms.swath_width_chroma_ub_all_states[j][k]; + mode_lib->ms.SwathWidthYThisState[k] = mode_lib->ms.SwathWidthYAllStates[j][k]; + mode_lib->ms.SwathWidthCThisState[k] = mode_lib->ms.SwathWidthCAllStates[j][k]; + mode_lib->ms.SwathHeightYThisState[k] = mode_lib->ms.SwathHeightYAllStates[j][k]; + mode_lib->ms.SwathHeightCThisState[k] = mode_lib->ms.SwathHeightCAllStates[j][k]; + mode_lib->ms.DETBufferSizeInKByteThisState[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[j][k]; + mode_lib->ms.DETBufferSizeYThisState[k] = mode_lib->ms.DETBufferSizeYAllStates[j][k]; + mode_lib->ms.DETBufferSizeCThisState[k] = mode_lib->ms.DETBufferSizeCAllStates[j][k]; + mode_lib->ms.RequiredDPPCLKThisState[k] = mode_lib->ms.RequiredDPPCLKPerSurface[j][k]; + mode_lib->ms.NoOfDPPThisState[k] = mode_lib->ms.NoOfDPP[j][k]; + } + + mode_lib->ms.TotalNumberOfDCCActiveDPP[j] = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { + mode_lib->ms.TotalNumberOfDCCActiveDPP[j] = mode_lib->ms.TotalNumberOfDCCActiveDPP[j] + mode_lib->ms.NoOfDPP[j][k]; + } + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->SurfParameters[k].PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; + s->SurfParameters[k].DPPPerSurface = mode_lib->ms.NoOfDPP[j][k]; + s->SurfParameters[k].SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; + s->SurfParameters[k].ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; + s->SurfParameters[k].ViewportHeightChroma = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; + s->SurfParameters[k].BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; + s->SurfParameters[k].BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; + s->SurfParameters[k].BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; + s->SurfParameters[k].BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; + s->SurfParameters[k].BlockWidthY = mode_lib->ms.MacroTileWidthY[k]; + s->SurfParameters[k].BlockHeightY = mode_lib->ms.MacroTileHeightY[k]; + s->SurfParameters[k].BlockWidthC = mode_lib->ms.MacroTileWidthC[k]; + s->SurfParameters[k].BlockHeightC = mode_lib->ms.MacroTileHeightC[k]; + s->SurfParameters[k].InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; + s->SurfParameters[k].HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; + s->SurfParameters[k].DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; + s->SurfParameters[k].SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; + s->SurfParameters[k].SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k]; + s->SurfParameters[k].BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; + s->SurfParameters[k].BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; + s->SurfParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + s->SurfParameters[k].VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + s->SurfParameters[k].VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; + s->SurfParameters[k].VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps[k]; + s->SurfParameters[k].VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]; + s->SurfParameters[k].PitchY = mode_lib->ms.cache_display_cfg.surface.PitchY[k]; + s->SurfParameters[k].DCCMetaPitchY = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; + s->SurfParameters[k].PitchC = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; + s->SurfParameters[k].DCCMetaPitchC = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; + s->SurfParameters[k].ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary[k]; + s->SurfParameters[k].ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart[k]; + s->SurfParameters[k].ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart[k]; + s->SurfParameters[k].ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC[k]; + s->SurfParameters[k].ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC[k]; + s->SurfParameters[k].FORCE_ONE_ROW_FOR_FRAME = mode_lib->ms.cache_display_cfg.plane.ForceOneRowForFrame[k]; + s->SurfParameters[k].SwathHeightY = mode_lib->ms.SwathHeightYThisState[k]; + s->SurfParameters[k].SwathHeightC = mode_lib->ms.SwathHeightCThisState[k]; + } + + CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; + CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; + CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; + CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; + CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthYThisState; + CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthCThisState; + CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; + CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; + CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceededPerState; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceededPerState; + CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[0]; + CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[1]; + CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; + CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; + CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[2]; // VBA_DELTA + CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[3]; // VBA_DELTA + CalculateVMRowAndSwath_params->meta_req_width = s->dummy_integer_array[4]; + CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[5]; + CalculateVMRowAndSwath_params->meta_req_height = s->dummy_integer_array[6]; + CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[7]; + CalculateVMRowAndSwath_params->meta_row_width = s->dummy_integer_array[8]; + CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[9]; + CalculateVMRowAndSwath_params->meta_row_height = mode_lib->ms.meta_row_height; + CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->ms.meta_row_height_chroma; + CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[10]; + CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[11]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[12]; + CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[13]; + CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[14]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[15]; + CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[16]; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[17]; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[18]; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[19]; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[20]; + CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesYThisState; + CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesCThisState; + CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; + CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; + CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwY; + CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwC; + CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bandwidth_this_state; + CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bandwidth_this_state; + CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRowThisState; + CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState; + CalculateVMRowAndSwath_params->MetaRowByte = mode_lib->ms.MetaRowBytesThisState; + CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame_this_state; + CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip_this_state; + CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = s->dummy_boolean_array[0]; + CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; + CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[21]; + + CalculateVMRowAndSwath(&mode_lib->scratch, + CalculateVMRowAndSwath_params); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.PrefetchLinesY[j][k] = mode_lib->ms.PrefetchLinesYThisState[k]; + mode_lib->ms.PrefetchLinesC[j][k] = mode_lib->ms.PrefetchLinesCThisState[k]; + mode_lib->ms.meta_row_bandwidth[j][k] = mode_lib->ms.meta_row_bandwidth_this_state[k]; + mode_lib->ms.dpte_row_bandwidth[j][k] = mode_lib->ms.dpte_row_bandwidth_this_state[k]; + mode_lib->ms.DPTEBytesPerRow[j][k] = mode_lib->ms.DPTEBytesPerRowThisState[k]; + mode_lib->ms.PDEAndMetaPTEBytesPerFrame[j][k] = mode_lib->ms.PDEAndMetaPTEBytesPerFrameThisState[k]; + mode_lib->ms.MetaRowBytes[j][k] = mode_lib->ms.MetaRowBytesThisState[k]; + mode_lib->ms.use_one_row_for_frame[j][k] = mode_lib->ms.use_one_row_for_frame_this_state[k]; + mode_lib->ms.use_one_row_for_frame_flip[j][k] = mode_lib->ms.use_one_row_for_frame_flip_this_state[k]; + } + + mode_lib->ms.support.PTEBufferSizeNotExceeded[j] = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.PTEBufferSizeNotExceededPerState[k] == false) + mode_lib->ms.support.PTEBufferSizeNotExceeded[j] = false; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: j=%u k=%u, PTEBufferSizeNotExceededPerState[%u] = %u\n", __func__, j, k, k, mode_lib->ms.PTEBufferSizeNotExceededPerState[k]); +#endif + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: PTEBufferSizeNotExceeded[%u] = %u\n", __func__, j, mode_lib->ms.support.PTEBufferSizeNotExceeded[j]); +#endif + + mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.DCCMetaBufferSizeNotExceededPerState[k] == false) + mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] = false; + } + + mode_lib->ms.UrgLatency = CalculateUrgentLatency(mode_lib->ms.state.urgent_latency_pixel_data_only_us, + mode_lib->ms.state.urgent_latency_pixel_mixed_with_vm_data_us, + mode_lib->ms.state.urgent_latency_vm_data_only_us, + mode_lib->ms.soc.do_urgent_latency_adjustment, + mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_component_us, + mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_reference_mhz, + mode_lib->ms.state.fabricclk_mhz); + + /* Getter functions work at mp interface so copy the urgent latency to mp*/ + mode_lib->mp.UrgentLatency = mode_lib->ms.UrgLatency; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateUrgentBurstFactor( + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + mode_lib->ms.swath_width_luma_ub_this_state[k], + mode_lib->ms.swath_width_chroma_ub_this_state[k], + mode_lib->ms.SwathHeightYThisState[k], + mode_lib->ms.SwathHeightCThisState[k], + (dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.UrgLatency, + mode_lib->ms.ip.cursor_buffer_size, + mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], + mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + mode_lib->ms.BytePerPixelInDETY[k], + mode_lib->ms.BytePerPixelInDETC[k], + mode_lib->ms.DETBufferSizeYThisState[k], + mode_lib->ms.DETBufferSizeCThisState[k], + /* Output */ + &mode_lib->ms.UrgentBurstFactorCursor[j][k], + &mode_lib->ms.UrgentBurstFactorLuma[j][k], + &mode_lib->ms.UrgentBurstFactorChroma[j][k], + &mode_lib->ms.NotUrgentLatencyHiding[k]); + } + + CalculateDCFCLKDeepSleep( + mode_lib->ms.num_active_planes, + mode_lib->ms.BytePerPixelY, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.cache_display_cfg.plane.VRatio, + mode_lib->ms.cache_display_cfg.plane.VRatioChroma, + mode_lib->ms.SwathWidthYThisState, + mode_lib->ms.SwathWidthCThisState, + mode_lib->ms.NoOfDPPThisState, + mode_lib->ms.cache_display_cfg.plane.HRatio, + mode_lib->ms.cache_display_cfg.plane.HRatioChroma, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + mode_lib->ms.PSCL_FACTOR, + mode_lib->ms.PSCL_FACTOR_CHROMA, + mode_lib->ms.RequiredDPPCLKThisState, + mode_lib->ms.ReadBandwidthLuma, + mode_lib->ms.ReadBandwidthChroma, + mode_lib->ms.soc.return_bus_width_bytes, + + /* Output */ + &mode_lib->ms.ProjectedDCFCLKDeepSleep[j]); + } + + //Calculate Return BW + for (j = 0; j < 2; ++j) { + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + mode_lib->ms.WritebackDelayTime[k] = mode_lib->ms.state.writeback_latency_us + CalculateWriteBackDelay( + mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / mode_lib->ms.RequiredDISPCLK[j]; + } else { + mode_lib->ms.WritebackDelayTime[k] = 0.0; + } + for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[m] == k && mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[m] == true) { + mode_lib->ms.WritebackDelayTime[k] = dml_max(mode_lib->ms.WritebackDelayTime[k], + mode_lib->ms.state.writeback_latency_us + CalculateWriteBackDelay( + mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[m], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[m], + mode_lib->ms.cache_display_cfg.timing.HTotal[m]) / mode_lib->ms.RequiredDISPCLK[j]); + } + } + } + } + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + for (m = 0; m <= mode_lib->ms.num_active_planes - 1; m++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == m) { + mode_lib->ms.WritebackDelayTime[k] = mode_lib->ms.WritebackDelayTime[m]; + } + } + } + s->MaxVStartupAllPlanes[j] = 0; // max vstartup among all planes + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + s->MaximumVStartup[j][k] = CalculateMaxVStartup(k, + mode_lib->ms.ip.ptoi_supported, + mode_lib->ms.ip.vblank_nom_default_us, + &mode_lib->ms.cache_display_cfg.timing, + mode_lib->ms.WritebackDelayTime[k]); + + s->MaxVStartupAllPlanes[j] = (dml_uint_t)(dml_max(s->MaxVStartupAllPlanes[j], s->MaximumVStartup[j][k])); +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, MaxVStartupAllPlanes[%u] = %u\n", __func__, k, j, s->MaxVStartupAllPlanes[j]); + dml_print("DML::%s: k=%u, MaximumVStartup[%u][%u] = %u\n", __func__, k, j, k, s->MaximumVStartup[j][k]); +#endif + } + } + + s->ReorderingBytes = (dml_uint_t)(mode_lib->ms.soc.num_chans * dml_max3(mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_only_bytes, + mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, + mode_lib->ms.soc.urgent_out_of_order_return_per_channel_vm_only_bytes)); + + for (j = 0; j < 2; ++j) { + mode_lib->ms.DCFCLKState[j] = mode_lib->ms.state.dcfclk_mhz; + } + + /* Immediate Flip and MALL parameters */ + s->ImmediateFlipRequiredFinal = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->ImmediateFlipRequiredFinal = s->ImmediateFlipRequiredFinal || (mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); + } + + mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified || + ((mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_required) && + (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required)); + } + mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified && s->ImmediateFlipRequiredFinal; + + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe || ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == true || mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) && + (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame || mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe)); + } + + mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen || + ((mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_enable || mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_optimize) && (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe)) || + ((mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_disable || mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen[k] == dml_use_mall_static_screen_optimize) && (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame)); + } + + s->FullFrameMALLPStateMethod = false; + s->SubViewportMALLPStateMethod = false; + s->PhantomPipeMALLPStateMethod = false; + s->SubViewportMALLRefreshGreaterThan120Hz = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_full_frame) + s->FullFrameMALLPStateMethod = true; + if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_sub_viewport) { + s->SubViewportMALLPStateMethod = true; + if (mode_lib->ms.cache_display_cfg.timing.RefreshRate[k] > 120) + s->SubViewportMALLRefreshGreaterThan120Hz = true; + } + if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] == dml_use_mall_pstate_change_phantom_pipe) + s->PhantomPipeMALLPStateMethod = true; + } + mode_lib->ms.support.InvalidCombinationOfMALLUseForPState = (s->SubViewportMALLPStateMethod != s->PhantomPipeMALLPStateMethod) + || (s->SubViewportMALLPStateMethod && s->FullFrameMALLPStateMethod) || s->SubViewportMALLRefreshGreaterThan120Hz; + + if (mode_lib->ms.policy.UseMinimumRequiredDCFCLK == true) { + UseMinimumDCFCLK_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + UseMinimumDCFCLK_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; + UseMinimumDCFCLK_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; + UseMinimumDCFCLK_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; + UseMinimumDCFCLK_params->MaxPrefetchMode = dml_prefetch_support_stutter; + UseMinimumDCFCLK_params->DRAMClockChangeLatencyFinal = mode_lib->ms.state.dram_clock_change_latency_us; + UseMinimumDCFCLK_params->FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; + UseMinimumDCFCLK_params->SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; + UseMinimumDCFCLK_params->ReturnBusWidth = mode_lib->ms.soc.return_bus_width_bytes; + UseMinimumDCFCLK_params->RoundTripPingLatencyCycles = mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles; + UseMinimumDCFCLK_params->ReorderingBytes = s->ReorderingBytes; + UseMinimumDCFCLK_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; + UseMinimumDCFCLK_params->MetaChunkSize = mode_lib->ms.ip.meta_chunk_size_kbytes; + UseMinimumDCFCLK_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + UseMinimumDCFCLK_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + UseMinimumDCFCLK_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + UseMinimumDCFCLK_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + UseMinimumDCFCLK_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + UseMinimumDCFCLK_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + UseMinimumDCFCLK_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; + UseMinimumDCFCLK_params->ImmediateFlipRequirement = s->ImmediateFlipRequiredFinal; + UseMinimumDCFCLK_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + UseMinimumDCFCLK_params->MaxAveragePercentOfIdealSDPPortBWDisplayCanUseInNormalSystemOperation = mode_lib->ms.soc.max_avg_sdp_bw_use_normal_percent; + UseMinimumDCFCLK_params->PercentOfIdealSDPPortBWReceivedAfterUrgLatency = mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent; + UseMinimumDCFCLK_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; + UseMinimumDCFCLK_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; + UseMinimumDCFCLK_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes; + UseMinimumDCFCLK_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired; + UseMinimumDCFCLK_params->Interlace = mode_lib->ms.cache_display_cfg.timing.Interlace; + UseMinimumDCFCLK_params->RequiredDPPCLKPerSurface = mode_lib->ms.RequiredDPPCLKPerSurface; + UseMinimumDCFCLK_params->RequiredDISPCLK = mode_lib->ms.RequiredDISPCLK; + UseMinimumDCFCLK_params->UrgLatency = mode_lib->ms.UrgLatency; + UseMinimumDCFCLK_params->NoOfDPP = mode_lib->ms.NoOfDPP; + UseMinimumDCFCLK_params->ProjectedDCFCLKDeepSleep = mode_lib->ms.ProjectedDCFCLKDeepSleep; + UseMinimumDCFCLK_params->MaximumVStartup = s->MaximumVStartup; + UseMinimumDCFCLK_params->TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP; + UseMinimumDCFCLK_params->TotalNumberOfDCCActiveDPP = mode_lib->ms.TotalNumberOfDCCActiveDPP; + UseMinimumDCFCLK_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + UseMinimumDCFCLK_params->PrefetchLinesY = mode_lib->ms.PrefetchLinesY; + UseMinimumDCFCLK_params->PrefetchLinesC = mode_lib->ms.PrefetchLinesC; + UseMinimumDCFCLK_params->swath_width_luma_ub_all_states = mode_lib->ms.swath_width_luma_ub_all_states; + UseMinimumDCFCLK_params->swath_width_chroma_ub_all_states = mode_lib->ms.swath_width_chroma_ub_all_states; + UseMinimumDCFCLK_params->BytePerPixelY = mode_lib->ms.BytePerPixelY; + UseMinimumDCFCLK_params->BytePerPixelC = mode_lib->ms.BytePerPixelC; + UseMinimumDCFCLK_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; + UseMinimumDCFCLK_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; + UseMinimumDCFCLK_params->PDEAndMetaPTEBytesPerFrame = mode_lib->ms.PDEAndMetaPTEBytesPerFrame; + UseMinimumDCFCLK_params->DPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow; + UseMinimumDCFCLK_params->MetaRowBytes = mode_lib->ms.MetaRowBytes; + UseMinimumDCFCLK_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable; + UseMinimumDCFCLK_params->ReadBandwidthLuma = mode_lib->ms.ReadBandwidthLuma; + UseMinimumDCFCLK_params->ReadBandwidthChroma = mode_lib->ms.ReadBandwidthChroma; + UseMinimumDCFCLK_params->DCFCLKPerState = mode_lib->ms.state.dcfclk_mhz; + UseMinimumDCFCLK_params->DCFCLKState = mode_lib->ms.DCFCLKState; + + UseMinimumDCFCLK(&mode_lib->scratch, + UseMinimumDCFCLK_params); + + } // UseMinimumRequiredDCFCLK == true + + for (j = 0; j < 2; ++j) { + mode_lib->ms.ReturnBWPerState[j] = dml_get_return_bw_mbps(&mode_lib->ms.soc, mode_lib->ms.state.use_ideal_dram_bw_strobe, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, mode_lib->ms.DCFCLKState[j], mode_lib->ms.state.fabricclk_mhz, + mode_lib->ms.state.dram_speed_mts); + mode_lib->ms.ReturnDRAMBWPerState[j] = dml_get_return_dram_bw_mbps(&mode_lib->ms.soc, mode_lib->ms.state.use_ideal_dram_bw_strobe, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.state.dram_speed_mts); + } + + //Re-ordering Buffer Support Check + for (j = 0; j < 2; ++j) { + if ((mode_lib->ms.ip.rob_buffer_size_kbytes - mode_lib->ms.ip.pixel_chunk_size_kbytes) * 1024 / mode_lib->ms.ReturnBWPerState[j] > + (mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles + 32) / mode_lib->ms.DCFCLKState[j] + s->ReorderingBytes / mode_lib->ms.ReturnBWPerState[j]) { + mode_lib->ms.support.ROBSupport[j] = true; + } else { + mode_lib->ms.support.ROBSupport[j] = false; + } + dml_print("DML::%s: DEBUG ROBSupport[%u] = %u (%u)\n", __func__, j, mode_lib->ms.support.ROBSupport[j], __LINE__); + } + + //Vertical Active BW support check + s->MaxTotalVActiveRDBandwidth = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->MaxTotalVActiveRDBandwidth = s->MaxTotalVActiveRDBandwidth + mode_lib->ms.ReadBandwidthLuma[k] + mode_lib->ms.ReadBandwidthChroma[k]; + } + + for (j = 0; j < 2; ++j) { + mode_lib->ms.support.MaxTotalVerticalActiveAvailableBandwidth[j] = dml_min3(mode_lib->ms.soc.return_bus_width_bytes * mode_lib->ms.DCFCLKState[j] * mode_lib->ms.soc.max_avg_sdp_bw_use_normal_percent / 100.0, + mode_lib->ms.state.fabricclk_mhz * mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes * mode_lib->ms.soc.max_avg_fabric_bw_use_normal_percent / 100.0, + mode_lib->ms.state.dram_speed_mts * mode_lib->ms.soc.num_chans * mode_lib->ms.soc.dram_channel_width_bytes * + ((mode_lib->ms.state.use_ideal_dram_bw_strobe && !mode_lib->ms.cache_display_cfg.plane.HostVMEnable) ? + mode_lib->ms.soc.max_avg_dram_bw_use_normal_strobe_percent : mode_lib->ms.soc.max_avg_dram_bw_use_normal_percent) / 100.0); + + if (s->MaxTotalVActiveRDBandwidth <= mode_lib->ms.support.MaxTotalVerticalActiveAvailableBandwidth[j]) { + mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] = true; + } else { + mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] = false; + } + } + + /* Prefetch Check */ + dml_prefetch_check(mode_lib); + + // End of Prefetch Check + dml_print("DML::%s: Done prefetch calculation\n", __func__); + + /*Cursor Support Check*/ + mode_lib->ms.support.CursorSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] > 0.0) { + if (mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] == 64 && mode_lib->ms.ip.cursor_64bpp_support == false) { + mode_lib->ms.support.CursorSupport = false; + } + } + } + + /*Valid Pitch Check*/ + mode_lib->ms.support.PitchSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.support.AlignedYPitch[k] = dml_ceil( + dml_max(mode_lib->ms.cache_display_cfg.surface.PitchY[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k]), + mode_lib->ms.MacroTileWidthY[k]); + if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { + mode_lib->ms.support.AlignedDCCMetaPitchY[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k]), 64.0 * mode_lib->ms.Read256BlockWidthY[k]); + } else { + mode_lib->ms.support.AlignedDCCMetaPitchY[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; + } + if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_16 + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe + && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_mono_8) { + mode_lib->ms.support.AlignedCPitch[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.PitchC[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k]), mode_lib->ms.MacroTileWidthC[k]); + if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k] == true) { + mode_lib->ms.support.AlignedDCCMetaPitchC[k] = dml_ceil(dml_max(mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k], mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k]), 64.0 * mode_lib->ms.Read256BlockWidthC[k]); + } else { + mode_lib->ms.support.AlignedDCCMetaPitchC[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; + } + } else { + mode_lib->ms.support.AlignedCPitch[k] = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; + mode_lib->ms.support.AlignedDCCMetaPitchC[k] = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; + } + if (mode_lib->ms.support.AlignedYPitch[k] > mode_lib->ms.cache_display_cfg.surface.PitchY[k] || mode_lib->ms.support.AlignedCPitch[k] > mode_lib->ms.cache_display_cfg.surface.PitchC[k] || + mode_lib->ms.support.AlignedDCCMetaPitchY[k] > mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k] || mode_lib->ms.support.AlignedDCCMetaPitchC[k] > mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]) { + mode_lib->ms.support.PitchSupport = false; + } + } + + mode_lib->ms.support.ViewportExceedsSurface = false; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.ViewportWidth[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k] || mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY[k]) { + mode_lib->ms.support.ViewportExceedsSurface = true; + if (mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_64 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_32 && + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_16 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_444_8 && mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k] != dml_rgbe) { + if (mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k] || mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k] > mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC[k]) { + mode_lib->ms.support.ViewportExceedsSurface = true; + } + } + } + } + + /*Mode Support, Voltage State and SOC Configuration*/ + for (j = 0; j < 2; j++) { // j iterator is for the combine mode off or on + dml_print("DML::%s: checking support for j=%u\n", __func__, j); + dml_print("DML::%s: state_idx=%0d max_state_idx=%0d\n", __func__, mode_lib->ms.state_idx, mode_lib->ms.max_state_idx); + + s->is_max_pwr_state = (mode_lib->ms.max_state_idx == mode_lib->ms.state_idx); + s->is_max_dram_pwr_state = (mode_lib->ms.max_state.dram_speed_mts == mode_lib->ms.state.dram_speed_mts); + + s->dram_clock_change_support = (!mode_lib->ms.policy.DRAMClockChangeRequirementFinal || + (s->is_max_dram_pwr_state && mode_lib->policy.AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported) || + mode_lib->ms.support.DRAMClockChangeSupport[j] != dml_dram_clock_change_unsupported); + s->f_clock_change_support = (!mode_lib->ms.policy.FCLKChangeRequirementFinal || + (s->is_max_pwr_state && mode_lib->policy.AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported) || + mode_lib->ms.support.FCLKChangeSupport[j] != dml_fclock_change_unsupported); + + if (mode_lib->ms.support.ScaleRatioAndTapsSupport == true + && mode_lib->ms.support.SourceFormatPixelAndScanSupport == true + && mode_lib->ms.support.ViewportSizeSupport[j] == true + && !mode_lib->ms.support.LinkRateDoesNotMatchDPVersion + && !mode_lib->ms.support.LinkRateForMultistreamNotIndicated + && !mode_lib->ms.support.BPPForMultistreamNotIndicated + && !mode_lib->ms.support.MultistreamWithHDMIOreDP + && !mode_lib->ms.support.ExceededMultistreamSlots + && !mode_lib->ms.support.MSOOrODMSplitWithNonDPLink + && !mode_lib->ms.support.NotEnoughLanesForMSO + && mode_lib->ms.support.LinkCapacitySupport == true + && !mode_lib->ms.support.P2IWith420 + && !mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP + && !mode_lib->ms.support.DSC422NativeNotSupported + && !mode_lib->ms.support.MPCCombineMethodIncompatible + && mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK == true + && mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK == true + && mode_lib->ms.support.NotEnoughDSCUnits == false + && !mode_lib->ms.support.NotEnoughDSCSlices + && !mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe + && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen + && mode_lib->ms.support.DSCCLKRequiredMoreThanSupported == false + && mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport + && mode_lib->ms.support.DTBCLKRequiredMoreThanSupported == false + && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPState + && !mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified + && mode_lib->ms.support.ROBSupport[j] == true + && mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] == true + && mode_lib->ms.support.TotalAvailablePipesSupport[j] == true + && mode_lib->ms.support.NumberOfOTGSupport == true + && mode_lib->ms.support.NumberOfHDMIFRLSupport == true + && mode_lib->ms.support.NumberOfDP2p0Support == true + && mode_lib->ms.support.EnoughWritebackUnits == true + && mode_lib->ms.support.WritebackLatencySupport == true + && mode_lib->ms.support.WritebackScaleRatioAndTapsSupport == true + && mode_lib->ms.support.CursorSupport == true + && mode_lib->ms.support.PitchSupport == true + && mode_lib->ms.support.ViewportExceedsSurface == false + && mode_lib->ms.support.PrefetchSupported[j] == true + && mode_lib->ms.support.VActiveBandwithSupport[j] == true + && mode_lib->ms.support.DynamicMetadataSupported[j] == true + && mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] == true + && mode_lib->ms.support.VRatioInPrefetchSupported[j] == true + && mode_lib->ms.support.PTEBufferSizeNotExceeded[j] == true + && mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] == true + && mode_lib->ms.support.NonsupportedDSCInputBPC == false + && !mode_lib->ms.support.ExceededMALLSize + && ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == false && !s->ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j]) + && s->dram_clock_change_support == true + && s->f_clock_change_support == true + && (!mode_lib->ms.policy.USRRetrainingRequiredFinal || mode_lib->ms.support.USRRetrainingSupport[j])) { + dml_print("DML::%s: mode is supported\n", __func__); + mode_lib->ms.support.ModeSupport[j] = true; + } else { + dml_print("DML::%s: mode is NOT supported\n", __func__); + mode_lib->ms.support.ModeSupport[j] = false; + dml_print_mode_support(mode_lib, j); + } + } + + mode_lib->ms.support.MaximumMPCCombine = 0; + mode_lib->ms.support.ModeIsSupported = 0; + if (mode_lib->ms.support.ModeSupport[0] == true || mode_lib->ms.support.ModeSupport[1] == true) { // if the mode is supported by either no combine or mpccombine + mode_lib->ms.support.ModeIsSupported = mode_lib->ms.support.ModeSupport[0] == true || mode_lib->ms.support.ModeSupport[1] == true; + + // Determine if MPC combine is necessary, depends on if using MPC combine will help dram clock change or fclk change, etc. + if ((mode_lib->ms.support.ModeSupport[0] == false && mode_lib->ms.support.ModeSupport[1] == true) || s->MPCCombineMethodAsPossible || + (s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && mode_lib->ms.policy.DRAMClockChangeRequirementFinal && + (((mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vactive_w_mall_sub_vp) && + !(mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive || mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive_w_mall_sub_vp)) || + ((mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr + || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_w_mall_full_frame || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr_w_mall_full_frame + || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_w_mall_sub_vp || mode_lib->ms.support.DRAMClockChangeSupport[1] == dml_dram_clock_change_vblank_drr_w_mall_sub_vp + ) && + mode_lib->ms.support.DRAMClockChangeSupport[0] == dml_dram_clock_change_unsupported))) + || (s->MPCCombineMethodAsNeededForPStateChangeAndVoltage && mode_lib->ms.policy.FCLKChangeRequirementFinal && + ((mode_lib->ms.support.FCLKChangeSupport[1] == dml_fclock_change_vactive && mode_lib->ms.support.FCLKChangeSupport[0] != dml_fclock_change_vactive) || + (mode_lib->ms.support.FCLKChangeSupport[1] == dml_fclock_change_vblank && mode_lib->ms.support.FCLKChangeSupport[0] == dml_fclock_change_unsupported)))) { + mode_lib->ms.support.MaximumMPCCombine = 1; + } else { + mode_lib->ms.support.MaximumMPCCombine = 0; + } + } + + // Since now the mode_support work on 1 particular power state, so there is only 1 state idx (index 0). + mode_lib->ms.support.ImmediateFlipSupport = mode_lib->ms.support.ImmediateFlipSupportedForState[mode_lib->ms.support.MaximumMPCCombine]; // Consider flip support if max combine support imm flip + mode_lib->ms.support.UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabledAllStates[mode_lib->ms.support.MaximumMPCCombine]; // Not used, informational + mode_lib->ms.support.CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByteAllStates[mode_lib->ms.support.MaximumMPCCombine]; // Not used, informational + + dml_print("DML::%s: ModeIsSupported = %u\n", __func__, mode_lib->ms.support.ModeIsSupported); + dml_print("DML::%s: MaximumMPCCombine = %u\n", __func__, mode_lib->ms.support.MaximumMPCCombine); + dml_print("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); + dml_print("DML::%s: UnboundedRequestEnabled = %u\n", __func__, mode_lib->ms.support.UnboundedRequestEnabled); + dml_print("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, mode_lib->ms.support.CompressedBufferSizeInkByte); + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.support.MPCCombineEnable[k] = mode_lib->ms.MPCCombine[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.support.DPPPerSurface[k] = mode_lib->ms.NoOfDPP[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.SwathHeightY[k] = mode_lib->ms.SwathHeightYAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.SwathHeightC[k] = mode_lib->ms.SwathHeightCAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.DETBufferSizeInKByte[k] = mode_lib->ms.DETBufferSizeInKByteAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.DETBufferSizeY[k] = mode_lib->ms.DETBufferSizeYAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; + mode_lib->ms.DETBufferSizeC[k] = mode_lib->ms.DETBufferSizeCAllStates[mode_lib->ms.support.MaximumMPCCombine][k]; + } + + mode_lib->ms.DRAMSpeed = mode_lib->ms.state.dram_speed_mts; + mode_lib->ms.FabricClock = mode_lib->ms.state.fabricclk_mhz; + mode_lib->ms.SOCCLK = mode_lib->ms.state.socclk_mhz; + mode_lib->ms.DCFCLK = mode_lib->ms.DCFCLKState[mode_lib->ms.support.MaximumMPCCombine]; + mode_lib->ms.ReturnBW = mode_lib->ms.ReturnBWPerState[mode_lib->ms.support.MaximumMPCCombine]; + mode_lib->ms.ReturnDRAMBW = mode_lib->ms.ReturnDRAMBWPerState[mode_lib->ms.support.MaximumMPCCombine]; + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + mode_lib->ms.support.ODMMode[k] = mode_lib->ms.ODMModePerState[k]; + } else { + mode_lib->ms.support.ODMMode[k] = dml_odm_mode_bypass; + } + + mode_lib->ms.support.DSCEnabled[k] = mode_lib->ms.RequiresDSC[k]; + mode_lib->ms.support.FECEnabled[k] = mode_lib->ms.RequiresFEC[k]; + mode_lib->ms.support.OutputBpp[k] = mode_lib->ms.OutputBppPerState[k]; + mode_lib->ms.support.OutputType[k] = mode_lib->ms.OutputTypePerState[k]; + mode_lib->ms.support.OutputRate[k] = mode_lib->ms.OutputRatePerState[k]; + mode_lib->ms.support.SubViewportLinesNeededInMALL[k] = mode_lib->ms.SubViewportLinesNeededInMALL[k]; + } + + return mode_lib->ms.support.ModeIsSupported; +} // dml_core_mode_support + +/// @brief This function calculates some parameters thats are needed ahead of the mode programming function all +void dml_core_mode_support_partial(struct display_mode_lib_st *mode_lib) +{ + CalculateMaxDETAndMinCompressedBufferSize( + mode_lib->ms.ip.config_return_buffer_size_in_kbytes, + mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes, + mode_lib->ms.ip.rob_buffer_size_kbytes, + mode_lib->ms.ip.max_num_dpp, + mode_lib->ms.policy.NomDETInKByteOverrideEnable, + mode_lib->ms.policy.NomDETInKByteOverrideValue, + + /* Output */ + &mode_lib->ms.MaxTotalDETInKByte, + &mode_lib->ms.NomDETInKByte, + &mode_lib->ms.MinCompressedBufferSizeInKByte); + + PixelClockAdjustmentForProgressiveToInterlaceUnit(&mode_lib->ms.cache_display_cfg, mode_lib->ms.ip.ptoi_supported); + + mode_lib->ms.ReturnBW = dml_get_return_bw_mbps(&mode_lib->ms.soc, + mode_lib->ms.state.use_ideal_dram_bw_strobe, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.DCFCLK, + mode_lib->ms.FabricClock, + mode_lib->ms.DRAMSpeed); + dml_print("DML::%s: ReturnBW = %f\n", __func__, mode_lib->ms.ReturnBW); + +} // dml_core_mode_support_partial + +/// @brief This is the mode programming function. It is assumed the display cfg is support at the given power state +void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struct dml_clk_cfg_st *clk_cfg) +{ + struct dml_core_mode_programming_locals_st *s = &mode_lib->scratch.dml_core_mode_programming_locals; + struct CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params_st *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; + struct CalculateVMRowAndSwath_params_st *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; + struct CalculateSwathAndDETConfiguration_params_st *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; + struct CalculateStutterEfficiency_params_st *CalculateStutterEfficiency_params = &mode_lib->scratch.CalculateStutterEfficiency_params; + struct CalculatePrefetchSchedule_params_st *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; + + struct mode_program_st *locals = &mode_lib->mp; + struct DmlPipe *myPipe; + dml_uint_t j = 0, k = 0; + dml_float_t TWait; + dml_bool_t isInterlaceTiming; + + mode_lib->ms.num_active_planes = dml_get_num_active_planes(&mode_lib->ms.cache_display_cfg); + mode_lib->mp.num_active_pipes = dml_get_num_active_pipes(&mode_lib->ms.cache_display_cfg); + dml_calc_pipe_plane_mapping(&mode_lib->ms.cache_display_cfg.hw, mode_lib->mp.pipe_plane); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: --- START --- \n", __func__); + dml_print("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); + dml_print("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); +#endif + + s->DSCFormatFactor = 0; + + // Unlike dppclk and dispclk which can be calculated in mode_programming + // DCFCLK is calculated in mode_support (which is the state bbox dcfclk or min dcfclk if min dcfclk option is used in mode support calculation) + if (clk_cfg->dcfclk_option != dml_use_override_freq) + locals->Dcfclk = mode_lib->ms.DCFCLK; + else + locals->Dcfclk = clk_cfg->dcfclk_mhz; + +#ifdef __DML_VBA_DEBUG__ + dml_print_dml_policy(&mode_lib->ms.policy); + dml_print_soc_state_bounding_box(&mode_lib->ms.state); + dml_print_soc_bounding_box(&mode_lib->ms.soc); + dml_print_clk_cfg(clk_cfg); + + dml_print("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); + dml_print("DML::%s: Using DCFCLK = %f\n", __func__, locals->Dcfclk); + dml_print("DML::%s: Using SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); +#endif + + locals->WritebackDISPCLK = 0.0; + locals->GlobalDPPCLK = 0.0; + + // DISPCLK and DPPCLK Calculation + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k]) { + locals->WritebackDISPCLK = + dml_max( + locals->WritebackDISPCLK, + CalculateWriteBackDISPCLK( + mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceWidth[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.ip.writeback_line_buffer_buffer_size, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz)); + } + } + + locals->Dispclk_calculated = locals->WritebackDISPCLK; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + locals->Dispclk_calculated = dml_max(locals->Dispclk_calculated, CalculateRequiredDispclk( + mode_lib->ms.cache_display_cfg.hw.ODMMode[k], + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.soc.dcn_downspread_percent, + mode_lib->ms.ip.dispclk_ramp_margin_percent, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, + mode_lib->ms.max_state.dispclk_mhz)); + } + } + if (clk_cfg->dispclk_option == dml_use_required_freq) + locals->Dispclk = locals->Dispclk_calculated; + else if (clk_cfg->dispclk_option == dml_use_override_freq) + locals->Dispclk = clk_cfg->dispclk_mhz; + else + locals->Dispclk = mode_lib->ms.state.dispclk_mhz; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Using Dispclk = %f\n", __func__, locals->Dispclk); +#endif + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateSinglePipeDPPCLKAndSCLThroughput( + mode_lib->ms.cache_display_cfg.plane.HRatio[k], + mode_lib->ms.cache_display_cfg.plane.HRatioChroma[k], + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + mode_lib->ms.ip.max_dchub_pscl_bw_pix_per_clk, + mode_lib->ms.ip.max_pscl_lb_bw_pix_per_clk, + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.plane.HTaps[k], + mode_lib->ms.cache_display_cfg.plane.HTapsChroma[k], + mode_lib->ms.cache_display_cfg.plane.VTaps[k], + mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k], + + /* Output */ + &locals->PSCL_THROUGHPUT[k], + &locals->PSCL_THROUGHPUT_CHROMA[k], + &locals->DPPCLKUsingSingleDPP[k]); + } + + CalculateDPPCLK(mode_lib->ms.num_active_planes, + mode_lib->ms.soc.dcn_downspread_percent, + mode_lib->ms.soc.dispclk_dppclk_vco_speed_mhz, + locals->DPPCLKUsingSingleDPP, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + /* Output */ + &locals->GlobalDPPCLK, + locals->Dppclk_calculated); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (clk_cfg->dppclk_option[k] == dml_use_required_freq) + locals->Dppclk[k] = locals->Dppclk_calculated[k]; + else if (clk_cfg->dppclk_option[k] == dml_use_override_freq) + locals->Dppclk[k] = clk_cfg->dppclk_mhz[k]; + else + locals->Dppclk[k] = mode_lib->ms.state.dppclk_mhz; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Using Dppclk[%0d] = %f\n", __func__, k, locals->Dppclk[k]); +#endif + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateBytePerPixelAndBlockSizes( + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], + + /* Output */ + &locals->BytePerPixelY[k], + &locals->BytePerPixelC[k], + &locals->BytePerPixelDETY[k], + &locals->BytePerPixelDETC[k], + &locals->BlockHeight256BytesY[k], + &locals->BlockHeight256BytesC[k], + &locals->BlockWidth256BytesY[k], + &locals->BlockWidth256BytesC[k], + &locals->BlockHeightY[k], + &locals->BlockHeightC[k], + &locals->BlockWidthY[k], + &locals->BlockWidthC[k]); + } + + + dml_print("DML::%s: %u\n", __func__, __LINE__); + CalculateSwathWidth( + false, // ForceSingleDPP + mode_lib->ms.num_active_planes, + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat, + mode_lib->ms.cache_display_cfg.plane.SourceScan, + mode_lib->ms.cache_display_cfg.plane.ViewportStationary, + mode_lib->ms.cache_display_cfg.plane.ViewportWidth, + mode_lib->ms.cache_display_cfg.plane.ViewportHeight, + mode_lib->ms.cache_display_cfg.plane.ViewportXStart, + mode_lib->ms.cache_display_cfg.plane.ViewportYStart, + mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, + mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, + mode_lib->ms.cache_display_cfg.hw.ODMMode, + locals->BytePerPixelY, + locals->BytePerPixelC, + locals->BlockHeight256BytesY, + locals->BlockHeight256BytesC, + locals->BlockWidth256BytesY, + locals->BlockWidth256BytesC, + mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming, + mode_lib->ms.cache_display_cfg.timing.HActive, + mode_lib->ms.cache_display_cfg.plane.HRatio, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + + /* Output */ + locals->SwathWidthSingleDPPY, + locals->SwathWidthSingleDPPC, + locals->SwathWidthY, + locals->SwathWidthC, + s->dummy_integer_array[0], // dml_uint_t MaximumSwathHeightY[] + s->dummy_integer_array[1], // dml_uint_t MaximumSwathHeightC[] + locals->swath_width_luma_ub, + locals->swath_width_chroma_ub); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + locals->ReadBandwidthSurfaceLuma[k] = locals->SwathWidthSingleDPPY[k] * locals->BytePerPixelY[k] / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + locals->ReadBandwidthSurfaceChroma[k] = locals->SwathWidthSingleDPPC[k] * locals->BytePerPixelC[k] / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; + dml_print("DML::%s: ReadBandwidthSurfaceLuma[%i] = %fBps\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); + dml_print("DML::%s: ReadBandwidthSurfaceChroma[%i] = %fBps\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); + } + + CalculateSwathAndDETConfiguration_params->DETSizeOverride = mode_lib->ms.cache_display_cfg.plane.DETSizeOverride; + CalculateSwathAndDETConfiguration_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ms.ip.config_return_buffer_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; + CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; + CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; + CalculateSwathAndDETConfiguration_params->UseUnboundedRequestingFinal = mode_lib->ms.policy.UseUnboundedRequesting; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ms.ip.config_return_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByteFinal = mode_lib->ms.ip.compressed_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->Output = s->dummy_output_encoder_array; + CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = locals->ReadBandwidthSurfaceLuma; + CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = locals->ReadBandwidthSurfaceChroma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = s->dummy_single_array[0]; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = s->dummy_single_array[1]; + CalculateSwathAndDETConfiguration_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; + CalculateSwathAndDETConfiguration_params->ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary; + CalculateSwathAndDETConfiguration_params->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat; + CalculateSwathAndDETConfiguration_params->SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling; + CalculateSwathAndDETConfiguration_params->ViewportWidth = mode_lib->ms.cache_display_cfg.plane.ViewportWidth; + CalculateSwathAndDETConfiguration_params->ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight; + CalculateSwathAndDETConfiguration_params->ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart; + CalculateSwathAndDETConfiguration_params->ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart; + CalculateSwathAndDETConfiguration_params->ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC; + CalculateSwathAndDETConfiguration_params->ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC; + CalculateSwathAndDETConfiguration_params->SurfaceWidthY = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY; + CalculateSwathAndDETConfiguration_params->SurfaceWidthC = mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC; + CalculateSwathAndDETConfiguration_params->SurfaceHeightY = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY; + CalculateSwathAndDETConfiguration_params->SurfaceHeightC = mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = locals->BlockHeight256BytesY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = locals->BlockHeight256BytesC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = locals->BlockWidth256BytesY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = locals->BlockWidth256BytesC; + CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.cache_display_cfg.hw.ODMMode; + CalculateSwathAndDETConfiguration_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateSwathAndDETConfiguration_params->BytePerPixY = locals->BytePerPixelY; + CalculateSwathAndDETConfiguration_params->BytePerPixC = locals->BytePerPixelC; + CalculateSwathAndDETConfiguration_params->BytePerPixDETY = locals->BytePerPixelDETY; + CalculateSwathAndDETConfiguration_params->BytePerPixDETC = locals->BytePerPixelDETC; + CalculateSwathAndDETConfiguration_params->HActive = mode_lib->ms.cache_display_cfg.timing.HActive; + CalculateSwathAndDETConfiguration_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; + CalculateSwathAndDETConfiguration_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_long_array[0]; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_long_array[1]; + CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_long_array[2]; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_long_array[3]; + CalculateSwathAndDETConfiguration_params->SwathHeightY = locals->SwathHeightY; + CalculateSwathAndDETConfiguration_params->SwathHeightC = locals->SwathHeightC; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = locals->DETBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = locals->DETBufferSizeY; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = locals->DETBufferSizeC; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &locals->UnboundedRequestEnabled; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &locals->compbuf_reserved_space_64b; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_zs = &locals->compbuf_reserved_space_zs; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &locals->CompressedBufferSizeInkByte; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = &s->dummy_boolean_array[0][0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[0]; + + // VBA_DELTA + // Calculate DET size, swath height here. In VBA, they are calculated in mode check stage + CalculateSwathAndDETConfiguration(&mode_lib->scratch, + CalculateSwathAndDETConfiguration_params); + + // DCFCLK Deep Sleep + CalculateDCFCLKDeepSleep( + mode_lib->ms.num_active_planes, + locals->BytePerPixelY, + locals->BytePerPixelC, + mode_lib->ms.cache_display_cfg.plane.VRatio, + mode_lib->ms.cache_display_cfg.plane.VRatioChroma, + locals->SwathWidthY, + locals->SwathWidthC, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + mode_lib->ms.cache_display_cfg.plane.HRatio, + mode_lib->ms.cache_display_cfg.plane.HRatioChroma, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + locals->PSCL_THROUGHPUT, + locals->PSCL_THROUGHPUT_CHROMA, + locals->Dppclk, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + mode_lib->ms.soc.return_bus_width_bytes, + + /* Output */ + &locals->DCFCLKDeepSleep); + + // DSCCLK + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if ((mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] != k) || !mode_lib->ms.cache_display_cfg.hw.DSCEnabled[k]) { + locals->DSCCLK_calculated[k] = 0.0; + } else { + if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_420) + s->DSCFormatFactor = 2; + else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_444) + s->DSCFormatFactor = 1; + else if (mode_lib->ms.cache_display_cfg.output.OutputFormat[k] == dml_n422 || mode_lib->ms.cache_display_cfg.output.OutputEncoder[k] == dml_hdmifrl) + s->DSCFormatFactor = 2; + else + s->DSCFormatFactor = 1; + if (mode_lib->ms.cache_display_cfg.hw.ODMMode[k] == dml_odm_mode_combine_4to1) + locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 12 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); + else if (mode_lib->ms.cache_display_cfg.hw.ODMMode[k] == dml_odm_mode_combine_2to1) + locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 6 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); + else + locals->DSCCLK_calculated[k] = mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k] / 3 / s->DSCFormatFactor / (1 - mode_lib->ms.soc.dcn_downspread_percent / 100); + } + } + + // DSC Delay + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + locals->DSCDelay[k] = DSCDelayRequirement(mode_lib->ms.cache_display_cfg.hw.DSCEnabled[k], + mode_lib->ms.cache_display_cfg.hw.ODMMode[k], + mode_lib->ms.cache_display_cfg.output.DSCInputBitPerComponent[k], + mode_lib->ms.cache_display_cfg.output.OutputBpp[k], + mode_lib->ms.cache_display_cfg.timing.HActive[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k], + mode_lib->ms.cache_display_cfg.hw.NumberOfDSCSlices[k], + mode_lib->ms.cache_display_cfg.output.OutputFormat[k], + mode_lib->ms.cache_display_cfg.output.OutputEncoder[k], + mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.output.PixelClockBackEnd[k]); + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) // NumberOfSurfaces + if (j != k && mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == j && mode_lib->ms.cache_display_cfg.hw.DSCEnabled[j]) + locals->DSCDelay[k] = locals->DSCDelay[j]; + + // Prefetch + CalculateSurfaceSizeInMall( + mode_lib->ms.num_active_planes, + mode_lib->ms.soc.mall_allocated_for_dcn_mbytes, + mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen, + mode_lib->ms.cache_display_cfg.surface.DCCEnable, + mode_lib->ms.cache_display_cfg.plane.ViewportStationary, + mode_lib->ms.cache_display_cfg.plane.ViewportXStart, + mode_lib->ms.cache_display_cfg.plane.ViewportYStart, + mode_lib->ms.cache_display_cfg.plane.ViewportXStartC, + mode_lib->ms.cache_display_cfg.plane.ViewportYStartC, + mode_lib->ms.cache_display_cfg.plane.ViewportWidth, + mode_lib->ms.cache_display_cfg.plane.ViewportHeight, + locals->BytePerPixelY, + mode_lib->ms.cache_display_cfg.plane.ViewportWidthChroma, + mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma, + locals->BytePerPixelC, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY, + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY, + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC, + locals->BlockWidth256BytesY, + locals->BlockWidth256BytesC, + locals->BlockHeight256BytesY, + locals->BlockHeight256BytesC, + locals->BlockWidthY, + locals->BlockWidthC, + locals->BlockHeightY, + locals->BlockHeightC, + + /* Output */ + locals->SurfaceSizeInTheMALL, + &s->dummy_boolean[0]); /* dml_bool_t *ExceededMALLSize */ + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->SurfaceParameters[k].PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; + s->SurfaceParameters[k].DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; + s->SurfaceParameters[k].SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; + s->SurfaceParameters[k].ViewportHeight = mode_lib->ms.cache_display_cfg.plane.ViewportHeight[k]; + s->SurfaceParameters[k].ViewportHeightChroma = mode_lib->ms.cache_display_cfg.plane.ViewportHeightChroma[k]; + s->SurfaceParameters[k].BlockWidth256BytesY = locals->BlockWidth256BytesY[k]; + s->SurfaceParameters[k].BlockHeight256BytesY = locals->BlockHeight256BytesY[k]; + s->SurfaceParameters[k].BlockWidth256BytesC = locals->BlockWidth256BytesC[k]; + s->SurfaceParameters[k].BlockHeight256BytesC = locals->BlockHeight256BytesC[k]; + s->SurfaceParameters[k].BlockWidthY = locals->BlockWidthY[k]; + s->SurfaceParameters[k].BlockHeightY = locals->BlockHeightY[k]; + s->SurfaceParameters[k].BlockWidthC = locals->BlockWidthC[k]; + s->SurfaceParameters[k].BlockHeightC = locals->BlockHeightC[k]; + s->SurfaceParameters[k].InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; + s->SurfaceParameters[k].HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; + s->SurfaceParameters[k].DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; + s->SurfaceParameters[k].SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; + s->SurfaceParameters[k].SurfaceTiling = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k]; + s->SurfaceParameters[k].BytePerPixelY = locals->BytePerPixelY[k]; + s->SurfaceParameters[k].BytePerPixelC = locals->BytePerPixelC[k]; + s->SurfaceParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + s->SurfaceParameters[k].VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + s->SurfaceParameters[k].VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k]; + s->SurfaceParameters[k].VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps[k]; + s->SurfaceParameters[k].VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma[k]; + s->SurfaceParameters[k].PitchY = mode_lib->ms.cache_display_cfg.surface.PitchY[k]; + s->SurfaceParameters[k].DCCMetaPitchY = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchY[k]; + s->SurfaceParameters[k].PitchC = mode_lib->ms.cache_display_cfg.surface.PitchC[k]; + s->SurfaceParameters[k].DCCMetaPitchC = mode_lib->ms.cache_display_cfg.surface.DCCMetaPitchC[k]; + s->SurfaceParameters[k].ViewportStationary = mode_lib->ms.cache_display_cfg.plane.ViewportStationary[k]; + s->SurfaceParameters[k].ViewportXStart = mode_lib->ms.cache_display_cfg.plane.ViewportXStart[k]; + s->SurfaceParameters[k].ViewportYStart = mode_lib->ms.cache_display_cfg.plane.ViewportYStart[k]; + s->SurfaceParameters[k].ViewportXStartC = mode_lib->ms.cache_display_cfg.plane.ViewportXStartC[k]; + s->SurfaceParameters[k].ViewportYStartC = mode_lib->ms.cache_display_cfg.plane.ViewportYStartC[k]; + s->SurfaceParameters[k].FORCE_ONE_ROW_FOR_FRAME = mode_lib->ms.cache_display_cfg.plane.ForceOneRowForFrame[k]; + s->SurfaceParameters[k].SwathHeightY = locals->SwathHeightY[k]; + s->SurfaceParameters[k].SwathHeightC = locals->SwathHeightC[k]; + } + + CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateVMRowAndSwath_params->myPipe = s->SurfaceParameters; + CalculateVMRowAndSwath_params->SurfaceSizeInMALL = locals->SurfaceSizeInTheMALL; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_luma; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ms.ip.dpte_buffer_size_in_pte_reqs_chroma; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ms.ip.dcc_meta_buffer_size_bytes; + CalculateVMRowAndSwath_params->UseMALLForStaticScreen = mode_lib->ms.cache_display_cfg.plane.UseMALLForStaticScreen; + CalculateVMRowAndSwath_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->ms.soc.mall_allocated_for_dcn_mbytes; + CalculateVMRowAndSwath_params->SwathWidthY = locals->SwathWidthY; + CalculateVMRowAndSwath_params->SwathWidthC = locals->SwathWidthC; + CalculateVMRowAndSwath_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + CalculateVMRowAndSwath_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + CalculateVMRowAndSwath_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMaxPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + CalculateVMRowAndSwath_params->GPUVMMinPageSizeKBytes = mode_lib->ms.cache_display_cfg.plane.GPUVMMinPageSizeKBytes; + CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideEn = mode_lib->ms.cache_display_cfg.plane.PTEBufferModeOverrideEn; + CalculateVMRowAndSwath_params->PTEBufferModeOverrideVal = mode_lib->ms.cache_display_cfg.plane.PTEBufferMode; + CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = s->dummy_boolean_array[0]; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = s->dummy_boolean_array[1]; + CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = locals->dpte_row_width_luma_ub; + CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = locals->dpte_row_width_chroma_ub; + CalculateVMRowAndSwath_params->dpte_row_height_luma = locals->dpte_row_height; + CalculateVMRowAndSwath_params->dpte_row_height_chroma = locals->dpte_row_height_chroma; + CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = locals->dpte_row_height_linear; + CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = locals->dpte_row_height_linear_chroma; + CalculateVMRowAndSwath_params->meta_req_width = locals->meta_req_width; + CalculateVMRowAndSwath_params->meta_req_width_chroma = locals->meta_req_width_chroma; + CalculateVMRowAndSwath_params->meta_req_height = locals->meta_req_height; + CalculateVMRowAndSwath_params->meta_req_height_chroma = locals->meta_req_height_chroma; + CalculateVMRowAndSwath_params->meta_row_width = locals->meta_row_width; + CalculateVMRowAndSwath_params->meta_row_width_chroma = locals->meta_row_width_chroma; + CalculateVMRowAndSwath_params->meta_row_height = locals->meta_row_height; + CalculateVMRowAndSwath_params->meta_row_height_chroma = locals->meta_row_height_chroma; + CalculateVMRowAndSwath_params->vm_group_bytes = locals->vm_group_bytes; + CalculateVMRowAndSwath_params->dpte_group_bytes = locals->dpte_group_bytes; + CalculateVMRowAndSwath_params->PixelPTEReqWidthY = locals->PixelPTEReqWidthY; + CalculateVMRowAndSwath_params->PixelPTEReqHeightY = locals->PixelPTEReqHeightY; + CalculateVMRowAndSwath_params->PTERequestSizeY = locals->PTERequestSizeY; + CalculateVMRowAndSwath_params->PixelPTEReqWidthC = locals->PixelPTEReqWidthC; + CalculateVMRowAndSwath_params->PixelPTEReqHeightC = locals->PixelPTEReqHeightC; + CalculateVMRowAndSwath_params->PTERequestSizeC = locals->PTERequestSizeC; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = locals->dpde0_bytes_per_frame_ub_l; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = locals->meta_pte_bytes_per_frame_ub_l; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = locals->dpde0_bytes_per_frame_ub_c; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = locals->meta_pte_bytes_per_frame_ub_c; + CalculateVMRowAndSwath_params->PrefetchSourceLinesY = locals->PrefetchSourceLinesY; + CalculateVMRowAndSwath_params->PrefetchSourceLinesC = locals->PrefetchSourceLinesC; + CalculateVMRowAndSwath_params->VInitPreFillY = locals->VInitPreFillY; + CalculateVMRowAndSwath_params->VInitPreFillC = locals->VInitPreFillC; + CalculateVMRowAndSwath_params->MaxNumSwathY = locals->MaxNumSwathY; + CalculateVMRowAndSwath_params->MaxNumSwathC = locals->MaxNumSwathC; + CalculateVMRowAndSwath_params->meta_row_bw = locals->meta_row_bw; + CalculateVMRowAndSwath_params->dpte_row_bw = locals->dpte_row_bw; + CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = locals->PixelPTEBytesPerRow; + CalculateVMRowAndSwath_params->PDEAndMetaPTEBytesFrame = locals->PDEAndMetaPTEBytesFrame; + CalculateVMRowAndSwath_params->MetaRowByte = locals->MetaRowByte; + CalculateVMRowAndSwath_params->use_one_row_for_frame = locals->use_one_row_for_frame; + CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = locals->use_one_row_for_frame_flip; + CalculateVMRowAndSwath_params->UsesMALLForStaticScreen = locals->UsesMALLForStaticScreen; + CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = locals->PTE_BUFFER_MODE; + CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = locals->BIGK_FRAGMENT_SIZE; + + CalculateVMRowAndSwath(&mode_lib->scratch, + CalculateVMRowAndSwath_params); + + s->ReorderBytes = (dml_uint_t)(mode_lib->ms.soc.num_chans * dml_max3( + mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_only_bytes, + mode_lib->ms.soc.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, + mode_lib->ms.soc.urgent_out_of_order_return_per_channel_vm_only_bytes)); + + s->VMDataOnlyReturnBW = dml_get_return_bw_mbps_vm_only(&mode_lib->ms.soc, + mode_lib->ms.state.use_ideal_dram_bw_strobe, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + locals->Dcfclk, + mode_lib->ms.FabricClock, + mode_lib->ms.DRAMSpeed); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: locals->Dcfclk = %f\n", __func__, locals->Dcfclk); + dml_print("DML::%s: mode_lib->ms.soc.return_bus_width_bytes = %u\n", __func__, mode_lib->ms.soc.return_bus_width_bytes); + dml_print("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); + dml_print("DML::%s: mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes = %u\n", __func__, mode_lib->ms.soc.fabric_datapath_to_dcn_data_return_bytes); + dml_print("DML::%s: mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent = %f\n", __func__, mode_lib->ms.soc.pct_ideal_sdp_bw_after_urgent); + dml_print("DML::%s: mode_lib->ms.DRAMSpeed = %f\n", __func__, mode_lib->ms.DRAMSpeed); + dml_print("DML::%s: mode_lib->ms.soc.num_chans = %u\n", __func__, mode_lib->ms.soc.num_chans); + dml_print("DML::%s: mode_lib->ms.soc.dram_channel_width_bytes = %u\n", __func__, mode_lib->ms.soc.dram_channel_width_bytes); + dml_print("DML::%s: mode_lib->ms.state_idx = %u\n", __func__, mode_lib->ms.state_idx); + dml_print("DML::%s: mode_lib->ms.max_state_idx = %u\n", __func__, mode_lib->ms.max_state_idx); + dml_print("DML::%s: mode_lib->ms.state.use_ideal_dram_bw_strobe = %u\n", __func__, mode_lib->ms.state.use_ideal_dram_bw_strobe); + dml_print("DML::%s: VMDataOnlyReturnBW = %f\n", __func__, s->VMDataOnlyReturnBW); + dml_print("DML::%s: ReturnBW = %f\n", __func__, mode_lib->ms.ReturnBW); +#endif + + s->HostVMInefficiencyFactor = 1.0; + if (mode_lib->ms.cache_display_cfg.plane.GPUVMEnable && mode_lib->ms.cache_display_cfg.plane.HostVMEnable) + s->HostVMInefficiencyFactor = mode_lib->ms.ReturnBW / s->VMDataOnlyReturnBW; + + s->TotalDCCActiveDPP = 0; + s->TotalActiveDPP = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->TotalActiveDPP = s->TotalActiveDPP + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; + if (mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]) + s->TotalDCCActiveDPP = s->TotalDCCActiveDPP + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; + } + + locals->UrgentExtraLatency = CalculateExtraLatency( + mode_lib->ms.soc.round_trip_ping_latency_dcfclk_cycles, + s->ReorderBytes, + locals->Dcfclk, + s->TotalActiveDPP, + mode_lib->ms.ip.pixel_chunk_size_kbytes, + s->TotalDCCActiveDPP, + mode_lib->ms.ip.meta_chunk_size_kbytes, + mode_lib->ms.ReturnBW, + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.num_active_planes, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + locals->dpte_group_bytes, + s->HostVMInefficiencyFactor, + mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, + mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels); + + locals->TCalc = 24.0 / locals->DCFCLKDeepSleep; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + locals->WritebackDelay[k] = + mode_lib->ms.state.writeback_latency_us + + CalculateWriteBackDelay( + mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / locals->Dispclk; + } else + locals->WritebackDelay[k] = 0; + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) { + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[j] == k + && mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[j] == true) { + locals->WritebackDelay[k] = + dml_max( + locals->WritebackDelay[k], + mode_lib->ms.state.writeback_latency_us + + CalculateWriteBackDelay( + mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackHRatio[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackVRatio[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackVTaps[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[j], + mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[j], + mode_lib->ms.cache_display_cfg.timing.HTotal[k]) / locals->Dispclk); + } + } + } + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) + for (j = 0; j < mode_lib->ms.num_active_planes; ++j) + if (mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming[k] == j) + locals->WritebackDelay[k] = locals->WritebackDelay[j]; + + locals->UrgentLatency = CalculateUrgentLatency(mode_lib->ms.state.urgent_latency_pixel_data_only_us, + mode_lib->ms.state.urgent_latency_pixel_mixed_with_vm_data_us, + mode_lib->ms.state.urgent_latency_vm_data_only_us, + mode_lib->ms.soc.do_urgent_latency_adjustment, + mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_component_us, + mode_lib->ms.state.urgent_latency_adjustment_fabric_clock_reference_mhz, + mode_lib->ms.FabricClock); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateUrgentBurstFactor(mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + locals->swath_width_luma_ub[k], + locals->swath_width_chroma_ub[k], + locals->SwathHeightY[k], + locals->SwathHeightC[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + locals->UrgentLatency, + mode_lib->ms.ip.cursor_buffer_size, + mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], + mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + locals->BytePerPixelDETY[k], + locals->BytePerPixelDETC[k], + locals->DETBufferSizeY[k], + locals->DETBufferSizeC[k], + + /* output */ + &locals->UrgBurstFactorCursor[k], + &locals->UrgBurstFactorLuma[k], + &locals->UrgBurstFactorChroma[k], + &locals->NoUrgentLatencyHiding[k]); + + locals->cursor_bw[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / + ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * mode_lib->ms.cache_display_cfg.plane.VRatio[k]; + } + + s->VStartupLines = __DML_VBA_MIN_VSTARTUP__; + s->MaxVStartupAllPlanes = 0; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->MaxVStartupLines[k] = CalculateMaxVStartup(k, + mode_lib->ms.ip.ptoi_supported, + mode_lib->ms.ip.vblank_nom_default_us, + &mode_lib->ms.cache_display_cfg.timing, + locals->WritebackDelay[k]); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + dml_print("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, locals->WritebackDelay[k]); +#endif + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) + s->MaxVStartupAllPlanes = (dml_uint_t)(dml_max(s->MaxVStartupAllPlanes, s->MaxVStartupLines[k])); + + s->ImmediateFlipRequirementFinal = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->ImmediateFlipRequirementFinal = s->ImmediateFlipRequirementFinal || (mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: ImmediateFlipRequirementFinal = %u\n", __func__, s->ImmediateFlipRequirementFinal); +#endif + + // The prefetch scheduling should only be calculated once as per AllowForPStateChangeOrStutterInVBlank requirement + // If the AllowForPStateChangeOrStutterInVBlank requirement is not strict (i.e. only try those power saving feature + // if possible, then will try to program for the best power saving features in order of difficulty (dram, fclk, stutter) + s->iteration = 0; + s->MaxTotalRDBandwidth = 0; + s->AllPrefetchModeTested = false; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculatePrefetchMode(mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k], &s->MinPrefetchMode[k], &s->MaxPrefetchMode[k]); + s->NextPrefetchMode[k] = s->MinPrefetchMode[k]; + } + + do { + s->MaxTotalRDBandwidthNoUrgentBurst = 0.0; + s->DestinationLineTimesForPrefetchLessThan2 = false; + s->VRatioPrefetchMoreThanMax = false; + + dml_print("DML::%s: Start one iteration: VStartupLines = %u\n", __func__, s->VStartupLines); + + s->AllPrefetchModeTested = true; + s->MaxTotalRDBandwidth = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + locals->PrefetchMode[k] = s->NextPrefetchMode[k]; + TWait = CalculateTWait( + locals->PrefetchMode[k], + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal, + mode_lib->ms.cache_display_cfg.timing.DRRDisplay[k], + mode_lib->ms.state.dram_clock_change_latency_us, + mode_lib->ms.state.fclk_change_latency_us, + locals->UrgentLatency, + mode_lib->ms.state.sr_enter_plus_exit_time_us); + + myPipe = &s->myPipe; + myPipe->Dppclk = locals->Dppclk[k]; + myPipe->Dispclk = locals->Dispclk; + myPipe->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; + myPipe->DCFClkDeepSleep = locals->DCFCLKDeepSleep; + myPipe->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]; + myPipe->ScalerEnabled = mode_lib->ms.cache_display_cfg.plane.ScalerEnabled[k]; + myPipe->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan[k]; + myPipe->BlockWidth256BytesY = locals->BlockWidth256BytesY[k]; + myPipe->BlockHeight256BytesY = locals->BlockHeight256BytesY[k]; + myPipe->BlockWidth256BytesC = locals->BlockWidth256BytesC[k]; + myPipe->BlockHeight256BytesC = locals->BlockHeight256BytesC[k]; + myPipe->InterlaceEnable = mode_lib->ms.cache_display_cfg.timing.Interlace[k]; + myPipe->NumberOfCursors = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k]; + myPipe->VBlank = mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]; + myPipe->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal[k]; + myPipe->HActive = mode_lib->ms.cache_display_cfg.timing.HActive[k]; + myPipe->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable[k]; + myPipe->ODMMode = mode_lib->ms.cache_display_cfg.hw.ODMMode[k]; + myPipe->SourcePixelFormat = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k]; + myPipe->BytePerPixelY = locals->BytePerPixelY[k]; + myPipe->BytePerPixelC = locals->BytePerPixelC[k]; + myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); + dml_print("DML::%s: AllowForPStateChangeOrStutterInVBlank = %u\n", __func__, mode_lib->ms.policy.AllowForPStateChangeOrStutterInVBlank[k]); + dml_print("DML::%s: PrefetchMode[k] = %u (Min=%u Max=%u)\n", __func__, locals->PrefetchMode[k], s->MinPrefetchMode[k], s->MaxPrefetchMode[k]); +#endif + + CalculatePrefetchSchedule_params->EnhancedPrefetchScheduleAccelerationFinal = mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal; + CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactor; + CalculatePrefetchSchedule_params->myPipe = myPipe; + CalculatePrefetchSchedule_params->DSCDelay = locals->DSCDelay[k]; + CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ms.ip.dppclk_delay_subtotal + mode_lib->ms.ip.dppclk_delay_cnvc_formatter; + CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ms.ip.dppclk_delay_scl; + CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ms.ip.dppclk_delay_scl_lb_only; + CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ms.ip.dppclk_delay_cnvc_cursor; + CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ms.ip.dispclk_delay_subtotal; + CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (dml_uint_t)(locals->SwathWidthY[k] / mode_lib->ms.cache_display_cfg.plane.HRatio[k]); + CalculatePrefetchSchedule_params->OutputFormat = mode_lib->ms.cache_display_cfg.output.OutputFormat[k]; + CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ms.ip.max_inter_dcn_tile_repeaters; + CalculatePrefetchSchedule_params->VStartup = (dml_uint_t)(dml_min(s->VStartupLines, s->MaxVStartupLines[k])); + CalculatePrefetchSchedule_params->MaxVStartup = s->MaxVStartupLines[k]; + CalculatePrefetchSchedule_params->GPUVMPageTableLevels = mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels; + CalculatePrefetchSchedule_params->GPUVMEnable = mode_lib->ms.cache_display_cfg.plane.GPUVMEnable; + CalculatePrefetchSchedule_params->HostVMEnable = mode_lib->ms.cache_display_cfg.plane.HostVMEnable; + CalculatePrefetchSchedule_params->HostVMMaxNonCachedPageTableLevels = mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels; + CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024; + CalculatePrefetchSchedule_params->DynamicMetadataEnable = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]; + CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ms.ip.dynamic_metadata_vm_enabled; + CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataLinesBeforeActiveRequired[k]; + CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = mode_lib->ms.cache_display_cfg.plane.DynamicMetadataTransmittedBytes[k]; + CalculatePrefetchSchedule_params->UrgentLatency = locals->UrgentLatency; + CalculatePrefetchSchedule_params->UrgentExtraLatency = locals->UrgentExtraLatency; + CalculatePrefetchSchedule_params->TCalc = locals->TCalc; + CalculatePrefetchSchedule_params->PDEAndMetaPTEBytesFrame = locals->PDEAndMetaPTEBytesFrame[k]; + CalculatePrefetchSchedule_params->MetaRowByte = locals->MetaRowByte[k]; + CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = locals->PixelPTEBytesPerRow[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesY = locals->PrefetchSourceLinesY[k]; + CalculatePrefetchSchedule_params->VInitPreFillY = locals->VInitPreFillY[k]; + CalculatePrefetchSchedule_params->MaxNumSwathY = locals->MaxNumSwathY[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesC = locals->PrefetchSourceLinesC[k]; + CalculatePrefetchSchedule_params->VInitPreFillC = locals->VInitPreFillC[k]; + CalculatePrefetchSchedule_params->MaxNumSwathC = locals->MaxNumSwathC[k]; + CalculatePrefetchSchedule_params->swath_width_luma_ub = locals->swath_width_luma_ub[k]; + CalculatePrefetchSchedule_params->swath_width_chroma_ub = locals->swath_width_chroma_ub[k]; + CalculatePrefetchSchedule_params->SwathHeightY = locals->SwathHeightY[k]; + CalculatePrefetchSchedule_params->SwathHeightC = locals->SwathHeightC[k]; + CalculatePrefetchSchedule_params->TWait = TWait; + CalculatePrefetchSchedule_params->DSTXAfterScaler = &locals->DSTXAfterScaler[k]; + CalculatePrefetchSchedule_params->DSTYAfterScaler = &locals->DSTYAfterScaler[k]; + CalculatePrefetchSchedule_params->DestinationLinesForPrefetch = &locals->DestinationLinesForPrefetch[k]; + CalculatePrefetchSchedule_params->DestinationLinesToRequestVMInVBlank = &locals->DestinationLinesToRequestVMInVBlank[k]; + CalculatePrefetchSchedule_params->DestinationLinesToRequestRowInVBlank = &locals->DestinationLinesToRequestRowInVBlank[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchY = &locals->VRatioPrefetchY[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchC = &locals->VRatioPrefetchC[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWLuma = &locals->RequiredPrefetchPixDataBWLuma[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixDataBWChroma = &locals->RequiredPrefetchPixDataBWChroma[k]; + CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &locals->NotEnoughTimeForDynamicMetadata[k]; + CalculatePrefetchSchedule_params->Tno_bw = &locals->Tno_bw[k]; + CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &locals->prefetch_vmrow_bw[k]; + CalculatePrefetchSchedule_params->Tdmdl_vm = &locals->Tdmdl_vm[k]; + CalculatePrefetchSchedule_params->Tdmdl = &locals->Tdmdl[k]; + CalculatePrefetchSchedule_params->TSetup = &locals->TSetup[k]; + CalculatePrefetchSchedule_params->VUpdateOffsetPix = &locals->VUpdateOffsetPix[k]; + CalculatePrefetchSchedule_params->VUpdateWidthPix = &locals->VUpdateWidthPix[k]; + CalculatePrefetchSchedule_params->VReadyOffsetPix = &locals->VReadyOffsetPix[k]; + + locals->NoTimeToPrefetch[k] = + CalculatePrefetchSchedule(&mode_lib->scratch, + CalculatePrefetchSchedule_params); + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, locals->NoTimeToPrefetch[k]); +#endif + locals->VStartup[k] = (dml_uint_t)(dml_min(s->VStartupLines, s->MaxVStartupLines[k])); + locals->VStartupMin[k] = locals->VStartup[k]; + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateUrgentBurstFactor( + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k], + locals->swath_width_luma_ub[k], + locals->swath_width_chroma_ub[k], + locals->SwathHeightY[k], + locals->SwathHeightC[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + locals->UrgentLatency, + mode_lib->ms.ip.cursor_buffer_size, + mode_lib->ms.cache_display_cfg.plane.CursorWidth[k], + mode_lib->ms.cache_display_cfg.plane.CursorBPP[k], + locals->VRatioPrefetchY[k], + locals->VRatioPrefetchC[k], + locals->BytePerPixelDETY[k], + locals->BytePerPixelDETC[k], + locals->DETBufferSizeY[k], + locals->DETBufferSizeC[k], + /* Output */ + &locals->UrgBurstFactorCursorPre[k], + &locals->UrgBurstFactorLumaPre[k], + &locals->UrgBurstFactorChromaPre[k], + &locals->NoUrgentLatencyHidingPre[k]); + + locals->cursor_bw_pre[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] * mode_lib->ms.cache_display_cfg.plane.CursorBPP[k] / 8.0 / (mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * locals->VRatioPrefetchY[k]; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]); + dml_print("DML::%s: k=%0u UrgBurstFactorLuma=%f\n", __func__, k, locals->UrgBurstFactorLuma[k]); + dml_print("DML::%s: k=%0u UrgBurstFactorChroma=%f\n", __func__, k, locals->UrgBurstFactorChroma[k]); + dml_print("DML::%s: k=%0u UrgBurstFactorLumaPre=%f\n", __func__, k, locals->UrgBurstFactorLumaPre[k]); + dml_print("DML::%s: k=%0u UrgBurstFactorChromaPre=%f\n", __func__, k, locals->UrgBurstFactorChromaPre[k]); + + dml_print("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, locals->VRatioPrefetchY[k]); + dml_print("DML::%s: k=%0u VRatioY=%f\n", __func__, k, mode_lib->ms.cache_display_cfg.plane.VRatio[k]); + + dml_print("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, locals->prefetch_vmrow_bw[k]); + dml_print("DML::%s: k=%0u ReadBandwidthSurfaceLuma=%f\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); + dml_print("DML::%s: k=%0u ReadBandwidthSurfaceChroma=%f\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); + dml_print("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, locals->cursor_bw[k]); + dml_print("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, locals->meta_row_bw[k]); + dml_print("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, locals->dpte_row_bw[k]); + dml_print("DML::%s: k=%0u RequiredPrefetchPixDataBWLuma=%f\n", __func__, k, locals->RequiredPrefetchPixDataBWLuma[k]); + dml_print("DML::%s: k=%0u RequiredPrefetchPixDataBWChroma=%f\n", __func__, k, locals->RequiredPrefetchPixDataBWChroma[k]); + dml_print("DML::%s: k=%0u cursor_bw_pre=%f\n", __func__, k, locals->cursor_bw_pre[k]); + dml_print("DML::%s: k=%0u MaxTotalRDBandwidthNoUrgentBurst=%f\n", __func__, k, s->MaxTotalRDBandwidthNoUrgentBurst); +#endif + if (locals->DestinationLinesForPrefetch[k] < 2) + s->DestinationLineTimesForPrefetchLessThan2 = true; + + if (locals->VRatioPrefetchY[k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || + locals->VRatioPrefetchC[k] > __DML_MAX_VRATIO_PRE_ENHANCE_PREFETCH_ACC__ || + ((s->VStartupLines < s->MaxVStartupLines[k] || mode_lib->ms.policy.EnhancedPrefetchScheduleAccelerationFinal == 0) && + (locals->VRatioPrefetchY[k] > __DML_MAX_VRATIO_PRE__ || locals->VRatioPrefetchC[k] > __DML_MAX_VRATIO_PRE__))) + s->VRatioPrefetchMoreThanMax = true; + + //dml_bool_t DestinationLinesToRequestVMInVBlankEqualOrMoreThan32 = false; + //dml_bool_t DestinationLinesToRequestRowInVBlankEqualOrMoreThan16 = false; + //if (locals->DestinationLinesToRequestVMInVBlank[k] >= 32) { + // DestinationLinesToRequestVMInVBlankEqualOrMoreThan32 = true; + //} + + //if (locals->DestinationLinesToRequestRowInVBlank[k] >= 16) { + // DestinationLinesToRequestRowInVBlankEqualOrMoreThan16 = true; + //} + } + + locals->FractionOfUrgentBandwidth = s->MaxTotalRDBandwidthNoUrgentBurst / mode_lib->ms.ReturnBW; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: MaxTotalRDBandwidthNoUrgentBurst=%f \n", __func__, s->MaxTotalRDBandwidthNoUrgentBurst); + dml_print("DML::%s: ReturnBW=%f \n", __func__, mode_lib->ms.ReturnBW); + dml_print("DML::%s: FractionOfUrgentBandwidth=%f \n", __func__, locals->FractionOfUrgentBandwidth); +#endif + + CalculatePrefetchBandwithSupport( + mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBW, + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + locals->NoUrgentLatencyHidingPre, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->RequiredPrefetchPixDataBWLuma, + locals->RequiredPrefetchPixDataBWChroma, + locals->cursor_bw, + locals->meta_row_bw, + locals->dpte_row_bw, + locals->cursor_bw_pre, + locals->prefetch_vmrow_bw, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + locals->UrgBurstFactorLuma, + locals->UrgBurstFactorChroma, + locals->UrgBurstFactorCursor, + locals->UrgBurstFactorLumaPre, + locals->UrgBurstFactorChromaPre, + locals->UrgBurstFactorCursorPre, + + /* output */ + &s->MaxTotalRDBandwidth, // dml_float_t *PrefetchBandwidth + &s->MaxTotalRDBandwidthNotIncludingMALLPrefetch, // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch + &s->dummy_single[0], // dml_float_t *FractionOfUrgentBandwidth + &locals->PrefetchModeSupported); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) + s->dummy_unit_vector[k] = 1.0; + + CalculatePrefetchBandwithSupport(mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBW, + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + locals->NoUrgentLatencyHidingPre, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->RequiredPrefetchPixDataBWLuma, + locals->RequiredPrefetchPixDataBWChroma, + locals->cursor_bw, + locals->meta_row_bw, + locals->dpte_row_bw, + locals->cursor_bw_pre, + locals->prefetch_vmrow_bw, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + + /* output */ + &s->NonUrgentMaxTotalRDBandwidth, // dml_float_t *PrefetchBandwidth + &s->NonUrgentMaxTotalRDBandwidthNotIncludingMALLPrefetch, // dml_float_t *PrefetchBandwidthNotIncludingMALLPrefetch + &locals->FractionOfUrgentBandwidth, + &s->dummy_boolean[0]); // dml_bool_t *PrefetchBandwidthSupport + + + + if (s->VRatioPrefetchMoreThanMax != false || s->DestinationLineTimesForPrefetchLessThan2 != false) { + dml_print("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); + dml_print("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); + locals->PrefetchModeSupported = false; + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (locals->NoTimeToPrefetch[k] == true || locals->NotEnoughTimeForDynamicMetadata[k]) { + dml_print("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, locals->NoTimeToPrefetch[k]); + dml_print("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, locals->NotEnoughTimeForDynamicMetadata[k]); + locals->PrefetchModeSupported = false; + } + } + + + if (locals->PrefetchModeSupported == true && mode_lib->ms.support.ImmediateFlipSupport == true) { + locals->BandwidthAvailableForImmediateFlip = CalculateBandwidthAvailableForImmediateFlip( + mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBW, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->RequiredPrefetchPixDataBWLuma, + locals->RequiredPrefetchPixDataBWChroma, + locals->cursor_bw, + locals->cursor_bw_pre, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + locals->UrgBurstFactorLuma, + locals->UrgBurstFactorChroma, + locals->UrgBurstFactorCursor, + locals->UrgBurstFactorLumaPre, + locals->UrgBurstFactorChromaPre, + locals->UrgBurstFactorCursorPre); + + locals->TotImmediateFlipBytes = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required) { + locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * (locals->PDEAndMetaPTEBytesFrame[k] + locals->MetaRowByte[k]); + if (locals->use_one_row_for_frame_flip[k]) { + locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * (2 * locals->PixelPTEBytesPerRow[k]); + } else { + locals->TotImmediateFlipBytes = locals->TotImmediateFlipBytes + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k] * locals->PixelPTEBytesPerRow[k]; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k = %u\n", __func__, k); + dml_print("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->ms.cache_display_cfg.hw.DPPPerSurface[k]); + dml_print("DML::%s: PDEAndMetaPTEBytesFrame = %u\n", __func__, locals->PDEAndMetaPTEBytesFrame[k]); + dml_print("DML::%s: MetaRowByte = %u\n", __func__, locals->MetaRowByte[k]); + dml_print("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, locals->PixelPTEBytesPerRow[k]); + dml_print("DML::%s: TotImmediateFlipBytes = %u\n", __func__, locals->TotImmediateFlipBytes); +#endif + } + } + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + CalculateFlipSchedule( + s->HostVMInefficiencyFactor, + locals->UrgentExtraLatency, + locals->UrgentLatency, + mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, + mode_lib->ms.cache_display_cfg.plane.HostVMEnable, + mode_lib->ms.cache_display_cfg.plane.HostVMMaxPageTableLevels, + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.soc.hostvm_min_page_size_kbytes * 1024, + locals->PDEAndMetaPTEBytesFrame[k], + locals->MetaRowByte[k], + locals->PixelPTEBytesPerRow[k], + locals->BandwidthAvailableForImmediateFlip, + locals->TotImmediateFlipBytes, + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k], + mode_lib->ms.cache_display_cfg.plane.VRatio[k], + mode_lib->ms.cache_display_cfg.plane.VRatioChroma[k], + locals->Tno_bw[k], + mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], + locals->dpte_row_height[k], + locals->meta_row_height[k], + locals->dpte_row_height_chroma[k], + locals->meta_row_height_chroma[k], + locals->use_one_row_for_frame_flip[k], + + /* Output */ + &locals->DestinationLinesToRequestVMInImmediateFlip[k], + &locals->DestinationLinesToRequestRowInImmediateFlip[k], + &locals->final_flip_bw[k], + &locals->ImmediateFlipSupportedForPipe[k]); + } + + CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBW, + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + mode_lib->ms.policy.ImmediateFlipRequirement, + locals->final_flip_bw, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->RequiredPrefetchPixDataBWLuma, + locals->RequiredPrefetchPixDataBWChroma, + locals->cursor_bw, + locals->meta_row_bw, + locals->dpte_row_bw, + locals->cursor_bw_pre, + locals->prefetch_vmrow_bw, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + locals->UrgBurstFactorLuma, + locals->UrgBurstFactorChroma, + locals->UrgBurstFactorCursor, + locals->UrgBurstFactorLumaPre, + locals->UrgBurstFactorChromaPre, + locals->UrgBurstFactorCursorPre, + + /* output */ + &locals->total_dcn_read_bw_with_flip, // dml_float_t *TotalBandwidth + &locals->total_dcn_read_bw_with_flip_not_including_MALL_prefetch, // dml_float_t TotalBandwidthNotIncludingMALLPrefetch + &s->dummy_single[0], // dml_float_t *FractionOfUrgentBandwidth + &locals->ImmediateFlipSupported); // dml_bool_t *ImmediateFlipBandwidthSupport + + CalculateImmediateFlipBandwithSupport(mode_lib->ms.num_active_planes, + mode_lib->ms.ReturnBW, + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + mode_lib->ms.policy.ImmediateFlipRequirement, + locals->final_flip_bw, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->RequiredPrefetchPixDataBWLuma, + locals->RequiredPrefetchPixDataBWChroma, + locals->cursor_bw, + locals->meta_row_bw, + locals->dpte_row_bw, + locals->cursor_bw_pre, + locals->prefetch_vmrow_bw, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + s->dummy_unit_vector, + + /* output */ + &locals->non_urgent_total_dcn_read_bw_with_flip, // dml_float_t *TotalBandwidth + &locals->non_urgent_total_dcn_read_bw_with_flip_not_including_MALL_prefetch, // dml_float_t TotalBandwidthNotIncludingMALLPrefetch + &locals->FractionOfUrgentBandwidthImmediateFlip, // dml_float_t *FractionOfUrgentBandwidth + &s->dummy_boolean[0]); // dml_bool_t *ImmediateFlipBandwidthSupport + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.policy.ImmediateFlipRequirement[k] != dml_immediate_flip_not_required && locals->ImmediateFlipSupportedForPipe[k] == false) { + locals->ImmediateFlipSupported = false; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Pipe %0d not supporting iflip\n", __func__, k); +#endif + } + } + } else { + locals->ImmediateFlipSupported = false; + locals->total_dcn_read_bw_with_flip = s->MaxTotalRDBandwidth; + locals->total_dcn_read_bw_with_flip_not_including_MALL_prefetch = s->MaxTotalRDBandwidthNotIncludingMALLPrefetch; + locals->non_urgent_total_dcn_read_bw_with_flip = s->NonUrgentMaxTotalRDBandwidth; + locals->non_urgent_total_dcn_read_bw_with_flip_not_including_MALL_prefetch = s->NonUrgentMaxTotalRDBandwidthNotIncludingMALLPrefetch; + } + + /* consider flip support is okay if the flip bw is ok or (when user does't require a iflip and there is no host vm) */ + locals->PrefetchAndImmediateFlipSupported = (locals->PrefetchModeSupported == true && + ((!mode_lib->ms.support.ImmediateFlipSupport && !mode_lib->ms.cache_display_cfg.plane.HostVMEnable && !s->ImmediateFlipRequirementFinal) || + locals->ImmediateFlipSupported)) ? true : false; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: PrefetchModeSupported = %u\n", __func__, locals->PrefetchModeSupported); + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) + dml_print("DML::%s: ImmediateFlipRequirement[%u] = %u\n", __func__, k, mode_lib->ms.policy.ImmediateFlipRequirement[k] == dml_immediate_flip_required); + dml_print("DML::%s: HostVMEnable = %u\n", __func__, mode_lib->ms.cache_display_cfg.plane.HostVMEnable); + dml_print("DML::%s: ImmediateFlipSupport = %u (from mode_support)\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); + dml_print("DML::%s: ImmediateFlipSupported = %u\n", __func__, locals->ImmediateFlipSupported); + dml_print("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, locals->PrefetchAndImmediateFlipSupported); +#endif + dml_print("DML::%s: Done one iteration: VStartupLines=%u, MaxVStartupAllPlanes=%u\n", __func__, s->VStartupLines, s->MaxVStartupAllPlanes); + + s->VStartupLines = s->VStartupLines + 1; + + if (s->VStartupLines > s->MaxVStartupAllPlanes) { + s->VStartupLines = __DML_VBA_MIN_VSTARTUP__; + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + s->NextPrefetchMode[k] = s->NextPrefetchMode[k] + 1; + + if (s->NextPrefetchMode[k] <= s->MaxPrefetchMode[k]) + s->AllPrefetchModeTested = false; + dml_print("DML::%s: VStartupLines=%u, reaches max vstartup, try next prefetch mode=%u\n", __func__, s->VStartupLines-1, s->AllPrefetchModeTested); + } + } else { + s->AllPrefetchModeTested = false; + } + s->iteration++; + if (s->iteration > 2500) { + dml_print("ERROR: DML::%s: Too many errors, exit now\n", __func__); + ASSERT(0); + } + } while (!(locals->PrefetchAndImmediateFlipSupported || s->AllPrefetchModeTested)); + + if (locals->PrefetchAndImmediateFlipSupported) { + dml_print("DML::%s: Good, Prefetch and flip scheduling solution found at VStartupLines=%u (MaxVStartupAllPlanes=%u)\n", __func__, s->VStartupLines-1, s->MaxVStartupAllPlanes); + } else { + dml_print("DML::%s: Bad, Prefetch and flip scheduling solution did NOT find solution! (MaxVStartupAllPlanes=%u)\n", __func__, s->MaxVStartupAllPlanes); + } + + //Watermarks and NB P-State/DRAM Clock Change Support + { + s->mmSOCParameters.UrgentLatency = locals->UrgentLatency; + s->mmSOCParameters.ExtraLatency = locals->UrgentExtraLatency; + s->mmSOCParameters.WritebackLatency = mode_lib->ms.state.writeback_latency_us; + s->mmSOCParameters.DRAMClockChangeLatency = mode_lib->ms.state.dram_clock_change_latency_us; + s->mmSOCParameters.FCLKChangeLatency = mode_lib->ms.state.fclk_change_latency_us; + s->mmSOCParameters.SRExitTime = mode_lib->ms.state.sr_exit_time_us; + s->mmSOCParameters.SREnterPlusExitTime = mode_lib->ms.state.sr_enter_plus_exit_time_us; + s->mmSOCParameters.SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; + s->mmSOCParameters.SREnterPlusExitZ8Time = mode_lib->ms.state.sr_enter_plus_exit_z8_time_us; + s->mmSOCParameters.USRRetrainingLatency = mode_lib->ms.state.usr_retraining_latency_us; + s->mmSOCParameters.SMNLatency = mode_lib->ms.soc.smn_latency_us; + + CalculateWatermarks_params->USRRetrainingRequiredFinal = mode_lib->ms.policy.USRRetrainingRequiredFinal; + CalculateWatermarks_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateWatermarks_params->PrefetchMode = locals->PrefetchMode; + CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ms.ip.max_line_buffer_lines; + CalculateWatermarks_params->LineBufferSize = mode_lib->ms.ip.line_buffer_size_bits; + CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ms.ip.writeback_interface_buffer_size_kbytes; + CalculateWatermarks_params->DCFCLK = locals->Dcfclk; + CalculateWatermarks_params->ReturnBW = mode_lib->ms.ReturnBW; + CalculateWatermarks_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; + CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = mode_lib->ms.policy.SynchronizeDRRDisplaysForUCLKPStateChangeFinal; + CalculateWatermarks_params->DRRDisplay = mode_lib->ms.cache_display_cfg.timing.DRRDisplay; + CalculateWatermarks_params->dpte_group_bytes = locals->dpte_group_bytes; + CalculateWatermarks_params->meta_row_height = locals->meta_row_height; + CalculateWatermarks_params->meta_row_height_chroma = locals->meta_row_height_chroma; + CalculateWatermarks_params->mmSOCParameters = s->mmSOCParameters; + CalculateWatermarks_params->WritebackChunkSize = mode_lib->ms.ip.writeback_chunk_size_kbytes; + CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; + CalculateWatermarks_params->DCFClkDeepSleep = locals->DCFCLKDeepSleep; + CalculateWatermarks_params->DETBufferSizeY = locals->DETBufferSizeY; + CalculateWatermarks_params->DETBufferSizeC = locals->DETBufferSizeC; + CalculateWatermarks_params->SwathHeightY = locals->SwathHeightY; + CalculateWatermarks_params->SwathHeightC = locals->SwathHeightC; + CalculateWatermarks_params->LBBitPerPixel = mode_lib->ms.cache_display_cfg.plane.LBBitPerPixel; + CalculateWatermarks_params->SwathWidthY = locals->SwathWidthY; + CalculateWatermarks_params->SwathWidthC = locals->SwathWidthC; + CalculateWatermarks_params->HRatio = mode_lib->ms.cache_display_cfg.plane.HRatio; + CalculateWatermarks_params->HRatioChroma = mode_lib->ms.cache_display_cfg.plane.HRatioChroma; + CalculateWatermarks_params->VTaps = mode_lib->ms.cache_display_cfg.plane.VTaps; + CalculateWatermarks_params->VTapsChroma = mode_lib->ms.cache_display_cfg.plane.VTapsChroma; + CalculateWatermarks_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; + CalculateWatermarks_params->VRatioChroma = mode_lib->ms.cache_display_cfg.plane.VRatioChroma; + CalculateWatermarks_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; + CalculateWatermarks_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; + CalculateWatermarks_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; + CalculateWatermarks_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; + CalculateWatermarks_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; + CalculateWatermarks_params->BytePerPixelDETY = locals->BytePerPixelDETY; + CalculateWatermarks_params->BytePerPixelDETC = locals->BytePerPixelDETC; + CalculateWatermarks_params->DSTXAfterScaler = locals->DSTXAfterScaler; + CalculateWatermarks_params->DSTYAfterScaler = locals->DSTYAfterScaler; + CalculateWatermarks_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; + CalculateWatermarks_params->WritebackPixelFormat = mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat; + CalculateWatermarks_params->WritebackDestinationWidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth; + CalculateWatermarks_params->WritebackDestinationHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight; + CalculateWatermarks_params->WritebackSourceHeight = mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight; + CalculateWatermarks_params->UnboundedRequestEnabled = locals->UnboundedRequestEnabled; + CalculateWatermarks_params->CompressedBufferSizeInkByte = locals->CompressedBufferSizeInkByte; + + // Output + CalculateWatermarks_params->Watermark = &locals->Watermark; // Watermarks *Watermark + CalculateWatermarks_params->DRAMClockChangeSupport = &locals->DRAMClockChangeSupport; + CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = locals->MaxActiveDRAMClockChangeLatencySupported; // dml_float_t *MaxActiveDRAMClockChangeLatencySupported[] + CalculateWatermarks_params->SubViewportLinesNeededInMALL = locals->SubViewportLinesNeededInMALL; // dml_uint_t SubViewportLinesNeededInMALL[] + CalculateWatermarks_params->FCLKChangeSupport = &locals->FCLKChangeSupport; + CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &locals->MaxActiveFCLKChangeLatencySupported; // dml_float_t *MaxActiveFCLKChangeLatencySupported + CalculateWatermarks_params->USRRetrainingSupport = &locals->USRRetrainingSupport; + + CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( + &mode_lib->scratch, + CalculateWatermarks_params); + + /* Copy the calculated watermarks to mp.Watermark as the getter functions are + * implemented by the DML team to copy the calculated values from the mp.Watermark interface. + * &mode_lib->mp.Watermark and &locals->Watermark are the same address, memcpy may lead to + * unexpected behavior. memmove should be used. + */ + memmove(&mode_lib->mp.Watermark, CalculateWatermarks_params->Watermark, sizeof(struct Watermarks)); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + locals->WritebackAllowDRAMClockChangeEndPosition[k] = dml_max(0, locals->VStartupMin[k] * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / + mode_lib->ms.cache_display_cfg.timing.PixelClock[k] - locals->Watermark.WritebackDRAMClockChangeWatermark); + locals->WritebackAllowFCLKChangeEndPosition[k] = dml_max(0, locals->VStartupMin[k] * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / + mode_lib->ms.cache_display_cfg.timing.PixelClock[k] - locals->Watermark.WritebackFCLKChangeWatermark); + } else { + locals->WritebackAllowDRAMClockChangeEndPosition[k] = 0; + locals->WritebackAllowFCLKChangeEndPosition[k] = 0; + } + } + } + + //Display Pipeline Delivery Time in Prefetch, Groups + CalculatePixelDeliveryTimes( + mode_lib->ms.num_active_planes, + mode_lib->ms.cache_display_cfg.plane.VRatio, + mode_lib->ms.cache_display_cfg.plane.VRatioChroma, + locals->VRatioPrefetchY, + locals->VRatioPrefetchC, + locals->swath_width_luma_ub, + locals->swath_width_chroma_ub, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + mode_lib->ms.cache_display_cfg.plane.HRatio, + mode_lib->ms.cache_display_cfg.plane.HRatioChroma, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + locals->PSCL_THROUGHPUT, + locals->PSCL_THROUGHPUT_CHROMA, + locals->Dppclk, + locals->BytePerPixelC, + mode_lib->ms.cache_display_cfg.plane.SourceScan, + mode_lib->ms.cache_display_cfg.plane.NumberOfCursors, + mode_lib->ms.cache_display_cfg.plane.CursorWidth, + mode_lib->ms.cache_display_cfg.plane.CursorBPP, + locals->BlockWidth256BytesY, + locals->BlockHeight256BytesY, + locals->BlockWidth256BytesC, + locals->BlockHeight256BytesC, + + /* Output */ + locals->DisplayPipeLineDeliveryTimeLuma, + locals->DisplayPipeLineDeliveryTimeChroma, + locals->DisplayPipeLineDeliveryTimeLumaPrefetch, + locals->DisplayPipeLineDeliveryTimeChromaPrefetch, + locals->DisplayPipeRequestDeliveryTimeLuma, + locals->DisplayPipeRequestDeliveryTimeChroma, + locals->DisplayPipeRequestDeliveryTimeLumaPrefetch, + locals->DisplayPipeRequestDeliveryTimeChromaPrefetch, + locals->CursorRequestDeliveryTime, + locals->CursorRequestDeliveryTimePrefetch); + + CalculateMetaAndPTETimes( + locals->use_one_row_for_frame, + mode_lib->ms.num_active_planes, + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.ip.meta_chunk_size_kbytes, + mode_lib->ms.ip.min_meta_chunk_size_bytes, + mode_lib->ms.cache_display_cfg.timing.HTotal, + mode_lib->ms.cache_display_cfg.plane.VRatio, + mode_lib->ms.cache_display_cfg.plane.VRatioChroma, + locals->DestinationLinesToRequestRowInVBlank, + locals->DestinationLinesToRequestRowInImmediateFlip, + mode_lib->ms.cache_display_cfg.surface.DCCEnable, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + locals->BytePerPixelY, + locals->BytePerPixelC, + mode_lib->ms.cache_display_cfg.plane.SourceScan, + locals->dpte_row_height, + locals->dpte_row_height_chroma, + locals->meta_row_width, + locals->meta_row_width_chroma, + locals->meta_row_height, + locals->meta_row_height_chroma, + locals->meta_req_width, + locals->meta_req_width_chroma, + locals->meta_req_height, + locals->meta_req_height_chroma, + locals->dpte_group_bytes, + locals->PTERequestSizeY, + locals->PTERequestSizeC, + locals->PixelPTEReqWidthY, + locals->PixelPTEReqHeightY, + locals->PixelPTEReqWidthC, + locals->PixelPTEReqHeightC, + locals->dpte_row_width_luma_ub, + locals->dpte_row_width_chroma_ub, + + /* Output */ + locals->DST_Y_PER_PTE_ROW_NOM_L, + locals->DST_Y_PER_PTE_ROW_NOM_C, + locals->DST_Y_PER_META_ROW_NOM_L, + locals->DST_Y_PER_META_ROW_NOM_C, + locals->TimePerMetaChunkNominal, + locals->TimePerChromaMetaChunkNominal, + locals->TimePerMetaChunkVBlank, + locals->TimePerChromaMetaChunkVBlank, + locals->TimePerMetaChunkFlip, + locals->TimePerChromaMetaChunkFlip, + locals->time_per_pte_group_nom_luma, + locals->time_per_pte_group_vblank_luma, + locals->time_per_pte_group_flip_luma, + locals->time_per_pte_group_nom_chroma, + locals->time_per_pte_group_vblank_chroma, + locals->time_per_pte_group_flip_chroma); + + CalculateVMGroupAndRequestTimes( + mode_lib->ms.num_active_planes, + mode_lib->ms.cache_display_cfg.plane.GPUVMEnable, + mode_lib->ms.cache_display_cfg.plane.GPUVMMaxPageTableLevels, + mode_lib->ms.cache_display_cfg.timing.HTotal, + locals->BytePerPixelC, + locals->DestinationLinesToRequestVMInVBlank, + locals->DestinationLinesToRequestVMInImmediateFlip, + mode_lib->ms.cache_display_cfg.surface.DCCEnable, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + locals->dpte_row_width_luma_ub, + locals->dpte_row_width_chroma_ub, + locals->vm_group_bytes, + locals->dpde0_bytes_per_frame_ub_l, + locals->dpde0_bytes_per_frame_ub_c, + locals->meta_pte_bytes_per_frame_ub_l, + locals->meta_pte_bytes_per_frame_ub_c, + + /* Output */ + locals->TimePerVMGroupVBlank, + locals->TimePerVMGroupFlip, + locals->TimePerVMRequestVBlank, + locals->TimePerVMRequestFlip); + + // Min TTUVBlank + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (locals->PrefetchMode[k] == 0) { + locals->MinTTUVBlank[k] = dml_max4( + locals->Watermark.DRAMClockChangeWatermark, + locals->Watermark.FCLKChangeWatermark, + locals->Watermark.StutterEnterPlusExitWatermark, + locals->Watermark.UrgentWatermark); + } else if (locals->PrefetchMode[k] == 1) { + locals->MinTTUVBlank[k] = dml_max3( + locals->Watermark.FCLKChangeWatermark, + locals->Watermark.StutterEnterPlusExitWatermark, + locals->Watermark.UrgentWatermark); + } else if (locals->PrefetchMode[k] == 2) { + locals->MinTTUVBlank[k] = dml_max( + locals->Watermark.StutterEnterPlusExitWatermark, + locals->Watermark.UrgentWatermark); + } else { + locals->MinTTUVBlank[k] = locals->Watermark.UrgentWatermark; + } + if (!mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k]) + locals->MinTTUVBlank[k] = locals->TCalc + locals->MinTTUVBlank[k]; + } + + // DCC Configuration + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); +#endif + CalculateDCCConfiguration( + mode_lib->ms.cache_display_cfg.surface.DCCEnable[k], + mode_lib->ms.policy.DCCProgrammingAssumesScanDirectionUnknownFinal, + mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthY[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceWidthC[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightY[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceHeightC[k], + mode_lib->ms.NomDETInKByte, + locals->BlockHeight256BytesY[k], + locals->BlockHeight256BytesC[k], + mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[k], + locals->BytePerPixelY[k], + locals->BytePerPixelC[k], + locals->BytePerPixelDETY[k], + locals->BytePerPixelDETC[k], + mode_lib->ms.cache_display_cfg.plane.SourceScan[k], + /* Output */ + &locals->DCCYMaxUncompressedBlock[k], + &locals->DCCCMaxUncompressedBlock[k], + &locals->DCCYMaxCompressedBlock[k], + &locals->DCCCMaxCompressedBlock[k], + &locals->DCCYIndependentBlock[k], + &locals->DCCCIndependentBlock[k]); + } + + // VStartup Adjustment + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->Tvstartup_margin = (s->MaxVStartupLines[k] - locals->VStartupMin[k]) * mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]; +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, locals->MinTTUVBlank[k]); +#endif + + locals->MinTTUVBlank[k] = locals->MinTTUVBlank[k] + s->Tvstartup_margin; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); + dml_print("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + dml_print("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, locals->MinTTUVBlank[k]); +#endif + + locals->Tdmdl[k] = locals->Tdmdl[k] + s->Tvstartup_margin; + if (mode_lib->ms.cache_display_cfg.plane.DynamicMetadataEnable[k] && mode_lib->ms.ip.dynamic_metadata_vm_enabled) { + locals->Tdmdl_vm[k] = locals->Tdmdl_vm[k] + s->Tvstartup_margin; + } + + isInterlaceTiming = (mode_lib->ms.cache_display_cfg.timing.Interlace[k] && !mode_lib->ms.ip.ptoi_supported); + + // The actual positioning of the vstartup + locals->VStartup[k] = (isInterlaceTiming ? (2 * s->MaxVStartupLines[k]) : s->MaxVStartupLines[k]); + + s->dlg_vblank_start = ((isInterlaceTiming ? dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) / 2.0, 1.0) : + mode_lib->ms.cache_display_cfg.timing.VTotal[k]) - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]); + s->LSetup = dml_floor(4.0 * locals->TSetup[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0) / 4.0; + s->blank_lines_remaining = (mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k]) - locals->VStartup[k]; + + if (s->blank_lines_remaining < 0) { + dml_print("ERROR: Vstartup is larger than vblank!?\n"); + s->blank_lines_remaining = 0; + ASSERT(0); + } + locals->MIN_DST_Y_NEXT_START[k] = s->dlg_vblank_start + s->blank_lines_remaining + s->LSetup; + + // debug only + s->old_MIN_DST_Y_NEXT_START = ((isInterlaceTiming ? dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) / 2.0, 1.0) : + mode_lib->ms.cache_display_cfg.timing.VTotal[k]) - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]) + + dml_max(1.0, dml_ceil((dml_float_t) locals->WritebackDelay[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0)) + + dml_floor(4.0 * locals->TSetup[k] / ((dml_float_t) mode_lib->ms.cache_display_cfg.timing.HTotal[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]), 1.0) / 4.0; + + if (((locals->VUpdateOffsetPix[k] + locals->VUpdateWidthPix[k] + locals->VReadyOffsetPix[k]) / (double) mode_lib->ms.cache_display_cfg.timing.HTotal[k]) <= + (isInterlaceTiming ? + dml_floor((mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k] - locals->VStartup[k]) / 2.0, 1.0) : + (int) (mode_lib->ms.cache_display_cfg.timing.VTotal[k] - mode_lib->ms.cache_display_cfg.timing.VActive[k] - mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k] - locals->VStartup[k]))) { + locals->VREADY_AT_OR_AFTER_VSYNC[k] = true; + } else { + locals->VREADY_AT_OR_AFTER_VSYNC[k] = false; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, locals->VStartup[k]); + dml_print("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, locals->VStartupMin[k]); + dml_print("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, locals->VUpdateOffsetPix[k]); + dml_print("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, locals->VUpdateWidthPix[k]); + dml_print("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, locals->VReadyOffsetPix[k]); + dml_print("DML::%s: k=%u, HTotal = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.HTotal[k]); + dml_print("DML::%s: k=%u, VTotal = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VTotal[k]); + dml_print("DML::%s: k=%u, VActive = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VActive[k]); + dml_print("DML::%s: k=%u, VFrontPorch = %u\n", __func__, k, mode_lib->ms.cache_display_cfg.timing.VFrontPorch[k]); + dml_print("DML::%s: k=%u, TSetup = %f\n", __func__, k, locals->TSetup[k]); + dml_print("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, locals->MIN_DST_Y_NEXT_START[k]); + dml_print("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f (old)\n", __func__, k, s->old_MIN_DST_Y_NEXT_START); + dml_print("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, locals->VREADY_AT_OR_AFTER_VSYNC[k]); +#endif + } + + //Maximum Bandwidth Used + s->TotalWRBandwidth = 0; + s->WRBandwidth = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true && mode_lib->ms.cache_display_cfg.writeback.WritebackPixelFormat[k] == dml_444_32) { + s->WRBandwidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] / + (mode_lib->ms.cache_display_cfg.timing.HTotal[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 4; + } else if (mode_lib->ms.cache_display_cfg.writeback.WritebackEnable[k] == true) { + s->WRBandwidth = mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationWidth[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackDestinationHeight[k] / + (mode_lib->ms.cache_display_cfg.timing.HTotal[k] * mode_lib->ms.cache_display_cfg.writeback.WritebackSourceHeight[k] / mode_lib->ms.cache_display_cfg.timing.PixelClock[k]) * 8; + } + s->TotalWRBandwidth = s->TotalWRBandwidth + s->WRBandwidth; + } + + locals->TotalDataReadBandwidth = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + locals->TotalDataReadBandwidth = locals->TotalDataReadBandwidth + locals->ReadBandwidthSurfaceLuma[k] + locals->ReadBandwidthSurfaceChroma[k]; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, locals->TotalDataReadBandwidth); + dml_print("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, locals->ReadBandwidthSurfaceLuma[k]); + dml_print("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, locals->ReadBandwidthSurfaceChroma[k]); +#endif + } + + locals->TotalDataReadBandwidthNotIncludingMALLPrefetch = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[k] != dml_use_mall_pstate_change_phantom_pipe) { + locals->TotalDataReadBandwidthNotIncludingMALLPrefetch = locals->TotalDataReadBandwidthNotIncludingMALLPrefetch + + locals->ReadBandwidthSurfaceLuma[k] + locals->ReadBandwidthSurfaceChroma[k]; + } + } + + CalculateStutterEfficiency_params->CompressedBufferSizeInkByte = locals->CompressedBufferSizeInkByte; + CalculateStutterEfficiency_params->UseMALLForPStateChange = mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange; + CalculateStutterEfficiency_params->UnboundedRequestEnabled = locals->UnboundedRequestEnabled; + CalculateStutterEfficiency_params->MetaFIFOSizeInKEntries = mode_lib->ms.ip.meta_fifo_size_in_kentries; + CalculateStutterEfficiency_params->ZeroSizeBufferEntries = mode_lib->ms.ip.zero_size_buffer_entries; + CalculateStutterEfficiency_params->PixelChunkSizeInKByte = mode_lib->ms.ip.pixel_chunk_size_kbytes; + CalculateStutterEfficiency_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateStutterEfficiency_params->ROBBufferSizeInKByte = mode_lib->ms.ip.rob_buffer_size_kbytes; + CalculateStutterEfficiency_params->TotalDataReadBandwidth = locals->TotalDataReadBandwidth; + CalculateStutterEfficiency_params->DCFCLK = locals->Dcfclk; + CalculateStutterEfficiency_params->ReturnBW = mode_lib->ms.ReturnBW; + CalculateStutterEfficiency_params->CompbufReservedSpace64B = locals->compbuf_reserved_space_64b; + CalculateStutterEfficiency_params->CompbufReservedSpaceZs = locals->compbuf_reserved_space_zs; + CalculateStutterEfficiency_params->SRExitTime = mode_lib->ms.state.sr_exit_time_us; + CalculateStutterEfficiency_params->SRExitZ8Time = mode_lib->ms.state.sr_exit_z8_time_us; + CalculateStutterEfficiency_params->SynchronizeTimingsFinal = mode_lib->ms.policy.SynchronizeTimingsFinal; + CalculateStutterEfficiency_params->BlendingAndTiming = mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming; + CalculateStutterEfficiency_params->StutterEnterPlusExitWatermark = locals->Watermark.StutterEnterPlusExitWatermark; + CalculateStutterEfficiency_params->Z8StutterEnterPlusExitWatermark = locals->Watermark.Z8StutterEnterPlusExitWatermark; + CalculateStutterEfficiency_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ms.ip.ptoi_supported; + CalculateStutterEfficiency_params->Interlace = mode_lib->ms.cache_display_cfg.timing.Interlace; + CalculateStutterEfficiency_params->MinTTUVBlank = locals->MinTTUVBlank; + CalculateStutterEfficiency_params->DPPPerSurface = mode_lib->ms.cache_display_cfg.hw.DPPPerSurface; + CalculateStutterEfficiency_params->DETBufferSizeY = locals->DETBufferSizeY; + CalculateStutterEfficiency_params->BytePerPixelY = locals->BytePerPixelY; + CalculateStutterEfficiency_params->BytePerPixelDETY = locals->BytePerPixelDETY; + CalculateStutterEfficiency_params->SwathWidthY = locals->SwathWidthY; + CalculateStutterEfficiency_params->SwathHeightY = locals->SwathHeightY; + CalculateStutterEfficiency_params->SwathHeightC = locals->SwathHeightC; + CalculateStutterEfficiency_params->NetDCCRateLuma = mode_lib->ms.cache_display_cfg.surface.DCCRateLuma; + CalculateStutterEfficiency_params->NetDCCRateChroma = mode_lib->ms.cache_display_cfg.surface.DCCRateChroma; + CalculateStutterEfficiency_params->DCCFractionOfZeroSizeRequestsLuma = mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsLuma; + CalculateStutterEfficiency_params->DCCFractionOfZeroSizeRequestsChroma = mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsChroma; + CalculateStutterEfficiency_params->HTotal = mode_lib->ms.cache_display_cfg.timing.HTotal; + CalculateStutterEfficiency_params->VTotal = mode_lib->ms.cache_display_cfg.timing.VTotal; + CalculateStutterEfficiency_params->PixelClock = mode_lib->ms.cache_display_cfg.timing.PixelClock; + CalculateStutterEfficiency_params->VRatio = mode_lib->ms.cache_display_cfg.plane.VRatio; + CalculateStutterEfficiency_params->SourceScan = mode_lib->ms.cache_display_cfg.plane.SourceScan; + CalculateStutterEfficiency_params->BlockHeight256BytesY = locals->BlockHeight256BytesY; + CalculateStutterEfficiency_params->BlockWidth256BytesY = locals->BlockWidth256BytesY; + CalculateStutterEfficiency_params->BlockHeight256BytesC = locals->BlockHeight256BytesC; + CalculateStutterEfficiency_params->BlockWidth256BytesC = locals->BlockWidth256BytesC; + CalculateStutterEfficiency_params->DCCYMaxUncompressedBlock = locals->DCCYMaxUncompressedBlock; + CalculateStutterEfficiency_params->DCCCMaxUncompressedBlock = locals->DCCCMaxUncompressedBlock; + CalculateStutterEfficiency_params->VActive = mode_lib->ms.cache_display_cfg.timing.VActive; + CalculateStutterEfficiency_params->DCCEnable = mode_lib->ms.cache_display_cfg.surface.DCCEnable; + CalculateStutterEfficiency_params->WritebackEnable = mode_lib->ms.cache_display_cfg.writeback.WritebackEnable; + CalculateStutterEfficiency_params->ReadBandwidthSurfaceLuma = locals->ReadBandwidthSurfaceLuma; + CalculateStutterEfficiency_params->ReadBandwidthSurfaceChroma = locals->ReadBandwidthSurfaceChroma; + CalculateStutterEfficiency_params->meta_row_bw = locals->meta_row_bw; + CalculateStutterEfficiency_params->dpte_row_bw = locals->dpte_row_bw; + CalculateStutterEfficiency_params->StutterEfficiencyNotIncludingVBlank = &locals->StutterEfficiencyNotIncludingVBlank; + CalculateStutterEfficiency_params->StutterEfficiency = &locals->StutterEfficiency; + CalculateStutterEfficiency_params->NumberOfStutterBurstsPerFrame = &locals->NumberOfStutterBurstsPerFrame; + CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &locals->Z8StutterEfficiencyNotIncludingVBlank; + CalculateStutterEfficiency_params->Z8StutterEfficiency = &locals->Z8StutterEfficiency; + CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &locals->Z8NumberOfStutterBurstsPerFrame; + CalculateStutterEfficiency_params->StutterPeriod = &locals->StutterPeriod; + CalculateStutterEfficiency_params->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = &locals->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE; + + // Stutter Efficiency + CalculateStutterEfficiency(&mode_lib->scratch, + CalculateStutterEfficiency_params); + +#ifdef __DML_VBA_ALLOW_DELTA__ + { + dml_float_t dummy_single[2]; + dml_uint_t dummy_integer[1]; + dml_bool_t dummy_boolean[1]; + + // Calculate z8 stutter eff assuming 0 reserved space + CalculateStutterEfficiency( + locals->CompressedBufferSizeInkByte, + mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange, + locals->UnboundedRequestEnabled, + mode_lib->ms.ip.meta_fifo_size_in_kentries, + mode_lib->ms.ip.zero_size_buffer_entries, + mode_lib->ms.ip.pixel_chunk_size_kbytes, + mode_lib->ms.num_active_planes, + mode_lib->ms.ip.rob_buffer_size_kbytes, + locals->TotalDataReadBandwidth, + locals->Dcfclk, + mode_lib->ms.ReturnBW, + 0, //mode_lib->ms.ip.compbuf_reserved_space_64b, + 0, //mode_lib->ms.ip.compbuf_reserved_space_zs, + mode_lib->ms.state.sr_exit_time_us, + mode_lib->ms.state.sr_exit_z8_time_us, + mode_lib->ms.policy.SynchronizeTimingsFinal, + mode_lib->ms.cache_display_cfg.plane.BlendingAndTiming, + locals->Watermark.StutterEnterPlusExitWatermark, + locals->Watermark.Z8StutterEnterPlusExitWatermark, + mode_lib->ms.ip.ptoi_supported, + mode_lib->ms.cache_display_cfg.timing.Interlace, + locals->MinTTUVBlank, + mode_lib->ms.cache_display_cfg.hw.DPPPerSurface, + mode_lib->ms.DETBufferSizeY, + locals->BytePerPixelY, + locals->BytePerPixelDETY, + locals->SwathWidthY, + mode_lib->ms.SwathHeightY, + mode_lib->ms.SwathHeightC, + mode_lib->ms.cache_display_cfg.surface.DCCRateLuma, + mode_lib->ms.cache_display_cfg.surface.DCCRateChroma, + mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsLuma, + mode_lib->ms.cache_display_cfg.surface.DCCFractionOfZeroSizeRequestsChroma, + mode_lib->ms.cache_display_cfg.timing.HTotal, + mode_lib->ms.cache_display_cfg.timing.VTotal, + mode_lib->ms.cache_display_cfg.timing.PixelClock, + mode_lib->ms.cache_display_cfg.plane.VRatio, + mode_lib->ms.cache_display_cfg.plane.SourceScan, + locals->BlockHeight256BytesY, + locals->BlockWidth256BytesY, + locals->BlockHeight256BytesC, + locals->BlockWidth256BytesC, + locals->DCCYMaxUncompressedBlock, + locals->DCCCMaxUncompressedBlock, + mode_lib->ms.cache_display_cfg.timing.VActive, + mode_lib->ms.cache_display_cfg.surface.DCCEnable, + mode_lib->ms.cache_display_cfg.writeback.WritebackEnable, + locals->ReadBandwidthSurfaceLuma, + locals->ReadBandwidthSurfaceChroma, + locals->meta_row_bw, + locals->dpte_row_bw, + + /* Output */ + &dummy_single[0], + &dummy_single[1], + &dummy_integer[0], + &locals->Z8StutterEfficiencyNotIncludingVBlankBestCase, + &locals->Z8StutterEfficiencyBestCase, + &locals->Z8NumberOfStutterBurstsPerFrameBestCase, + &locals->StutterPeriodBestCase, + &dummy_boolean[0]); + } +#else + locals->Z8StutterEfficiencyNotIncludingVBlankBestCase = locals->Z8StutterEfficiencyNotIncludingVBlank; + locals->Z8StutterEfficiencyBestCase = locals->Z8StutterEfficiency; + locals->Z8NumberOfStutterBurstsPerFrameBestCase = locals->Z8NumberOfStutterBurstsPerFrame; + locals->StutterPeriodBestCase = locals->StutterPeriod; +#endif + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: --- END --- \n", __func__); +#endif +} // dml_core_mode_programming + +/// Function: dml_core_get_row_heights +/// @brief Get row height for DPTE and META with minimal input. +void dml_core_get_row_heights( + dml_uint_t *dpte_row_height, + dml_uint_t *meta_row_height, + const struct display_mode_lib_st *mode_lib, + dml_bool_t is_plane1, + enum dml_source_format_class SourcePixelFormat, + enum dml_swizzle_mode SurfaceTiling, + enum dml_rotation_angle ScanDirection, + dml_uint_t pitch, + dml_uint_t GPUVMMinPageSizeKBytes) +{ + dml_uint_t BytePerPixelY; + dml_uint_t BytePerPixelC; + dml_float_t BytePerPixelInDETY; + dml_float_t BytePerPixelInDETC; + dml_uint_t BlockHeight256BytesY; + dml_uint_t BlockHeight256BytesC; + dml_uint_t BlockWidth256BytesY; + dml_uint_t BlockWidth256BytesC; + dml_uint_t MacroTileWidthY; + dml_uint_t MacroTileWidthC; + dml_uint_t MacroTileHeightY; + dml_uint_t MacroTileHeightC; + + dml_uint_t BytePerPixel; + dml_uint_t BlockHeight256Bytes; + dml_uint_t BlockWidth256Bytes; + dml_uint_t MacroTileWidth; + dml_uint_t MacroTileHeight; + dml_uint_t PTEBufferSizeInRequests; + + dml_uint_t dummy_integer[16]; + + CalculateBytePerPixelAndBlockSizes( + SourcePixelFormat, + SurfaceTiling, + + /* Output */ + &BytePerPixelY, + &BytePerPixelC, + &BytePerPixelInDETY, + &BytePerPixelInDETC, + &BlockHeight256BytesY, + &BlockHeight256BytesC, + &BlockWidth256BytesY, + &BlockWidth256BytesC, + &MacroTileHeightY, + &MacroTileHeightC, + &MacroTileWidthY, + &MacroTileWidthC); + + BytePerPixel = is_plane1 ? BytePerPixelC : BytePerPixelY; + BlockHeight256Bytes = is_plane1 ? BlockHeight256BytesC : BlockHeight256BytesY; + BlockWidth256Bytes = is_plane1 ? BlockWidth256BytesC : BlockWidth256BytesY; + MacroTileWidth = is_plane1 ? MacroTileWidthC : MacroTileWidthY; + MacroTileHeight = is_plane1 ? MacroTileHeightC : MacroTileHeightY; + PTEBufferSizeInRequests = is_plane1 ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; +#ifdef __DML_RQ_DLG_CALC_DEBUG__ + dml_print("DML_DLG: %s: is_plane1 = %u\n", __func__, is_plane1); + dml_print("DML_DLG: %s: BytePerPixel = %u\n", __func__, BytePerPixel); + dml_print("DML_DLG: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); + dml_print("DML_DLG: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); + dml_print("DML_DLG: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); + dml_print("DML_DLG: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); + dml_print("DML_DLG: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); + dml_print("DML_DLG: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); + dml_print("DML_DLG: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); + dml_print("DML_DLG: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); +#endif + + // just supply with enough parameters to calculate meta and dte + CalculateVMAndRowBytes( + 0, // dml_bool_t ViewportStationary, + 1, // dml_bool_t DCCEnable, + 1, // dml_uint_t NumberOfDPPs, + BlockHeight256Bytes, + BlockWidth256Bytes, + SourcePixelFormat, + SurfaceTiling, + BytePerPixel, + ScanDirection, + 0, // dml_uint_t SwathWidth, + 0, // dml_uint_t ViewportHeight, (Note: DML calculates one_row_for_frame height regardless, would need test input if that height is useful) + 0, // dml_uint_t ViewportXStart, + 0, // dml_uint_t ViewportYStart, + 1, // dml_bool_t GPUVMEnable, + 4, // dml_uint_t GPUVMMaxPageTableLevels, + GPUVMMinPageSizeKBytes, + PTEBufferSizeInRequests, + pitch, + 0, // dml_uint_t DCCMetaPitch, + MacroTileWidth, + MacroTileHeight, + + // /* Output */ + &dummy_integer[0], // dml_uint_t *MetaRowByte, + &dummy_integer[1], // dml_uint_t *PixelPTEBytesPerRow, + &dummy_integer[2], // dml_uint_t *PixelPTEBytesPerRowStorage, + &dummy_integer[3], // dml_uint_t *dpte_row_width_ub, + dpte_row_height, + &dummy_integer[4], // dml_uint_t *dpte_row_height_linear + &dummy_integer[5], // dml_uint_t *PixelPTEBytesPerRow_one_row_per_frame, + &dummy_integer[6], // dml_uint_t *dpte_row_width_ub_one_row_per_frame, + &dummy_integer[7], // dml_uint_t *dpte_row_height_one_row_per_frame, + &dummy_integer[8], // dml_uint_t *MetaRequestWidth, + &dummy_integer[9], // dml_uint_t *MetaRequestHeight, + &dummy_integer[10], // dml_uint_t *meta_row_width, + meta_row_height, + &dummy_integer[11], // dml_uint_t *PixelPTEReqWidth, + &dummy_integer[12], // dml_uint_t *PixelPTEReqHeight, + &dummy_integer[13], // dml_uint_t *PTERequestSize, + &dummy_integer[14], // dml_uint_t *DPDE0BytesFrame, + &dummy_integer[15]); // dml_uint_t *MetaPTEBytesFrame) + +#ifdef __DML_RQ_DLG_CALC_DEBUG__ + dml_print("DML_DLG: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); + dml_print("DML_DLG: %s: meta_row_height = %u\n", __func__, *meta_row_height); +#endif +} + +static struct soc_state_bounding_box_st dml_get_soc_state_bounding_box( + const struct soc_states_st *states, + dml_uint_t state_idx) +{ + dml_print("DML::%s: state_idx=%u (num_states=%u)\n", __func__, state_idx, states->num_states); + + if (state_idx >= (dml_uint_t)states->num_states) { + dml_print("DML::%s: ERROR: Invalid state_idx=%u! num_states=%u\n", __func__, state_idx, states->num_states); + ASSERT(0); + } + return (states->state_array[state_idx]); +} + +/// @brief Copy the parameters to a calculation struct, it actually only need when the DML needs to have +/// the intelligence to re-calculate when any of display cfg, bbox, or policy changes since last calculated. +/// +static void cache_ip_soc_cfg(struct display_mode_lib_st *mode_lib, + dml_uint_t state_idx) +{ + mode_lib->ms.state_idx = state_idx; + mode_lib->ms.max_state_idx = mode_lib->states.num_states - 1; + mode_lib->ms.soc = mode_lib->soc; + mode_lib->ms.ip = mode_lib->ip; + mode_lib->ms.policy = mode_lib->policy; + mode_lib->ms.state = dml_get_soc_state_bounding_box(&mode_lib->states, state_idx); + mode_lib->ms.max_state = dml_get_soc_state_bounding_box(&mode_lib->states, mode_lib->states.num_states - 1); +} + +static void cache_display_cfg(struct display_mode_lib_st *mode_lib, + const struct dml_display_cfg_st *display_cfg) +{ + mode_lib->ms.cache_display_cfg = *display_cfg; +} + +static void fetch_socbb_params(struct display_mode_lib_st *mode_lib) +{ + struct soc_state_bounding_box_st *state = &mode_lib->ms.state; + + // Default values, SOCCLK, DRAMSpeed, and FabricClock will be reassigned to the same state value in mode_check step + // If UseMinimumRequiredDCFCLK is used, the DCFCLK will be the min dcflk for the mode support + mode_lib->ms.SOCCLK = (dml_float_t)state->socclk_mhz; + mode_lib->ms.DRAMSpeed = (dml_float_t)state->dram_speed_mts; + mode_lib->ms.FabricClock = (dml_float_t)state->fabricclk_mhz; + mode_lib->ms.DCFCLK = (dml_float_t)state->dcfclk_mhz; +} + +/// @brief Use display_cfg directly for mode_support calculation +/// Calculated values and informational output are stored in mode_lib.vba data struct +/// The display configuration is described with pipes struct and num_pipes +/// This function is used when physical resource mapping is not finalized (for example, +/// don't know how many pipes to represent a surface) +/// @param mode_lib Contains the bounding box and policy setting. +/// @param state_idx Power state index +/// @param display_cfg Display configurations. A display +dml_bool_t dml_mode_support( + struct display_mode_lib_st *mode_lib, + dml_uint_t state_idx, + const struct dml_display_cfg_st *display_cfg) +{ + dml_bool_t is_mode_support; + + dml_print("DML::%s: ------------- START ----------\n", __func__); + cache_ip_soc_cfg(mode_lib, state_idx); + cache_display_cfg(mode_lib, display_cfg); + + fetch_socbb_params(mode_lib); + + dml_print("DML::%s: state_idx = %u\n", __func__, state_idx); + + is_mode_support = dml_core_mode_support(mode_lib); + + dml_print("DML::%s: is_mode_support = %u\n", __func__, is_mode_support); + dml_print("DML::%s: ------------- DONE ----------\n", __func__); + return is_mode_support; +} + +/// @Brief A function to calculate the programming values for DCN DCHUB (Assume mode is supported) +/// The output will be stored in the mode_lib.mp (mode_program_st) data struct and those can be accessed via the getter functions +/// Calculated values include: watermarks, dlg, rq reg, different clock frequency +/// This function returns 1 when there is no error. +/// Note: In this function, it is assumed that DCFCLK, SOCCLK freq are the state values, and mode_program will just use the DML calculated DPPCLK and DISPCLK +/// @param mode_lib mode_lib data struct that house all the input/output/bbox and calculation values. +/// @param state_idx Power state idx chosen +/// @param display_cfg Display Configuration +/// @param call_standalone Calling mode_programming without calling mode support. Some of the "support" struct member will be pre-calculated before doing mode programming +/// TODO: Add clk_cfg input, could be useful for standalone mode +dml_bool_t dml_mode_programming( + struct display_mode_lib_st *mode_lib, + dml_uint_t state_idx, + const struct dml_display_cfg_st *display_cfg, + bool call_standalone) +{ + struct dml_clk_cfg_st clk_cfg; + memset(&clk_cfg, 0, sizeof(clk_cfg)); + + clk_cfg.dcfclk_option = dml_use_required_freq; + clk_cfg.dispclk_option = dml_use_required_freq; + for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; ++k) + clk_cfg.dppclk_option[k] = dml_use_required_freq; + + dml_print("DML::%s: ------------- START ----------\n", __func__); + dml_print("DML::%s: state_idx = %u\n", __func__, state_idx); + dml_print("DML::%s: call_standalone = %u\n", __func__, call_standalone); + + cache_ip_soc_cfg(mode_lib, state_idx); + cache_display_cfg(mode_lib, display_cfg); + + fetch_socbb_params(mode_lib); + if (call_standalone) { + mode_lib->ms.support.ImmediateFlipSupport = 1; // assume mode support say immediate flip ok at max state/combine + dml_core_mode_support_partial(mode_lib); + } + + dml_core_mode_programming(mode_lib, &clk_cfg); + + dml_print("DML::%s: ------------- DONE ----------\n", __func__); + dml_print("DML::%s: PrefetchAndImmediateFlipSupported = %0d\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); + return mode_lib->mp.PrefetchAndImmediateFlipSupported; +} + +static dml_uint_t mode_support_pwr_states( + dml_uint_t *lowest_state_idx, + struct display_mode_lib_st *mode_lib, + const struct dml_display_cfg_st *display_cfg, + dml_uint_t start_state_idx, + dml_uint_t end_state_idx) +{ + dml_uint_t state_idx = 0; + dml_bool_t mode_is_supported = 0; + *lowest_state_idx = end_state_idx; + + if (end_state_idx < start_state_idx) + ASSERT(0); + + if (end_state_idx >= mode_lib->states.num_states) // idx is 0-based + ASSERT(0); + + for (state_idx = start_state_idx; state_idx <= end_state_idx; state_idx++) { + if (dml_mode_support(mode_lib, state_idx, display_cfg)) { + dml_print("DML::%s: Mode is supported at power state_idx = %u\n", __func__, state_idx); + mode_is_supported = 1; + *lowest_state_idx = state_idx; + break; + } + } + + return mode_is_supported; +} + +dml_uint_t dml_mode_support_ex(struct dml_mode_support_ex_params_st *in_out_params) +{ + dml_uint_t result; + + result = mode_support_pwr_states(&in_out_params->out_lowest_state_idx, + in_out_params->mode_lib, + in_out_params->in_display_cfg, + in_out_params->in_start_state_idx, + in_out_params->mode_lib->states.num_states - 1); + + if (result) + *in_out_params->out_evaluation_info = in_out_params->mode_lib->ms.support; + + return result; +} + +dml_bool_t dml_get_is_phantom_pipe(struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) +{ + dml_uint_t plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; + dml_print("DML::%s: pipe_idx=%d UseMALLForPStateChange=%0d\n", __func__, pipe_idx, mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[plane_idx]); + return (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[plane_idx] == dml_use_mall_pstate_change_phantom_pipe); +} + + +#define dml_get_per_surface_var_func(variable, type, interval_var) type dml_get_##variable(struct display_mode_lib_st *mode_lib, dml_uint_t surface_idx) \ +{ \ + dml_uint_t plane_idx; \ + plane_idx = mode_lib->mp.pipe_plane[surface_idx]; \ + return (type) interval_var[plane_idx]; \ +} + +#define dml_get_var_func(var, type, internal_var) type dml_get_##var(struct display_mode_lib_st *mode_lib) \ +{ \ + return (type) internal_var; \ +} + +dml_get_var_func(wm_urgent, dml_float_t, mode_lib->mp.Watermark.UrgentWatermark); +dml_get_var_func(wm_stutter_exit, dml_float_t, mode_lib->mp.Watermark.StutterExitWatermark); +dml_get_var_func(wm_stutter_enter_exit, dml_float_t, mode_lib->mp.Watermark.StutterEnterPlusExitWatermark); +dml_get_var_func(wm_memory_trip, dml_float_t, mode_lib->mp.UrgentLatency); +dml_get_var_func(wm_fclk_change, dml_float_t, mode_lib->mp.Watermark.FCLKChangeWatermark); +dml_get_var_func(wm_usr_retraining, dml_float_t, mode_lib->mp.Watermark.USRRetrainingWatermark); +dml_get_var_func(wm_dram_clock_change, dml_float_t, mode_lib->mp.Watermark.DRAMClockChangeWatermark); +dml_get_var_func(wm_z8_stutter_enter_exit, dml_float_t, mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark); +dml_get_var_func(wm_z8_stutter, dml_float_t, mode_lib->mp.Watermark.Z8StutterExitWatermark); +dml_get_var_func(fraction_of_urgent_bandwidth, dml_float_t, mode_lib->mp.FractionOfUrgentBandwidth); +dml_get_var_func(fraction_of_urgent_bandwidth_imm_flip, dml_float_t, mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip); +dml_get_var_func(urgent_latency, dml_float_t, mode_lib->mp.UrgentLatency); +dml_get_var_func(clk_dcf_deepsleep, dml_float_t, mode_lib->mp.DCFCLKDeepSleep); +dml_get_var_func(wm_writeback_dram_clock_change, dml_float_t, mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); +dml_get_var_func(wm_writeback_urgent, dml_float_t, mode_lib->mp.Watermark.WritebackUrgentWatermark); +dml_get_var_func(stutter_efficiency, dml_float_t, mode_lib->mp.StutterEfficiency); +dml_get_var_func(stutter_efficiency_no_vblank, dml_float_t, mode_lib->mp.StutterEfficiencyNotIncludingVBlank); +dml_get_var_func(stutter_efficiency_z8, dml_float_t, mode_lib->mp.Z8StutterEfficiency); +dml_get_var_func(stutter_num_bursts_z8, dml_float_t, mode_lib->mp.Z8NumberOfStutterBurstsPerFrame); +dml_get_var_func(stutter_period, dml_float_t, mode_lib->mp.StutterPeriod); +dml_get_var_func(stutter_efficiency_z8_bestcase, dml_float_t, mode_lib->mp.Z8StutterEfficiencyBestCase); +dml_get_var_func(stutter_num_bursts_z8_bestcase, dml_float_t, mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase); +dml_get_var_func(stutter_period_bestcase, dml_float_t, mode_lib->mp.StutterPeriodBestCase); +dml_get_var_func(urgent_extra_latency, dml_float_t, mode_lib->mp.UrgentExtraLatency); +dml_get_var_func(fclk_change_latency, dml_float_t, mode_lib->mp.MaxActiveFCLKChangeLatencySupported); +dml_get_var_func(dispclk_calculated, dml_float_t, mode_lib->mp.Dispclk_calculated); +dml_get_var_func(total_data_read_bw, dml_float_t, mode_lib->mp.TotalDataReadBandwidth); +dml_get_var_func(return_bw, dml_float_t, mode_lib->ms.ReturnBW); +dml_get_var_func(return_dram_bw, dml_float_t, mode_lib->ms.ReturnDRAMBW); +dml_get_var_func(tcalc, dml_float_t, mode_lib->mp.TCalc); +dml_get_var_func(comp_buffer_size_kbytes, dml_uint_t, mode_lib->mp.CompressedBufferSizeInkByte); +dml_get_var_func(pixel_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.pixel_chunk_size_kbytes); +dml_get_var_func(alpha_pixel_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.alpha_pixel_chunk_size_kbytes); +dml_get_var_func(meta_chunk_size_in_kbyte, dml_uint_t, mode_lib->ms.ip.meta_chunk_size_kbytes); +dml_get_var_func(min_pixel_chunk_size_in_byte, dml_uint_t, mode_lib->ms.ip.min_pixel_chunk_size_bytes); +dml_get_var_func(min_meta_chunk_size_in_byte, dml_uint_t, mode_lib->ms.ip.min_meta_chunk_size_bytes); +dml_get_var_func(total_immediate_flip_bytes, dml_uint_t, mode_lib->mp.TotImmediateFlipBytes); + +dml_get_per_surface_var_func(dsc_delay, dml_uint_t, mode_lib->mp.DSCDelay); // this is the dsc latency +dml_get_per_surface_var_func(dppclk_calculated, dml_float_t, mode_lib->mp.Dppclk_calculated); +dml_get_per_surface_var_func(dscclk_calculated, dml_float_t, mode_lib->mp.DSCCLK_calculated); +dml_get_per_surface_var_func(min_ttu_vblank_in_us, dml_float_t, mode_lib->mp.MinTTUVBlank); +dml_get_per_surface_var_func(vratio_prefetch_l, dml_float_t, mode_lib->mp.VRatioPrefetchY); +dml_get_per_surface_var_func(vratio_prefetch_c, dml_float_t, mode_lib->mp.VRatioPrefetchC); +dml_get_per_surface_var_func(dst_x_after_scaler, dml_uint_t, mode_lib->mp.DSTXAfterScaler); +dml_get_per_surface_var_func(dst_y_after_scaler, dml_uint_t, mode_lib->mp.DSTYAfterScaler); +dml_get_per_surface_var_func(dst_y_per_vm_vblank, dml_float_t, mode_lib->mp.DestinationLinesToRequestVMInVBlank); +dml_get_per_surface_var_func(dst_y_per_row_vblank, dml_float_t, mode_lib->mp.DestinationLinesToRequestRowInVBlank); +dml_get_per_surface_var_func(dst_y_prefetch, dml_float_t, mode_lib->mp.DestinationLinesForPrefetch); +dml_get_per_surface_var_func(dst_y_per_vm_flip, dml_float_t, mode_lib->mp.DestinationLinesToRequestVMInImmediateFlip); +dml_get_per_surface_var_func(dst_y_per_row_flip, dml_float_t, mode_lib->mp.DestinationLinesToRequestRowInImmediateFlip); +dml_get_per_surface_var_func(dst_y_per_pte_row_nom_l, dml_float_t, mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L); +dml_get_per_surface_var_func(dst_y_per_pte_row_nom_c, dml_float_t, mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C); +dml_get_per_surface_var_func(dst_y_per_meta_row_nom_l, dml_float_t, mode_lib->mp.DST_Y_PER_META_ROW_NOM_L); +dml_get_per_surface_var_func(dst_y_per_meta_row_nom_c, dml_float_t, mode_lib->mp.DST_Y_PER_META_ROW_NOM_C); +dml_get_per_surface_var_func(refcyc_per_vm_group_vblank_in_us, dml_float_t, mode_lib->mp.TimePerVMGroupVBlank); +dml_get_per_surface_var_func(refcyc_per_vm_group_flip_in_us, dml_float_t, mode_lib->mp.TimePerVMGroupFlip); +dml_get_per_surface_var_func(refcyc_per_vm_req_vblank_in_us, dml_float_t, mode_lib->mp.TimePerVMRequestVBlank); +dml_get_per_surface_var_func(refcyc_per_vm_req_flip_in_us, dml_float_t, mode_lib->mp.TimePerVMRequestFlip); +dml_get_per_surface_var_func(refcyc_per_vm_dmdata_in_us, dml_float_t, mode_lib->mp.Tdmdl_vm); +dml_get_per_surface_var_func(dmdata_dl_delta_in_us, dml_float_t, mode_lib->mp.Tdmdl); +dml_get_per_surface_var_func(refcyc_per_line_delivery_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeLuma); +dml_get_per_surface_var_func(refcyc_per_line_delivery_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeChroma); +dml_get_per_surface_var_func(refcyc_per_line_delivery_pre_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch); +dml_get_per_surface_var_func(refcyc_per_line_delivery_pre_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch); +dml_get_per_surface_var_func(refcyc_per_req_delivery_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma); +dml_get_per_surface_var_func(refcyc_per_req_delivery_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma); +dml_get_per_surface_var_func(refcyc_per_req_delivery_pre_l_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch); +dml_get_per_surface_var_func(refcyc_per_req_delivery_pre_c_in_us, dml_float_t, mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch); +dml_get_per_surface_var_func(refcyc_per_cursor_req_delivery_in_us, dml_float_t, mode_lib->mp.CursorRequestDeliveryTime); +dml_get_per_surface_var_func(refcyc_per_cursor_req_delivery_pre_in_us, dml_float_t, mode_lib->mp.CursorRequestDeliveryTimePrefetch); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_nom_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkNominal); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_nom_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkNominal); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_vblank_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkVBlank); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_vblank_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkVBlank); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_flip_l_in_us, dml_float_t, mode_lib->mp.TimePerMetaChunkFlip); +dml_get_per_surface_var_func(refcyc_per_meta_chunk_flip_c_in_us, dml_float_t, mode_lib->mp.TimePerChromaMetaChunkFlip); +dml_get_per_surface_var_func(refcyc_per_pte_group_nom_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_nom_luma); +dml_get_per_surface_var_func(refcyc_per_pte_group_nom_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_nom_chroma); +dml_get_per_surface_var_func(refcyc_per_pte_group_vblank_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_vblank_luma); +dml_get_per_surface_var_func(refcyc_per_pte_group_vblank_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_vblank_chroma); +dml_get_per_surface_var_func(refcyc_per_pte_group_flip_l_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_flip_luma); +dml_get_per_surface_var_func(refcyc_per_pte_group_flip_c_in_us, dml_float_t, mode_lib->mp.time_per_pte_group_flip_chroma); +dml_get_per_surface_var_func(dpte_group_size_in_bytes, dml_uint_t, mode_lib->mp.dpte_group_bytes); +dml_get_per_surface_var_func(vm_group_size_in_bytes, dml_uint_t, mode_lib->mp.vm_group_bytes); +dml_get_per_surface_var_func(swath_height_l, dml_uint_t, mode_lib->ms.SwathHeightY); +dml_get_per_surface_var_func(swath_height_c, dml_uint_t, mode_lib->ms.SwathHeightC); +dml_get_per_surface_var_func(dpte_row_height_l, dml_uint_t, mode_lib->mp.dpte_row_height); +dml_get_per_surface_var_func(dpte_row_height_c, dml_uint_t, mode_lib->mp.dpte_row_height_chroma); +dml_get_per_surface_var_func(dpte_row_height_linear_l, dml_uint_t, mode_lib->mp.dpte_row_height_linear); +dml_get_per_surface_var_func(dpte_row_height_linear_c, dml_uint_t, mode_lib->mp.dpte_row_height_linear_chroma); +dml_get_per_surface_var_func(meta_row_height_l, dml_uint_t, mode_lib->mp.meta_row_height); +dml_get_per_surface_var_func(meta_row_height_c, dml_uint_t, mode_lib->mp.meta_row_height_chroma); + +dml_get_per_surface_var_func(vstartup_calculated, dml_uint_t, mode_lib->mp.VStartup); +dml_get_per_surface_var_func(vupdate_offset, dml_uint_t, mode_lib->mp.VUpdateOffsetPix); +dml_get_per_surface_var_func(vupdate_width, dml_uint_t, mode_lib->mp.VUpdateWidthPix); +dml_get_per_surface_var_func(vready_offset, dml_uint_t, mode_lib->mp.VReadyOffsetPix); +dml_get_per_surface_var_func(vready_at_or_after_vsync, dml_uint_t, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC); +dml_get_per_surface_var_func(min_dst_y_next_start, dml_uint_t, mode_lib->mp.MIN_DST_Y_NEXT_START); +dml_get_per_surface_var_func(det_stored_buffer_size_l_bytes, dml_uint_t, mode_lib->ms.DETBufferSizeY); +dml_get_per_surface_var_func(det_stored_buffer_size_c_bytes, dml_uint_t, mode_lib->ms.DETBufferSizeC); +dml_get_per_surface_var_func(use_mall_for_static_screen, dml_uint_t, mode_lib->mp.UsesMALLForStaticScreen); +dml_get_per_surface_var_func(surface_size_for_mall, dml_uint_t, mode_lib->mp.SurfaceSizeInTheMALL); +dml_get_per_surface_var_func(dcc_max_uncompressed_block_l, dml_uint_t, mode_lib->mp.DCCYMaxUncompressedBlock); +dml_get_per_surface_var_func(dcc_max_compressed_block_l, dml_uint_t, mode_lib->mp.DCCYMaxCompressedBlock); +dml_get_per_surface_var_func(dcc_independent_block_l, dml_uint_t, mode_lib->mp.DCCYIndependentBlock); +dml_get_per_surface_var_func(dcc_max_uncompressed_block_c, dml_uint_t, mode_lib->mp.DCCCMaxUncompressedBlock); +dml_get_per_surface_var_func(dcc_max_compressed_block_c, dml_uint_t, mode_lib->mp.DCCCMaxCompressedBlock); +dml_get_per_surface_var_func(dcc_independent_block_c, dml_uint_t, mode_lib->mp.DCCCIndependentBlock); +dml_get_per_surface_var_func(max_active_dram_clock_change_latency_supported, dml_uint_t, mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported); +dml_get_per_surface_var_func(pte_buffer_mode, dml_uint_t, mode_lib->mp.PTE_BUFFER_MODE); +dml_get_per_surface_var_func(bigk_fragment_size, dml_uint_t, mode_lib->mp.BIGK_FRAGMENT_SIZE); +dml_get_per_surface_var_func(dpte_bytes_per_row, dml_uint_t, mode_lib->mp.PixelPTEBytesPerRow); +dml_get_per_surface_var_func(meta_bytes_per_row, dml_uint_t, mode_lib->mp.MetaRowByte); +dml_get_per_surface_var_func(det_buffer_size_kbytes, dml_uint_t, mode_lib->ms.DETBufferSizeInKByte); + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h new file mode 100644 index 000000000000..a38ed89c47a9 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h @@ -0,0 +1,204 @@ +/* + * Copyright 2022 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DISPLAY_MODE_CORE_H__ +#define __DISPLAY_MODE_CORE_H__ + +#include "display_mode_core_structs.h" + +struct display_mode_lib_st; + +dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib); +void dml_core_mode_support_partial(struct display_mode_lib_st *mode_lib); +void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struct dml_clk_cfg_st *clk_cfg); + +void dml_core_get_row_heights( + dml_uint_t *dpte_row_height, + dml_uint_t *meta_row_height, + const struct display_mode_lib_st *mode_lib, + dml_bool_t is_plane1, + enum dml_source_format_class SourcePixelFormat, + enum dml_swizzle_mode SurfaceTiling, + enum dml_rotation_angle ScanDirection, + dml_uint_t pitch, + dml_uint_t GPUVMMinPageSizeKBytes); + +dml_float_t dml_get_return_bw_mbps_vm_only( + const struct soc_bounding_box_st *soc, + dml_bool_t use_ideal_dram_bw_strobe, + dml_bool_t HostVMEnable, + dml_float_t DCFCLK, + dml_float_t FabricClock, + dml_float_t DRAMSpeed); + +dml_float_t dml_get_return_bw_mbps( + const struct soc_bounding_box_st *soc, + dml_bool_t use_ideal_dram_bw_strobe, + dml_bool_t HostVMEnable, + dml_float_t DCFCLK, + dml_float_t FabricClock, + dml_float_t DRAMSpeed); + +dml_bool_t dml_mode_support( + struct display_mode_lib_st *mode_lib, + dml_uint_t state_idx, + const struct dml_display_cfg_st *display_cfg); + +dml_bool_t dml_mode_programming( + struct display_mode_lib_st *mode_lib, + dml_uint_t state_idx, + const struct dml_display_cfg_st *display_cfg, + bool call_standalone); + +dml_uint_t dml_mode_support_ex( + struct dml_mode_support_ex_params_st *in_out_params); + +dml_bool_t dml_get_is_phantom_pipe(struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx); + +#define dml_get_per_surface_var_decl(variable, type) type dml_get_##variable(struct display_mode_lib_st *mode_lib, dml_uint_t surface_idx) +#define dml_get_var_decl(var, type) type dml_get_##var(struct display_mode_lib_st *mode_lib) + +dml_get_var_decl(wm_urgent, dml_float_t); +dml_get_var_decl(wm_stutter_exit, dml_float_t); +dml_get_var_decl(wm_stutter_enter_exit, dml_float_t); +dml_get_var_decl(wm_memory_trip, dml_float_t); +dml_get_var_decl(wm_dram_clock_change, dml_float_t); +dml_get_var_decl(wm_z8_stutter_enter_exit, dml_float_t); +dml_get_var_decl(wm_z8_stutter, dml_float_t); +dml_get_var_decl(urgent_latency, dml_float_t); +dml_get_var_decl(clk_dcf_deepsleep, dml_float_t); +dml_get_var_decl(wm_fclk_change, dml_float_t); +dml_get_var_decl(wm_usr_retraining, dml_float_t); +dml_get_var_decl(urgent_latency, dml_float_t); + +dml_get_var_decl(wm_writeback_dram_clock_change, dml_float_t); +dml_get_var_decl(wm_writeback_urgent, dml_float_t); +dml_get_var_decl(stutter_efficiency_no_vblank, dml_float_t); +dml_get_var_decl(stutter_efficiency, dml_float_t); +dml_get_var_decl(stutter_efficiency_z8, dml_float_t); +dml_get_var_decl(stutter_num_bursts_z8, dml_float_t); +dml_get_var_decl(stutter_period, dml_float_t); +dml_get_var_decl(stutter_efficiency_z8_bestcase, dml_float_t); +dml_get_var_decl(stutter_num_bursts_z8_bestcase, dml_float_t); +dml_get_var_decl(stutter_period_bestcase, dml_float_t); +dml_get_var_decl(urgent_latency, dml_float_t); +dml_get_var_decl(urgent_extra_latency, dml_float_t); +dml_get_var_decl(fclk_change_latency, dml_float_t); +dml_get_var_decl(nonurgent_latency, dml_float_t); +dml_get_var_decl(dispclk_calculated, dml_float_t); +dml_get_var_decl(total_data_read_bw, dml_float_t); +dml_get_var_decl(return_bw, dml_float_t); +dml_get_var_decl(return_dram_bw, dml_float_t); +dml_get_var_decl(tcalc, dml_float_t); +dml_get_var_decl(fraction_of_urgent_bandwidth, dml_float_t); +dml_get_var_decl(fraction_of_urgent_bandwidth_imm_flip, dml_float_t); +dml_get_var_decl(comp_buffer_size_kbytes, dml_uint_t); +dml_get_var_decl(pixel_chunk_size_in_kbyte, dml_uint_t); +dml_get_var_decl(alpha_pixel_chunk_size_in_kbyte, dml_uint_t); +dml_get_var_decl(meta_chunk_size_in_kbyte, dml_uint_t); +dml_get_var_decl(min_pixel_chunk_size_in_byte, dml_uint_t); +dml_get_var_decl(min_meta_chunk_size_in_byte, dml_uint_t); +dml_get_var_decl(total_immediate_flip_bytes, dml_uint_t); + +dml_get_per_surface_var_decl(dsc_delay, dml_uint_t); +dml_get_per_surface_var_decl(dppclk_calculated, dml_float_t); +dml_get_per_surface_var_decl(dscclk_calculated, dml_float_t); +dml_get_per_surface_var_decl(min_ttu_vblank_in_us, dml_float_t); +dml_get_per_surface_var_decl(vratio_prefetch_l, dml_float_t); +dml_get_per_surface_var_decl(vratio_prefetch_c, dml_float_t); +dml_get_per_surface_var_decl(dst_x_after_scaler, dml_uint_t); +dml_get_per_surface_var_decl(dst_y_after_scaler, dml_uint_t); +dml_get_per_surface_var_decl(dst_y_per_vm_vblank, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_row_vblank, dml_float_t); +dml_get_per_surface_var_decl(dst_y_prefetch, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_vm_flip, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_row_flip, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_pte_row_nom_l, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_pte_row_nom_c, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_meta_row_nom_l, dml_float_t); +dml_get_per_surface_var_decl(dst_y_per_meta_row_nom_c, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_vm_group_vblank_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_vm_group_flip_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_vm_req_vblank_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_vm_req_flip_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_vm_dmdata_in_us, dml_float_t); +dml_get_per_surface_var_decl(dmdata_dl_delta_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_line_delivery_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_line_delivery_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_line_delivery_pre_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_line_delivery_pre_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_req_delivery_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_req_delivery_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_req_delivery_pre_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_req_delivery_pre_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_cursor_req_delivery_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_cursor_req_delivery_pre_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_nom_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_nom_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_vblank_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_vblank_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_flip_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_meta_chunk_flip_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_nom_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_nom_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_vblank_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_vblank_c_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_flip_l_in_us, dml_float_t); +dml_get_per_surface_var_decl(refcyc_per_pte_group_flip_c_in_us, dml_float_t); + +dml_get_per_surface_var_decl(dpte_group_size_in_bytes, dml_uint_t); +dml_get_per_surface_var_decl(vm_group_size_in_bytes, dml_uint_t); +dml_get_per_surface_var_decl(swath_height_l, dml_uint_t); +dml_get_per_surface_var_decl(swath_height_c, dml_uint_t); +dml_get_per_surface_var_decl(dpte_row_height_l, dml_uint_t); +dml_get_per_surface_var_decl(dpte_row_height_c, dml_uint_t); +dml_get_per_surface_var_decl(dpte_row_height_linear_l, dml_uint_t); +dml_get_per_surface_var_decl(dpte_row_height_linear_c, dml_uint_t); +dml_get_per_surface_var_decl(meta_row_height_l, dml_uint_t); +dml_get_per_surface_var_decl(meta_row_height_c, dml_uint_t); +dml_get_per_surface_var_decl(vstartup_calculated, dml_uint_t); +dml_get_per_surface_var_decl(vupdate_offset, dml_uint_t); +dml_get_per_surface_var_decl(vupdate_width, dml_uint_t); +dml_get_per_surface_var_decl(vready_offset, dml_uint_t); +dml_get_per_surface_var_decl(vready_at_or_after_vsync, dml_uint_t); +dml_get_per_surface_var_decl(min_dst_y_next_start, dml_uint_t); +dml_get_per_surface_var_decl(det_stored_buffer_size_l_bytes, dml_uint_t); +dml_get_per_surface_var_decl(det_stored_buffer_size_c_bytes, dml_uint_t); +dml_get_per_surface_var_decl(use_mall_for_static_screen, dml_uint_t); +dml_get_per_surface_var_decl(surface_size_for_mall, dml_uint_t); +dml_get_per_surface_var_decl(dcc_max_uncompressed_block_l, dml_uint_t); +dml_get_per_surface_var_decl(dcc_max_uncompressed_block_c, dml_uint_t); +dml_get_per_surface_var_decl(dcc_max_compressed_block_l, dml_uint_t); +dml_get_per_surface_var_decl(dcc_max_compressed_block_c, dml_uint_t); +dml_get_per_surface_var_decl(dcc_independent_block_l, dml_uint_t); +dml_get_per_surface_var_decl(dcc_independent_block_c, dml_uint_t); +dml_get_per_surface_var_decl(max_active_dram_clock_change_latency_supported, dml_uint_t); +dml_get_per_surface_var_decl(pte_buffer_mode, dml_uint_t); +dml_get_per_surface_var_decl(bigk_fragment_size, dml_uint_t); +dml_get_per_surface_var_decl(dpte_bytes_per_row, dml_uint_t); +dml_get_per_surface_var_decl(meta_bytes_per_row, dml_uint_t); +dml_get_per_surface_var_decl(det_buffer_size_kbytes, dml_uint_t); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h new file mode 100644 index 000000000000..3b1d92e7697f --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h @@ -0,0 +1,2032 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DISPLAY_MODE_CORE_STRUCT_H__ +#define __DISPLAY_MODE_CORE_STRUCT_H__ + +#include "display_mode_lib_defines.h" +#include "dml_top_display_cfg_types.h" + +enum dml_project_id { + dml_project_invalid = 0, + dml_project_default = 1, + dml_project_dcn32 = dml_project_default, + dml_project_dcn321 = 2, + dml_project_dcn35 = 3, + dml_project_dcn351 = 4, + dml_project_dcn401 = 5, + dml_project_dcn36 = 6, +}; +enum dml_prefetch_modes { + dml_prefetch_support_uclk_fclk_and_stutter_if_possible = 0, + dml_prefetch_support_uclk_fclk_and_stutter = 1, + dml_prefetch_support_fclk_and_stutter = 2, + dml_prefetch_support_stutter = 3, + dml_prefetch_support_none = 4 +}; +enum dml_use_mall_for_pstate_change_mode { + dml_use_mall_pstate_change_disable = 0, + dml_use_mall_pstate_change_full_frame = 1, + dml_use_mall_pstate_change_sub_viewport = 2, + dml_use_mall_pstate_change_phantom_pipe = 3, + dml_use_mall_pstate_change_phantom_pipe_no_data_return = 4, + dml_use_mall_pstate_change_imall = 5 +}; +enum dml_use_mall_for_static_screen_mode { + dml_use_mall_static_screen_disable = 0, + dml_use_mall_static_screen_enable = 1, + dml_use_mall_static_screen_optimize = 2 +}; +enum dml_output_encoder_class { + dml_dp = 0, + dml_edp = 1, + dml_dp2p0 = 2, + dml_hdmi = 3, + dml_hdmifrl = 4, + dml_none = 5 +}; +enum dml_output_link_dp_rate{ + dml_dp_rate_na = 0, + dml_dp_rate_hbr = 1, + dml_dp_rate_hbr2 = 2, + dml_dp_rate_hbr3 = 3, + dml_dp_rate_uhbr10 = 4, + dml_dp_rate_uhbr13p5 = 5, + dml_dp_rate_uhbr20 = 6 +}; +enum dml_output_type_and_rate__type{ + dml_output_type_unknown = 0, + dml_output_type_dp = 1, + dml_output_type_edp = 2, + dml_output_type_dp2p0 = 3, + dml_output_type_hdmi = 4, + dml_output_type_hdmifrl = 5 +}; +enum dml_output_type_and_rate__rate { + dml_output_rate_unknown = 0, + dml_output_rate_dp_rate_hbr = 1, + dml_output_rate_dp_rate_hbr2 = 2, + dml_output_rate_dp_rate_hbr3 = 3, + dml_output_rate_dp_rate_uhbr10 = 4, + dml_output_rate_dp_rate_uhbr13p5 = 5, + dml_output_rate_dp_rate_uhbr20 = 6, + dml_output_rate_hdmi_rate_3x3 = 7, + dml_output_rate_hdmi_rate_6x3 = 8, + dml_output_rate_hdmi_rate_6x4 = 9, + dml_output_rate_hdmi_rate_8x4 = 10, + dml_output_rate_hdmi_rate_10x4 = 11, + dml_output_rate_hdmi_rate_12x4 = 12 +}; +enum dml_output_format_class { + dml_444 = 0, + dml_s422 = 1, + dml_n422 = 2, + dml_420 = 3 +}; +enum dml_source_format_class { + dml_444_8 = 0, + dml_444_16 = 1, + dml_444_32 = 2, + dml_444_64 = 3, + dml_420_8 = 4, + dml_420_10 = 5, + dml_420_12 = 6, + dml_422_8 = 7, + dml_422_10 = 8, + dml_rgbe_alpha = 9, + dml_rgbe = 10, + dml_mono_8 = 11, + dml_mono_16 = 12 +}; +enum dml_output_bpc_class { + dml_out_6 = 0, + dml_out_8 = 1, + dml_out_10 = 2, + dml_out_12 = 3, + dml_out_16 = 4 +}; +enum dml_output_standard_class { + dml_std_cvt = 0, + dml_std_cea = 1, + dml_std_cvtr2 = 2 +}; +enum dml_rotation_angle { + dml_rotation_0 = 0, + dml_rotation_90 = 1, + dml_rotation_180 = 2, + dml_rotation_270 = 3, + dml_rotation_0m = 4, + dml_rotation_90m = 5, + dml_rotation_180m = 6, + dml_rotation_270m = 7 +}; +enum dml_swizzle_mode { + dml_sw_linear = 0, + dml_sw_256b_s = 1, + dml_sw_256b_d = 2, + dml_sw_256b_r = 3, + dml_sw_4kb_z = 4, + dml_sw_4kb_s = 5, + dml_sw_4kb_d = 6, + dml_sw_4kb_r = 7, + dml_sw_64kb_z = 8, + dml_sw_64kb_s = 9, + dml_sw_64kb_d = 10, + dml_sw_64kb_r = 11, + dml_sw_256kb_z = 12, + dml_sw_256kb_s = 13, + dml_sw_256kb_d = 14, + dml_sw_256kb_r = 15, + dml_sw_64kb_z_t = 16, + dml_sw_64kb_s_t = 17, + dml_sw_64kb_d_t = 18, + dml_sw_64kb_r_t = 19, + dml_sw_4kb_z_x = 20, + dml_sw_4kb_s_x = 21, + dml_sw_4kb_d_x = 22, + dml_sw_4kb_r_x = 23, + dml_sw_64kb_z_x = 24, + dml_sw_64kb_s_x = 25, + dml_sw_64kb_d_x = 26, + dml_sw_64kb_r_x = 27, + dml_sw_256kb_z_x = 28, + dml_sw_256kb_s_x = 29, + dml_sw_256kb_d_x = 30, + dml_sw_256kb_r_x = 31, + dml_sw_256b_2d = 32, + dml_sw_4kb_2d = 33, + dml_sw_64kb_2d = 34, + dml_sw_256kb_2d = 35 +}; +enum dml_lb_depth { + dml_lb_6 = 0, + dml_lb_8 = 1, + dml_lb_10 = 2, + dml_lb_12 = 3, + dml_lb_16 = 4 +}; +enum dml_voltage_state { + dml_vmin_lv = 0, + dml_vmin = 1, + dml_vmid = 2, + dml_vnom = 3, + dml_vmax = 4 +}; +enum dml_source_macro_tile_size { + dml_4k_tile = 0, + dml_64k_tile = 1, + dml_256k_tile = 2 +}; +enum dml_cursor_bpp { + dml_cur_2bit = 0, + dml_cur_32bit = 1, + dml_cur_64bit = 2 +}; +enum dml_dram_clock_change_support { + dml_dram_clock_change_vactive = 0, + dml_dram_clock_change_vblank = 1, + dml_dram_clock_change_vblank_drr = 2, + dml_dram_clock_change_vactive_w_mall_full_frame = 3, + dml_dram_clock_change_vactive_w_mall_sub_vp = 4, + dml_dram_clock_change_vblank_w_mall_full_frame = 5, + dml_dram_clock_change_vblank_drr_w_mall_full_frame = 6, + dml_dram_clock_change_vblank_w_mall_sub_vp = 7, + dml_dram_clock_change_vblank_drr_w_mall_sub_vp = 8, + dml_dram_clock_change_unsupported = 9 +}; +enum dml_fclock_change_support { + dml_fclock_change_vactive = 0, + dml_fclock_change_vblank = 1, + dml_fclock_change_unsupported = 2 +}; +enum dml_dsc_enable { + dml_dsc_disable = 0, + dml_dsc_enable = 1, + dml_dsc_enable_if_necessary = 2 +}; +enum dml_mpc_use_policy { + dml_mpc_disabled = 0, + dml_mpc_as_possible = 1, + dml_mpc_as_needed_for_voltage = 2, + dml_mpc_as_needed_for_pstate_and_voltage = 3, + dml_mpc_as_needed = 4, + dml_mpc_2to1 = 5 +}; +enum dml_odm_use_policy { + dml_odm_use_policy_bypass = 0, + dml_odm_use_policy_combine_as_needed = 1, + dml_odm_use_policy_combine_2to1 = 2, + dml_odm_use_policy_combine_3to1 = 3, + dml_odm_use_policy_combine_4to1 = 4, + dml_odm_use_policy_split_1to2 = 5, + dml_odm_use_policy_mso_1to2 = 6, + dml_odm_use_policy_mso_1to4 = 7 +}; +enum dml_odm_mode { + dml_odm_mode_bypass = 0, + dml_odm_mode_combine_2to1 = 1, + dml_odm_mode_combine_3to1 = 2, + dml_odm_mode_combine_4to1 = 3, + dml_odm_mode_split_1to2 = 4, + dml_odm_mode_mso_1to2 = 5, + dml_odm_mode_mso_1to4 = 6 +}; +enum dml_writeback_configuration { + dml_whole_buffer_for_single_stream_no_interleave = 0, + dml_whole_buffer_for_single_stream_interleave = 1 +}; +enum dml_immediate_flip_requirement { + dml_immediate_flip_not_required = 0, + dml_immediate_flip_required = 1, + dml_immediate_flip_if_possible = 2 +}; +enum dml_unbounded_requesting_policy { + dml_unbounded_requesting_enable = 0, + dml_unbounded_requesting_edp_only = 1, + dml_unbounded_requesting_disable = 2 +}; +enum dml_clk_cfg_policy { + dml_use_required_freq = 0, + dml_use_override_freq = 1, + dml_use_state_freq = 2 +}; + +struct soc_state_bounding_box_st { + dml_float_t socclk_mhz; + dml_float_t dscclk_mhz; + dml_float_t phyclk_mhz; + dml_float_t phyclk_d18_mhz; + dml_float_t phyclk_d32_mhz; + dml_float_t dtbclk_mhz; + dml_float_t fabricclk_mhz; + dml_float_t dcfclk_mhz; + dml_float_t dispclk_mhz; + dml_float_t dppclk_mhz; + dml_float_t dram_speed_mts; + dml_float_t urgent_latency_pixel_data_only_us; + dml_float_t urgent_latency_pixel_mixed_with_vm_data_us; + dml_float_t urgent_latency_vm_data_only_us; + dml_float_t writeback_latency_us; + dml_float_t urgent_latency_adjustment_fabric_clock_component_us; + dml_float_t urgent_latency_adjustment_fabric_clock_reference_mhz; + dml_float_t sr_exit_time_us; + dml_float_t sr_enter_plus_exit_time_us; + dml_float_t sr_exit_z8_time_us; + dml_float_t sr_enter_plus_exit_z8_time_us; + dml_float_t dram_clock_change_latency_us; + dml_float_t fclk_change_latency_us; + dml_float_t usr_retraining_latency_us; + dml_bool_t use_ideal_dram_bw_strobe; + dml_float_t g6_temp_read_blackout_us; + + struct { + dml_uint_t urgent_ramp_uclk_cycles; + dml_uint_t trip_to_memory_uclk_cycles; + dml_uint_t meta_trip_to_memory_uclk_cycles; + dml_uint_t maximum_latency_when_urgent_uclk_cycles; + dml_uint_t average_latency_when_urgent_uclk_cycles; + dml_uint_t maximum_latency_when_non_urgent_uclk_cycles; + dml_uint_t average_latency_when_non_urgent_uclk_cycles; + } dml_dcn401_uclk_dpm_dependent_soc_qos_params; +}; + +struct soc_bounding_box_st { + dml_float_t dprefclk_mhz; + dml_float_t xtalclk_mhz; + dml_float_t pcierefclk_mhz; + dml_float_t refclk_mhz; + dml_float_t amclk_mhz; + dml_uint_t max_outstanding_reqs; + dml_float_t pct_ideal_sdp_bw_after_urgent; + dml_float_t pct_ideal_fabric_bw_after_urgent; + dml_float_t pct_ideal_dram_bw_after_urgent_pixel_only; + dml_float_t pct_ideal_dram_bw_after_urgent_pixel_and_vm; + dml_float_t pct_ideal_dram_bw_after_urgent_vm_only; + dml_float_t pct_ideal_dram_bw_after_urgent_strobe; + dml_float_t max_avg_sdp_bw_use_normal_percent; + dml_float_t max_avg_fabric_bw_use_normal_percent; + dml_float_t max_avg_dram_bw_use_normal_percent; + dml_float_t max_avg_dram_bw_use_normal_strobe_percent; + + dml_float_t svp_prefetch_pct_ideal_sdp_bw_after_urgent; + dml_float_t svp_prefetch_pct_ideal_fabric_bw_after_urgent; + dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_pixel_only; + dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_pixel_and_vm; + dml_float_t svp_prefetch_pct_ideal_dram_bw_after_urgent_vm_only; + dml_float_t svp_prefetch_max_avg_sdp_bw_use_normal_percent; + dml_float_t svp_prefetch_max_avg_fabric_bw_use_normal_percent; + dml_float_t svp_prefetch_max_avg_dram_bw_use_normal_percent; + + dml_uint_t round_trip_ping_latency_dcfclk_cycles; + dml_uint_t urgent_out_of_order_return_per_channel_pixel_only_bytes; + dml_uint_t urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; + dml_uint_t urgent_out_of_order_return_per_channel_vm_only_bytes; + dml_uint_t num_chans; + dml_uint_t return_bus_width_bytes; + dml_uint_t dram_channel_width_bytes; + dml_uint_t fabric_datapath_to_dcn_data_return_bytes; + dml_uint_t hostvm_min_page_size_kbytes; + dml_uint_t gpuvm_min_page_size_kbytes; + dml_float_t phy_downspread_percent; + dml_float_t dcn_downspread_percent; + dml_float_t smn_latency_us; + dml_uint_t mall_allocated_for_dcn_mbytes; + dml_float_t dispclk_dppclk_vco_speed_mhz; + dml_bool_t do_urgent_latency_adjustment; + + dml_uint_t mem_word_bytes; + dml_uint_t num_dcc_mcaches; + dml_uint_t mcache_size_bytes; + dml_uint_t mcache_line_size_bytes; + + struct { + dml_bool_t UseNewDCN401SOCParameters; + dml_uint_t df_qos_response_time_fclk_cycles; + dml_uint_t max_round_trip_to_furthest_cs_fclk_cycles; + dml_uint_t mall_overhead_fclk_cycles; + dml_uint_t meta_trip_adder_fclk_cycles; + dml_uint_t average_transport_distance_fclk_cycles; + dml_float_t umc_urgent_ramp_latency_margin; + dml_float_t umc_max_latency_margin; + dml_float_t umc_average_latency_margin; + dml_float_t fabric_max_transport_latency_margin; + dml_float_t fabric_average_transport_latency_margin; + } dml_dcn401_soc_qos_params; + +}; + +struct ip_params_st { + dml_uint_t vblank_nom_default_us; + dml_uint_t rob_buffer_size_kbytes; + dml_uint_t config_return_buffer_size_in_kbytes; + dml_uint_t config_return_buffer_segment_size_in_kbytes; + dml_uint_t compressed_buffer_segment_size_in_kbytes; + dml_uint_t meta_fifo_size_in_kentries; + dml_uint_t zero_size_buffer_entries; + dml_uint_t dpte_buffer_size_in_pte_reqs_luma; + dml_uint_t dpte_buffer_size_in_pte_reqs_chroma; + dml_uint_t dcc_meta_buffer_size_bytes; + dml_bool_t gpuvm_enable; + dml_bool_t hostvm_enable; + dml_uint_t gpuvm_max_page_table_levels; + dml_uint_t hostvm_max_page_table_levels; + dml_uint_t pixel_chunk_size_kbytes; + dml_uint_t alpha_pixel_chunk_size_kbytes; + dml_uint_t min_pixel_chunk_size_bytes; + dml_uint_t meta_chunk_size_kbytes; + dml_uint_t min_meta_chunk_size_bytes; + dml_uint_t writeback_chunk_size_kbytes; + dml_uint_t line_buffer_size_bits; + dml_uint_t max_line_buffer_lines; + dml_uint_t writeback_interface_buffer_size_kbytes; + dml_uint_t max_num_dpp; + dml_uint_t max_num_otg; + dml_uint_t max_num_wb; + dml_uint_t max_dchub_pscl_bw_pix_per_clk; + dml_uint_t max_pscl_lb_bw_pix_per_clk; + dml_uint_t max_lb_vscl_bw_pix_per_clk; + dml_uint_t max_vscl_hscl_bw_pix_per_clk; + dml_float_t max_hscl_ratio; + dml_float_t max_vscl_ratio; + dml_uint_t max_hscl_taps; + dml_uint_t max_vscl_taps; + dml_uint_t num_dsc; + dml_uint_t maximum_dsc_bits_per_component; + dml_uint_t maximum_pixels_per_line_per_dsc_unit; + dml_bool_t dsc422_native_support; + dml_bool_t cursor_64bpp_support; + dml_float_t dispclk_ramp_margin_percent; + dml_uint_t dppclk_delay_subtotal; + dml_uint_t dppclk_delay_scl; + dml_uint_t dppclk_delay_scl_lb_only; + dml_uint_t dppclk_delay_cnvc_formatter; + dml_uint_t dppclk_delay_cnvc_cursor; + dml_uint_t cursor_buffer_size; + dml_uint_t cursor_chunk_size; + dml_uint_t dispclk_delay_subtotal; + dml_bool_t dynamic_metadata_vm_enabled; + dml_uint_t max_inter_dcn_tile_repeaters; + dml_uint_t max_num_hdmi_frl_outputs; + dml_uint_t max_num_dp2p0_outputs; + dml_uint_t max_num_dp2p0_streams; + dml_bool_t dcc_supported; + dml_bool_t ptoi_supported; + dml_float_t writeback_max_hscl_ratio; + dml_float_t writeback_max_vscl_ratio; + dml_float_t writeback_min_hscl_ratio; + dml_float_t writeback_min_vscl_ratio; + dml_uint_t writeback_max_hscl_taps; + dml_uint_t writeback_max_vscl_taps; + dml_uint_t writeback_line_buffer_buffer_size; +}; + +struct DmlPipe { + dml_float_t Dppclk; + dml_float_t Dispclk; + dml_float_t PixelClock; + dml_float_t DCFClkDeepSleep; + dml_uint_t DPPPerSurface; + dml_bool_t ScalerEnabled; + enum dml_rotation_angle SourceScan; + dml_uint_t ViewportHeight; + dml_uint_t ViewportHeightChroma; + dml_uint_t BlockWidth256BytesY; + dml_uint_t BlockHeight256BytesY; + dml_uint_t BlockWidth256BytesC; + dml_uint_t BlockHeight256BytesC; + dml_uint_t BlockWidthY; + dml_uint_t BlockHeightY; + dml_uint_t BlockWidthC; + dml_uint_t BlockHeightC; + dml_uint_t InterlaceEnable; + dml_uint_t NumberOfCursors; + dml_uint_t VBlank; + dml_uint_t HTotal; + dml_uint_t HActive; + dml_bool_t DCCEnable; + enum dml_odm_mode ODMMode; + enum dml_source_format_class SourcePixelFormat; + enum dml_swizzle_mode SurfaceTiling; + dml_uint_t BytePerPixelY; + dml_uint_t BytePerPixelC; + dml_bool_t ProgressiveToInterlaceUnitInOPP; + dml_float_t VRatio; + dml_float_t VRatioChroma; + dml_uint_t VTaps; + dml_uint_t VTapsChroma; + dml_uint_t PitchY; + dml_uint_t DCCMetaPitchY; + dml_uint_t PitchC; + dml_uint_t DCCMetaPitchC; + dml_bool_t ViewportStationary; + dml_uint_t ViewportXStart; + dml_uint_t ViewportYStart; + dml_uint_t ViewportXStartC; + dml_uint_t ViewportYStartC; + dml_bool_t FORCE_ONE_ROW_FOR_FRAME; + dml_uint_t SwathHeightY; + dml_uint_t SwathHeightC; +}; + +struct Watermarks { + dml_float_t UrgentWatermark; + dml_float_t WritebackUrgentWatermark; + dml_float_t DRAMClockChangeWatermark; + dml_float_t FCLKChangeWatermark; + dml_float_t WritebackDRAMClockChangeWatermark; + dml_float_t WritebackFCLKChangeWatermark; + dml_float_t StutterExitWatermark; + dml_float_t StutterEnterPlusExitWatermark; + dml_float_t Z8StutterExitWatermark; + dml_float_t Z8StutterEnterPlusExitWatermark; + dml_float_t USRRetrainingWatermark; +}; + +struct SOCParametersList { + dml_float_t UrgentLatency; + dml_float_t ExtraLatency; + dml_float_t WritebackLatency; + dml_float_t DRAMClockChangeLatency; + dml_float_t FCLKChangeLatency; + dml_float_t SRExitTime; + dml_float_t SREnterPlusExitTime; + dml_float_t SRExitZ8Time; + dml_float_t SREnterPlusExitZ8Time; + dml_float_t USRRetrainingLatency; + dml_float_t SMNLatency; +}; + +/// @brief Struct that represent Plane configration of a display cfg +struct dml_plane_cfg_st { + // + // Pipe/Surface Parameters + // + dml_bool_t GPUVMEnable; /// > 23) & 255) - 128; + + x &= ~(255 << 23); + x += 127 << 23; + *exp_ptr = x; + + in = ((-1.0f / 3) * in + 2) * in - 2.0f / 3; + + return (in + log_2); +} + +dml_bool_t dml_util_is_420(enum dml_source_format_class source_format) +{ + dml_bool_t val = false; + + switch (source_format) { + case dml_444_16: + val = 0; + break; + case dml_444_32: + val = 0; + break; + case dml_444_64: + val = 0; + break; + case dml_420_8: + val = 1; + break; + case dml_420_10: + val = 1; + break; + case dml_422_8: + val = 0; + break; + case dml_422_10: + val = 0; + break; + default: + ASSERT(0); + break; + } + return val; +} + +static inline float dcn_bw_pow(float a, float exp) +{ + float temp; + /*ASSERT(exp == (int)exp);*/ + if ((int)exp == 0) + return 1; + temp = dcn_bw_pow(a, (int)(exp / 2)); + if (((int)exp % 2) == 0) { + return temp * temp; + } else { + if ((int)exp > 0) + return a * temp * temp; + else + return (temp * temp) / a; + } +} + +static inline float dcn_bw_ceil2(const float arg, const float significance) +{ + ASSERT(significance != 0); + + return ((int)(arg / significance + 0.99999)) * significance; +} + +static inline float dcn_bw_floor2(const float arg, const float significance) +{ + ASSERT(significance != 0); + + return ((int)(arg / significance)) * significance; +} + +dml_float_t dml_ceil(dml_float_t x, dml_float_t granularity) +{ + if (granularity == 0) + return 0; + //return (dml_float_t) (ceil(x / granularity) * granularity); + return (dml_float_t)dcn_bw_ceil2(x, granularity); +} + +dml_float_t dml_floor(dml_float_t x, dml_float_t granularity) +{ + if (granularity == 0) + return 0; + //return (dml_float_t) (floor(x / granularity) * granularity); + return (dml_float_t)dcn_bw_floor2(x, granularity); +} + +dml_float_t dml_min(dml_float_t x, dml_float_t y) +{ + if (x != x) + return y; + if (y != y) + return x; + if (x < y) + return x; + else + return y; +} + +dml_float_t dml_min3(dml_float_t x, dml_float_t y, dml_float_t z) +{ + return dml_min(dml_min(x, y), z); +} + +dml_float_t dml_min4(dml_float_t x, dml_float_t y, dml_float_t z, dml_float_t w) +{ + return dml_min(dml_min(dml_min(x, y), z), w); +} + +dml_float_t dml_max(dml_float_t x, dml_float_t y) +{ + if (x != x) + return y; + if (y != y) + return x; +if (x > y) + return x; + else + return y; +} +dml_float_t dml_max3(dml_float_t x, dml_float_t y, dml_float_t z) +{ + return dml_max(dml_max(x, y), z); +} +dml_float_t dml_max4(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d) +{ + return dml_max(dml_max(a, b), dml_max(c, d)); +} +dml_float_t dml_max5(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d, dml_float_t e) +{ + return dml_max(dml_max4(a, b, c, d), e); +} +dml_float_t dml_log(dml_float_t x, dml_float_t base) +{ + return (dml_float_t) (_log(x) / _log(base)); +} + +dml_float_t dml_log2(dml_float_t x) +{ + return (dml_float_t) (_log(x) / _log(2)); +} + +dml_float_t dml_round(dml_float_t val, dml_bool_t bankers_rounding) +{ +// if (bankers_rounding) +// return (dml_float_t) lrint(val); +// else { +// return round(val); + double round_pt = 0.5; + double ceil = dml_ceil(val, 1); + double floor = dml_floor(val, 1); + + if (val - floor >= round_pt) + return ceil; + else + return floor; +// } +} + +dml_float_t dml_pow(dml_float_t base, int exp) +{ + return (dml_float_t) dcn_bw_pow(base, exp); +} + +dml_uint_t dml_round_to_multiple(dml_uint_t num, dml_uint_t multiple, dml_bool_t up) +{ + dml_uint_t remainder; + + if (multiple == 0) + return num; + + remainder = num % multiple; + if (remainder == 0) + return num; + + if (up) + return (num + multiple - remainder); + else + return (num - remainder); +} + +void dml_print_data_rq_regs_st(const dml_display_plane_rq_regs_st *rq_regs) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DISPLAY_PLANE_RQ_REGS_ST\n"); + dml_print("DML: chunk_size = 0x%x\n", rq_regs->chunk_size); + dml_print("DML: min_chunk_size = 0x%x\n", rq_regs->min_chunk_size); + dml_print("DML: meta_chunk_size = 0x%x\n", rq_regs->meta_chunk_size); + dml_print("DML: min_meta_chunk_size = 0x%x\n", rq_regs->min_meta_chunk_size); + dml_print("DML: dpte_group_size = 0x%x\n", rq_regs->dpte_group_size); + dml_print("DML: mpte_group_size = 0x%x\n", rq_regs->mpte_group_size); + dml_print("DML: swath_height = 0x%x\n", rq_regs->swath_height); + dml_print("DML: pte_row_height_linear = 0x%x\n", rq_regs->pte_row_height_linear); + dml_print("DML: ===================================== \n"); +} + +void dml_print_rq_regs_st(const dml_display_rq_regs_st *rq_regs) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DISPLAY_RQ_REGS_ST\n"); + dml_print("DML: \n"); + dml_print_data_rq_regs_st(&rq_regs->rq_regs_l); + dml_print("DML: \n"); + dml_print_data_rq_regs_st(&rq_regs->rq_regs_c); + dml_print("DML: drq_expansion_mode = 0x%x\n", rq_regs->drq_expansion_mode); + dml_print("DML: prq_expansion_mode = 0x%x\n", rq_regs->prq_expansion_mode); + dml_print("DML: mrq_expansion_mode = 0x%x\n", rq_regs->mrq_expansion_mode); + dml_print("DML: crq_expansion_mode = 0x%x\n", rq_regs->crq_expansion_mode); + dml_print("DML: plane1_base_address = 0x%x\n", rq_regs->plane1_base_address); + dml_print("DML: ===================================== \n"); +} + +void dml_print_dlg_regs_st(const dml_display_dlg_regs_st *dlg_regs) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DISPLAY_DLG_REGS_ST \n"); + dml_print("DML: refcyc_h_blank_end = 0x%x\n", dlg_regs->refcyc_h_blank_end); + dml_print("DML: dlg_vblank_end = 0x%x\n", dlg_regs->dlg_vblank_end); + dml_print("DML: min_dst_y_next_start = 0x%x\n", dlg_regs->min_dst_y_next_start); + dml_print("DML: refcyc_per_htotal = 0x%x\n", dlg_regs->refcyc_per_htotal); + dml_print("DML: refcyc_x_after_scaler = 0x%x\n", dlg_regs->refcyc_x_after_scaler); + dml_print("DML: dst_y_after_scaler = 0x%x\n", dlg_regs->dst_y_after_scaler); + dml_print("DML: dst_y_prefetch = 0x%x\n", dlg_regs->dst_y_prefetch); + dml_print("DML: dst_y_per_vm_vblank = 0x%x\n", dlg_regs->dst_y_per_vm_vblank); + dml_print("DML: dst_y_per_row_vblank = 0x%x\n", dlg_regs->dst_y_per_row_vblank); + dml_print("DML: dst_y_per_vm_flip = 0x%x\n", dlg_regs->dst_y_per_vm_flip); + dml_print("DML: dst_y_per_row_flip = 0x%x\n", dlg_regs->dst_y_per_row_flip); + dml_print("DML: ref_freq_to_pix_freq = 0x%x\n", dlg_regs->ref_freq_to_pix_freq); + dml_print("DML: vratio_prefetch = 0x%x\n", dlg_regs->vratio_prefetch); + dml_print("DML: vratio_prefetch_c = 0x%x\n", dlg_regs->vratio_prefetch_c); + dml_print("DML: refcyc_per_pte_group_vblank_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_vblank_l); + dml_print("DML: refcyc_per_pte_group_vblank_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_vblank_c); + dml_print("DML: refcyc_per_meta_chunk_vblank_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_vblank_l); + dml_print("DML: refcyc_per_meta_chunk_vblank_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_vblank_c); + dml_print("DML: refcyc_per_pte_group_flip_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_flip_l); + dml_print("DML: refcyc_per_pte_group_flip_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_flip_c); + dml_print("DML: refcyc_per_meta_chunk_flip_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_flip_l); + dml_print("DML: refcyc_per_meta_chunk_flip_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_flip_c); + dml_print("DML: dst_y_per_pte_row_nom_l = 0x%x\n", dlg_regs->dst_y_per_pte_row_nom_l); + dml_print("DML: dst_y_per_pte_row_nom_c = 0x%x\n", dlg_regs->dst_y_per_pte_row_nom_c); + dml_print("DML: refcyc_per_pte_group_nom_l = 0x%x\n", dlg_regs->refcyc_per_pte_group_nom_l); + dml_print("DML: refcyc_per_pte_group_nom_c = 0x%x\n", dlg_regs->refcyc_per_pte_group_nom_c); + dml_print("DML: dst_y_per_meta_row_nom_l = 0x%x\n", dlg_regs->dst_y_per_meta_row_nom_l); + dml_print("DML: dst_y_per_meta_row_nom_c = 0x%x\n", dlg_regs->dst_y_per_meta_row_nom_c); + dml_print("DML: refcyc_per_meta_chunk_nom_l = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_nom_l); + dml_print("DML: refcyc_per_meta_chunk_nom_c = 0x%x\n", dlg_regs->refcyc_per_meta_chunk_nom_c); + dml_print("DML: refcyc_per_line_delivery_pre_l = 0x%x\n", dlg_regs->refcyc_per_line_delivery_pre_l); + dml_print("DML: refcyc_per_line_delivery_pre_c = 0x%x\n", dlg_regs->refcyc_per_line_delivery_pre_c); + dml_print("DML: refcyc_per_line_delivery_l = 0x%x\n", dlg_regs->refcyc_per_line_delivery_l); + dml_print("DML: refcyc_per_line_delivery_c = 0x%x\n", dlg_regs->refcyc_per_line_delivery_c); + dml_print("DML: refcyc_per_vm_group_vblank = 0x%x\n", dlg_regs->refcyc_per_vm_group_vblank); + dml_print("DML: refcyc_per_vm_group_flip = 0x%x\n", dlg_regs->refcyc_per_vm_group_flip); + dml_print("DML: refcyc_per_vm_req_vblank = 0x%x\n", dlg_regs->refcyc_per_vm_req_vblank); + dml_print("DML: refcyc_per_vm_req_flip = 0x%x\n", dlg_regs->refcyc_per_vm_req_flip); + dml_print("DML: chunk_hdl_adjust_cur0 = 0x%x\n", dlg_regs->chunk_hdl_adjust_cur0); + dml_print("DML: dst_y_offset_cur1 = 0x%x\n", dlg_regs->dst_y_offset_cur1); + dml_print("DML: chunk_hdl_adjust_cur1 = 0x%x\n", dlg_regs->chunk_hdl_adjust_cur1); + dml_print("DML: vready_after_vcount0 = 0x%x\n", dlg_regs->vready_after_vcount0); + dml_print("DML: dst_y_delta_drq_limit = 0x%x\n", dlg_regs->dst_y_delta_drq_limit); + dml_print("DML: refcyc_per_vm_dmdata = 0x%x\n", dlg_regs->refcyc_per_vm_dmdata); + dml_print("DML: ===================================== \n"); +} + +void dml_print_ttu_regs_st(const dml_display_ttu_regs_st *ttu_regs) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DISPLAY_TTU_REGS_ST \n"); + dml_print("DML: qos_level_low_wm = 0x%x\n", ttu_regs->qos_level_low_wm); + dml_print("DML: qos_level_high_wm = 0x%x\n", ttu_regs->qos_level_high_wm); + dml_print("DML: min_ttu_vblank = 0x%x\n", ttu_regs->min_ttu_vblank); + dml_print("DML: qos_level_flip = 0x%x\n", ttu_regs->qos_level_flip); + dml_print("DML: refcyc_per_req_delivery_pre_l = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_l); + dml_print("DML: refcyc_per_req_delivery_l = 0x%x\n", ttu_regs->refcyc_per_req_delivery_l); + dml_print("DML: refcyc_per_req_delivery_pre_c = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_c); + dml_print("DML: refcyc_per_req_delivery_c = 0x%x\n", ttu_regs->refcyc_per_req_delivery_c); + dml_print("DML: refcyc_per_req_delivery_cur0 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_cur0); + dml_print("DML: refcyc_per_req_delivery_pre_cur0 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_cur0); + dml_print("DML: refcyc_per_req_delivery_cur1 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_cur1); + dml_print("DML: refcyc_per_req_delivery_pre_cur1 = 0x%x\n", ttu_regs->refcyc_per_req_delivery_pre_cur1); + dml_print("DML: qos_level_fixed_l = 0x%x\n", ttu_regs->qos_level_fixed_l); + dml_print("DML: qos_ramp_disable_l = 0x%x\n", ttu_regs->qos_ramp_disable_l); + dml_print("DML: qos_level_fixed_c = 0x%x\n", ttu_regs->qos_level_fixed_c); + dml_print("DML: qos_ramp_disable_c = 0x%x\n", ttu_regs->qos_ramp_disable_c); + dml_print("DML: qos_level_fixed_cur0 = 0x%x\n", ttu_regs->qos_level_fixed_cur0); + dml_print("DML: qos_ramp_disable_cur0 = 0x%x\n", ttu_regs->qos_ramp_disable_cur0); + dml_print("DML: qos_level_fixed_cur1 = 0x%x\n", ttu_regs->qos_level_fixed_cur1); + dml_print("DML: qos_ramp_disable_cur1 = 0x%x\n", ttu_regs->qos_ramp_disable_cur1); + dml_print("DML: ===================================== \n"); +} + +void dml_print_dml_policy(const struct dml_mode_eval_policy_st *policy) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DML_MODE_EVAL_POLICY_ST\n"); + dml_print("DML: Policy: UseUnboundedRequesting = 0x%x\n", policy->UseUnboundedRequesting); + dml_print("DML: Policy: UseMinimumRequiredDCFCLK = 0x%x\n", policy->UseMinimumRequiredDCFCLK); + dml_print("DML: Policy: DRAMClockChangeRequirementFinal = 0x%x\n", policy->DRAMClockChangeRequirementFinal); + dml_print("DML: Policy: FCLKChangeRequirementFinal = 0x%x\n", policy->FCLKChangeRequirementFinal); + dml_print("DML: Policy: USRRetrainingRequiredFinal = 0x%x\n", policy->USRRetrainingRequiredFinal); + dml_print("DML: Policy: EnhancedPrefetchScheduleAccelerationFinal = 0x%x\n", policy->EnhancedPrefetchScheduleAccelerationFinal); + dml_print("DML: Policy: NomDETInKByteOverrideEnable = 0x%x\n", policy->NomDETInKByteOverrideEnable); + dml_print("DML: Policy: NomDETInKByteOverrideValue = 0x%x\n", policy->NomDETInKByteOverrideValue); + dml_print("DML: Policy: DCCProgrammingAssumesScanDirectionUnknownFinal = 0x%x\n", policy->DCCProgrammingAssumesScanDirectionUnknownFinal); + dml_print("DML: Policy: SynchronizeTimingsFinal = 0x%x\n", policy->SynchronizeTimingsFinal); + dml_print("DML: Policy: SynchronizeDRRDisplaysForUCLKPStateChangeFinal = 0x%x\n", policy->SynchronizeDRRDisplaysForUCLKPStateChangeFinal); + dml_print("DML: Policy: AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported = 0x%x\n", policy->AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported); + dml_print("DML: Policy: AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported = 0x%x\n", policy->AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported); + + for (dml_uint_t i = 0; i < DCN_DML__NUM_PLANE; i++) { + dml_print("DML: i=%0d, Policy: MPCCombineUse = 0x%x\n", i, policy->MPCCombineUse[i]); + dml_print("DML: i=%0d, Policy: ODMUse = 0x%x\n", i, policy->ODMUse[i]); + dml_print("DML: i=%0d, Policy: ImmediateFlipRequirement = 0x%x\n", i, policy->ImmediateFlipRequirement[i]); + dml_print("DML: i=%0d, Policy: AllowForPStateChangeOrStutterInVBlank = 0x%x\n", i, policy->AllowForPStateChangeOrStutterInVBlank[i]); + } + dml_print("DML: ===================================== \n"); +} + +void dml_print_mode_support(struct display_mode_lib_st *mode_lib, dml_uint_t j) +{ + dml_print("DML: MODE SUPPORT: ===============================================\n"); + dml_print("DML: MODE SUPPORT: Voltage State %d\n", j); + dml_print("DML: MODE SUPPORT: Mode Supported : %s\n", mode_lib->ms.support.ModeSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Scale Ratio And Taps : %s\n", mode_lib->ms.support.ScaleRatioAndTapsSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Source Format Pixel And Scan : %s\n", mode_lib->ms.support.SourceFormatPixelAndScanSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Viewport Size : %s\n", mode_lib->ms.support.ViewportSizeSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Link Rate Does Not Match DP Version : %s\n", mode_lib->ms.support.LinkRateDoesNotMatchDPVersion == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Link Rate For Multistream Not Indicated : %s\n", mode_lib->ms.support.LinkRateForMultistreamNotIndicated == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: BPP For Multi stream Not Indicated : %s\n", mode_lib->ms.support.BPPForMultistreamNotIndicated == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Multistream With HDMI Or eDP : %s\n", mode_lib->ms.support.MultistreamWithHDMIOreDP == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Exceeded Multistream Slots : %s\n", mode_lib->ms.support.ExceededMultistreamSlots == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: MSO Or ODM Split With Non DP Link : %s\n", mode_lib->ms.support.MSOOrODMSplitWithNonDPLink == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Not Enough Lanes For MSO : %s\n", mode_lib->ms.support.NotEnoughLanesForMSO == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: LinkCapacitySupport : %s\n", mode_lib->ms.support.LinkCapacitySupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: P2IWith420 : %s\n", mode_lib->ms.support.P2IWith420 == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DSCOnlyIfNecessaryWithBPP : %s\n", mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DSC422NativeNotSupported : %s\n", mode_lib->ms.support.DSC422NativeNotSupported == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: MPCCombineMethodIncompatible : %s\n", mode_lib->ms.support.MPCCombineMethodIncompatible == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ODMCombineTwoToOneSupportCheckOK : %s\n", mode_lib->ms.support.ODMCombineTwoToOneSupportCheckOK == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ODMCombineFourToOneSupportCheckOK : %s\n", mode_lib->ms.support.ODMCombineFourToOneSupportCheckOK == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: NotEnoughDSCUnits : %s\n", mode_lib->ms.support.NotEnoughDSCUnits == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: NotEnoughDSCSlices : %s\n", mode_lib->ms.support.NotEnoughDSCSlices == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe : %s\n", mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: InvalidCombinationOfMALLUseForPStateAndStaticScreen : %s\n", mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DSCCLKRequiredMoreThanSupported : %s\n", mode_lib->ms.support.DSCCLKRequiredMoreThanSupported == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: PixelsPerLinePerDSCUnitSupport : %s\n", mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DTBCLKRequiredMoreThanSupported : %s\n", mode_lib->ms.support.DTBCLKRequiredMoreThanSupported == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: InvalidCombinationOfMALLUseForPState : %s\n", mode_lib->ms.support.InvalidCombinationOfMALLUseForPState == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified : %s\n", mode_lib->ms.support.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ROB Support : %s\n", mode_lib->ms.support.ROBSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DISPCLK DPPCLK Support : %s\n", mode_lib->ms.support.DISPCLK_DPPCLK_Support[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Total Available Pipes Support : %s\n", mode_lib->ms.support.TotalAvailablePipesSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Number Of OTG Support : %s\n", mode_lib->ms.support.NumberOfOTGSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Number Of DP2p0 Support : %s\n", mode_lib->ms.support.NumberOfDP2p0Support == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Writeback Latency Support : %s\n", mode_lib->ms.support.WritebackLatencySupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Writeback Scale Ratio And Taps Support : %s\n", mode_lib->ms.support.WritebackScaleRatioAndTapsSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Cursor Support : %s\n", mode_lib->ms.support.CursorSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Pitch Support : %s\n", mode_lib->ms.support.PitchSupport == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Viewport Exceeds Surface : %s\n", mode_lib->ms.support.ViewportExceedsSurface == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Prefetch Supported : %s\n", mode_lib->ms.support.PrefetchSupported[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: VActive Bandwith Support : %s\n", mode_lib->ms.support.VActiveBandwithSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Dynamic Metadata Supported : %s\n", mode_lib->ms.support.DynamicMetadataSupported[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Total Vertical Active Bandwidth Support : %s\n", mode_lib->ms.support.TotalVerticalActiveBandwidthSupport[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: VRatio In Prefetch Supported : %s\n", mode_lib->ms.support.VRatioInPrefetchSupported[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: PTE Buffer Size Not Exceeded : %s\n", mode_lib->ms.support.PTEBufferSizeNotExceeded[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: DCC Meta Buffer Size Not Exceeded : %s\n", mode_lib->ms.support.DCCMetaBufferSizeNotExceeded[j] == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Non supported DSC Input BPC : %s\n", mode_lib->ms.support.NonsupportedDSCInputBPC == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Exceeded MALL Size : %s\n", mode_lib->ms.support.ExceededMALLSize == false ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: Host VM or Immediate Flip Supported : %s\n", ((mode_lib->ms.cache_display_cfg.plane.HostVMEnable == false && !mode_lib->scratch.dml_core_mode_support_locals.ImmediateFlipRequiredFinal) || mode_lib->ms.support.ImmediateFlipSupportedForState[j]) ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: dram clock change support : %s\n", mode_lib->scratch.dml_core_mode_support_locals.dram_clock_change_support == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: f_clock change support : %s\n", mode_lib->scratch.dml_core_mode_support_locals.f_clock_change_support == true ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: USR Retraining Support : %s\n", (!mode_lib->ms.policy.USRRetrainingRequiredFinal || &mode_lib->ms.support.USRRetrainingSupport[j]) ? "Supported" : "NOT Supported"); + dml_print("DML: MODE SUPPORT: ===============================================\n"); +} + +void dml_print_dml_mode_support_info(const struct dml_mode_support_info_st *support, dml_bool_t fail_only) +{ + dml_print("DML: ===================================== \n"); + dml_print("DML: DML_MODE_SUPPORT_INFO_ST\n"); + if (!fail_only || support->ModeIsSupported == 0) + dml_print("DML: support: ModeIsSupported = 0x%x\n", support->ModeIsSupported); + if (!fail_only || support->ImmediateFlipSupport == 0) + dml_print("DML: support: ImmediateFlipSupport = 0x%x\n", support->ImmediateFlipSupport); + if (!fail_only || support->WritebackLatencySupport == 0) + dml_print("DML: support: WritebackLatencySupport = 0x%x\n", support->WritebackLatencySupport); + if (!fail_only || support->ScaleRatioAndTapsSupport == 0) + dml_print("DML: support: ScaleRatioAndTapsSupport = 0x%x\n", support->ScaleRatioAndTapsSupport); + if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) + dml_print("DML: support: SourceFormatPixelAndScanSupport = 0x%x\n", support->SourceFormatPixelAndScanSupport); + if (!fail_only || support->MPCCombineMethodIncompatible == 1) + dml_print("DML: support: MPCCombineMethodIncompatible = 0x%x\n", support->MPCCombineMethodIncompatible); + if (!fail_only || support->P2IWith420 == 1) + dml_print("DML: support: P2IWith420 = 0x%x\n", support->P2IWith420); + if (!fail_only || support->DSCOnlyIfNecessaryWithBPP == 1) + dml_print("DML: support: DSCOnlyIfNecessaryWithBPP = 0x%x\n", support->DSCOnlyIfNecessaryWithBPP); + if (!fail_only || support->DSC422NativeNotSupported == 1) + dml_print("DML: support: DSC422NativeNotSupported = 0x%x\n", support->DSC422NativeNotSupported); + if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) + dml_print("DML: support: LinkRateDoesNotMatchDPVersion = 0x%x\n", support->LinkRateDoesNotMatchDPVersion); + if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) + dml_print("DML: support: LinkRateForMultistreamNotIndicated = 0x%x\n", support->LinkRateForMultistreamNotIndicated); + if (!fail_only || support->BPPForMultistreamNotIndicated == 1) + dml_print("DML: support: BPPForMultistreamNotIndicated = 0x%x\n", support->BPPForMultistreamNotIndicated); + if (!fail_only || support->MultistreamWithHDMIOreDP == 1) + dml_print("DML: support: MultistreamWithHDMIOreDP = 0x%x\n", support->MultistreamWithHDMIOreDP); + if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) + dml_print("DML: support: MSOOrODMSplitWithNonDPLink = 0x%x\n", support->MSOOrODMSplitWithNonDPLink); + if (!fail_only || support->NotEnoughLanesForMSO == 1) + dml_print("DML: support: NotEnoughLanesForMSO = 0x%x\n", support->NotEnoughLanesForMSO); + if (!fail_only || support->NumberOfOTGSupport == 0) + dml_print("DML: support: NumberOfOTGSupport = 0x%x\n", support->NumberOfOTGSupport); + if (!fail_only || support->NumberOfDP2p0Support == 0) + dml_print("DML: support: NumberOfDP2p0Support = 0x%x\n", support->NumberOfDP2p0Support); + if (!fail_only || support->NonsupportedDSCInputBPC == 1) + dml_print("DML: support: NonsupportedDSCInputBPC = 0x%x\n", support->NonsupportedDSCInputBPC); + if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) + dml_print("DML: support: WritebackScaleRatioAndTapsSupport = 0x%x\n", support->WritebackScaleRatioAndTapsSupport); + if (!fail_only || support->CursorSupport == 0) + dml_print("DML: support: CursorSupport = 0x%x\n", support->CursorSupport); + if (!fail_only || support->PitchSupport == 0) + dml_print("DML: support: PitchSupport = 0x%x\n", support->PitchSupport); + if (!fail_only || support->ViewportExceedsSurface == 1) + dml_print("DML: support: ViewportExceedsSurface = 0x%x\n", support->ViewportExceedsSurface); + if (!fail_only || support->ExceededMALLSize == 1) + dml_print("DML: support: ExceededMALLSize = 0x%x\n", support->ExceededMALLSize); + if (!fail_only || support->EnoughWritebackUnits == 0) + dml_print("DML: support: EnoughWritebackUnits = 0x%x\n", support->EnoughWritebackUnits); + if (!fail_only || support->ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified == 1) + dml_print("DML: support: ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = 0x%x\n", support->ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified); + if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) + dml_print("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = 0x%x\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); + if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) + dml_print("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = 0x%x\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); + if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) + dml_print("DML: support: InvalidCombinationOfMALLUseForPState = 0x%x\n", support->InvalidCombinationOfMALLUseForPState); + + if (!fail_only || support->ExceededMultistreamSlots == 1) + dml_print("DML: support: ExceededMultistreamSlots = 0x%x\n", support->ExceededMultistreamSlots); + if (!fail_only || support->ODMCombineTwoToOneSupportCheckOK == 0) + dml_print("DML: support: ODMCombineTwoToOneSupportCheckOK = 0x%x\n", support->ODMCombineTwoToOneSupportCheckOK); + if (!fail_only || support->ODMCombineFourToOneSupportCheckOK == 0) + dml_print("DML: support: ODMCombineFourToOneSupportCheckOK = 0x%x\n", support->ODMCombineFourToOneSupportCheckOK); + if (!fail_only || support->NotEnoughDSCUnits == 1) + dml_print("DML: support: NotEnoughDSCUnits = 0x%x\n", support->NotEnoughDSCUnits); + if (!fail_only || support->NotEnoughDSCSlices == 1) + dml_print("DML: support: NotEnoughDSCSlices = 0x%x\n", support->NotEnoughDSCSlices); + if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) + dml_print("DML: support: PixelsPerLinePerDSCUnitSupport = 0x%x\n", support->PixelsPerLinePerDSCUnitSupport); + if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) + dml_print("DML: support: DSCCLKRequiredMoreThanSupported = 0x%x\n", support->DSCCLKRequiredMoreThanSupported); + if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) + dml_print("DML: support: DTBCLKRequiredMoreThanSupported = 0x%x\n", support->DTBCLKRequiredMoreThanSupported); + if (!fail_only || support->LinkCapacitySupport == 0) + dml_print("DML: support: LinkCapacitySupport = 0x%x\n", support->LinkCapacitySupport); + + for (dml_uint_t j = 0; j < 2; j++) { + if (!fail_only || support->DRAMClockChangeSupport[j] == dml_dram_clock_change_unsupported) + dml_print("DML: support: combine=%d, DRAMClockChangeSupport = %d\n", j, support->DRAMClockChangeSupport[j]); + if (!fail_only || support->FCLKChangeSupport[j] == dml_fclock_change_unsupported) + dml_print("DML: support: combine=%d, FCLKChangeSupport = %d\n", j, support->FCLKChangeSupport[j]); + if (!fail_only || support->ROBSupport[j] == 0) + dml_print("DML: support: combine=%d, ROBSupport = %d\n", j, support->ROBSupport[j]); + if (!fail_only || support->PTEBufferSizeNotExceeded[j] == 0) + dml_print("DML: support: combine=%d, PTEBufferSizeNotExceeded = %d\n", j, support->PTEBufferSizeNotExceeded[j]); + if (!fail_only || support->DCCMetaBufferSizeNotExceeded[j] == 0) + dml_print("DML: support: combine=%d, DCCMetaBufferSizeNotExceeded = %d\n", j, support->DCCMetaBufferSizeNotExceeded[j]); + if (!fail_only || support->TotalVerticalActiveBandwidthSupport[j] == 0) + dml_print("DML: support: combine=%d, TotalVerticalActiveBandwidthSupport = %d\n", j, support->TotalVerticalActiveBandwidthSupport[j]); + if (!fail_only || support->USRRetrainingSupport[j] == 0) + dml_print("DML: support: combine=%d, USRRetrainingSupport = %d\n", j, support->USRRetrainingSupport[j]); + if (!fail_only || support->VActiveBandwithSupport[j] == 0) + dml_print("DML: support: combine=%d, VActiveBandwithSupport = %d\n", j, support->VActiveBandwithSupport[j]); + if (!fail_only || support->PrefetchSupported[j] == 0) + dml_print("DML: support: combine=%d, PrefetchSupported = %d\n", j, support->PrefetchSupported[j]); + if (!fail_only || support->DynamicMetadataSupported[j] == 0) + dml_print("DML: support: combine=%d, DynamicMetadataSupported = %d\n", j, support->DynamicMetadataSupported[j]); + if (!fail_only || support->VRatioInPrefetchSupported[j] == 0) + dml_print("DML: support: combine=%d, VRatioInPrefetchSupported = %d\n", j, support->VRatioInPrefetchSupported[j]); + if (!fail_only || support->DISPCLK_DPPCLK_Support[j] == 0) + dml_print("DML: support: combine=%d, DISPCLK_DPPCLK_Support = %d\n", j, support->DISPCLK_DPPCLK_Support[j]); + if (!fail_only || support->TotalAvailablePipesSupport[j] == 0) + dml_print("DML: support: combine=%d, TotalAvailablePipesSupport = %d\n", j, support->TotalAvailablePipesSupport[j]); + if (!fail_only || support->ModeSupport[j] == 0) + dml_print("DML: support: combine=%d, ModeSupport = %d\n", j, support->ModeSupport[j]); + if (!fail_only || support->ViewportSizeSupport[j] == 0) + dml_print("DML: support: combine=%d, ViewportSizeSupport = %d\n", j, support->ViewportSizeSupport[j]); + if (!fail_only || support->ImmediateFlipSupportedForState[j] == 0) + dml_print("DML: support: combine=%d, ImmediateFlipSupportedForState = %d\n", j, support->ImmediateFlipSupportedForState[j]); + } +} + +void dml_print_dml_display_cfg_timing(const struct dml_timing_cfg_st *timing, dml_uint_t num_plane) +{ + for (dml_uint_t i = 0; i < num_plane; i++) { + dml_print("DML: timing_cfg: plane=%d, HTotal = %d\n", i, timing->HTotal[i]); + dml_print("DML: timing_cfg: plane=%d, VTotal = %d\n", i, timing->VTotal[i]); + dml_print("DML: timing_cfg: plane=%d, HActive = %d\n", i, timing->HActive[i]); + dml_print("DML: timing_cfg: plane=%d, VActive = %d\n", i, timing->VActive[i]); + dml_print("DML: timing_cfg: plane=%d, VFrontPorch = %d\n", i, timing->VFrontPorch[i]); + dml_print("DML: timing_cfg: plane=%d, VBlankNom = %d\n", i, timing->VBlankNom[i]); + dml_print("DML: timing_cfg: plane=%d, RefreshRate = %d\n", i, timing->RefreshRate[i]); + dml_print("DML: timing_cfg: plane=%d, PixelClock = %f\n", i, timing->PixelClock[i]); + dml_print("DML: timing_cfg: plane=%d, Interlace = %d\n", i, timing->Interlace[i]); + dml_print("DML: timing_cfg: plane=%d, DRRDisplay = %d\n", i, timing->DRRDisplay[i]); + } +} + +void dml_print_dml_display_cfg_plane(const struct dml_plane_cfg_st *plane, dml_uint_t num_plane) +{ + dml_print("DML: plane_cfg: num_plane = %d\n", num_plane); + dml_print("DML: plane_cfg: GPUVMEnable = %d\n", plane->GPUVMEnable); + dml_print("DML: plane_cfg: HostVMEnable = %d\n", plane->HostVMEnable); + dml_print("DML: plane_cfg: GPUVMMaxPageTableLevels = %d\n", plane->GPUVMMaxPageTableLevels); + dml_print("DML: plane_cfg: HostVMMaxPageTableLevels = %d\n", plane->HostVMMaxPageTableLevels); + + for (dml_uint_t i = 0; i < num_plane; i++) { + dml_print("DML: plane_cfg: plane=%d, GPUVMMinPageSizeKBytes = %d\n", i, plane->GPUVMMinPageSizeKBytes[i]); + dml_print("DML: plane_cfg: plane=%d, ForceOneRowForFrame = %d\n", i, plane->ForceOneRowForFrame[i]); + dml_print("DML: plane_cfg: plane=%d, PTEBufferModeOverrideEn = %d\n", i, plane->PTEBufferModeOverrideEn[i]); + dml_print("DML: plane_cfg: plane=%d, PTEBufferMode = %d\n", i, plane->PTEBufferMode[i]); + dml_print("DML: plane_cfg: plane=%d, DETSizeOverride = %d\n", i, plane->DETSizeOverride[i]); + dml_print("DML: plane_cfg: plane=%d, UseMALLForStaticScreen = %d\n", i, plane->UseMALLForStaticScreen[i]); + dml_print("DML: plane_cfg: plane=%d, UseMALLForPStateChange = %d\n", i, plane->UseMALLForPStateChange[i]); + dml_print("DML: plane_cfg: plane=%d, BlendingAndTiming = %d\n", i, plane->BlendingAndTiming[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportWidth = %d\n", i, plane->ViewportWidth[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportHeight = %d\n", i, plane->ViewportHeight[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportWidthChroma = %d\n", i, plane->ViewportWidthChroma[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportHeightChroma = %d\n", i, plane->ViewportHeightChroma[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportXStart = %d\n", i, plane->ViewportXStart[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportXStartC = %d\n", i, plane->ViewportXStartC[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportYStart = %d\n", i, plane->ViewportYStart[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportYStartC = %d\n", i, plane->ViewportYStartC[i]); + dml_print("DML: plane_cfg: plane=%d, ViewportStationary = %d\n", i, plane->ViewportStationary[i]); + dml_print("DML: plane_cfg: plane=%d, ScalerEnabled = %d\n", i, plane->ScalerEnabled[i]); + dml_print("DML: plane_cfg: plane=%d, HRatio = %3.2f\n", i, plane->HRatio[i]); + dml_print("DML: plane_cfg: plane=%d, VRatio = %3.2f\n", i, plane->VRatio[i]); + dml_print("DML: plane_cfg: plane=%d, HRatioChroma = %3.2f\n", i, plane->HRatioChroma[i]); + dml_print("DML: plane_cfg: plane=%d, VRatioChroma = %3.2f\n", i, plane->VRatioChroma[i]); + dml_print("DML: plane_cfg: plane=%d, HTaps = %d\n", i, plane->HTaps[i]); + dml_print("DML: plane_cfg: plane=%d, VTaps = %d\n", i, plane->VTaps[i]); + dml_print("DML: plane_cfg: plane=%d, HTapsChroma = %d\n", i, plane->HTapsChroma[i]); + dml_print("DML: plane_cfg: plane=%d, VTapsChroma = %d\n", i, plane->VTapsChroma[i]); + dml_print("DML: plane_cfg: plane=%d, LBBitPerPixel = %d\n", i, plane->LBBitPerPixel[i]); + dml_print("DML: plane_cfg: plane=%d, SourceScan = %d\n", i, plane->SourceScan[i]); + dml_print("DML: plane_cfg: plane=%d, ScalerRecoutWidth = %d\n", i, plane->ScalerRecoutWidth[i]); + dml_print("DML: plane_cfg: plane=%d, NumberOfCursors = %d\n", i, plane->NumberOfCursors[i]); + dml_print("DML: plane_cfg: plane=%d, CursorWidth = %d\n", i, plane->CursorWidth[i]); + dml_print("DML: plane_cfg: plane=%d, CursorBPP = %d\n", i, plane->CursorBPP[i]); + + dml_print("DML: plane_cfg: plane=%d, DynamicMetadataEnable = %d\n", i, plane->DynamicMetadataEnable[i]); + dml_print("DML: plane_cfg: plane=%d, DynamicMetadataLinesBeforeActiveRequired = %d\n", i, plane->DynamicMetadataLinesBeforeActiveRequired[i]); + dml_print("DML: plane_cfg: plane=%d, DynamicMetadataTransmittedBytes = %d\n", i, plane->DynamicMetadataTransmittedBytes[i]); + } +} + +void dml_print_dml_display_cfg_surface(const struct dml_surface_cfg_st *surface, dml_uint_t num_plane) +{ + for (dml_uint_t i = 0; i < num_plane; i++) { + dml_print("DML: surface_cfg: plane=%d, PitchY = %d\n", i, surface->PitchY[i]); + dml_print("DML: surface_cfg: plane=%d, SurfaceWidthY = %d\n", i, surface->SurfaceWidthY[i]); + dml_print("DML: surface_cfg: plane=%d, SurfaceHeightY = %d\n", i, surface->SurfaceHeightY[i]); + dml_print("DML: surface_cfg: plane=%d, PitchC = %d\n", i, surface->PitchC[i]); + dml_print("DML: surface_cfg: plane=%d, SurfaceWidthC = %d\n", i, surface->SurfaceWidthC[i]); + dml_print("DML: surface_cfg: plane=%d, SurfaceHeightC = %d\n", i, surface->SurfaceHeightC[i]); + dml_print("DML: surface_cfg: plane=%d, DCCEnable = %d\n", i, surface->DCCEnable[i]); + dml_print("DML: surface_cfg: plane=%d, DCCMetaPitchY = %d\n", i, surface->DCCMetaPitchY[i]); + dml_print("DML: surface_cfg: plane=%d, DCCMetaPitchC = %d\n", i, surface->DCCMetaPitchC[i]); + dml_print("DML: surface_cfg: plane=%d, DCCRateLuma = %f\n", i, surface->DCCRateLuma[i]); + dml_print("DML: surface_cfg: plane=%d, DCCRateChroma = %f\n", i, surface->DCCRateChroma[i]); + dml_print("DML: surface_cfg: plane=%d, DCCFractionOfZeroSizeRequestsLuma = %f\n", i, surface->DCCFractionOfZeroSizeRequestsLuma[i]); + dml_print("DML: surface_cfg: plane=%d, DCCFractionOfZeroSizeRequestsChroma= %f\n", i, surface->DCCFractionOfZeroSizeRequestsChroma[i]); + } +} + +void dml_print_dml_display_cfg_hw_resource(const struct dml_hw_resource_st *hw, dml_uint_t num_plane) +{ + for (dml_uint_t i = 0; i < num_plane; i++) { + dml_print("DML: hw_resource: plane=%d, ODMMode = %d\n", i, hw->ODMMode[i]); + dml_print("DML: hw_resource: plane=%d, DPPPerSurface = %d\n", i, hw->DPPPerSurface[i]); + dml_print("DML: hw_resource: plane=%d, DSCEnabled = %d\n", i, hw->DSCEnabled[i]); + dml_print("DML: hw_resource: plane=%d, NumberOfDSCSlices = %d\n", i, hw->NumberOfDSCSlices[i]); + } + dml_print("DML: hw_resource: DLGRefClkFreqMHz = %f\n", hw->DLGRefClkFreqMHz); +} + +__DML_DLL_EXPORT__ void dml_print_soc_state_bounding_box(const struct soc_state_bounding_box_st *state) +{ + dml_print("DML: state_bbox: socclk_mhz = %f\n", state->socclk_mhz); + dml_print("DML: state_bbox: dscclk_mhz = %f\n", state->dscclk_mhz); + dml_print("DML: state_bbox: phyclk_mhz = %f\n", state->phyclk_mhz); + dml_print("DML: state_bbox: phyclk_d18_mhz = %f\n", state->phyclk_d18_mhz); + dml_print("DML: state_bbox: phyclk_d32_mhz = %f\n", state->phyclk_d32_mhz); + dml_print("DML: state_bbox: dtbclk_mhz = %f\n", state->dtbclk_mhz); + dml_print("DML: state_bbox: dispclk_mhz = %f\n", state->dispclk_mhz); + dml_print("DML: state_bbox: dppclk_mhz = %f\n", state->dppclk_mhz); + dml_print("DML: state_bbox: fabricclk_mhz = %f\n", state->fabricclk_mhz); + dml_print("DML: state_bbox: dcfclk_mhz = %f\n", state->dcfclk_mhz); + dml_print("DML: state_bbox: dram_speed_mts = %f\n", state->dram_speed_mts); + dml_print("DML: state_bbox: urgent_latency_pixel_data_only_us = %f\n", state->urgent_latency_pixel_data_only_us); + dml_print("DML: state_bbox: urgent_latency_pixel_mixed_with_vm_data_us = %f\n", state->urgent_latency_pixel_mixed_with_vm_data_us); + dml_print("DML: state_bbox: urgent_latency_vm_data_only_us = %f\n", state->urgent_latency_vm_data_only_us); + dml_print("DML: state_bbox: writeback_latency_us = %f\n", state->writeback_latency_us); + dml_print("DML: state_bbox: urgent_latency_adjustment_fabric_clock_component_us = %f\n", state->urgent_latency_adjustment_fabric_clock_component_us); + dml_print("DML: state_bbox: urgent_latency_adjustment_fabric_clock_reference_mhz= %f\n", state->urgent_latency_adjustment_fabric_clock_reference_mhz); + dml_print("DML: state_bbox: sr_exit_time_us = %f\n", state->sr_exit_time_us); + dml_print("DML: state_bbox: sr_enter_plus_exit_time_us = %f\n", state->sr_enter_plus_exit_time_us); + dml_print("DML: state_bbox: sr_exit_z8_time_us = %f\n", state->sr_exit_z8_time_us); + dml_print("DML: state_bbox: sr_enter_plus_exit_z8_time_us = %f\n", state->sr_enter_plus_exit_z8_time_us); + dml_print("DML: state_bbox: dram_clock_change_latency_us = %f\n", state->dram_clock_change_latency_us); + dml_print("DML: state_bbox: fclk_change_latency_us = %f\n", state->fclk_change_latency_us); + dml_print("DML: state_bbox: usr_retraining_latency_us = %f\n", state->usr_retraining_latency_us); + dml_print("DML: state_bbox: use_ideal_dram_bw_strobe = %d\n", state->use_ideal_dram_bw_strobe); +} + +__DML_DLL_EXPORT__ void dml_print_soc_bounding_box(const struct soc_bounding_box_st *soc) +{ + dml_print("DML: soc_bbox: dprefclk_mhz = %f\n", soc->dprefclk_mhz); + dml_print("DML: soc_bbox: xtalclk_mhz = %f\n", soc->xtalclk_mhz); + dml_print("DML: soc_bbox: pcierefclk_mhz = %f\n", soc->pcierefclk_mhz); + dml_print("DML: soc_bbox: refclk_mhz = %f\n", soc->refclk_mhz); + dml_print("DML: soc_bbox: amclk_mhz = %f\n", soc->amclk_mhz); + + dml_print("DML: soc_bbox: max_outstanding_reqs = %f\n", soc->max_outstanding_reqs); + dml_print("DML: soc_bbox: pct_ideal_sdp_bw_after_urgent = %f\n", soc->pct_ideal_sdp_bw_after_urgent); + dml_print("DML: soc_bbox: pct_ideal_fabric_bw_after_urgent = %f\n", soc->pct_ideal_fabric_bw_after_urgent); + dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_pixel_only = %f\n", soc->pct_ideal_dram_bw_after_urgent_pixel_only); + dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_pixel_and_vm = %f\n", soc->pct_ideal_dram_bw_after_urgent_pixel_and_vm); + dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_vm_only = %f\n", soc->pct_ideal_dram_bw_after_urgent_vm_only); + dml_print("DML: soc_bbox: pct_ideal_dram_bw_after_urgent_strobe = %f\n", soc->pct_ideal_dram_bw_after_urgent_strobe); + dml_print("DML: soc_bbox: max_avg_sdp_bw_use_normal_percent = %f\n", soc->max_avg_sdp_bw_use_normal_percent); + dml_print("DML: soc_bbox: max_avg_fabric_bw_use_normal_percent = %f\n", soc->max_avg_fabric_bw_use_normal_percent); + dml_print("DML: soc_bbox: max_avg_dram_bw_use_normal_percent = %f\n", soc->max_avg_dram_bw_use_normal_percent); + dml_print("DML: soc_bbox: max_avg_dram_bw_use_normal_strobe_percent = %f\n", soc->max_avg_dram_bw_use_normal_strobe_percent); + dml_print("DML: soc_bbox: round_trip_ping_latency_dcfclk_cycles = %d\n", soc->round_trip_ping_latency_dcfclk_cycles); + dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_pixel_only_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_pixel_only_bytes); + dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes); + dml_print("DML: soc_bbox: urgent_out_of_order_return_per_channel_vm_only_bytes = %d\n", soc->urgent_out_of_order_return_per_channel_vm_only_bytes); + dml_print("DML: soc_bbox: num_chans = %d\n", soc->num_chans); + dml_print("DML: soc_bbox: return_bus_width_bytes = %d\n", soc->return_bus_width_bytes); + dml_print("DML: soc_bbox: dram_channel_width_bytes = %d\n", soc->dram_channel_width_bytes); + dml_print("DML: soc_bbox: fabric_datapath_to_dcn_data_return_bytes = %d\n", soc->fabric_datapath_to_dcn_data_return_bytes); + dml_print("DML: soc_bbox: hostvm_min_page_size_kbytes = %d\n", soc->hostvm_min_page_size_kbytes); + dml_print("DML: soc_bbox: gpuvm_min_page_size_kbytes = %d\n", soc->gpuvm_min_page_size_kbytes); + dml_print("DML: soc_bbox: phy_downspread_percent = %f\n", soc->phy_downspread_percent); + dml_print("DML: soc_bbox: dcn_downspread_percent = %f\n", soc->dcn_downspread_percent); + dml_print("DML: soc_bbox: smn_latency_us = %f\n", soc->smn_latency_us); + dml_print("DML: soc_bbox: mall_allocated_for_dcn_mbytes = %d\n", soc->mall_allocated_for_dcn_mbytes); + dml_print("DML: soc_bbox: dispclk_dppclk_vco_speed_mhz = %f\n", soc->dispclk_dppclk_vco_speed_mhz); + dml_print("DML: soc_bbox: do_urgent_latency_adjustment = %d\n", soc->do_urgent_latency_adjustment); +} + +__DML_DLL_EXPORT__ void dml_print_clk_cfg(const struct dml_clk_cfg_st *clk_cfg) +{ + dml_print("DML: clk_cfg: 0-use_required, 1-use pipe.clks_cfg, 2-use state bbox\n"); + dml_print("DML: clk_cfg: dcfclk_option = %d\n", clk_cfg->dcfclk_option); + dml_print("DML: clk_cfg: dispclk_option = %d\n", clk_cfg->dispclk_option); + + dml_print("DML: clk_cfg: dcfclk_mhz = %f\n", clk_cfg->dcfclk_mhz); + dml_print("DML: clk_cfg: dispclk_mhz = %f\n", clk_cfg->dispclk_mhz); + + for (dml_uint_t i = 0; i < DCN_DML__NUM_PLANE; i++) { + dml_print("DML: clk_cfg: i=%d, dppclk_option = %d\n", i, clk_cfg->dppclk_option[i]); + dml_print("DML: clk_cfg: i=%d, dppclk_mhz = %f\n", i, clk_cfg->dppclk_mhz[i]); + } +} + +dml_bool_t dml_is_vertical_rotation(enum dml_rotation_angle Scan) +{ + dml_bool_t is_vert = false; + if (Scan == dml_rotation_90 || Scan == dml_rotation_90m || Scan == dml_rotation_270 || Scan == dml_rotation_270m) { + is_vert = true; + } else { + is_vert = false; + } + return is_vert; +} // dml_is_vertical_rotation + +dml_uint_t dml_get_cursor_bit_per_pixel(enum dml_cursor_bpp ebpp) +{ + switch (ebpp) { + case dml_cur_2bit: + return 2; + case dml_cur_32bit: + return 32; + case dml_cur_64bit: + return 64; + default: + return 0; + } +} + +/// @brief Determine the physical pipe to logical plane mapping using the display_cfg +dml_uint_t dml_get_num_active_planes(const struct dml_display_cfg_st *display_cfg) +{ + dml_uint_t num_active_planes = 0; + + for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; k++) { + if (display_cfg->plane.ViewportWidth[k] > 0) + num_active_planes = num_active_planes + 1; + } +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: num_active_planes = %d\n", __func__, num_active_planes); +#endif + return num_active_planes; +} + +/// @brief Determine the physical pipe to logical plane mapping using the display_cfg +dml_uint_t dml_get_num_active_pipes(const struct dml_display_cfg_st *display_cfg) +{ + dml_uint_t num_active_pipes = 0; + + for (dml_uint_t j = 0; j < dml_get_num_active_planes(display_cfg); j++) { + num_active_pipes = num_active_pipes + display_cfg->hw.DPPPerSurface[j]; + } + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); +#endif + return num_active_pipes; +} + +dml_uint_t dml_get_plane_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) +{ + dml_uint_t plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; + return plane_idx; +} + +dml_uint_t dml_get_pipe_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t plane_idx) +{ + dml_uint_t pipe_idx = 0; + dml_bool_t pipe_found = 0; + + ASSERT(plane_idx < __DML_NUM_PLANES__); + + for (dml_uint_t i = 0; i < __DML_NUM_PLANES__; i++) { + if (plane_idx == mode_lib->mp.pipe_plane[i]) { + pipe_idx = i; + pipe_found = 1; + break; + } + } + ASSERT(pipe_found != 0); + + return pipe_idx; +} + +void dml_calc_pipe_plane_mapping(const struct dml_hw_resource_st *hw, dml_uint_t *pipe_plane) +{ + dml_uint_t pipe_idx = 0; + + for (dml_uint_t k = 0; k < __DML_NUM_PLANES__; ++k) { + pipe_plane[k] = __DML_PIPE_NO_PLANE__; + } + + for (dml_uint_t plane_idx = 0; plane_idx < __DML_NUM_PLANES__; plane_idx++) { + for (dml_uint_t i = 0; i < hw->DPPPerSurface[plane_idx]; i++) { + pipe_plane[pipe_idx] = plane_idx; + pipe_idx++; + } + } +} + + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h new file mode 100644 index 000000000000..a82b49cf7fb0 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h @@ -0,0 +1,74 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DISPLAY_MODE_UTIL_H__ +#define __DISPLAY_MODE_UTIL_H__ + +#include "display_mode_core_structs.h" +#include "cmntypes.h" + +#include "dml_assert.h" +#include "dml_logging.h" + +__DML_DLL_EXPORT__ dml_bool_t dml_util_is_420(enum dml_source_format_class source_format); +__DML_DLL_EXPORT__ dml_float_t dml_ceil(dml_float_t x, dml_float_t granularity); +__DML_DLL_EXPORT__ dml_float_t dml_floor(dml_float_t x, dml_float_t granularity); +__DML_DLL_EXPORT__ dml_float_t dml_min(dml_float_t x, dml_float_t y); +__DML_DLL_EXPORT__ dml_float_t dml_min3(dml_float_t x, dml_float_t y, dml_float_t z); +__DML_DLL_EXPORT__ dml_float_t dml_min4(dml_float_t x, dml_float_t y, dml_float_t z, dml_float_t w); +__DML_DLL_EXPORT__ dml_float_t dml_max(dml_float_t x, dml_float_t y); +__DML_DLL_EXPORT__ dml_float_t dml_max3(dml_float_t x, dml_float_t y, dml_float_t z); +__DML_DLL_EXPORT__ dml_float_t dml_max4(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d); +__DML_DLL_EXPORT__ dml_float_t dml_max5(dml_float_t a, dml_float_t b, dml_float_t c, dml_float_t d, dml_float_t e); +__DML_DLL_EXPORT__ dml_float_t dml_log(dml_float_t x, dml_float_t base); +__DML_DLL_EXPORT__ dml_float_t dml_log2(dml_float_t x); +__DML_DLL_EXPORT__ dml_float_t dml_round(dml_float_t val, dml_bool_t bankers_rounding); +__DML_DLL_EXPORT__ dml_float_t dml_pow(dml_float_t base, int exp); +__DML_DLL_EXPORT__ dml_uint_t dml_round_to_multiple(dml_uint_t num, dml_uint_t multiple, dml_bool_t up); +__DML_DLL_EXPORT__ dml_bool_t dml_is_vertical_rotation(enum dml_rotation_angle scan); +__DML_DLL_EXPORT__ dml_uint_t dml_get_cursor_bit_per_pixel(enum dml_cursor_bpp ebpp); +__DML_DLL_EXPORT__ void dml_print_data_rq_regs_st(const dml_display_plane_rq_regs_st *data_rq_regs); +__DML_DLL_EXPORT__ void dml_print_rq_regs_st(const dml_display_rq_regs_st *rq_regs); +__DML_DLL_EXPORT__ void dml_print_dlg_regs_st(const dml_display_dlg_regs_st *dlg_regs); +__DML_DLL_EXPORT__ void dml_print_ttu_regs_st(const dml_display_ttu_regs_st *ttu_regs); +__DML_DLL_EXPORT__ void dml_print_dml_policy(const struct dml_mode_eval_policy_st *policy); +__DML_DLL_EXPORT__ void dml_print_mode_support(struct display_mode_lib_st *mode_lib, dml_uint_t j); +__DML_DLL_EXPORT__ void dml_print_dml_mode_support_info(const struct dml_mode_support_info_st *support, dml_bool_t fail_only); +__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_timing(const struct dml_timing_cfg_st *timing, dml_uint_t num_plane); +__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_plane(const struct dml_plane_cfg_st *plane, dml_uint_t num_plane); +__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_surface(const struct dml_surface_cfg_st *surface, dml_uint_t num_plane); +__DML_DLL_EXPORT__ void dml_print_dml_display_cfg_hw_resource(const struct dml_hw_resource_st *hw, dml_uint_t num_plane); +__DML_DLL_EXPORT__ void dml_print_soc_state_bounding_box(const struct soc_state_bounding_box_st *state); +__DML_DLL_EXPORT__ void dml_print_soc_bounding_box(const struct soc_bounding_box_st *soc); +__DML_DLL_EXPORT__ void dml_print_clk_cfg(const struct dml_clk_cfg_st *clk_cfg); + +__DML_DLL_EXPORT__ dml_uint_t dml_get_num_active_planes(const struct dml_display_cfg_st *display_cfg); +__DML_DLL_EXPORT__ dml_uint_t dml_get_num_active_pipes(const struct dml_display_cfg_st *display_cfg); +__DML_DLL_EXPORT__ dml_uint_t dml_get_plane_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx); +__DML_DLL_EXPORT__ dml_uint_t dml_get_pipe_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t plane_idx); +__DML_DLL_EXPORT__ void dml_calc_pipe_plane_mapping(const struct dml_hw_resource_st *hw, dml_uint_t *pipe_plane); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c new file mode 100644 index 000000000000..bf5e7f4e0416 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c @@ -0,0 +1,929 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml21_wrapper.h" +#include "dml2_core_dcn4_calcs.h" +#include "dml2_internal_shared_types.h" +#include "dml2_internal_types.h" +#include "dml21_utils.h" +#include "dml21_translation_helper.h" +#include "soc_and_ip_translator.h" + +static void dml21_populate_pmo_options(struct dml2_pmo_options *pmo_options, + const struct dc *in_dc, + const struct dml2_configuration_options *config) +{ + bool disable_fams2 = !in_dc->debug.fams2_config.bits.enable; + + /* ODM options */ + pmo_options->disable_dyn_odm = !config->minimize_dispclk_using_odm; + pmo_options->disable_dyn_odm_for_multi_stream = true; + pmo_options->disable_dyn_odm_for_stream_with_svp = true; + + pmo_options->disable_vblank = ((in_dc->debug.dml21_disable_pstate_method_mask >> 1) & 1); + + /* NOTE: DRR and SubVP Require FAMS2 */ + pmo_options->disable_svp = ((in_dc->debug.dml21_disable_pstate_method_mask >> 2) & 1) || + in_dc->debug.force_disable_subvp || + disable_fams2; + pmo_options->disable_drr_clamped = ((in_dc->debug.dml21_disable_pstate_method_mask >> 3) & 1) || + disable_fams2; + pmo_options->disable_drr_var = ((in_dc->debug.dml21_disable_pstate_method_mask >> 4) & 1) || + disable_fams2; + pmo_options->disable_fams2 = disable_fams2; + + pmo_options->disable_drr_var_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE || + in_dc->debug.disable_fams_gaming == INGAME_FAMS_MULTI_DISP_CLAMPED_ONLY; + pmo_options->disable_drr_clamped_when_var_active = in_dc->debug.disable_fams_gaming == INGAME_FAMS_DISABLE; +} + +static enum dml2_project_id dml21_dcn_revision_to_dml2_project_id(enum dce_version dcn_version) +{ + enum dml2_project_id project_id; + switch (dcn_version) { + case DCN_VERSION_4_01: + project_id = dml2_project_dcn4x_stage2_auto_drr_svp; + break; + default: + project_id = dml2_project_invalid; + DC_ERR("unsupported dcn version for DML21!"); + break; + } + + return project_id; +} + +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, + const struct dml2_configuration_options *config, + const struct dc *in_dc) +{ + dml_init->options.project_id = dml21_dcn_revision_to_dml2_project_id(in_dc->ctx->dce_version); + + if (config->use_native_soc_bb_construction) { + in_dc->soc_and_ip_translator->translator_funcs->get_soc_bb(&dml_init->soc_bb, in_dc, config); + in_dc->soc_and_ip_translator->translator_funcs->get_ip_caps(&dml_init->ip_caps); + } else { + dml_init->soc_bb = config->external_socbb_ip_params->soc_bb; + dml_init->ip_caps = config->external_socbb_ip_params->ip_params; + } + + dml21_populate_pmo_options(&dml_init->options.pmo_options, in_dc, config); +} + +static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream) +{ + unsigned int max_hw_v_total = stream->ctx->dc->caps.max_v_total; + + if (stream->ctx->dc->caps.vtotal_limited_by_fp2) { + max_hw_v_total -= stream->timing.v_front_porch + 1; + } + + return max_hw_v_total; +} + +static void populate_dml21_timing_config_from_stream_state(struct dml2_timing_cfg *timing, + struct dc_stream_state *stream, + struct pipe_ctx *pipe_ctx, + struct dml2_context *dml_ctx) +{ + unsigned int hblank_start, vblank_start, min_hardware_refresh_in_uhz; + uint32_t pix_clk_100hz; + + timing->h_active = stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right + pipe_ctx->dsc_padding_params.dsc_hactive_padding; + timing->v_active = stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top; + timing->h_front_porch = stream->timing.h_front_porch; + timing->v_front_porch = stream->timing.v_front_porch; + timing->pixel_clock_khz = stream->timing.pix_clk_100hz / 10; + if (pipe_ctx->dsc_padding_params.dsc_hactive_padding != 0) + timing->pixel_clock_khz = pipe_ctx->dsc_padding_params.dsc_pix_clk_100hz / 10; + if (stream->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING) + timing->pixel_clock_khz *= 2; + timing->h_total = stream->timing.h_total + pipe_ctx->dsc_padding_params.dsc_htotal_padding; + timing->v_total = stream->timing.v_total; + timing->h_sync_width = stream->timing.h_sync_width; + timing->interlaced = stream->timing.flags.INTERLACE; + + hblank_start = stream->timing.h_total - stream->timing.h_front_porch; + + timing->h_blank_end = hblank_start - stream->timing.h_addressable - pipe_ctx->dsc_padding_params.dsc_hactive_padding + - stream->timing.h_border_left - stream->timing.h_border_right; + + if (hblank_start < stream->timing.h_addressable) + timing->h_blank_end = 0; + + vblank_start = stream->timing.v_total - stream->timing.v_front_porch; + + timing->v_blank_end = vblank_start - stream->timing.v_addressable + - stream->timing.v_border_top - stream->timing.v_border_bottom; + + timing->drr_config.enabled = stream->ignore_msa_timing_param; + timing->drr_config.drr_active_variable = stream->vrr_active_variable; + timing->drr_config.drr_active_fixed = stream->vrr_active_fixed; + timing->drr_config.disallowed = !stream->allow_freesync; + + /* limit min refresh rate to DC cap */ + min_hardware_refresh_in_uhz = stream->timing.min_refresh_in_uhz; + if (stream->ctx->dc->caps.max_v_total != 0) { + if (pipe_ctx->dsc_padding_params.dsc_hactive_padding != 0) { + pix_clk_100hz = pipe_ctx->dsc_padding_params.dsc_pix_clk_100hz; + } else { + pix_clk_100hz = stream->timing.pix_clk_100hz; + } + min_hardware_refresh_in_uhz = div64_u64((pix_clk_100hz * 100000000ULL), + (timing->h_total * (long long)calc_max_hardware_v_total(stream))); + } + + timing->drr_config.min_refresh_uhz = max(stream->timing.min_refresh_in_uhz, min_hardware_refresh_in_uhz); + + if (dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase && + stream->ctx->dc->config.enable_fpo_flicker_detection == 1) + timing->drr_config.max_instant_vtotal_delta = dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase(stream, false); + else + timing->drr_config.max_instant_vtotal_delta = 0; + + if (stream->timing.flags.DSC) { + timing->dsc.enable = dml2_dsc_enable; + timing->dsc.overrides.num_slices = stream->timing.dsc_cfg.num_slices_h; + timing->dsc.dsc_compressed_bpp_x16 = stream->timing.dsc_cfg.bits_per_pixel; + } else + timing->dsc.enable = dml2_dsc_disable; + + switch (stream->timing.display_color_depth) { + case COLOR_DEPTH_666: + timing->bpc = 6; + break; + case COLOR_DEPTH_888: + timing->bpc = 8; + break; + case COLOR_DEPTH_101010: + timing->bpc = 10; + break; + case COLOR_DEPTH_121212: + timing->bpc = 12; + break; + case COLOR_DEPTH_141414: + timing->bpc = 14; + break; + case COLOR_DEPTH_161616: + timing->bpc = 16; + break; + case COLOR_DEPTH_999: + timing->bpc = 9; + break; + case COLOR_DEPTH_111111: + timing->bpc = 11; + break; + default: + timing->bpc = 8; + break; + } + + timing->vblank_nom = timing->v_total - timing->v_active; +} + +static void populate_dml21_output_config_from_stream_state(struct dml2_link_output_cfg *output, + struct dc_stream_state *stream, const struct pipe_ctx *pipe) +{ + output->output_dp_lane_count = 4; + + switch (stream->signal) { + case SIGNAL_TYPE_DISPLAY_PORT_MST: + case SIGNAL_TYPE_DISPLAY_PORT: + output->output_encoder = dml2_dp; + if (check_dp2p0_output_encoder(pipe)) + output->output_encoder = dml2_dp2p0; + break; + case SIGNAL_TYPE_EDP: + output->output_encoder = dml2_edp; + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + output->output_encoder = dml2_hdmi; + break; + default: + output->output_encoder = dml2_dp; + } + + switch (stream->timing.pixel_encoding) { + case PIXEL_ENCODING_RGB: + case PIXEL_ENCODING_YCBCR444: + output->output_format = dml2_444; + break; + case PIXEL_ENCODING_YCBCR420: + output->output_format = dml2_420; + break; + case PIXEL_ENCODING_YCBCR422: + if (stream->timing.flags.DSC && !stream->timing.dsc_cfg.ycbcr422_simple) + output->output_format = dml2_n422; + else + output->output_format = dml2_s422; + break; + default: + output->output_format = dml2_444; + break; + } + + switch (stream->signal) { + case SIGNAL_TYPE_NONE: + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + case SIGNAL_TYPE_HDMI_TYPE_A: + case SIGNAL_TYPE_LVDS: + case SIGNAL_TYPE_RGB: + case SIGNAL_TYPE_DISPLAY_PORT: + case SIGNAL_TYPE_DISPLAY_PORT_MST: + case SIGNAL_TYPE_EDP: + case SIGNAL_TYPE_VIRTUAL: + default: + output->output_dp_link_rate = dml2_dp_rate_na; + break; + } + + output->audio_sample_layout = stream->audio_info.modes->sample_size; + output->audio_sample_rate = stream->audio_info.modes->max_bit_rate; + output->output_disabled = true; + + //TODO : New to DML2.1. How do we populate this ? + // output->validate_output +} + +static void populate_dml21_stream_overrides_from_stream_state( + struct dml2_stream_parameters *stream_desc, + struct dc_stream_state *stream, + struct dc_stream_status *stream_status) +{ + switch (stream->debug.force_odm_combine_segments) { + case 0: + stream_desc->overrides.odm_mode = dml2_odm_mode_auto; + break; + case 1: + stream_desc->overrides.odm_mode = dml2_odm_mode_bypass; + break; + case 2: + stream_desc->overrides.odm_mode = dml2_odm_mode_combine_2to1; + break; + case 3: + stream_desc->overrides.odm_mode = dml2_odm_mode_combine_3to1; + break; + case 4: + stream_desc->overrides.odm_mode = dml2_odm_mode_combine_4to1; + break; + default: + stream_desc->overrides.odm_mode = dml2_odm_mode_auto; + break; + } + if (!stream->ctx->dc->debug.enable_single_display_2to1_odm_policy || + stream->debug.force_odm_combine_segments > 0) + stream_desc->overrides.disable_dynamic_odm = true; + stream_desc->overrides.disable_subvp = stream->ctx->dc->debug.force_disable_subvp || + stream->hw_cursor_req || + stream_status->mall_stream_config.cursor_size_limit_subvp; +} + +static enum dml2_swizzle_mode gfx_addr3_to_dml2_swizzle_mode(enum swizzle_mode_addr3_values addr3_mode) +{ + enum dml2_swizzle_mode dml2_mode = dml2_sw_linear; + + switch (addr3_mode) { + case DC_ADDR3_SW_LINEAR: + dml2_mode = dml2_sw_linear; + break; + case DC_ADDR3_SW_256B_2D: + dml2_mode = dml2_sw_256b_2d; + break; + case DC_ADDR3_SW_4KB_2D: + dml2_mode = dml2_sw_4kb_2d; + break; + case DC_ADDR3_SW_64KB_2D: + dml2_mode = dml2_sw_64kb_2d; + break; + case DC_ADDR3_SW_256KB_2D: + dml2_mode = dml2_sw_256kb_2d; + break; + default: + /* invalid swizzle mode for DML2.1 */ + ASSERT(false); + dml2_mode = dml2_sw_linear; + } + + return dml2_mode; +} + +static enum dml2_swizzle_mode gfx9_to_dml2_swizzle_mode(enum swizzle_mode_values gfx9_mode) +{ + enum dml2_swizzle_mode dml2_mode = dml2_sw_64kb_2d; + + switch (gfx9_mode) { + case DC_SW_LINEAR: + dml2_mode = dml2_sw_linear; + break; + case DC_SW_256_D: + case DC_SW_256_R: + dml2_mode = dml2_sw_256b_2d; + break; + case DC_SW_4KB_D: + case DC_SW_4KB_R: + case DC_SW_4KB_R_X: + dml2_mode = dml2_sw_4kb_2d; + break; + case DC_SW_64KB_D: + case DC_SW_64KB_D_X: + case DC_SW_64KB_R: + case DC_SW_64KB_R_X: + dml2_mode = dml2_sw_64kb_2d; + break; + case DC_SW_256B_S: + case DC_SW_4KB_S: + case DC_SW_64KB_S: + case DC_SW_VAR_S: + case DC_SW_VAR_D: + case DC_SW_VAR_R: + case DC_SW_64KB_S_T: + case DC_SW_64KB_D_T: + case DC_SW_4KB_S_X: + case DC_SW_4KB_D_X: + case DC_SW_64KB_S_X: + case DC_SW_VAR_S_X: + case DC_SW_VAR_D_X: + case DC_SW_VAR_R_X: + default: + /* + * invalid swizzle mode for DML2.1. This could happen because + * DML21 is not intended to be used by N-1 in production. To + * properly filter out unsupported swizzle modes, we will need + * to fix capability reporting when DML2.1 is used for N-1 in + * dc. So DML will only receive DML21 supported swizzle modes. + * This implementation is not added and has a low value because + * the supported swizzle modes should already cover most of our + * N-1 test cases. + */ + return dml2_sw_64kb_2d; + } + + return dml2_mode; +} + +static void populate_dml21_dummy_surface_cfg(struct dml2_surface_cfg *surface, const struct dc_stream_state *stream) +{ + surface->plane0.width = stream->timing.h_addressable; + surface->plane0.height = stream->timing.v_addressable; + surface->plane1.width = stream->timing.h_addressable; + surface->plane1.height = stream->timing.v_addressable; + surface->plane0.pitch = ((surface->plane0.width + 127) / 128) * 128; + surface->plane1.pitch = 0; + surface->dcc.enable = false; + surface->dcc.informative.dcc_rate_plane0 = 1.0; + surface->dcc.informative.dcc_rate_plane1 = 1.0; + surface->dcc.informative.fraction_of_zero_size_request_plane0 = 0; + surface->dcc.informative.fraction_of_zero_size_request_plane1 = 0; + surface->tiling = dml2_sw_64kb_2d; +} + +static void populate_dml21_dummy_plane_cfg(struct dml2_plane_parameters *plane, const struct dc_stream_state *stream) +{ + unsigned int width, height; + + if (stream->timing.h_addressable > 3840) + width = 3840; + else + width = stream->timing.h_addressable; // 4K max + + if (stream->timing.v_addressable > 2160) + height = 2160; + else + height = stream->timing.v_addressable; // 4K max + + plane->cursor.cursor_bpp = 32; + + plane->cursor.cursor_width = 256; + plane->cursor.num_cursors = 1; + + plane->composition.viewport.plane0.width = width; + plane->composition.viewport.plane0.height = height; + plane->composition.viewport.plane1.width = 0; + plane->composition.viewport.plane1.height = 0; + + plane->composition.viewport.stationary = false; + plane->composition.viewport.plane0.x_start = 0; + plane->composition.viewport.plane0.y_start = 0; + plane->composition.viewport.plane1.x_start = 0; + plane->composition.viewport.plane1.y_start = 0; + + plane->composition.scaler_info.enabled = false; + plane->composition.rotation_angle = dml2_rotation_0; + plane->composition.scaler_info.plane0.h_ratio = 1.0; + plane->composition.scaler_info.plane0.v_ratio = 1.0; + plane->composition.scaler_info.plane1.h_ratio = 0; + plane->composition.scaler_info.plane1.v_ratio = 0; + plane->composition.scaler_info.plane0.h_taps = 1; + plane->composition.scaler_info.plane0.v_taps = 1; + plane->composition.scaler_info.plane1.h_taps = 0; + plane->composition.scaler_info.plane1.v_taps = 0; + plane->composition.scaler_info.rect_out_width = width; + plane->pixel_format = dml2_444_32; + + plane->dynamic_meta_data.enable = false; + plane->overrides.gpuvm_min_page_size_kbytes = 256; +} + +static void populate_dml21_surface_config_from_plane_state( + const struct dc *in_dc, + struct dml2_surface_cfg *surface, + const struct dc_plane_state *plane_state) +{ + surface->plane0.pitch = plane_state->plane_size.surface_pitch; + surface->plane1.pitch = plane_state->plane_size.chroma_pitch; + surface->plane0.height = plane_state->plane_size.surface_size.height; + surface->plane0.width = plane_state->plane_size.surface_size.width; + surface->plane1.height = plane_state->plane_size.chroma_size.height; + surface->plane1.width = plane_state->plane_size.chroma_size.width; + surface->dcc.enable = plane_state->dcc.enable; + surface->dcc.informative.dcc_rate_plane0 = 1.0; + surface->dcc.informative.dcc_rate_plane1 = 1.0; + surface->dcc.informative.fraction_of_zero_size_request_plane0 = plane_state->dcc.independent_64b_blks; + surface->dcc.informative.fraction_of_zero_size_request_plane1 = plane_state->dcc.independent_64b_blks_c; + surface->dcc.plane0.pitch = plane_state->dcc.meta_pitch; + surface->dcc.plane1.pitch = plane_state->dcc.meta_pitch_c; + + // Update swizzle / array mode based on the gfx_format + switch (plane_state->tiling_info.gfxversion) { + case DcGfxVersion7: + case DcGfxVersion8: + break; + case DcGfxVersion9: + case DcGfxVersion10: + case DcGfxVersion11: + surface->tiling = gfx9_to_dml2_swizzle_mode(plane_state->tiling_info.gfx9.swizzle); + break; + case DcGfxAddr3: + surface->tiling = gfx_addr3_to_dml2_swizzle_mode(plane_state->tiling_info.gfx_addr3.swizzle); + break; + } +} + +static const struct scaler_data *get_scaler_data_for_plane( + struct dml2_context *dml_ctx, + const struct dc_plane_state *in, + const struct dc_state *context) +{ + int i; + struct pipe_ctx *temp_pipe = &dml_ctx->v21.scratch.temp_pipe; + + memset(temp_pipe, 0, sizeof(struct pipe_ctx)); + + for (i = 0; i < MAX_PIPES; i++) { + const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state == in && !pipe->prev_odm_pipe) { + temp_pipe->stream = pipe->stream; + temp_pipe->plane_state = pipe->plane_state; + temp_pipe->plane_res.scl_data.taps = pipe->plane_res.scl_data.taps; + temp_pipe->stream_res = pipe->stream_res; + temp_pipe->dsc_padding_params.dsc_hactive_padding = pipe->dsc_padding_params.dsc_hactive_padding; + temp_pipe->dsc_padding_params.dsc_htotal_padding = pipe->dsc_padding_params.dsc_htotal_padding; + temp_pipe->dsc_padding_params.dsc_pix_clk_100hz = pipe->dsc_padding_params.dsc_pix_clk_100hz; + dml_ctx->config.callbacks.build_scaling_params(temp_pipe); + break; + } + } + + ASSERT(i < MAX_PIPES); + return &temp_pipe->plane_res.scl_data; +} + +static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dml_ctx, + struct dml2_plane_parameters *plane, const struct dc_plane_state *plane_state, + const struct dc_state *context, unsigned int stream_index) +{ + const struct scaler_data *scaler_data = get_scaler_data_for_plane(dml_ctx, plane_state, context); + struct dc_stream_state *stream = context->streams[stream_index]; + + plane->cursor.cursor_bpp = 32; + plane->cursor.cursor_width = 256; + plane->cursor.num_cursors = 1; + + switch (plane_state->format) { + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr: + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: + plane->pixel_format = dml2_420_8; + break; + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: + plane->pixel_format = dml2_420_10; + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: + plane->pixel_format = dml2_444_64; + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555: + case SURFACE_PIXEL_FORMAT_GRPH_RGB565: + plane->pixel_format = dml2_444_16; + break; + case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS: + plane->pixel_format = dml2_444_8; + break; + case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA: + plane->pixel_format = dml2_rgbe_alpha; + break; + default: + plane->pixel_format = dml2_444_32; + break; + } + + plane->composition.viewport.plane0.height = scaler_data->viewport.height; + plane->composition.viewport.plane0.width = scaler_data->viewport.width; + plane->composition.viewport.plane1.height = scaler_data->viewport_c.height; + plane->composition.viewport.plane1.width = scaler_data->viewport_c.width; + plane->composition.viewport.plane0.x_start = scaler_data->viewport.x; + plane->composition.viewport.plane0.y_start = scaler_data->viewport.y; + plane->composition.viewport.plane1.x_start = scaler_data->viewport_c.x; + plane->composition.viewport.plane1.y_start = scaler_data->viewport_c.y; + plane->composition.viewport.stationary = false; + plane->composition.scaler_info.enabled = scaler_data->ratios.horz.value != dc_fixpt_one.value || + scaler_data->ratios.horz_c.value != dc_fixpt_one.value || + scaler_data->ratios.vert.value != dc_fixpt_one.value || + scaler_data->ratios.vert_c.value != dc_fixpt_one.value; + + if (!scaler_data->taps.h_taps) { + /* Above logic determines scaling should be enabled even when there are no taps for + * certain cases. Hence do corrective active and disable scaling. + */ + plane->composition.scaler_info.enabled = false; + } else if ((plane_state->ctx->dc->config.use_spl == true) && + (plane->composition.scaler_info.enabled == false)) { + /* To enable sharpener for 1:1, scaler must be enabled. If use_spl is set, then + * allow case where ratio is 1 but taps > 1 + */ + if ((scaler_data->taps.h_taps > 1) || (scaler_data->taps.v_taps > 1) || + (scaler_data->taps.h_taps_c > 1) || (scaler_data->taps.v_taps_c > 1)) + plane->composition.scaler_info.enabled = true; + } + + /* always_scale is only used for debug purposes not used in production but has to be + * maintained for certain complainces. */ + if (plane_state->ctx->dc->debug.always_scale == true) { + plane->composition.scaler_info.enabled = true; + } + + if (plane->composition.scaler_info.enabled == false) { + plane->composition.scaler_info.plane0.h_ratio = 1.0; + plane->composition.scaler_info.plane0.v_ratio = 1.0; + plane->composition.scaler_info.plane1.h_ratio = 1.0; + plane->composition.scaler_info.plane1.v_ratio = 1.0; + } else { + plane->composition.scaler_info.plane0.h_ratio = (double)scaler_data->ratios.horz.value / (1ULL << 32); + plane->composition.scaler_info.plane0.v_ratio = (double)scaler_data->ratios.vert.value / (1ULL << 32); + plane->composition.scaler_info.plane1.h_ratio = (double)scaler_data->ratios.horz_c.value / (1ULL << 32); + plane->composition.scaler_info.plane1.v_ratio = (double)scaler_data->ratios.vert_c.value / (1ULL << 32); + } + + if (!scaler_data->taps.h_taps) { + plane->composition.scaler_info.plane0.h_taps = 1; + plane->composition.scaler_info.plane1.h_taps = 1; + } else { + plane->composition.scaler_info.plane0.h_taps = scaler_data->taps.h_taps; + plane->composition.scaler_info.plane1.h_taps = scaler_data->taps.h_taps_c; + } + if (!scaler_data->taps.v_taps) { + plane->composition.scaler_info.plane0.v_taps = 1; + plane->composition.scaler_info.plane1.v_taps = 1; + } else { + plane->composition.scaler_info.plane0.v_taps = scaler_data->taps.v_taps; + plane->composition.scaler_info.plane1.v_taps = scaler_data->taps.v_taps_c; + } + + plane->composition.viewport.stationary = false; + + if (plane_state->mcm_luts.lut3d_data.lut3d_src == DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { + plane->tdlut.setup_for_tdlut = true; + + switch (plane_state->mcm_luts.lut3d_data.gpu_mem_params.layout) { + case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_RGB: + case DC_CM2_GPU_MEM_LAYOUT_3D_SWIZZLE_LINEAR_BGR: + plane->tdlut.tdlut_addressing_mode = dml2_tdlut_sw_linear; + break; + case DC_CM2_GPU_MEM_LAYOUT_1D_PACKED_LINEAR: + plane->tdlut.tdlut_addressing_mode = dml2_tdlut_simple_linear; + break; + } + + switch (plane_state->mcm_luts.lut3d_data.gpu_mem_params.size) { + case DC_CM2_GPU_MEM_SIZE_171717: + plane->tdlut.tdlut_width_mode = dml2_tdlut_width_17_cube; + break; + case DC_CM2_GPU_MEM_SIZE_TRANSFORMED: + default: + //plane->tdlut.tdlut_width_mode = dml2_tdlut_width_flatten; // dml2_tdlut_width_flatten undefined + break; + } + } + plane->tdlut.setup_for_tdlut |= dml_ctx->config.force_tdlut_enable; + + plane->dynamic_meta_data.enable = false; + plane->dynamic_meta_data.lines_before_active_required = 0; + plane->dynamic_meta_data.transmitted_bytes = 0; + + plane->composition.scaler_info.rect_out_width = plane_state->dst_rect.width; + plane->composition.rotation_angle = (enum dml2_rotation_angle) plane_state->rotation; + plane->stream_index = stream_index; + + plane->overrides.gpuvm_min_page_size_kbytes = 256; + + plane->immediate_flip = plane_state->flip_immediate; + + plane->composition.rect_out_height_spans_vactive = + plane_state->dst_rect.height >= stream->src.height && + stream->dst.height >= stream->timing.v_addressable; +} + +//TODO : Could be possibly moved to a common helper layer. +static bool dml21_wrapper_get_plane_id(const struct dc_state *context, unsigned int stream_id, const struct dc_plane_state *plane, unsigned int *plane_id) +{ + int i, j; + + if (!plane_id) + return false; + + for (i = 0; i < context->stream_count; i++) { + if (context->streams[i]->stream_id == stream_id) { + for (j = 0; j < context->stream_status[i].plane_count; j++) { + if (context->stream_status[i].plane_states[j] == plane) { + *plane_id = (i << 16) | j; + return true; + } + } + } + } + + return false; +} + +static unsigned int map_stream_to_dml21_display_cfg(const struct dml2_context *dml_ctx, const struct dc_stream_state *stream) +{ + int i = 0; + int location = -1; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] && dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == stream->stream_id) { + location = i; + break; + } + } + + return location; +} + +unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, + const struct dc_plane_state *plane, const struct dc_state *context) +{ + unsigned int plane_id; + int i = 0; + int location = -1; + + if (!dml21_wrapper_get_plane_id(context, stream_id, plane, &plane_id)) { + ASSERT(false); + return -1; + } + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] && dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i] == plane_id) { + location = i; + break; + } + } + + return location; +} + +static enum dml2_uclk_pstate_change_strategy dml21_force_pstate_method_to_uclk_state_change_strategy(enum dml2_force_pstate_methods force_pstate_method) +{ + enum dml2_uclk_pstate_change_strategy val = dml2_uclk_pstate_change_strategy_auto; + + switch (force_pstate_method) { + case dml2_force_pstate_method_vactive: + val = dml2_uclk_pstate_change_strategy_force_vactive; + break; + case dml2_force_pstate_method_vblank: + val = dml2_uclk_pstate_change_strategy_force_vblank; + break; + case dml2_force_pstate_method_drr: + val = dml2_uclk_pstate_change_strategy_force_drr; + break; + case dml2_force_pstate_method_subvp: + val = dml2_uclk_pstate_change_strategy_force_mall_svp; + break; + case dml2_force_pstate_method_auto: + default: + val = dml2_uclk_pstate_change_strategy_auto; + } + + return val; +} + +bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) +{ + int stream_index, plane_index; + int disp_cfg_stream_location, disp_cfg_plane_location; + struct dml2_display_cfg *dml_dispcfg = &dml_ctx->v21.display_config; + unsigned int plane_count = 0; + + memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); + + dml_dispcfg->gpuvm_enable = dml_ctx->config.gpuvm_enable; + dml_dispcfg->gpuvm_max_page_table_levels = 4; + dml_dispcfg->hostvm_enable = false; + dml_dispcfg->minimize_det_reallocation = true; + dml_dispcfg->overrides.enable_subvp_implicit_pmo = true; + + if (in_dc->debug.disable_unbounded_requesting) { + dml_dispcfg->overrides.hw.force_unbounded_requesting.enable = true; + dml_dispcfg->overrides.hw.force_unbounded_requesting.value = false; + } + + for (stream_index = 0; stream_index < context->stream_count; stream_index++) { + disp_cfg_stream_location = map_stream_to_dml21_display_cfg(dml_ctx, context->streams[stream_index]); + + if (disp_cfg_stream_location < 0) + disp_cfg_stream_location = dml_dispcfg->num_streams++; + + ASSERT(disp_cfg_stream_location >= 0 && disp_cfg_stream_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); + populate_dml21_timing_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index], dml_ctx); + populate_dml21_output_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].output, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index]); + populate_dml21_stream_overrides_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location], context->streams[stream_index], &context->stream_status[stream_index]); + + dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.fclk_pstate = dml2_twait_budgeting_setting_if_needed; + dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.uclk_pstate = dml2_twait_budgeting_setting_if_needed; + dml_dispcfg->stream_descriptors[disp_cfg_stream_location].overrides.hw.twait_budgeting.stutter_enter_exit = dml2_twait_budgeting_setting_if_needed; + + dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_stream_location] = context->streams[stream_index]->stream_id; + dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_stream_location] = true; + + if (context->stream_status[stream_index].plane_count == 0) { + disp_cfg_plane_location = dml_dispcfg->num_planes++; + populate_dml21_dummy_surface_cfg(&dml_dispcfg->plane_descriptors[disp_cfg_plane_location].surface, context->streams[stream_index]); + populate_dml21_dummy_plane_cfg(&dml_dispcfg->plane_descriptors[disp_cfg_plane_location], context->streams[stream_index]); + dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; + } else { + for (plane_index = 0; plane_index < context->stream_status[stream_index].plane_count; plane_index++) { + disp_cfg_plane_location = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], context); + + if (disp_cfg_plane_location < 0) + disp_cfg_plane_location = dml_dispcfg->num_planes++; + + ASSERT(disp_cfg_plane_location >= 0 && disp_cfg_plane_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); + + populate_dml21_surface_config_from_plane_state(in_dc, &dml_dispcfg->plane_descriptors[disp_cfg_plane_location].surface, context->stream_status[stream_index].plane_states[plane_index]); + populate_dml21_plane_config_from_plane_state(dml_ctx, &dml_dispcfg->plane_descriptors[disp_cfg_plane_location], context->stream_status[stream_index].plane_states[plane_index], context, stream_index); + dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; + + if (dml21_wrapper_get_plane_id(context, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], &dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) + dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; + + /* apply forced pstate policy */ + if (dml_ctx->config.pmo.force_pstate_method_enable) { + dml_dispcfg->plane_descriptors[disp_cfg_plane_location].overrides.uclk_pstate_change_strategy = + dml21_force_pstate_method_to_uclk_state_change_strategy(dml_ctx->config.pmo.force_pstate_method_values[stream_index]); + } + + plane_count++; + } + } + } + + if (plane_count == 0) { + dml_dispcfg->overrides.all_streams_blanked = true; + } + + return true; +} + +void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context) +{ + /* TODO these should be the max of active, svp prefetch and idle should be tracked seperately */ + context->bw_ctx.bw.dcn.clk.dispclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dispclk_khz; + context->bw_ctx.bw.dcn.clk.dcfclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.dcfclk_khz; + context->bw_ctx.bw.dcn.clk.dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.uclk_khz; + context->bw_ctx.bw.dcn.clk.fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.active.fclk_khz; + context->bw_ctx.bw.dcn.clk.idle_dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.idle.uclk_khz; + context->bw_ctx.bw.dcn.clk.idle_fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.idle.fclk_khz; + context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.deepsleep_dcfclk_khz; + context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = in_ctx->v21.mode_programming.programming->fclk_pstate_supported; + context->bw_ctx.bw.dcn.clk.p_state_change_support = in_ctx->v21.mode_programming.programming->uclk_pstate_supported; + context->bw_ctx.bw.dcn.clk.dtbclk_en = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dtbrefclk_khz > 0; + context->bw_ctx.bw.dcn.clk.ref_dtbclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.dtbrefclk_khz; + context->bw_ctx.bw.dcn.clk.socclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.socclk_khz; + context->bw_ctx.bw.dcn.clk.subvp_prefetch_dramclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz; + context->bw_ctx.bw.dcn.clk.subvp_prefetch_fclk_khz = in_ctx->v21.mode_programming.programming->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz; + context->bw_ctx.bw.dcn.clk.stutter_efficiency.base_efficiency = in_ctx->v21.mode_programming.programming->stutter.base_percent_efficiency; + context->bw_ctx.bw.dcn.clk.stutter_efficiency.low_power_efficiency = in_ctx->v21.mode_programming.programming->stutter.low_power_percent_efficiency; +} + +static struct dml2_dchub_watermark_regs *wm_set_index_to_dc_wm_set(union dcn_watermark_set *watermarks, const enum dml2_dchub_watermark_reg_set_index wm_index) +{ + struct dml2_dchub_watermark_regs *wm_regs = NULL; + + switch (wm_index) { + case DML2_DCHUB_WATERMARK_SET_A: + wm_regs = &watermarks->dcn4x.a; + break; + case DML2_DCHUB_WATERMARK_SET_B: + wm_regs = &watermarks->dcn4x.b; + break; + case DML2_DCHUB_WATERMARK_SET_C: + wm_regs = &watermarks->dcn4x.c; + break; + case DML2_DCHUB_WATERMARK_SET_D: + wm_regs = &watermarks->dcn4x.d; + break; + case DML2_DCHUB_WATERMARK_SET_NUM: + default: + /* invalid wm set index */ + wm_regs = NULL; + } + + return wm_regs; +} + +void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx) +{ + const struct dml2_display_cfg_programming *programming = in_ctx->v21.mode_programming.programming; + + unsigned int wm_index; + + /* copy watermark sets from DML */ + for (wm_index = 0; wm_index < programming->global_regs.num_watermark_sets; wm_index++) { + struct dml2_dchub_watermark_regs *wm_regs = wm_set_index_to_dc_wm_set(watermarks, wm_index); + + if (wm_regs) + memcpy(wm_regs, + &programming->global_regs.wm_regs[wm_index], + sizeof(struct dml2_dchub_watermark_regs)); + } +} + +void dml21_map_hw_resources(struct dml2_context *dml_ctx) +{ + unsigned int i = 0; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] = dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i]; + dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] = true; + dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] = dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i]; + dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] = true; + } + +} + +void dml21_get_pipe_mcache_config( + struct dc_state *context, + struct pipe_ctx *pipe_ctx, + struct dml2_per_plane_programming *pln_prog, + struct dml2_pipe_configuration_descriptor *mcache_pipe_config) +{ + mcache_pipe_config->plane0.viewport_x_start = pipe_ctx->plane_res.scl_data.viewport.x; + mcache_pipe_config->plane0.viewport_width = pipe_ctx->plane_res.scl_data.viewport.width; + + mcache_pipe_config->plane1.viewport_x_start = pipe_ctx->plane_res.scl_data.viewport_c.x; + mcache_pipe_config->plane1.viewport_width = pipe_ctx->plane_res.scl_data.viewport_c.width; + + mcache_pipe_config->plane1_enabled = + dml21_is_plane1_enabled(pln_prog->plane_descriptor->pixel_format); +} + +void dml21_set_dc_p_state_type( + struct pipe_ctx *pipe_ctx, + struct dml2_per_stream_programming *stream_programming, + bool sub_vp_enabled) +{ + switch (stream_programming->uclk_pstate_method) { + case dml2_pstate_method_vactive: + case dml2_pstate_method_fw_vactive_drr: + pipe_ctx->p_state_type = P_STATE_V_ACTIVE; + break; + case dml2_pstate_method_vblank: + case dml2_pstate_method_fw_vblank_drr: + if (sub_vp_enabled) + pipe_ctx->p_state_type = P_STATE_V_BLANK_SUB_VP; + else + pipe_ctx->p_state_type = P_STATE_V_BLANK; + break; + case dml2_pstate_method_fw_svp: + case dml2_pstate_method_fw_svp_drr: + pipe_ctx->p_state_type = P_STATE_SUB_VP; + break; + case dml2_pstate_method_fw_drr: + if (sub_vp_enabled) + pipe_ctx->p_state_type = P_STATE_DRR_SUB_VP; + else + pipe_ctx->p_state_type = P_STATE_FPO; + break; + default: + pipe_ctx->p_state_type = P_STATE_UNKNOWN; + break; + } +} + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h new file mode 100644 index 000000000000..9880d3e0398e --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h @@ -0,0 +1,28 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + + +#ifndef _DML21_TRANSLATION_HELPER_H_ +#define _DML21_TRANSLATION_HELPER_H_ + +struct dc; +struct dc_state; +struct dcn_watermarks; +union dcn_watermark_set; +struct pipe_ctx; +struct dc_plane_state; + +struct dml2_context; +struct dml2_configuration_options; +struct dml2_initialize_instance_in_out; + +void dml21_populate_dml_init_params(struct dml2_initialize_instance_in_out *dml_init, const struct dml2_configuration_options *config, const struct dc *in_dc); +bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); +void dml21_copy_clocks_to_dc_state(struct dml2_context *in_ctx, struct dc_state *context); +void dml21_extract_watermark_sets(const struct dc *in_dc, union dcn_watermark_set *watermarks, struct dml2_context *in_ctx); +void dml21_map_hw_resources(struct dml2_context *dml_ctx); +void dml21_get_pipe_mcache_config(struct dc_state *context, struct pipe_ctx *pipe_ctx, struct dml2_per_plane_programming *pln_prog, struct dml2_pipe_configuration_descriptor *mcache_pipe_config); +void dml21_set_dc_p_state_type(struct pipe_ctx *pipe_ctx, struct dml2_per_stream_programming *stream_programming, bool sub_vp_enabled); +unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, const struct dc_plane_state *plane, const struct dc_state *context); +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c new file mode 100644 index 000000000000..ee721606b883 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c @@ -0,0 +1,516 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + + +#include "dml2_internal_shared_types.h" +#include "dml21_translation_helper.h" +#include "dml2_internal_types.h" +#include "dml21_utils.h" +#include "dml2_dc_resource_mgmt.h" + +#include "dml2_core_dcn4_calcs.h" + +int dml21_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id) +{ + int i; + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] && ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] == stream_id) + return i; + } + + return -1; +} + +int dml21_find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id) +{ + int i; + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] && ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] == plane_id) + return i; + } + + return -1; +} + +bool dml21_get_plane_id(const struct dc_state *state, const struct dc_plane_state *plane, unsigned int *plane_id) +{ + int i, j; + + if (!plane_id) + return false; + + for (i = 0; i < state->stream_count; i++) { + for (j = 0; j < state->stream_status[i].plane_count; j++) { + if (state->stream_status[i].plane_states[j] == plane) { + *plane_id = (i << 16) | j; + return true; + } + } + } + + return false; +} + +unsigned int dml21_get_dc_plane_idx_from_plane_id(unsigned int plane_id) +{ + return 0xffff & plane_id; +} + +void find_valid_pipe_idx_for_stream_index(const struct dml2_context *dml_ctx, unsigned int *dml_pipe_idx, unsigned int stream_index) +{ + unsigned int i = 0; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (dml_ctx->v21.mode_programming.programming->plane_programming[i].plane_descriptor->stream_index == stream_index) { + *dml_pipe_idx = i; + return; + } + } +} + +void find_pipe_regs_idx(const struct dml2_context *dml_ctx, + struct pipe_ctx *pipe, unsigned int *pipe_regs_idx) +{ + struct pipe_ctx *opp_head = dml_ctx->config.callbacks.get_opp_head(pipe); + + *pipe_regs_idx = dml_ctx->config.callbacks.get_odm_slice_index(opp_head); + + if (pipe->plane_state) + *pipe_regs_idx += dml_ctx->config.callbacks.get_mpc_slice_index(pipe); +} + +/* places pipe references into pipes arrays and returns number of pipes */ +int dml21_find_dc_pipes_for_plane(const struct dc *in_dc, + struct dc_state *context, + struct dml2_context *dml_ctx, + struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], + struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], + int dml_plane_idx) +{ + unsigned int dml_stream_index; + unsigned int main_stream_id; + unsigned int dc_plane_index; + struct dc_stream_state *dc_main_stream; + struct dc_stream_status *dc_main_stream_status; + struct dc_plane_state *dc_main_plane; + struct dc_stream_state *dc_phantom_stream; + struct dc_stream_status *dc_phantom_stream_status; + struct dc_plane_state *dc_phantom_plane; + int num_pipes = 0; + + memset(dc_main_pipes, 0, sizeof(struct pipe_ctx *) * __DML2_WRAPPER_MAX_STREAMS_PLANES__); + memset(dc_phantom_pipes, 0, sizeof(struct pipe_ctx *) * __DML2_WRAPPER_MAX_STREAMS_PLANES__); + + dml_stream_index = dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_idx].plane_descriptor->stream_index; + main_stream_id = dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_stream_index]; + + dc_main_stream = dml_ctx->config.callbacks.get_stream_from_id(context, main_stream_id); + dc_main_stream_status = dml_ctx->config.callbacks.get_stream_status(context, dc_main_stream); + if (!dc_main_stream_status) + return num_pipes; + + /* find main plane based on id */ + dc_plane_index = dml21_get_dc_plane_idx_from_plane_id(dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[dml_plane_idx]); + dc_main_plane = dc_main_stream_status->plane_states[dc_plane_index]; + + if (dc_main_plane) { + num_pipes = dml_ctx->config.callbacks.get_dpp_pipes_for_plane(dc_main_plane, &context->res_ctx, dc_main_pipes); + } else { + /* stream was configured with dummy plane, so get pipes from opp head */ + struct pipe_ctx *otg_master_pipe = dml_ctx->config.callbacks.get_otg_master_for_stream(&context->res_ctx, dc_main_stream); + if (otg_master_pipe != NULL) + num_pipes = dml_ctx->config.callbacks.get_opp_heads_for_otg_master(otg_master_pipe, &context->res_ctx, dc_main_pipes); + } + + /* if phantom exists, find associated pipes */ + dc_phantom_stream = dml_ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, dc_main_stream); + if (dc_phantom_stream && num_pipes > 0) { + dc_phantom_stream_status = dml_ctx->config.callbacks.get_stream_status(context, dc_phantom_stream); + + if (dc_phantom_stream_status) { + /* phantom plane will have same index as main */ + dc_phantom_plane = dc_phantom_stream_status->plane_states[dc_plane_index]; + + if (dc_phantom_plane) { + /* only care about phantom pipes if they contain the phantom plane */ + dml_ctx->config.callbacks.get_dpp_pipes_for_plane(dc_phantom_plane, &context->res_ctx, dc_phantom_pipes); + } + } + } + + return num_pipes; +} + +void dml21_pipe_populate_global_sync(struct dml2_context *dml_ctx, + struct dc_state *context, + struct pipe_ctx *pipe_ctx, + struct dml2_per_stream_programming *stream_programming) +{ + union dml2_global_sync_programming *global_sync = &stream_programming->global_sync; + + if (dml_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe_ctx) == SUBVP_PHANTOM) { + /* phantom has its own global sync */ + global_sync = &stream_programming->phantom_stream.global_sync; + } + + memcpy(&pipe_ctx->global_sync, + global_sync, + sizeof(union dml2_global_sync_programming)); +} + +void dml21_populate_mall_allocation_size(struct dc_state *context, + struct dml2_context *in_ctx, + struct dml2_per_plane_programming *pln_prog, + struct pipe_ctx *dc_pipe) +{ + + /* Reuse MALL Allocation Sizes logic from dcn32_fpu.c */ + /* Count from active, top pipes per plane only. Only add mall_ss_size_bytes for each unique plane. */ + if (dc_pipe->stream && dc_pipe->plane_state && + (dc_pipe->top_pipe == NULL || + dc_pipe->plane_state != dc_pipe->top_pipe->plane_state) && + dc_pipe->prev_odm_pipe == NULL) { + /* SS: all active surfaces stored in MALL */ + if (in_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, dc_pipe) != SUBVP_PHANTOM) { + dc_pipe->surface_size_in_mall_bytes = pln_prog->surface_size_mall_bytes; + context->bw_ctx.bw.dcn.mall_ss_size_bytes += dc_pipe->surface_size_in_mall_bytes; + } else { + /* SUBVP: phantom surfaces only stored in MALL */ + dc_pipe->surface_size_in_mall_bytes = pln_prog->svp_size_mall_bytes; + context->bw_ctx.bw.dcn.mall_subvp_size_bytes += dc_pipe->surface_size_in_mall_bytes; + } + } +} + +bool check_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx) +{ + /* If this assert is hit then we have a link encoder dynamic management issue */ + ASSERT(pipe_ctx->stream_res.hpo_dp_stream_enc ? pipe_ctx->link_res.hpo_dp_link_enc != NULL : true); + return (pipe_ctx->stream_res.hpo_dp_stream_enc && + pipe_ctx->link_res.hpo_dp_link_enc && + dc_is_dp_signal(pipe_ctx->stream->signal)); +} + + +static bool is_sub_vp_enabled(struct dc *dc, struct dc_state *context) +{ + int i; + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i]; + + if (pipe_ctx->stream && dc_state_get_paired_subvp_stream(context, pipe_ctx->stream) && + dc_state_get_pipe_subvp_type(context, pipe_ctx) == SUBVP_MAIN) { + return true; + } + } + return false; +} + + +void dml21_program_dc_pipe(struct dml2_context *dml_ctx, struct dc_state *context, struct pipe_ctx *pipe_ctx, struct dml2_per_plane_programming *pln_prog, + struct dml2_per_stream_programming *stream_prog) +{ + unsigned int pipe_reg_index = 0; + + dml21_pipe_populate_global_sync(dml_ctx, context, pipe_ctx, stream_prog); + find_pipe_regs_idx(dml_ctx, pipe_ctx, &pipe_reg_index); + + if (dml_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe_ctx) == SUBVP_PHANTOM) { + memcpy(&pipe_ctx->hubp_regs, pln_prog->phantom_plane.pipe_regs[pipe_reg_index], sizeof(struct dml2_dchub_per_pipe_register_set)); + pipe_ctx->unbounded_req = false; + pipe_ctx->det_buffer_size_kb = 0; + } else { + memcpy(&pipe_ctx->hubp_regs, pln_prog->pipe_regs[pipe_reg_index], sizeof(struct dml2_dchub_per_pipe_register_set)); + pipe_ctx->unbounded_req = pln_prog->pipe_regs[pipe_reg_index]->rq_regs.unbounded_request_enabled; + pipe_ctx->det_buffer_size_kb = pln_prog->pipe_regs[pipe_reg_index]->det_size * 64; + } + + pipe_ctx->plane_res.bw.dppclk_khz = pln_prog->min_clocks.dcn4x.dppclk_khz; + if (context->bw_ctx.bw.dcn.clk.dppclk_khz < pipe_ctx->plane_res.bw.dppclk_khz) + context->bw_ctx.bw.dcn.clk.dppclk_khz = pipe_ctx->plane_res.bw.dppclk_khz; + + dml21_populate_mall_allocation_size(context, dml_ctx, pln_prog, pipe_ctx); + + bool sub_vp_enabled = is_sub_vp_enabled(pipe_ctx->stream->ctx->dc, context); + + dml21_set_dc_p_state_type(pipe_ctx, stream_prog, sub_vp_enabled); +} + +static struct dc_stream_state *dml21_add_phantom_stream(struct dml2_context *dml_ctx, + const struct dc *dc, + struct dc_state *context, + struct dc_stream_state *main_stream, + struct dml2_per_stream_programming *stream_programming) +{ + struct dc_stream_state *phantom_stream; + struct dml2_stream_parameters *phantom_stream_descriptor = &stream_programming->phantom_stream.descriptor; + + phantom_stream = dml_ctx->config.svp_pstate.callbacks.create_phantom_stream(dc, context, main_stream); + if (!phantom_stream) + return NULL; + + /* copy details of phantom stream from main */ + memcpy(&phantom_stream->timing, &main_stream->timing, sizeof(phantom_stream->timing)); + memcpy(&phantom_stream->src, &main_stream->src, sizeof(phantom_stream->src)); + memcpy(&phantom_stream->dst, &main_stream->dst, sizeof(phantom_stream->dst)); + + /* modify timing for phantom */ + phantom_stream->timing.v_front_porch = phantom_stream_descriptor->timing.v_front_porch; + phantom_stream->timing.v_addressable = phantom_stream_descriptor->timing.v_active; + phantom_stream->timing.v_total = phantom_stream_descriptor->timing.v_total; + phantom_stream->timing.flags.DSC = 0; // phantom always has DSC disabled + + phantom_stream->dst.y = 0; + phantom_stream->dst.height = stream_programming->phantom_stream.descriptor.timing.v_active; + + phantom_stream->src.y = 0; + phantom_stream->src.height = (double)phantom_stream_descriptor->timing.v_active * (double)main_stream->src.height / (double)main_stream->dst.height; + + phantom_stream->use_dynamic_meta = false; + + dml_ctx->config.svp_pstate.callbacks.add_phantom_stream(dc, context, phantom_stream, main_stream); + + return phantom_stream; +} + +static struct dc_plane_state *dml21_add_phantom_plane(struct dml2_context *dml_ctx, + const struct dc *dc, + struct dc_state *context, + struct dc_stream_state *phantom_stream, + struct dc_plane_state *main_plane, + struct dml2_per_plane_programming *plane_programming) +{ + struct dc_plane_state *phantom_plane; + + phantom_plane = dml_ctx->config.svp_pstate.callbacks.create_phantom_plane(dc, context, main_plane); + if (!phantom_plane) + return NULL; + + phantom_plane->format = main_plane->format; + phantom_plane->rotation = main_plane->rotation; + phantom_plane->visible = main_plane->visible; + + memcpy(&phantom_plane->address, &main_plane->address, sizeof(phantom_plane->address)); + memcpy(&phantom_plane->scaling_quality, &main_plane->scaling_quality, + sizeof(phantom_plane->scaling_quality)); + memcpy(&phantom_plane->src_rect, &main_plane->src_rect, sizeof(phantom_plane->src_rect)); + memcpy(&phantom_plane->dst_rect, &main_plane->dst_rect, sizeof(phantom_plane->dst_rect)); + memcpy(&phantom_plane->clip_rect, &main_plane->clip_rect, sizeof(phantom_plane->clip_rect)); + memcpy(&phantom_plane->plane_size, &main_plane->plane_size, + sizeof(phantom_plane->plane_size)); + memcpy(&phantom_plane->tiling_info, &main_plane->tiling_info, + sizeof(phantom_plane->tiling_info)); + memcpy(&phantom_plane->dcc, &main_plane->dcc, sizeof(phantom_plane->dcc)); + + phantom_plane->format = main_plane->format; + phantom_plane->rotation = main_plane->rotation; + phantom_plane->visible = main_plane->visible; + + /* Shadow pipe has small viewport. */ + phantom_plane->clip_rect.y = 0; + phantom_plane->clip_rect.height = phantom_stream->src.height; + + dml_ctx->config.svp_pstate.callbacks.add_phantom_plane(dc, phantom_stream, phantom_plane, context); + + return phantom_plane; +} + +void dml21_handle_phantom_streams_planes(const struct dc *dc, struct dc_state *context, struct dml2_context *dml_ctx) +{ + unsigned int dml_stream_index, dml_plane_index, dc_plane_index; + struct dc_stream_state *main_stream; + struct dc_stream_status *main_stream_status; + struct dc_stream_state *phantom_stream; + struct dc_plane_state *main_plane; + bool phantoms_added = false; + + /* create phantom streams and planes and add to context */ + for (dml_stream_index = 0; dml_stream_index < dml_ctx->v21.mode_programming.programming->display_config.num_streams; dml_stream_index++) { + /* iterate through DML streams looking for phantoms */ + if (dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_index].phantom_stream.enabled) { + /* find associated dc stream */ + main_stream = dml_ctx->config.callbacks.get_stream_from_id(context, + dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_stream_index]); + + main_stream_status = dml_ctx->config.callbacks.get_stream_status(context, main_stream); + + if (!main_stream_status || main_stream_status->plane_count == 0) + continue; + + /* create phantom stream for subvp enabled stream */ + phantom_stream = dml21_add_phantom_stream(dml_ctx, + dc, + context, + main_stream, + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_index]); + + if (!phantom_stream) + continue; + + /* iterate through DML planes associated with this stream */ + for (dml_plane_index = 0; dml_plane_index < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_plane_index++) { + if (dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_index].plane_descriptor->stream_index == dml_stream_index) { + /* find associated dc plane */ + dc_plane_index = dml21_get_dc_plane_idx_from_plane_id(dml_ctx->v21.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[dml_plane_index]); + main_plane = main_stream_status->plane_states[dc_plane_index]; + + /* create phantom planes for subvp enabled plane */ + dml21_add_phantom_plane(dml_ctx, + dc, + context, + phantom_stream, + main_plane, + &dml_ctx->v21.mode_programming.programming->plane_programming[dml_plane_index]); + + phantoms_added = true; + } + } + } + } + + if (phantoms_added) + dml2_map_dc_pipes(dml_ctx, context, NULL, &dml_ctx->v21.dml_to_dc_pipe_mapping, dc->current_state); +} + +void dml21_build_fams2_programming(const struct dc *dc, + struct dc_state *context, + struct dml2_context *dml_ctx) +{ + int i, j, k; + unsigned int num_fams2_streams = 0; + + /* reset fams2 data */ + memset(&context->bw_ctx.bw.dcn.fams2_stream_base_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); + memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params, 0, sizeof(union dmub_cmd_fams2_config) * DML2_MAX_PLANES); + memset(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2, 0, sizeof(union dmub_fams2_stream_static_sub_state_v2) * DML2_MAX_PLANES); + memset(&context->bw_ctx.bw.dcn.fams2_global_config, 0, sizeof(struct dmub_cmd_fams2_global_config)); + + if (dml_ctx->v21.mode_programming.programming->fams2_required) { + for (i = 0; i < context->stream_count; i++) { + int dml_stream_idx; + struct dc_stream_state *phantom_stream; + struct dc_stream_status *phantom_status; + enum fams2_stream_type type = 0; + + union dmub_cmd_fams2_config *static_base_state = &context->bw_ctx.bw.dcn.fams2_stream_base_params[num_fams2_streams]; + union dmub_cmd_fams2_config *static_sub_state = &context->bw_ctx.bw.dcn.fams2_stream_sub_params[num_fams2_streams]; + + struct dc_stream_state *stream = context->streams[i]; + + if (context->stream_status[i].plane_count == 0 || + dml_ctx->config.svp_pstate.callbacks.get_stream_subvp_type(context, stream) == SUBVP_PHANTOM) { + /* can ignore blanked or phantom streams */ + continue; + } + + dml_stream_idx = dml21_helper_find_dml_pipe_idx_by_stream_id(dml_ctx, stream->stream_id); + if (dml_stream_idx < 0) { + ASSERT(dml_stream_idx >= 0); + continue; + } + + /* copy static state from PMO */ + memcpy(static_base_state, + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_base_params, + sizeof(union dmub_cmd_fams2_config)); + + if (dc->debug.fams_version.major == 3) { + memcpy(&context->bw_ctx.bw.dcn.fams2_stream_sub_params_v2[num_fams2_streams], + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params_v2, + sizeof(union dmub_fams2_stream_static_sub_state_v2)); + } else { + memcpy(static_sub_state, + &dml_ctx->v21.mode_programming.programming->stream_programming[dml_stream_idx].fams2_sub_params, + sizeof(union dmub_cmd_fams2_config)); + } + + switch (dc->debug.fams_version.minor) { + case 1: + default: + type = static_base_state->stream_v1.base.type; + + /* get information from context */ + static_base_state->stream_v1.base.num_planes = context->stream_status[i].plane_count; + static_base_state->stream_v1.base.otg_inst = context->stream_status[i].primary_otg_inst; + + /* populate pipe masks for planes */ + for (j = 0; j < context->stream_status[i].plane_count; j++) { + for (k = 0; k < dc->res_pool->pipe_count; k++) { + if (context->res_ctx.pipe_ctx[k].stream && + context->res_ctx.pipe_ctx[k].stream->stream_id == stream->stream_id && + context->res_ctx.pipe_ctx[k].plane_state == context->stream_status[i].plane_states[j]) { + static_base_state->stream_v1.base.pipe_mask |= (1 << k); + static_base_state->stream_v1.base.plane_pipe_masks[j] |= (1 << k); + } + } + } + } + + + /* get per method programming */ + switch (type) { + case FAMS2_STREAM_TYPE_VBLANK: + case FAMS2_STREAM_TYPE_VACTIVE: + case FAMS2_STREAM_TYPE_DRR: + break; + case FAMS2_STREAM_TYPE_SUBVP: + phantom_stream = dml_ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, stream); + if (!phantom_stream) + break; + + phantom_status = dml_ctx->config.callbacks.get_stream_status(context, phantom_stream); + + /* phantom status should always be present */ + ASSERT(phantom_status); + if (!phantom_status) + break; + + switch (dc->debug.fams_version.minor) { + case 1: + default: + static_sub_state->stream_v1.sub_state.subvp.phantom_otg_inst = phantom_status->primary_otg_inst; + + /* populate pipe masks for phantom planes */ + for (j = 0; j < phantom_status->plane_count; j++) { + for (k = 0; k < dc->res_pool->pipe_count; k++) { + if (context->res_ctx.pipe_ctx[k].stream && + context->res_ctx.pipe_ctx[k].stream->stream_id == phantom_stream->stream_id && + context->res_ctx.pipe_ctx[k].plane_state == phantom_status->plane_states[j]) { + switch (dc->debug.fams_version.minor) { + case 1: + default: + static_sub_state->stream_v1.sub_state.subvp.phantom_pipe_mask |= (1 << k); + static_sub_state->stream_v1.sub_state.subvp.phantom_plane_pipe_masks[j] |= (1 << k); + } + } + } + } + } + break; + default: + ASSERT(false); + break; + } + + num_fams2_streams++; + } + } + + if (num_fams2_streams > 0) { + /* copy FAMS2 configuration */ + memcpy(&context->bw_ctx.bw.dcn.fams2_global_config, + &dml_ctx->v21.mode_programming.programming->fams2_global_config, + sizeof(struct dmub_cmd_fams2_global_config)); + + context->bw_ctx.bw.dcn.fams2_global_config.num_streams = num_fams2_streams; + } + + context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching = context->bw_ctx.bw.dcn.fams2_global_config.features.bits.enable; +} + +bool dml21_is_plane1_enabled(enum dml2_source_format_class source_format) +{ + return source_format >= dml2_420_8 && source_format <= dml2_rgbe_alpha; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h new file mode 100644 index 000000000000..4bff52eaaef8 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h @@ -0,0 +1,50 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + + +#ifndef _DML21_UTILS_H_ +#define _DML21_UTILS_H_ + +struct dc_state; +struct dc_plane_state; +struct pipe_ctx; + +struct dml2_context; +struct dml2_display_rq_regs; +struct dml2_display_dlg_regs; +struct dml2_display_ttu_regs; + +int dml21_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id); +int dml21_find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id); +bool dml21_get_plane_id(const struct dc_state *state, const struct dc_plane_state *plane, unsigned int *plane_id); +void dml21_pipe_populate_global_sync(struct dml2_context *dml_ctx, + struct dc_state *context, + struct pipe_ctx *pipe_ctx, + struct dml2_per_stream_programming *stream_programming); +void dml21_populate_mall_allocation_size(struct dc_state *context, + struct dml2_context *in_ctx, + struct dml2_per_plane_programming *pln_prog, + struct pipe_ctx *dc_pipe); +bool check_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx); +void find_valid_pipe_idx_for_stream_index(const struct dml2_context *dml_ctx, unsigned int *dml_pipe_idx, unsigned int stream_index); +void find_pipe_regs_idx(const struct dml2_context *dml_ctx, + struct pipe_ctx *pipe, unsigned int *pipe_regs_idx); +int dml21_find_dc_pipes_for_plane(const struct dc *in_dc, + struct dc_state *context, + struct dml2_context *dml_ctx, + struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], + struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__], + int dml_plane_idx); +void dml21_program_dc_pipe(struct dml2_context *dml_ctx, + struct dc_state *context, + struct pipe_ctx *pipe_ctx, + struct dml2_per_plane_programming *pln_prog, + struct dml2_per_stream_programming *stream_prog); +void dml21_handle_phantom_streams_planes(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); +unsigned int dml21_get_dc_plane_idx_from_plane_id(unsigned int plane_id); +void dml21_build_fams2_programming(const struct dc *dc, + struct dc_state *context, + struct dml2_context *dml_ctx); +bool dml21_is_plane1_enabled(enum dml2_source_format_class source_format); +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c new file mode 100644 index 000000000000..798abb2b2e67 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c @@ -0,0 +1,466 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_internal_types.h" +#include "dml_top.h" +#include "dml2_core_dcn4_calcs.h" +#include "dml2_internal_shared_types.h" +#include "dml21_utils.h" +#include "dml21_translation_helper.h" +#include "dml2_dc_resource_mgmt.h" + +#define INVALID -1 + +static bool dml21_allocate_memory(struct dml2_context **dml_ctx) +{ + *dml_ctx = vzalloc(sizeof(struct dml2_context)); + if (!(*dml_ctx)) + return false; + + (*dml_ctx)->v21.dml_init.dml2_instance = vzalloc(sizeof(struct dml2_instance)); + if (!((*dml_ctx)->v21.dml_init.dml2_instance)) + return false; + + (*dml_ctx)->v21.mode_support.dml2_instance = (*dml_ctx)->v21.dml_init.dml2_instance; + (*dml_ctx)->v21.mode_programming.dml2_instance = (*dml_ctx)->v21.dml_init.dml2_instance; + + (*dml_ctx)->v21.mode_support.display_config = &(*dml_ctx)->v21.display_config; + (*dml_ctx)->v21.mode_programming.display_config = (*dml_ctx)->v21.mode_support.display_config; + + (*dml_ctx)->v21.mode_programming.programming = vzalloc(sizeof(struct dml2_display_cfg_programming)); + if (!((*dml_ctx)->v21.mode_programming.programming)) + return false; + + return true; +} + +static void dml21_populate_configuration_options(const struct dc *in_dc, + struct dml2_context *dml_ctx, + const struct dml2_configuration_options *config) +{ + dml_ctx->config = *config; + + /* UCLK P-State options */ + if (in_dc->debug.dml21_force_pstate_method) { + dml_ctx->config.pmo.force_pstate_method_enable = true; + for (int i = 0; i < MAX_PIPES; i++) + dml_ctx->config.pmo.force_pstate_method_values[i] = in_dc->debug.dml21_force_pstate_method_values[i]; + } else { + dml_ctx->config.pmo.force_pstate_method_enable = false; + } +} + +static void dml21_init(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) +{ + + dml_ctx->architecture = dml2_architecture_21; + + dml21_populate_configuration_options(in_dc, dml_ctx, config); + + DC_FP_START(); + + dml21_populate_dml_init_params(&dml_ctx->v21.dml_init, &dml_ctx->config, in_dc); + + dml2_initialize_instance(&dml_ctx->v21.dml_init); + + DC_FP_END(); +} + +bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config) +{ + /* Allocate memory for initializing DML21 instance */ + if (!dml21_allocate_memory(dml_ctx)) + return false; + + dml21_init(in_dc, *dml_ctx, config); + + return true; +} + +void dml21_destroy(struct dml2_context *dml2) +{ + vfree(dml2->v21.dml_init.dml2_instance); + vfree(dml2->v21.mode_programming.programming); +} + +static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, + struct dml2_context *in_ctx, unsigned int pipe_cnt) +{ + unsigned int dml_prog_idx = 0, dc_pipe_index = 0, num_dpps_required = 0; + struct dml2_per_plane_programming *pln_prog = NULL; + struct dml2_per_stream_programming *stream_prog = NULL; + struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; + int num_pipes; + unsigned int dml_phantom_prog_idx; + + context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; + + /* copy global DCHUBBUB arbiter registers */ + memcpy(&context->bw_ctx.bw.dcn.arb_regs, &in_ctx->v21.mode_programming.programming->global_regs.arb_regs, sizeof(struct dml2_display_arb_regs)); + + /* legacy only */ + context->bw_ctx.bw.dcn.compbuf_size_kb = (int)in_ctx->v21.mode_programming.programming->global_regs.arb_regs.compbuf_size * 64; + + context->bw_ctx.bw.dcn.mall_ss_size_bytes = 0; + context->bw_ctx.bw.dcn.mall_ss_psr_active_size_bytes = 0; + context->bw_ctx.bw.dcn.mall_subvp_size_bytes = 0; + + /* phantom's start after main planes */ + dml_phantom_prog_idx = in_ctx->v21.mode_programming.programming->display_config.num_planes; + + for (dml_prog_idx = 0; dml_prog_idx < DML2_MAX_PLANES; dml_prog_idx++) { + pln_prog = &in_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; + + if (!pln_prog->plane_descriptor) + continue; + + stream_prog = &in_ctx->v21.mode_programming.programming->stream_programming[pln_prog->plane_descriptor->stream_index]; + num_dpps_required = pln_prog->num_dpps_required; + + if (num_dpps_required == 0) { + continue; + } + num_pipes = dml21_find_dc_pipes_for_plane(dc, context, in_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); + + if (num_pipes <= 0) + continue; + + /* program each pipe */ + for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { + dml21_program_dc_pipe(in_ctx, context, dc_main_pipes[dc_pipe_index], pln_prog, stream_prog); + + if (pln_prog->phantom_plane.valid && dc_phantom_pipes[dc_pipe_index]) { + dml21_program_dc_pipe(in_ctx, context, dc_phantom_pipes[dc_pipe_index], pln_prog, stream_prog); + } + } + + /* copy per plane mcache allocation */ + memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_prog_idx], &pln_prog->mcache_allocation, sizeof(struct dml2_mcache_surface_allocation)); + if (pln_prog->phantom_plane.valid) { + memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_phantom_prog_idx], + &pln_prog->phantom_plane.mcache_allocation, + sizeof(struct dml2_mcache_surface_allocation)); + + dml_phantom_prog_idx++; + } + } + + /* assign global clocks */ + context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; + context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; + if (in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.num_clk_values > 1) { + context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = + in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.clk_values_khz[in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.num_clk_values] * 1000; + } else { + context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = in_ctx->v21.dml_init.soc_bb.clk_table.dispclk.clk_values_khz[0] * 1000; + } + + if (in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.num_clk_values > 1) { + context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = + in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.clk_values_khz[in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.num_clk_values] * 1000; + } else { + context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = in_ctx->v21.dml_init.soc_bb.clk_table.dppclk.clk_values_khz[0] * 1000; + } + + /* get global mall allocation */ + if (dc->res_pool->funcs->calculate_mall_ways_from_bytes) { + context->bw_ctx.bw.dcn.clk.num_ways = dc->res_pool->funcs->calculate_mall_ways_from_bytes(dc, context->bw_ctx.bw.dcn.mall_subvp_size_bytes); + } else { + context->bw_ctx.bw.dcn.clk.num_ways = 0; + } +} + +static void dml21_prepare_mcache_params(struct dml2_context *dml_ctx, struct dc_state *context, struct dc_mcache_params *mcache_params) +{ + int dc_plane_idx = 0; + int dml_prog_idx, stream_idx, plane_idx; + struct dml2_per_plane_programming *pln_prog = NULL; + + for (stream_idx = 0; stream_idx < context->stream_count; stream_idx++) { + for (plane_idx = 0; plane_idx < context->stream_status[stream_idx].plane_count; plane_idx++) { + dml_prog_idx = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_idx]->stream_id, context->stream_status[stream_idx].plane_states[plane_idx], context); + if (dml_prog_idx == INVALID) { + continue; + } + pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; + mcache_params[dc_plane_idx].valid = pln_prog->mcache_allocation.valid; + mcache_params[dc_plane_idx].num_mcaches_plane0 = pln_prog->mcache_allocation.num_mcaches_plane0; + mcache_params[dc_plane_idx].num_mcaches_plane1 = pln_prog->mcache_allocation.num_mcaches_plane1; + mcache_params[dc_plane_idx].requires_dedicated_mall_mcache = pln_prog->mcache_allocation.requires_dedicated_mall_mcache; + mcache_params[dc_plane_idx].last_slice_sharing.plane0_plane1 = pln_prog->mcache_allocation.last_slice_sharing.plane0_plane1; + memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane0, + pln_prog->mcache_allocation.mcache_x_offsets_plane0, + sizeof(int) * (DML2_MAX_MCACHES + 1)); + memcpy(mcache_params[dc_plane_idx].mcache_x_offsets_plane1, + pln_prog->mcache_allocation.mcache_x_offsets_plane1, + sizeof(int) * (DML2_MAX_MCACHES + 1)); + dc_plane_idx++; + } + } +} + +static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) +{ + bool result = false; + struct dml2_build_mode_programming_in_out *mode_programming = &dml_ctx->v21.mode_programming; + struct dc_mcache_params mcache_params[MAX_PLANES] = {0}; + + memset(&dml_ctx->v21.display_config, 0, sizeof(struct dml2_display_cfg)); + memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); + memset(&dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params, 0, sizeof(struct dml2_core_mode_programming_in_out)); + + if (!context) + return true; + + if (context->stream_count == 0) { + dml21_build_fams2_programming(in_dc, context, dml_ctx); + return true; + } + + /* scrub phantom's from current dc_state */ + dml_ctx->config.svp_pstate.callbacks.remove_phantom_streams_and_planes(in_dc, context); + dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); + + /* Populate stream, plane mappings and other fields in display config. */ + result = dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); + if (!result) + return false; + + DC_FP_START(); + result = dml2_build_mode_programming(mode_programming); + DC_FP_END(); + if (!result) + return false; + + /* Check and map HW resources */ + if (result && !dml_ctx->config.skip_hw_state_mapping) { + dml21_map_hw_resources(dml_ctx); + dml2_map_dc_pipes(dml_ctx, context, NULL, &dml_ctx->v21.dml_to_dc_pipe_mapping, in_dc->current_state); + /* if subvp phantoms are present, expand them into dc context */ + dml21_handle_phantom_streams_planes(in_dc, context, dml_ctx); + + if (in_dc->res_pool->funcs->program_mcache_pipe_config) { + //Prepare mcache params for each plane based on mcache output from DML + dml21_prepare_mcache_params(dml_ctx, context, mcache_params); + + //populate mcache regs to each pipe + dml_ctx->config.callbacks.allocate_mcache(context, mcache_params); + } + } + + /* Copy DML CLK, WM and REG outputs to bandwidth context */ + if (result && !dml_ctx->config.skip_hw_state_mapping) { + dml21_calculate_rq_and_dlg_params(in_dc, context, &context->res_ctx, dml_ctx, in_dc->res_pool->pipe_count); + dml21_copy_clocks_to_dc_state(dml_ctx, context); + dml21_extract_watermark_sets(in_dc, &context->bw_ctx.bw.dcn.watermarks, dml_ctx); + dml21_build_fams2_programming(in_dc, context, dml_ctx); + } + + return true; +} + +static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) +{ + bool is_supported = false; + struct dml2_initialize_instance_in_out *dml_init = &dml_ctx->v21.dml_init; + struct dml2_check_mode_supported_in_out *mode_support = &dml_ctx->v21.mode_support; + + memset(&dml_ctx->v21.display_config, 0, sizeof(struct dml2_display_cfg)); + memset(&dml_ctx->v21.dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); + memset(&dml_ctx->v21.mode_programming.dml2_instance->scratch.check_mode_supported_locals.mode_support_params, 0, sizeof(struct dml2_core_mode_support_in_out)); + + if (!context || context->stream_count == 0) + return true; + + /* Scrub phantom's from current dc_state */ + dml_ctx->config.svp_pstate.callbacks.remove_phantom_streams_and_planes(in_dc, context); + dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); + + mode_support->dml2_instance = dml_init->dml2_instance; + dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); + dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params.programming = dml_ctx->v21.mode_programming.programming; + DC_FP_START(); + is_supported = dml2_check_mode_supported(mode_support); + DC_FP_END(); + if (!is_supported) + return false; + + return true; +} + +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode) +{ + bool out = false; + + /* Use dml21_check_mode_support for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) + out = dml21_check_mode_support(in_dc, context, dml_ctx); + else + out = dml21_mode_check_and_programming(in_dc, context, dml_ctx); + + return out; +} + +void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx) +{ + unsigned int dml_prog_idx, dml_phantom_prog_idx, dc_pipe_index; + int num_pipes; + struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; + + struct dml2_per_plane_programming *pln_prog = NULL; + struct dml2_plane_mcache_configuration_descriptor *mcache_config = NULL; + struct prepare_mcache_programming_locals *l = &dml_ctx->v21.scratch.prepare_mcache_locals; + + if (context->stream_count == 0) { + return; + } + + memset(&l->build_mcache_programming_params, 0, sizeof(struct dml2_build_mcache_programming_in_out)); + l->build_mcache_programming_params.dml2_instance = dml_ctx->v21.dml_init.dml2_instance; + + /* phantom's start after main planes */ + dml_phantom_prog_idx = dml_ctx->v21.mode_programming.programming->display_config.num_planes; + + /* Build mcache programming parameters per plane per pipe */ + for (dml_prog_idx = 0; dml_prog_idx < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_prog_idx++) { + pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; + + mcache_config = &l->build_mcache_programming_params.mcache_configurations[dml_prog_idx]; + memset(mcache_config, 0, sizeof(struct dml2_plane_mcache_configuration_descriptor)); + mcache_config->plane_descriptor = pln_prog->plane_descriptor; + mcache_config->mcache_allocation = &context->bw_ctx.bw.dcn.mcache_allocations[dml_prog_idx]; + mcache_config->num_pipes = pln_prog->num_dpps_required; + l->build_mcache_programming_params.num_configurations++; + + if (pln_prog->num_dpps_required == 0) { + continue; + } + + num_pipes = dml21_find_dc_pipes_for_plane(in_dc, context, dml_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); + if (num_pipes <= 0 || dc_main_pipes[0]->stream == NULL || + dc_main_pipes[0]->plane_state == NULL) + continue; + + /* get config for each pipe */ + for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { + ASSERT(dc_main_pipes[dc_pipe_index]); + dml21_get_pipe_mcache_config(context, dc_main_pipes[dc_pipe_index], pln_prog, &mcache_config->pipe_configurations[dc_pipe_index]); + } + + /* get config for each phantom pipe */ + if (pln_prog->phantom_plane.valid && + dc_phantom_pipes[0] && + dc_main_pipes[0]->stream && + dc_phantom_pipes[0]->plane_state) { + mcache_config = &l->build_mcache_programming_params.mcache_configurations[dml_phantom_prog_idx]; + memset(mcache_config, 0, sizeof(struct dml2_plane_mcache_configuration_descriptor)); + mcache_config->plane_descriptor = pln_prog->plane_descriptor; + mcache_config->mcache_allocation = &context->bw_ctx.bw.dcn.mcache_allocations[dml_phantom_prog_idx]; + mcache_config->num_pipes = pln_prog->num_dpps_required; + l->build_mcache_programming_params.num_configurations++; + + for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { + ASSERT(dc_phantom_pipes[dc_pipe_index]); + dml21_get_pipe_mcache_config(context, dc_phantom_pipes[dc_pipe_index], pln_prog, &mcache_config->pipe_configurations[dc_pipe_index]); + } + + /* increment phantom index */ + dml_phantom_prog_idx++; + } + } + + /* Call to generate mcache programming per plane per pipe for the given display configuration */ + dml2_build_mcache_programming(&l->build_mcache_programming_params); + + /* get per plane per pipe mcache programming */ + for (dml_prog_idx = 0; dml_prog_idx < dml_ctx->v21.mode_programming.programming->display_config.num_planes; dml_prog_idx++) { + pln_prog = &dml_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; + + num_pipes = dml21_find_dc_pipes_for_plane(in_dc, context, dml_ctx, dc_main_pipes, dc_phantom_pipes, dml_prog_idx); + if (num_pipes <= 0 || dc_main_pipes[0]->stream == NULL || + dc_main_pipes[0]->plane_state == NULL) + continue; + + /* get config for each pipe */ + for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { + ASSERT(dc_main_pipes[dc_pipe_index]); + if (l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_prog_idx][dc_pipe_index]) { + memcpy(&dc_main_pipes[dc_pipe_index]->mcache_regs, + l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_prog_idx][dc_pipe_index], + sizeof(struct dml2_hubp_pipe_mcache_regs)); + } + } + + /* get config for each phantom pipe */ + if (pln_prog->phantom_plane.valid && + dc_phantom_pipes[0] && + dc_main_pipes[0]->stream && + dc_phantom_pipes[0]->plane_state) { + for (dc_pipe_index = 0; dc_pipe_index < num_pipes; dc_pipe_index++) { + ASSERT(dc_phantom_pipes[dc_pipe_index]); + if (l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_phantom_prog_idx][dc_pipe_index]) { + memcpy(&dc_phantom_pipes[dc_pipe_index]->mcache_regs, + l->build_mcache_programming_params.per_plane_pipe_mcache_regs[dml_phantom_prog_idx][dc_pipe_index], + sizeof(struct dml2_hubp_pipe_mcache_regs)); + } + } + /* increment phantom index */ + dml_phantom_prog_idx++; + } + } +} + +void dml21_copy(struct dml2_context *dst_dml_ctx, + struct dml2_context *src_dml_ctx) +{ + /* Preserve references to internals */ + struct dml2_instance *dst_dml2_instance = dst_dml_ctx->v21.dml_init.dml2_instance; + struct dml2_display_cfg_programming *dst_dml2_programming = dst_dml_ctx->v21.mode_programming.programming; + + /* Copy context */ + memcpy(dst_dml_ctx, src_dml_ctx, sizeof(struct dml2_context)); + + /* Copy Internals */ + memcpy(dst_dml2_instance, src_dml_ctx->v21.dml_init.dml2_instance, sizeof(struct dml2_instance)); + memcpy(dst_dml2_programming, src_dml_ctx->v21.mode_programming.programming, sizeof(struct dml2_display_cfg_programming)); + + /* Restore references to internals */ + dst_dml_ctx->v21.dml_init.dml2_instance = dst_dml2_instance; + + dst_dml_ctx->v21.mode_support.dml2_instance = dst_dml2_instance; + dst_dml_ctx->v21.mode_programming.dml2_instance = dst_dml2_instance; + + dst_dml_ctx->v21.mode_support.display_config = &dst_dml_ctx->v21.display_config; + dst_dml_ctx->v21.mode_programming.display_config = dst_dml_ctx->v21.mode_support.display_config; + + dst_dml_ctx->v21.mode_programming.programming = dst_dml2_programming; + + DC_FP_START(); + + /* need to initialize copied instance for internal references to be correct */ + dml2_initialize_instance(&dst_dml_ctx->v21.dml_init); + + DC_FP_END(); +} + +bool dml21_create_copy(struct dml2_context **dst_dml_ctx, + struct dml2_context *src_dml_ctx) +{ + /* Allocate memory for initializing DML21 instance */ + if (!dml21_allocate_memory(dst_dml_ctx)) + return false; + + dml21_copy(*dst_dml_ctx, src_dml_ctx); + + return true; +} + +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config) +{ + dml21_init(in_dc, dml_ctx, config); +} + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h new file mode 100644 index 000000000000..15f92029d2e5 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h @@ -0,0 +1,135 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + + +#ifndef _DML21_WRAPPER_H_ +#define _DML21_WRAPPER_H_ + +#include "os_types.h" +#include "dml_top_soc_parameter_types.h" +#include "dml_top_display_cfg_types.h" + +struct dc; +struct dc_state; +struct dml2_configuration_options; +struct dml2_context; +enum dc_validate_mode; + +/** + * dml2_create - Creates dml21_context. + * @in_dc: dc. + * @dml2: Created dml21 context. + * @config: dml21 configuration options. + * + * Create of DML21 is done as part of dc_state creation. + * DML21 IP, SOC and STATES are initialized at + * creation time. + * + * Return: True if dml2 is successfully created, false otherwise. + */ +bool dml21_create(const struct dc *in_dc, struct dml2_context **dml_ctx, const struct dml2_configuration_options *config); +void dml21_destroy(struct dml2_context *dml2); +void dml21_copy(struct dml2_context *dst_dml_ctx, + struct dml2_context *src_dml_ctx); +bool dml21_create_copy(struct dml2_context **dst_dml_ctx, + struct dml2_context *src_dml_ctx); +void dml21_reinit(const struct dc *in_dc, struct dml2_context *dml_ctx, const struct dml2_configuration_options *config); + +/** + * dml21_validate - Determines if a display configuration is supported or not. + * @in_dc: dc. + * @context: dc_state to be validated. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX + * will not populate context.res_ctx. + * + * Based on fast_validate option internally would call: + * + * -dml21_mode_check_and_programming - for DC_VALIDATE_MODE_AND_PROGRAMMING option + * Calculates if dc_state can be supported on the input display + * configuration. If supported, generates the necessary HW + * programming for the new dc_state. + * + * -dml21_check_mode_support - for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX option + * Calculates if dc_state can be supported for the input display + * config. + + * Context: Two threads may not invoke this function concurrently unless they reference + * separate dc_states for validation. + * Return: True if mode is supported, false otherwise. + */ +bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx, + enum dc_validate_mode validate_mode); + +/* Prepare hubp mcache_regs for hubp mcache ID and split coordinate programming */ +void dml21_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml_ctx); + +/* Structure for inputting external SOCBB and DCNIP values for tool based debugging. */ +struct socbb_ip_params_external { + struct dml2_ip_capabilities ip_params; + struct dml2_soc_bb soc_bb; +}; + +/*mcache parameters decided by dml*/ +struct dc_mcache_params { + bool valid; + /* + * For iMALL, dedicated mall mcaches are required (sharing of last + * slice possible), for legacy phantom or phantom without return + * the only mall mcaches need to be valid. + */ + bool requires_dedicated_mall_mcache; + unsigned int num_mcaches_plane0; + unsigned int num_mcaches_plane1; + /* + * Generally, plane0/1 slices must use a disjoint set of caches + * but in some cases the final segement of the two planes can + * use the same cache. If plane0_plane1 is set, then this is + * allowed. + * + * Similarly, the caches allocated to MALL prefetcher are generally + * disjoint, but if mall_prefetch is set, then the final segment + * between the main and the mall pixel requestor can use the same + * cache. + * + * Note that both bits may be set at the same time. + */ + struct { + bool mall_comb_mcache_p0; + bool mall_comb_mcache_p1; + bool plane0_plane1; + } last_slice_sharing; + /* + * A plane is divided into vertical slices of mcaches, + * which wrap on the surface width. + * + * For example, if the surface width is 7680, and split into + * three slices of equal width, the boundary array would contain + * [2560, 5120, 7680] + * + * The assignments are + * 0 = [0 .. 2559] + * 1 = [2560 .. 5119] + * 2 = [5120 .. 7679] + * 0 = [7680 .. INF] + * The final element implicitly is the same as the first, and + * at first seems invalid since it is never referenced (since) + * it is outside the surface. However, its useful when shifting + * (see below). + * + * For any given valid mcache assignment, a shifted version, wrapped + * on the surface width boundary is also assumed to be valid. + * + * For example, shifting [2560, 5120, 7680] by -50 results in + * [2510, 5170, 7630]. + * + * The assignments are now: + * 0 = [0 .. 2509] + * 1 = [2510 .. 5169] + * 2 = [5170 .. 7629] + * 0 = [7630 .. INF] + */ + int mcache_x_offsets_plane0[DML2_MAX_MCACHES + 1]; + int mcache_x_offsets_plane1[DML2_MAX_MCACHES + 1]; +}; +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h new file mode 100644 index 000000000000..16a4f97bca4e --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h @@ -0,0 +1,372 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_DML_DCN4_SOC_BB__ +#define __DML_DML_DCN4_SOC_BB__ + +#include "dml_top_soc_parameter_types.h" + +static const struct dml2_soc_qos_parameters dml_dcn4_variant_a_soc_qos_params = { + .derate_table = { + .system_active_urgent = { + .dram_derate_percent_pixel = 22, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 76, + .dcfclk_derate_percent = 100, + }, + .system_active_average = { + .dram_derate_percent_pixel = 17, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 57, + .dcfclk_derate_percent = 75, + }, + .dcn_mall_prefetch_urgent = { + .dram_derate_percent_pixel = 40, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 83, + .dcfclk_derate_percent = 100, + }, + .dcn_mall_prefetch_average = { + .dram_derate_percent_pixel = 33, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 62, + .dcfclk_derate_percent = 83, + }, + .system_idle_average = { + .dram_derate_percent_pixel = 70, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 83, + .dcfclk_derate_percent = 100, + }, + }, + .writeback = { + .base_latency_us = 12, + .scaling_factor_us = 0, + .scaling_factor_mhz = 0, + }, + .qos_params = { + .dcn4x = { + .df_qos_response_time_fclk_cycles = 300, + .max_round_trip_to_furthest_cs_fclk_cycles = 350, + .mall_overhead_fclk_cycles = 50, + .meta_trip_adder_fclk_cycles = 36, + .average_transport_distance_fclk_cycles = 257, + .umc_urgent_ramp_latency_margin = 50, + .umc_max_latency_margin = 30, + .umc_average_latency_margin = 20, + .fabric_max_transport_latency_margin = 20, + .fabric_average_transport_latency_margin = 10, + + .per_uclk_dpm_params = { + { + .minimum_uclk_khz = 97 * 1000, + .urgent_ramp_uclk_cycles = 472, + .trip_to_memory_uclk_cycles = 827, + .meta_trip_to_memory_uclk_cycles = 827, + .maximum_latency_when_urgent_uclk_cycles = 72, + .average_latency_when_urgent_uclk_cycles = 61, + .maximum_latency_when_non_urgent_uclk_cycles = 827, + .average_latency_when_non_urgent_uclk_cycles = 118, + }, + }, + }, + }, + .qos_type = dml2_qos_param_type_dcn4x, +}; + +static const struct dml2_soc_bb dml2_socbb_dcn401 = { + .clk_table = { + .uclk = { + .clk_values_khz = {97000}, + .num_clk_values = 1, + }, + .fclk = { + .clk_values_khz = {300000, 2500000}, + .num_clk_values = 2, + }, + .dcfclk = { + .clk_values_khz = {200000, 1564000}, + .num_clk_values = 2, + }, + .dispclk = { + .clk_values_khz = {100000, 2000000}, + .num_clk_values = 2, + }, + .dppclk = { + .clk_values_khz = {100000, 2000000}, + .num_clk_values = 2, + }, + .dtbclk = { + .clk_values_khz = {100000, 1564000}, + .num_clk_values = 2, + }, + .phyclk = { + .clk_values_khz = {810000, 810000}, + .num_clk_values = 2, + }, + .socclk = { + .clk_values_khz = {300000, 1200000}, + .num_clk_values = 2, + }, + .dscclk = { + .clk_values_khz = {666667, 666667}, + .num_clk_values = 2, + }, + .phyclk_d18 = { + .clk_values_khz = {625000, 625000}, + .num_clk_values = 2, + }, + .phyclk_d32 = { + .clk_values_khz = {625000, 625000}, + .num_clk_values = 2, + }, + .dram_config = { + .channel_width_bytes = 2, + .channel_count = 16, + .transactions_per_clock = 16, + }, + }, + + .qos_parameters = { + .derate_table = { + .system_active_urgent = { + .dram_derate_percent_pixel = 22, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 76, + .dcfclk_derate_percent = 100, + }, + .system_active_average = { + .dram_derate_percent_pixel = 15, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 57, + .dcfclk_derate_percent = 75, + }, + .dcn_mall_prefetch_urgent = { + .dram_derate_percent_pixel = 40, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 83, + .dcfclk_derate_percent = 100, + }, + .dcn_mall_prefetch_average = { + .dram_derate_percent_pixel = 30, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 62, + .dcfclk_derate_percent = 83, + }, + .system_idle_average = { + .dram_derate_percent_pixel = 70, + .dram_derate_percent_vm = 0, + .dram_derate_percent_pixel_and_vm = 0, + .fclk_derate_percent = 83, + .dcfclk_derate_percent = 100, + }, + }, + .writeback = { + .base_latency_us = 0, + .scaling_factor_us = 0, + .scaling_factor_mhz = 0, + }, + .qos_params = { + .dcn4x = { + .df_qos_response_time_fclk_cycles = 300, + .max_round_trip_to_furthest_cs_fclk_cycles = 350, + .mall_overhead_fclk_cycles = 50, + .meta_trip_adder_fclk_cycles = 36, + .average_transport_distance_fclk_cycles = 260, + .umc_urgent_ramp_latency_margin = 50, + .umc_max_latency_margin = 30, + .umc_average_latency_margin = 20, + .fabric_max_transport_latency_margin = 20, + .fabric_average_transport_latency_margin = 10, + + .per_uclk_dpm_params = { + { + // State 1 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 472, + .trip_to_memory_uclk_cycles = 827, + .meta_trip_to_memory_uclk_cycles = 827, + .maximum_latency_when_urgent_uclk_cycles = 72, + .average_latency_when_urgent_uclk_cycles = 72, + .maximum_latency_when_non_urgent_uclk_cycles = 827, + .average_latency_when_non_urgent_uclk_cycles = 117, + }, + { + // State 2 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 546, + .trip_to_memory_uclk_cycles = 848, + .meta_trip_to_memory_uclk_cycles = 848, + .maximum_latency_when_urgent_uclk_cycles = 146, + .average_latency_when_urgent_uclk_cycles = 146, + .maximum_latency_when_non_urgent_uclk_cycles = 848, + .average_latency_when_non_urgent_uclk_cycles = 133, + }, + { + // State 3 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 564, + .trip_to_memory_uclk_cycles = 853, + .meta_trip_to_memory_uclk_cycles = 853, + .maximum_latency_when_urgent_uclk_cycles = 164, + .average_latency_when_urgent_uclk_cycles = 164, + .maximum_latency_when_non_urgent_uclk_cycles = 853, + .average_latency_when_non_urgent_uclk_cycles = 136, + }, + { + // State 4 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 613, + .trip_to_memory_uclk_cycles = 869, + .meta_trip_to_memory_uclk_cycles = 869, + .maximum_latency_when_urgent_uclk_cycles = 213, + .average_latency_when_urgent_uclk_cycles = 213, + .maximum_latency_when_non_urgent_uclk_cycles = 869, + .average_latency_when_non_urgent_uclk_cycles = 149, + }, + { + // State 5 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 632, + .trip_to_memory_uclk_cycles = 874, + .meta_trip_to_memory_uclk_cycles = 874, + .maximum_latency_when_urgent_uclk_cycles = 232, + .average_latency_when_urgent_uclk_cycles = 232, + .maximum_latency_when_non_urgent_uclk_cycles = 874, + .average_latency_when_non_urgent_uclk_cycles = 153, + }, + { + // State 6 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 665, + .trip_to_memory_uclk_cycles = 885, + .meta_trip_to_memory_uclk_cycles = 885, + .maximum_latency_when_urgent_uclk_cycles = 265, + .average_latency_when_urgent_uclk_cycles = 265, + .maximum_latency_when_non_urgent_uclk_cycles = 885, + .average_latency_when_non_urgent_uclk_cycles = 161, + }, + { + // State 7 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 689, + .trip_to_memory_uclk_cycles = 895, + .meta_trip_to_memory_uclk_cycles = 895, + .maximum_latency_when_urgent_uclk_cycles = 289, + .average_latency_when_urgent_uclk_cycles = 289, + .maximum_latency_when_non_urgent_uclk_cycles = 895, + .average_latency_when_non_urgent_uclk_cycles = 167, + }, + { + // State 8 + .minimum_uclk_khz = 0, + .urgent_ramp_uclk_cycles = 716, + .trip_to_memory_uclk_cycles = 902, + .meta_trip_to_memory_uclk_cycles = 902, + .maximum_latency_when_urgent_uclk_cycles = 316, + .average_latency_when_urgent_uclk_cycles = 316, + .maximum_latency_when_non_urgent_uclk_cycles = 902, + .average_latency_when_non_urgent_uclk_cycles = 174, + }, + }, + }, + }, + .qos_type = dml2_qos_param_type_dcn4x, + }, + + .power_management_parameters = { + .dram_clk_change_blackout_us = 400, + .fclk_change_blackout_us = 0, + .g7_ppt_blackout_us = 0, + .stutter_enter_plus_exit_latency_us = 54, + .stutter_exit_latency_us = 41, + .z8_stutter_enter_plus_exit_latency_us = 0, + .z8_stutter_exit_latency_us = 0, + /* + .g6_temp_read_blackout_us = { + 23.00, + 10.00, + 10.00, + 8.00, + 8.00, + 5.00, + 5.00, + 5.00, + }, + */ + }, + + .vmin_limit = { + .dispclk_khz = 600 * 1000, + }, + + .dprefclk_mhz = 720, + .xtalclk_mhz = 100, + .pcie_refclk_mhz = 100, + .dchub_refclk_mhz = 50, + .mall_allocated_for_dcn_mbytes = 64, + .max_outstanding_reqs = 512, + .fabric_datapath_to_dcn_data_return_bytes = 64, + .return_bus_width_bytes = 64, + .hostvm_min_page_size_kbytes = 0, + .gpuvm_min_page_size_kbytes = 256, + .phy_downspread_percent = 0.38, + .dcn_downspread_percent = 0.38, + .dispclk_dppclk_vco_speed_mhz = 4500, + .do_urgent_latency_adjustment = 0, + .mem_word_bytes = 32, + .num_dcc_mcaches = 8, + .mcache_size_bytes = 2048, + .mcache_line_size_bytes = 32, + .max_fclk_for_uclk_dpm_khz = 1250 * 1000, +}; + +static const struct dml2_ip_capabilities dml2_dcn401_max_ip_caps = { + .pipe_count = 4, + .otg_count = 4, + .num_dsc = 4, + .max_num_dp2p0_streams = 4, + .max_num_hdmi_frl_outputs = 1, + .max_num_dp2p0_outputs = 4, + .rob_buffer_size_kbytes = 192, + .config_return_buffer_size_in_kbytes = 1344, + .config_return_buffer_segment_size_in_kbytes = 64, + .meta_fifo_size_in_kentries = 22, + .compressed_buffer_segment_size_in_kbytes = 64, + .cursor_buffer_size = 24, + .max_flip_time_us = 80, + .max_flip_time_lines = 32, + .hostvm_mode = 0, + .subvp_drr_scheduling_margin_us = 100, + .subvp_prefetch_end_to_mall_start_us = 15, + .subvp_fw_processing_delay = 15, + .max_vactive_det_fill_delay_us = 400, + + .fams2 = { + .max_allow_delay_us = 100 * 1000, + .scheduling_delay_us = 550, + .vertical_interrupt_ack_delay_us = 40, + .allow_programming_delay_us = 18, + .min_allow_width_us = 20, + .subvp_df_throttle_delay_us = 100, + .subvp_programming_delay_us = 200, + .subvp_prefetch_to_mall_delay_us = 18, + .drr_programming_delay_us = 35, + + .lock_timeout_us = 5000, + .recovery_timeout_us = 5000, + .flip_programming_delay_us = 300, + }, +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h new file mode 100644 index 000000000000..281d7ad230d8 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h @@ -0,0 +1,10 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_EXTERNAL_LIB_DEPS__ +#define __DML2_EXTERNAL_LIB_DEPS__ + +#include "os_types.h" + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h new file mode 100644 index 000000000000..a64ec4dcf11a --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_TOP_H__ +#define __DML_TOP_H__ + +#include "dml_top_types.h" + +/* + * Top Level Interface for DML2 + */ + +/* + * Returns the size of the DML instance for the caller to allocate + */ +unsigned int dml2_get_instance_size_bytes(void); + +/* + * Initializes the DML instance (i.e. with configuration, soc BB, IP params, etc...) + */ +bool dml2_initialize_instance(struct dml2_initialize_instance_in_out *in_out); + +/* + * Determines if the input mode is supported (boolean) on the SoC at all. Does not return + * information on how mode should be programmed. + */ +bool dml2_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out); + +/* + * Determines the full (optimized) programming for the input mode. Returns minimum + * clocks as well as dchub register programming values for all pipes, additional meta + * such as ODM or MPCC combine factors. + */ +bool dml2_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out); + +/* + * Determines the correct per pipe mcache register programming for a valid mode. + * The mcache allocation must have been calculated (successfully) in a previous + * call to dml2_build_mode_programming. + * The actual hubp viewport dimensions be what the actual registers will be + * programmed to (i.e. based on scaler setup). + */ +bool dml2_build_mcache_programming(struct dml2_build_mcache_programming_in_out *in_out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h new file mode 100644 index 000000000000..8e5a30287220 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h @@ -0,0 +1,189 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __dml2_TOP_DCHUB_REGISTERS_H__ +#define __dml2_TOP_DCHUB_REGISTERS_H__ + +#include "dml2_external_lib_deps.h" +// These types are uint32_t as they represent actual calculated register values for HW + +struct dml2_display_dlg_regs { + uint32_t refcyc_h_blank_end; + uint32_t dlg_vblank_end; + uint32_t min_dst_y_next_start; + uint32_t refcyc_per_htotal; + uint32_t refcyc_x_after_scaler; + uint32_t dst_y_after_scaler; + uint32_t dst_y_prefetch; + uint32_t dst_y_per_vm_vblank; + uint32_t dst_y_per_row_vblank; + uint32_t dst_y_per_vm_flip; + uint32_t dst_y_per_row_flip; + uint32_t ref_freq_to_pix_freq; + uint32_t vratio_prefetch; + uint32_t vratio_prefetch_c; + uint32_t refcyc_per_tdlut_group; + uint32_t refcyc_per_pte_group_vblank_l; + uint32_t refcyc_per_pte_group_vblank_c; + uint32_t refcyc_per_pte_group_flip_l; + uint32_t refcyc_per_pte_group_flip_c; + uint32_t dst_y_per_pte_row_nom_l; + uint32_t dst_y_per_pte_row_nom_c; + uint32_t refcyc_per_pte_group_nom_l; + uint32_t refcyc_per_pte_group_nom_c; + uint32_t refcyc_per_line_delivery_pre_l; + uint32_t refcyc_per_line_delivery_pre_c; + uint32_t refcyc_per_line_delivery_l; + uint32_t refcyc_per_line_delivery_c; + uint32_t refcyc_per_vm_group_vblank; + uint32_t refcyc_per_vm_group_flip; + uint32_t refcyc_per_vm_req_vblank; + uint32_t refcyc_per_vm_req_flip; + uint32_t dst_y_offset_cur0; + uint32_t chunk_hdl_adjust_cur0; + uint32_t vready_after_vcount0; + uint32_t dst_y_delta_drq_limit; + uint32_t refcyc_per_vm_dmdata; + uint32_t dmdata_dl_delta; + + // MRQ + uint32_t refcyc_per_meta_chunk_vblank_l; + uint32_t refcyc_per_meta_chunk_vblank_c; + uint32_t refcyc_per_meta_chunk_flip_l; + uint32_t refcyc_per_meta_chunk_flip_c; + uint32_t dst_y_per_meta_row_nom_l; + uint32_t dst_y_per_meta_row_nom_c; + uint32_t refcyc_per_meta_chunk_nom_l; + uint32_t refcyc_per_meta_chunk_nom_c; +}; + +struct dml2_display_ttu_regs { + uint32_t qos_level_low_wm; + uint32_t qos_level_high_wm; + uint32_t min_ttu_vblank; + uint32_t qos_level_flip; + uint32_t refcyc_per_req_delivery_l; + uint32_t refcyc_per_req_delivery_c; + uint32_t refcyc_per_req_delivery_cur0; + uint32_t refcyc_per_req_delivery_pre_l; + uint32_t refcyc_per_req_delivery_pre_c; + uint32_t refcyc_per_req_delivery_pre_cur0; + uint32_t qos_level_fixed_l; + uint32_t qos_level_fixed_c; + uint32_t qos_level_fixed_cur0; + uint32_t qos_ramp_disable_l; + uint32_t qos_ramp_disable_c; + uint32_t qos_ramp_disable_cur0; +}; + +struct dml2_display_arb_regs { + uint32_t max_req_outstanding; + uint32_t min_req_outstanding; + uint32_t sat_level_us; + uint32_t hvm_max_qos_commit_threshold; + uint32_t hvm_min_req_outstand_commit_threshold; + uint32_t compbuf_reserved_space_kbytes; + uint32_t compbuf_size; + uint32_t sdpif_request_rate_limit; + uint32_t allow_sdpif_rate_limit_when_cstate_req; + uint32_t dcfclk_deep_sleep_hysteresis; + uint32_t pstate_stall_threshold; +}; + +struct dml2_cursor_dlg_regs{ + uint32_t dst_x_offset; // CURSOR0_DST_X_OFFSET + uint32_t dst_y_offset; // CURSOR0_DST_Y_OFFSET + uint32_t chunk_hdl_adjust; // CURSOR0_CHUNK_HDL_ADJUST + + uint32_t qos_level_fixed; + uint32_t qos_ramp_disable; +}; + +struct dml2_display_plane_rq_regs { + uint32_t chunk_size; + uint32_t min_chunk_size; + uint32_t dpte_group_size; + uint32_t mpte_group_size; + uint32_t swath_height; + uint32_t pte_row_height_linear; + + // MRQ + uint32_t meta_chunk_size; + uint32_t min_meta_chunk_size; +}; + +struct dml2_display_rq_regs { + struct dml2_display_plane_rq_regs rq_regs_l; + struct dml2_display_plane_rq_regs rq_regs_c; + uint32_t drq_expansion_mode; + uint32_t prq_expansion_mode; + uint32_t crq_expansion_mode; + uint32_t plane1_base_address; + uint32_t unbounded_request_enabled; + + // MRQ + uint32_t mrq_expansion_mode; +}; + +struct dml2_display_mcache_regs { + uint32_t mcache_id_first; + uint32_t mcache_id_second; + uint32_t split_location; +}; + +struct dml2_hubp_pipe_mcache_regs { + struct { + struct dml2_display_mcache_regs p0; + struct dml2_display_mcache_regs p1; + } main; + struct { + struct dml2_display_mcache_regs p0; + struct dml2_display_mcache_regs p1; + } mall; +}; + +struct dml2_dchub_per_pipe_register_set { + struct dml2_display_rq_regs rq_regs; + struct dml2_display_ttu_regs ttu_regs; + struct dml2_display_dlg_regs dlg_regs; + + uint32_t det_size; +}; + +struct dml2_dchub_watermark_regs { + /* watermarks */ + uint32_t urgent; + uint32_t sr_enter; + uint32_t sr_exit; + uint32_t sr_enter_z8; + uint32_t sr_exit_z8; + uint32_t sr_enter_low_power; + uint32_t sr_exit_low_power; + uint32_t uclk_pstate; + uint32_t fclk_pstate; + uint32_t temp_read_or_ppt; + uint32_t usr; + /* qos */ + uint32_t refcyc_per_trip_to_mem; + uint32_t refcyc_per_meta_trip_to_mem; + uint32_t frac_urg_bw_flip; + uint32_t frac_urg_bw_nom; + uint32_t frac_urg_bw_mall; +}; + +enum dml2_dchub_watermark_reg_set_index { + DML2_DCHUB_WATERMARK_SET_A = 0, + DML2_DCHUB_WATERMARK_SET_B = 1, + DML2_DCHUB_WATERMARK_SET_C = 2, + DML2_DCHUB_WATERMARK_SET_D = 3, + DML2_DCHUB_WATERMARK_SET_NUM = 4, +}; + +struct dml2_dchub_global_register_set { + struct dml2_display_arb_regs arb_regs; + struct dml2_dchub_watermark_regs wm_regs[DML2_DCHUB_WATERMARK_SET_NUM]; + unsigned int num_watermark_sets; +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h new file mode 100644 index 000000000000..13749c9fcf18 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h @@ -0,0 +1,520 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_TOP_DISPLAY_CFG_TYPES_H__ +#define __DML_TOP_DISPLAY_CFG_TYPES_H__ + +#include "dml2_external_lib_deps.h" + +#define DML2_MAX_PLANES 8 +#define DML2_MAX_DCN_PIPES 8 +#define DML2_MAX_MCACHES 8 // assume plane is going to be supported by a max of 8 mcaches +#define DML2_MAX_WRITEBACK 3 + +enum dml2_swizzle_mode { + dml2_sw_linear, // SW_LINEAR accepts 256 byte aligned pitch and also 128 byte aligned pitch if DCC is not enabled + dml2_sw_256b_2d, + dml2_sw_4kb_2d, + dml2_sw_64kb_2d, + dml2_sw_256kb_2d, + + dml2_gfx11_sw_linear, + dml2_gfx11_sw_64kb_d, + dml2_gfx11_sw_64kb_d_t, + dml2_gfx11_sw_64kb_d_x, + dml2_gfx11_sw_64kb_r_x, + dml2_gfx11_sw_256kb_d_x, + dml2_gfx11_sw_256kb_r_x, + +}; + +enum dml2_source_format_class { + dml2_444_8 = 0, + dml2_444_16 = 1, + dml2_444_32 = 2, + dml2_444_64 = 3, + dml2_420_8 = 4, + dml2_420_10 = 5, + dml2_420_12 = 6, + dml2_rgbe_alpha = 9, + dml2_rgbe = 10, + dml2_mono_8 = 11, + dml2_mono_16 = 12, + dml2_422_planar_8 = 13, + dml2_422_planar_10 = 14, + dml2_422_planar_12 = 15, + dml2_422_packed_8 = 16, + dml2_422_packed_10 = 17, + dml2_422_packed_12 = 18 +}; + +enum dml2_sample_positioning { + dml2_interstitial = 0, + dml2_cosited = 1 +}; + +enum dml2_rotation_angle { + dml2_rotation_0 = 0, + dml2_rotation_90 = 1, + dml2_rotation_180 = 2, + dml2_rotation_270 = 3 +}; + +enum dml2_output_format_class { + dml2_444 = 0, + dml2_s422 = 1, + dml2_n422 = 2, + dml2_420 = 3 +}; + +enum dml2_output_encoder_class { + dml2_dp = 0, + dml2_edp = 1, + dml2_dp2p0 = 2, + dml2_hdmi = 3, + dml2_hdmifrl = 4, + dml2_none = 5 +}; + +enum dml2_output_link_dp_rate { + dml2_dp_rate_na = 0, + dml2_dp_rate_hbr = 1, + dml2_dp_rate_hbr2 = 2, + dml2_dp_rate_hbr3 = 3, + dml2_dp_rate_uhbr10 = 4, + dml2_dp_rate_uhbr13p5 = 5, + dml2_dp_rate_uhbr20 = 6 +}; + +enum dml2_uclk_pstate_change_strategy { + dml2_uclk_pstate_change_strategy_auto = 0, + dml2_uclk_pstate_change_strategy_force_vactive = 1, + dml2_uclk_pstate_change_strategy_force_vblank = 2, + dml2_uclk_pstate_change_strategy_force_drr = 3, + dml2_uclk_pstate_change_strategy_force_mall_svp = 4, + dml2_uclk_pstate_change_strategy_force_mall_full_frame = 5, +}; + +enum dml2_svp_mode_override { + dml2_svp_mode_override_auto = 0, + dml2_svp_mode_override_main_pipe = 1, + dml2_svp_mode_override_phantom_pipe = 2, //does not need to be defined explicitly, main overrides result in implicit phantom additions + dml2_svp_mode_override_phantom_pipe_no_data_return = 3, + dml2_svp_mode_override_imall = 4 +}; + +enum dml2_refresh_from_mall_mode_override { + dml2_refresh_from_mall_mode_override_auto = 0, + dml2_refresh_from_mall_mode_override_force_disable = 1, + dml2_refresh_from_mall_mode_override_force_enable = 2 +}; + +enum dml2_odm_mode { + dml2_odm_mode_auto = 0, + dml2_odm_mode_bypass, + dml2_odm_mode_combine_2to1, + dml2_odm_mode_combine_3to1, + dml2_odm_mode_combine_4to1, + dml2_odm_mode_split_1to2, + dml2_odm_mode_mso_1to2, + dml2_odm_mode_mso_1to4 +}; + +enum dml2_scaling_transform { + dml2_scaling_transform_explicit = 0, + dml2_scaling_transform_fullscreen, + dml2_scaling_transform_aspect_ratio, + dml2_scaling_transform_centered +}; + +enum dml2_dsc_enable_option { + dml2_dsc_disable = 0, + dml2_dsc_enable = 1, + dml2_dsc_enable_if_necessary = 2 +}; + +enum dml2_tdlut_addressing_mode { + dml2_tdlut_sw_linear = 0, + dml2_tdlut_simple_linear = 1 +}; + +enum dml2_tdlut_width_mode { + dml2_tdlut_width_17_cube = 0, + dml2_tdlut_width_33_cube = 1 +}; + +enum dml2_twait_budgeting_setting { + dml2_twait_budgeting_setting_ignore = 0,// Ignore this budget in twait + + dml2_twait_budgeting_setting_if_needed, // Budget for it only if needed + //(i.e. UCLK/FCLK DPM cannot be supported in active) + + dml2_twait_budgeting_setting_try, // Budget for it as long as there is an SoC state that + // can support it +}; + +struct dml2_get_cursor_dlg_reg{ + unsigned int cursor_x_position; + unsigned int cursor_hotspot_x; + unsigned int cursor_primary_offset; + unsigned int cursor_secondary_offset; + bool cursor_stereo_en; + bool cursor_2x_magnify; + double hratio; + double pixel_rate_mhz; + double dlg_refclk_mhz; +}; + +/// @brief Surface Parameters +struct dml2_surface_cfg { + enum dml2_swizzle_mode tiling; + + struct { + unsigned long pitch; // In elements, two pixels per element in 422 packed format + unsigned long width; + unsigned long height; + } plane0; + + + struct { + unsigned long pitch; + unsigned long width; + unsigned long height; + } plane1; + + struct { + bool enable; + struct { + unsigned long pitch; + } plane0; + struct { + unsigned long pitch; + } plane1; + + struct { + double dcc_rate_plane0; + double dcc_rate_plane1; + double fraction_of_zero_size_request_plane0; + double fraction_of_zero_size_request_plane1; + } informative; + } dcc; +}; + + +struct dml2_composition_cfg { + enum dml2_rotation_angle rotation_angle; + bool mirrored; + enum dml2_scaling_transform scaling_transform; + bool rect_out_height_spans_vactive; + + struct { + bool stationary; + struct { + unsigned long width; + unsigned long height; + unsigned long x_start; + unsigned long y_start; + } plane0; + + struct { + unsigned long width; + unsigned long height; + unsigned long x_start; + unsigned long y_start; + } plane1; + } viewport; + + struct { + bool enabled; + bool easf_enabled; + bool isharp_enabled; + bool upsp_enabled; + enum dml2_sample_positioning upsp_sample_positioning; + unsigned int upsp_vtaps; + struct { + double h_ratio; + double v_ratio; + unsigned int h_taps; + unsigned int v_taps; + } plane0; + + struct { + double h_ratio; + double v_ratio; + unsigned int h_taps; + unsigned int v_taps; + } plane1; + + unsigned long rect_out_width; + } scaler_info; +}; + +struct dml2_timing_cfg { + unsigned long h_total; + unsigned long v_total; + unsigned long h_blank_end; + unsigned long v_blank_end; + unsigned long h_front_porch; + unsigned long v_front_porch; + unsigned long h_sync_width; + unsigned long pixel_clock_khz; + unsigned long h_active; + unsigned long v_active; + unsigned int bpc; //FIXME: review with Jun + struct { + enum dml2_dsc_enable_option enable; + unsigned int dsc_compressed_bpp_x16; + struct { + // for dv to specify num dsc slices to use + unsigned int num_slices; + } overrides; + } dsc; + bool interlaced; + struct { + /* static */ + bool enabled; + unsigned long min_refresh_uhz; + unsigned int max_instant_vtotal_delta; + /* dynamic */ + bool disallowed; + bool drr_active_variable; + bool drr_active_fixed; + } drr_config; + unsigned long vblank_nom; +}; + +struct dml2_link_output_cfg { + enum dml2_output_format_class output_format; + enum dml2_output_encoder_class output_encoder; + unsigned int output_dp_lane_count; + enum dml2_output_link_dp_rate output_dp_link_rate; + unsigned long audio_sample_rate; + unsigned long audio_sample_layout; + bool output_disabled; // The stream does not go to a backend for output to a physical + //connector (e.g. writeback only, phantom pipe) goes to writeback + bool validate_output; // Do not validate the link configuration for this display stream. +}; + +struct dml2_writeback_info { + enum dml2_source_format_class pixel_format; + unsigned long input_width; + unsigned long input_height; + unsigned long output_width; + unsigned long output_height; + unsigned long v_taps; + unsigned long h_taps; + unsigned long v_taps_chroma; + unsigned long h_taps_chroma; + double h_ratio; + double v_ratio; +}; + +struct dml2_writeback_cfg { + unsigned int active_writebacks_per_stream; + struct dml2_writeback_info writeback_stream[DML2_MAX_WRITEBACK]; +}; + +struct dml2_plane_parameters { + unsigned int stream_index; // Identifies which plane will be composed + + enum dml2_source_format_class pixel_format; + /* + * The surface and composition structures use + * the terms plane0 and plane1. These planes + * are expected to hold the following data based + * on the pixel format. + * + * RGB or YUV Non-Planar Types: + * dml2_444_8 + * dml2_444_16 + * dml2_444_32 + * dml2_444_64 + * dml2_rgbe + * + * plane0 = argb or rgbe + * plane1 = not used + * + * YUV Planar-Types: + * dml2_420_8 + * dml2_420_10 + * dml2_420_12 + * + * plane0 = luma + * plane1 = chroma + * + * RGB Planar Types: + * dml2_rgbe_alpha + * + * plane0 = rgbe + * plane1 = alpha + * + * Mono Non-Planar Types: + * dml2_mono_8 + * dml2_mono_16 + * + * plane0 = luma + * plane1 = not used + */ + + struct dml2_surface_cfg surface; + struct dml2_composition_cfg composition; + + struct { + bool enable; + unsigned long lines_before_active_required; + unsigned long transmitted_bytes; + } dynamic_meta_data; + + struct { + unsigned int num_cursors; + unsigned long cursor_width; + unsigned long cursor_bpp; + } cursor; + + // For TDLUT, SW would assume TDLUT is setup and enable all the time and + // budget for worst case addressing/width mode + struct { + bool setup_for_tdlut; + enum dml2_tdlut_addressing_mode tdlut_addressing_mode; + enum dml2_tdlut_width_mode tdlut_width_mode; + bool tdlut_mpc_width_flag; + } tdlut; + + bool immediate_flip; + + struct { + // Logical overrides to power management policies (usually) + enum dml2_uclk_pstate_change_strategy uclk_pstate_change_strategy; + enum dml2_refresh_from_mall_mode_override refresh_from_mall; + unsigned int det_size_override_kb; + unsigned int mpcc_combine_factor; + + // reserved_vblank_time_ns is the minimum time to reserve in vblank for Twait + // The actual reserved vblank time used for the corresponding stream in mode_programming would be at least as much as this per-plane override. + long reserved_vblank_time_ns; + unsigned int max_vactive_det_fill_delay_us; // 0 = no reserved time, +ve = explicit max delay + unsigned int vactive_latency_to_hide_for_pstate_admissibility_us; + unsigned int gpuvm_min_page_size_kbytes; + unsigned int hostvm_min_page_size_kbytes; + + enum dml2_svp_mode_override legacy_svp_config; //TODO remove in favor of svp_config + + struct { + // HW specific overrides, there's almost no reason to mess with these + // generally used for debugging or simulation + bool force_one_row_for_frame; + struct { + bool enable; + bool value; + } force_pte_buffer_mode; + double dppclk_mhz; + } hw; + } overrides; +}; + +struct dml2_stream_parameters { + struct dml2_timing_cfg timing; + struct dml2_link_output_cfg output; + struct dml2_writeback_cfg writeback; + + struct { + enum dml2_odm_mode odm_mode; + bool disable_dynamic_odm; + bool disable_subvp; + int minimum_vblank_idle_requirement_us; + bool minimize_active_latency_hiding; + + struct { + struct { + enum dml2_twait_budgeting_setting uclk_pstate; + enum dml2_twait_budgeting_setting fclk_pstate; + enum dml2_twait_budgeting_setting stutter_enter_exit; + } twait_budgeting; + } hw; + } overrides; +}; + +struct dml2_display_cfg { + bool gpuvm_enable; + bool ffbm_enable; + bool hostvm_enable; + + // Allocate DET proportionally between streams based on pixel rate + // and then allocate proportionally between planes. + bool minimize_det_reallocation; + + unsigned int gpuvm_max_page_table_levels; + unsigned int hostvm_max_non_cached_page_table_levels; + + struct dml2_plane_parameters plane_descriptors[DML2_MAX_PLANES]; + struct dml2_stream_parameters stream_descriptors[DML2_MAX_PLANES]; + + unsigned int num_planes; + unsigned int num_streams; + + struct { + struct { + // HW specific overrides, there's almost no reason to mess with these + // generally used for debugging or simulation + struct { + bool enable; + bool value; + } force_unbounded_requesting; + + struct { + bool enable; + bool value; + } force_nom_det_size_kbytes; + + bool mode_support_check_disable; + bool mcache_admissibility_check_disable; + bool surface_viewport_size_check_disable; + double dlg_ref_clk_mhz; + double dispclk_mhz; + double dcfclk_mhz; + bool optimize_tdlut_scheduling; // TBD: for DV, will set this to 1, to ensure tdlut schedule is calculated based on address/width mode + } hw; + + struct { + bool uclk_pstate_change_disable; + bool fclk_pstate_change_disable; + bool g6_temp_read_pstate_disable; + bool g7_ppt_pstate_disable; + } power_management; + + bool enhanced_prefetch_schedule_acceleration; + bool dcc_programming_assumes_scan_direction_unknown; + bool synchronize_timings; + bool synchronize_ddr_displays_for_uclk_pstate_change; + bool max_outstanding_when_urgent_expected_disable; + bool enable_subvp_implicit_pmo; //enables PMO to switch pipe uclk strategy to subvp, and generate phantom programming + unsigned int best_effort_min_active_latency_hiding_us; + bool all_streams_blanked; + } overrides; +}; + +struct dml2_pipe_configuration_descriptor { + struct { + unsigned int viewport_x_start; + unsigned int viewport_width; + } plane0; + + struct { + unsigned int viewport_x_start; + unsigned int viewport_width; + } plane1; + + bool plane1_enabled; + bool imall_enabled; +}; + +struct dml2_plane_mcache_configuration_descriptor { + const struct dml2_plane_parameters *plane_descriptor; + const struct dml2_mcache_surface_allocation *mcache_allocation; + + struct dml2_pipe_configuration_descriptor pipe_configurations[DML2_MAX_DCN_PIPES]; + char num_pipes; +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h new file mode 100644 index 000000000000..8f624a912e78 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_TOP_POLICY_TYPES_H__ +#define __DML_TOP_POLICY_TYPES_H__ + +struct dml2_policy_parameters { + unsigned long odm_combine_dispclk_threshold_khz; + unsigned int max_immediate_flip_latency; +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h new file mode 100644 index 000000000000..4a9a0d5a09b7 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h @@ -0,0 +1,215 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_TOP_SOC_PARAMETER_TYPES_H__ +#define __DML_TOP_SOC_PARAMETER_TYPES_H__ + +#include "dml2_external_lib_deps.h" + +#define DML_MAX_CLK_TABLE_SIZE 20 + +struct dml2_soc_derate_values { + unsigned int dram_derate_percent_pixel; + unsigned int dram_derate_percent_vm; + unsigned int dram_derate_percent_pixel_and_vm; + + unsigned int fclk_derate_percent; + unsigned int dcfclk_derate_percent; +}; + +struct dml2_soc_derates { + struct dml2_soc_derate_values system_active_urgent; + struct dml2_soc_derate_values system_active_average; + struct dml2_soc_derate_values dcn_mall_prefetch_urgent; + struct dml2_soc_derate_values dcn_mall_prefetch_average; + struct dml2_soc_derate_values system_idle_average; +}; + +struct dml2_dcn32x_soc_qos_params { + struct { + unsigned int base_latency_us; + unsigned int base_latency_pixel_vm_us; + unsigned int base_latency_vm_us; + unsigned int scaling_factor_fclk_us; + unsigned int scaling_factor_mhz; + } urgent_latency_us; + + unsigned int loaded_round_trip_latency_fclk_cycles; + unsigned int urgent_out_of_order_return_per_channel_pixel_only_bytes; + unsigned int urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; + unsigned int urgent_out_of_order_return_per_channel_vm_only_bytes; +}; + +struct dml2_dcn4_uclk_dpm_dependent_qos_params { + unsigned long minimum_uclk_khz; + unsigned int urgent_ramp_uclk_cycles; + unsigned int trip_to_memory_uclk_cycles; + unsigned int meta_trip_to_memory_uclk_cycles; + unsigned int maximum_latency_when_urgent_uclk_cycles; + unsigned int average_latency_when_urgent_uclk_cycles; + unsigned int maximum_latency_when_non_urgent_uclk_cycles; + unsigned int average_latency_when_non_urgent_uclk_cycles; +}; + +struct dml2_dcn4x_soc_qos_params { + unsigned int df_qos_response_time_fclk_cycles; + unsigned int max_round_trip_to_furthest_cs_fclk_cycles; + unsigned int mall_overhead_fclk_cycles; + unsigned int meta_trip_adder_fclk_cycles; + unsigned int average_transport_distance_fclk_cycles; + double umc_urgent_ramp_latency_margin; + double umc_max_latency_margin; + double umc_average_latency_margin; + double fabric_max_transport_latency_margin; + double fabric_average_transport_latency_margin; + struct dml2_dcn4_uclk_dpm_dependent_qos_params per_uclk_dpm_params[DML_MAX_CLK_TABLE_SIZE]; +}; + +enum dml2_qos_param_type { + dml2_qos_param_type_dcn3, + dml2_qos_param_type_dcn4x +}; + +struct dml2_soc_qos_parameters { + struct dml2_soc_derates derate_table; + struct { + unsigned int base_latency_us; + unsigned int scaling_factor_us; + unsigned int scaling_factor_mhz; + } writeback; + + union { + struct dml2_dcn32x_soc_qos_params dcn32x; + struct dml2_dcn4x_soc_qos_params dcn4x; + } qos_params; + + enum dml2_qos_param_type qos_type; +}; + +struct dml2_soc_power_management_parameters { + double dram_clk_change_blackout_us; + double dram_clk_change_read_only_us; + double dram_clk_change_write_only_us; + double fclk_change_blackout_us; + double g7_ppt_blackout_us; + double g7_temperature_read_blackout_us; + double stutter_enter_plus_exit_latency_us; + double stutter_exit_latency_us; + double low_power_stutter_enter_plus_exit_latency_us; + double low_power_stutter_exit_latency_us; + double z8_stutter_enter_plus_exit_latency_us; + double z8_stutter_exit_latency_us; + double z8_min_idle_time; + double g6_temp_read_blackout_us[DML_MAX_CLK_TABLE_SIZE]; + double type_b_dram_clk_change_blackout_us; + double type_b_ppt_blackout_us; +}; + +struct dml2_clk_table { + unsigned long clk_values_khz[DML_MAX_CLK_TABLE_SIZE]; + unsigned char num_clk_values; +}; + +struct dml2_dram_params { + unsigned int channel_width_bytes; + unsigned int channel_count; + unsigned int transactions_per_clock; +}; + +struct dml2_soc_state_table { + struct dml2_clk_table uclk; + struct dml2_clk_table fclk; + struct dml2_clk_table dcfclk; + struct dml2_clk_table dispclk; + struct dml2_clk_table dppclk; + struct dml2_clk_table dtbclk; + struct dml2_clk_table phyclk; + struct dml2_clk_table socclk; + struct dml2_clk_table dscclk; + struct dml2_clk_table phyclk_d18; + struct dml2_clk_table phyclk_d32; + + struct dml2_dram_params dram_config; +}; + +struct dml2_soc_vmin_clock_limits { + unsigned long dispclk_khz; + unsigned long dcfclk_khz; +}; + +struct dml2_soc_bb { + struct dml2_soc_state_table clk_table; + struct dml2_soc_qos_parameters qos_parameters; + struct dml2_soc_power_management_parameters power_management_parameters; + struct dml2_soc_vmin_clock_limits vmin_limit; + + double lower_bound_bandwidth_dchub; + double fraction_of_urgent_bandwidth_nominal_target; + double fraction_of_urgent_bandwidth_flip_target; + unsigned int dprefclk_mhz; + unsigned int xtalclk_mhz; + unsigned int pcie_refclk_mhz; + unsigned int dchub_refclk_mhz; + unsigned int mall_allocated_for_dcn_mbytes; + unsigned int max_outstanding_reqs; + unsigned long fabric_datapath_to_dcn_data_return_bytes; + unsigned long return_bus_width_bytes; + unsigned long hostvm_min_page_size_kbytes; + unsigned long gpuvm_min_page_size_kbytes; + double phy_downspread_percent; + double dcn_downspread_percent; + double dispclk_dppclk_vco_speed_mhz; + bool no_dfs; + bool do_urgent_latency_adjustment; + unsigned int mem_word_bytes; + unsigned int num_dcc_mcaches; + unsigned int mcache_size_bytes; + unsigned int mcache_line_size_bytes; + unsigned long max_fclk_for_uclk_dpm_khz; +}; + +struct dml2_ip_capabilities { + unsigned int pipe_count; + unsigned int otg_count; + unsigned int TDLUT_33cube_count; + unsigned int num_dsc; + unsigned int max_num_dp2p0_streams; + unsigned int max_num_hdmi_frl_outputs; + unsigned int max_num_dp2p0_outputs; + unsigned int max_num_wb; + unsigned int rob_buffer_size_kbytes; + unsigned int config_return_buffer_size_in_kbytes; + unsigned int config_return_buffer_segment_size_in_kbytes; + unsigned int meta_fifo_size_in_kentries; + unsigned int compressed_buffer_segment_size_in_kbytes; + unsigned int cursor_buffer_size; + unsigned int max_flip_time_us; + unsigned int max_flip_time_lines; + unsigned int hostvm_mode; + unsigned int subvp_drr_scheduling_margin_us; + unsigned int subvp_prefetch_end_to_mall_start_us; + unsigned int subvp_fw_processing_delay; + unsigned int max_vactive_det_fill_delay_us; + unsigned int ppt_max_allow_delay_ns; + unsigned int temp_read_max_allow_delay_us; + unsigned int dummy_pstate_max_allow_delay_us; + /* FAMS2 delays */ + struct { + unsigned int max_allow_delay_us; + unsigned int scheduling_delay_us; + unsigned int vertical_interrupt_ack_delay_us; // delay to acknowledge vline int + unsigned int allow_programming_delay_us; // time requires to program allow + unsigned int min_allow_width_us; + unsigned int subvp_df_throttle_delay_us; + unsigned int subvp_programming_delay_us; + unsigned int subvp_prefetch_to_mall_delay_us; + unsigned int drr_programming_delay_us; + + unsigned int lock_timeout_us; + unsigned int recovery_timeout_us; + unsigned int flip_programming_delay_us; + } fams2; +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h new file mode 100644 index 000000000000..8646ce5f1c01 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h @@ -0,0 +1,750 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML_TOP_TYPES_H__ +#define __DML_TOP_TYPES_H__ + +#include "dml_top_display_cfg_types.h" +#include "dml_top_soc_parameter_types.h" +#include "dml_top_policy_types.h" +#include "dml_top_dchub_registers.h" + +#include "dmub_cmd.h" + +struct dml2_instance; + +enum dml2_project_id { + dml2_project_invalid = 0, + dml2_project_dcn4x_stage1, + dml2_project_dcn4x_stage2, + dml2_project_dcn4x_stage2_auto_drr_svp, +}; + +enum dml2_pstate_change_support { + dml2_pstate_change_vactive = 0, + dml2_pstate_change_vblank = 1, + dml2_pstate_change_vblank_and_vactive = 2, + dml2_pstate_change_drr = 3, + dml2_pstate_change_mall_svp = 4, + dml2_pstate_change_mall_full_frame = 6, + dml2_pstate_change_unsupported = 7 +}; + +enum dml2_output_type_and_rate__type { + dml2_output_type_unknown = 0, + dml2_output_type_dp = 1, + dml2_output_type_edp = 2, + dml2_output_type_dp2p0 = 3, + dml2_output_type_hdmi = 4, + dml2_output_type_hdmifrl = 5 +}; + +enum dml2_output_type_and_rate__rate { + dml2_output_rate_unknown = 0, + dml2_output_rate_dp_rate_hbr = 1, + dml2_output_rate_dp_rate_hbr2 = 2, + dml2_output_rate_dp_rate_hbr3 = 3, + dml2_output_rate_dp_rate_uhbr10 = 4, + dml2_output_rate_dp_rate_uhbr13p5 = 5, + dml2_output_rate_dp_rate_uhbr20 = 6, + dml2_output_rate_hdmi_rate_3x3 = 7, + dml2_output_rate_hdmi_rate_6x3 = 8, + dml2_output_rate_hdmi_rate_6x4 = 9, + dml2_output_rate_hdmi_rate_8x4 = 10, + dml2_output_rate_hdmi_rate_10x4 = 11, + dml2_output_rate_hdmi_rate_12x4 = 12, + dml2_output_rate_hdmi_rate_16x4 = 13, + dml2_output_rate_hdmi_rate_20x4 = 14 +}; + +struct dml2_pmo_options { + bool disable_vblank; + bool disable_svp; + bool disable_drr_var; + bool disable_drr_clamped; + bool disable_drr_var_when_var_active; + bool disable_drr_clamped_when_var_active; + bool disable_fams2; + bool disable_vactive_det_fill_bw_pad; /* dml2_project_dcn4x_stage2_auto_drr_svp and above only */ + bool disable_dyn_odm; + bool disable_dyn_odm_for_multi_stream; + bool disable_dyn_odm_for_stream_with_svp; + struct dml2_pmo_pstate_strategy *override_strategy_lists[DML2_MAX_PLANES]; + unsigned int num_override_strategies_per_list[DML2_MAX_PLANES]; +}; + +struct dml2_options { + enum dml2_project_id project_id; + struct dml2_pmo_options pmo_options; +}; + +struct dml2_initialize_instance_in_out { + struct dml2_instance *dml2_instance; + struct dml2_options options; + struct dml2_soc_bb soc_bb; + struct dml2_ip_capabilities ip_caps; + + struct { + void *explicit_ip_bb; + unsigned int explicit_ip_bb_size; + } overrides; +}; + +struct dml2_reset_instance_in_out { + struct dml2_instance *dml2_instance; +}; + +struct dml2_check_mode_supported_in_out { + /* + * Inputs + */ + struct dml2_instance *dml2_instance; + const struct dml2_display_cfg *display_config; + + /* + * Outputs + */ + bool is_supported; +}; + +struct dml2_mcache_surface_allocation { + bool valid; + /* + * For iMALL, dedicated mall mcaches are required (sharing of last + * slice possible), for legacy phantom or phantom without return + * the only mall mcaches need to be valid. + */ + bool requires_dedicated_mall_mcache; + + unsigned int num_mcaches_plane0; + unsigned int num_mcaches_plane1; + /* + * A plane is divided into vertical slices of mcaches, + * which wrap on the surface width. + * + * For example, if the surface width is 7680, and split into + * three slices of equal width, the boundary array would contain + * [2560, 5120, 7680] + * + * The assignments are + * 0 = [0 .. 2559] + * 1 = [2560 .. 5119] + * 2 = [5120 .. 7679] + * 0 = [7680 .. INF] + * The final element implicitly is the same as the first, and + * at first seems invalid since it is never referenced (since) + * it is outside the surface. However, its useful when shifting + * (see below). + * + * For any given valid mcache assignment, a shifted version, wrapped + * on the surface width boundary is also assumed to be valid. + * + * For example, shifting [2560, 5120, 7680] by -50 results in + * [2510, 5170, 7630]. + * + * The assignments are now: + * 0 = [0 .. 2509] + * 1 = [2510 .. 5169] + * 2 = [5170 .. 7629] + * 0 = [7630 .. INF] + */ + int mcache_x_offsets_plane0[DML2_MAX_MCACHES + 1]; + int mcache_x_offsets_plane1[DML2_MAX_MCACHES + 1]; + + /* + * Shift grainularity is not necessarily 1 + */ + struct { + int p0; + int p1; + } shift_granularity; + + /* + * MCacheIDs have global scope in the SoC, and they are stored here. + * These IDs are generally not valid until all planes in a display + * configuration have had their mcache requirements calculated. + */ + int global_mcache_ids_plane0[DML2_MAX_MCACHES + 1]; + int global_mcache_ids_plane1[DML2_MAX_MCACHES + 1]; + int global_mcache_ids_mall_plane0[DML2_MAX_MCACHES + 1]; + int global_mcache_ids_mall_plane1[DML2_MAX_MCACHES + 1]; + + /* + * Generally, plane0/1 slices must use a disjoint set of caches + * but in some cases the final segement of the two planes can + * use the same cache. If plane0_plane1 is set, then this is + * allowed. + * + * Similarly, the caches allocated to MALL prefetcher are generally + * disjoint, but if mall_prefetch is set, then the final segment + * between the main and the mall pixel requestor can use the same + * cache. + * + * Note that both bits may be set at the same time. + */ + struct { + bool mall_comb_mcache_p0; + bool mall_comb_mcache_p1; + bool plane0_plane1; + } last_slice_sharing; + + struct { + int meta_row_bytes_plane0; + int meta_row_bytes_plane1; + } informative; +}; + +enum dml2_pstate_type { + dml2_pstate_type_uclk, + dml2_pstate_type_ppt, + dml2_pstate_type_temp_read, + dml2_pstate_type_dummy_pstate, + dml2_pstate_type_count +}; + +enum dml2_pstate_method { + dml2_pstate_method_na = 0, + /* hw exclusive modes */ + dml2_pstate_method_vactive = 1, + dml2_pstate_method_vblank = 2, + dml2_pstate_method_reserved_hw = 5, + /* fw assisted exclusive modes */ + dml2_pstate_method_fw_svp = 6, + dml2_pstate_method_reserved_fw = 10, + /* fw assisted modes requiring drr modulation */ + dml2_pstate_method_fw_vactive_drr = 11, + dml2_pstate_method_fw_vblank_drr = 12, + dml2_pstate_method_fw_svp_drr = 13, + dml2_pstate_method_reserved_fw_drr_clamped = 20, + dml2_pstate_method_fw_drr = 21, + dml2_pstate_method_reserved_fw_drr_var = 22, + dml2_pstate_method_count +}; + +struct dml2_per_plane_programming { + const struct dml2_plane_parameters *plane_descriptor; + + union { + struct { + unsigned long dppclk_khz; + } dcn4x; + } min_clocks; + + struct dml2_mcache_surface_allocation mcache_allocation; + + // If a stream is using automatic or forced odm combine + // and the stream for this plane has num_odms_required > 1 + // num_dpps_required is always equal to num_odms_required for + // ALL planes of the stream + + // If a stream is using odm split, then this value is always 1 + unsigned int num_dpps_required; + + enum dml2_pstate_method uclk_pstate_support_method; + + // MALL size requirements for MALL SS and SubVP + unsigned int surface_size_mall_bytes; + unsigned int svp_size_mall_bytes; + + struct dml2_dchub_per_pipe_register_set *pipe_regs[DML2_MAX_PLANES]; + + struct { + bool valid; + struct dml2_plane_parameters descriptor; + struct dml2_mcache_surface_allocation mcache_allocation; + struct dml2_dchub_per_pipe_register_set *pipe_regs[DML2_MAX_PLANES]; + } phantom_plane; +}; + +union dml2_global_sync_programming { + struct { + unsigned int vstartup_lines; + unsigned int vupdate_offset_pixels; + unsigned int vupdate_vupdate_width_pixels; + unsigned int vready_offset_pixels; + unsigned int pstate_keepout_start_lines; + } dcn4x; +}; + +struct dml2_per_stream_programming { + const struct dml2_stream_parameters *stream_descriptor; + + union { + struct { + unsigned long dscclk_khz; + unsigned long dtbclk_khz; + unsigned long phyclk_khz; + } dcn4x; + } min_clocks; + + union dml2_global_sync_programming global_sync; + + unsigned int num_odms_required; + + enum dml2_pstate_method uclk_pstate_method; + + struct { + bool enabled; + struct dml2_stream_parameters descriptor; + union dml2_global_sync_programming global_sync; + } phantom_stream; + + union dmub_cmd_fams2_config fams2_base_params; + union { + union dmub_cmd_fams2_config fams2_sub_params; + union dmub_fams2_stream_static_sub_state_v2 fams2_sub_params_v2; + }; +}; + +//----------------- +// Mode Support Information +//----------------- + +struct dml2_mode_support_info { + bool ModeIsSupported; //pipe_count = ip_params->max_num_dpp; + ip_caps->otg_count = ip_params->max_num_otg; + ip_caps->num_dsc = ip_params->num_dsc; + ip_caps->max_num_dp2p0_streams = ip_params->max_num_dp2p0_streams; + ip_caps->max_num_dp2p0_outputs = ip_params->max_num_dp2p0_outputs; + ip_caps->max_num_hdmi_frl_outputs = ip_params->max_num_hdmi_frl_outputs; + ip_caps->rob_buffer_size_kbytes = ip_params->rob_buffer_size_kbytes; + ip_caps->config_return_buffer_size_in_kbytes = ip_params->config_return_buffer_size_in_kbytes; + ip_caps->config_return_buffer_segment_size_in_kbytes = ip_params->config_return_buffer_segment_size_in_kbytes; + ip_caps->meta_fifo_size_in_kentries = ip_params->meta_fifo_size_in_kentries; + ip_caps->compressed_buffer_segment_size_in_kbytes = ip_params->compressed_buffer_segment_size_in_kbytes; + ip_caps->cursor_buffer_size = ip_params->cursor_buffer_size; + ip_caps->max_flip_time_us = ip_params->max_flip_time_us; + ip_caps->max_flip_time_lines = ip_params->max_flip_time_lines; + ip_caps->hostvm_mode = ip_params->hostvm_mode; + + // FIXME_STAGE2: cleanup after adding all dv override to ip_caps + ip_caps->subvp_drr_scheduling_margin_us = 100; + ip_caps->subvp_prefetch_end_to_mall_start_us = 15; + ip_caps->subvp_fw_processing_delay = 16; + +} + +static void patch_ip_params_with_ip_caps(struct dml2_core_ip_params *ip_params, const struct dml2_ip_capabilities *ip_caps) +{ + ip_params->max_num_dpp = ip_caps->pipe_count; + ip_params->max_num_otg = ip_caps->otg_count; + ip_params->num_dsc = ip_caps->num_dsc; + ip_params->max_num_dp2p0_streams = ip_caps->max_num_dp2p0_streams; + ip_params->max_num_dp2p0_outputs = ip_caps->max_num_dp2p0_outputs; + ip_params->max_num_hdmi_frl_outputs = ip_caps->max_num_hdmi_frl_outputs; + ip_params->rob_buffer_size_kbytes = ip_caps->rob_buffer_size_kbytes; + ip_params->config_return_buffer_size_in_kbytes = ip_caps->config_return_buffer_size_in_kbytes; + ip_params->config_return_buffer_segment_size_in_kbytes = ip_caps->config_return_buffer_segment_size_in_kbytes; + ip_params->meta_fifo_size_in_kentries = ip_caps->meta_fifo_size_in_kentries; + ip_params->compressed_buffer_segment_size_in_kbytes = ip_caps->compressed_buffer_segment_size_in_kbytes; + ip_params->cursor_buffer_size = ip_caps->cursor_buffer_size; + ip_params->max_flip_time_us = ip_caps->max_flip_time_us; + ip_params->max_flip_time_lines = ip_caps->max_flip_time_lines; + ip_params->hostvm_mode = ip_caps->hostvm_mode; +} + +bool core_dcn4_initialize(struct dml2_core_initialize_in_out *in_out) +{ + struct dml2_core_instance *core = in_out->instance; + + if (!in_out->minimum_clock_table) + return false; + else + core->minimum_clock_table = in_out->minimum_clock_table; + + if (in_out->explicit_ip_bb && in_out->explicit_ip_bb_size > 0) { + memcpy(&core->clean_me_up.mode_lib.ip, in_out->explicit_ip_bb, in_out->explicit_ip_bb_size); + + // FIXME_STAGE2: + // DV still uses stage1 ip_param_st for each variant, need to patch the ip_caps with ip_param info + // Should move DV to use ip_caps but need move more overrides to ip_caps + patch_ip_caps_with_explicit_ip_params(in_out->ip_caps, in_out->explicit_ip_bb); + core->clean_me_up.mode_lib.ip.subvp_pstate_allow_width_us = core_dcn4_ip_caps_base.subvp_pstate_allow_width_us; + core->clean_me_up.mode_lib.ip.subvp_fw_processing_delay_us = core_dcn4_ip_caps_base.subvp_pstate_allow_width_us; + core->clean_me_up.mode_lib.ip.subvp_swath_height_margin_lines = core_dcn4_ip_caps_base.subvp_swath_height_margin_lines; + } else { + memcpy(&core->clean_me_up.mode_lib.ip, &core_dcn4_ip_caps_base, sizeof(struct dml2_core_ip_params)); + patch_ip_params_with_ip_caps(&core->clean_me_up.mode_lib.ip, in_out->ip_caps); + core->clean_me_up.mode_lib.ip.imall_supported = false; + } + + memcpy(&core->clean_me_up.mode_lib.soc, in_out->soc_bb, sizeof(struct dml2_soc_bb)); + memcpy(&core->clean_me_up.mode_lib.ip_caps, in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); + + return true; +} + +static void create_phantom_stream_from_main_stream(struct dml2_stream_parameters *phantom, const struct dml2_stream_parameters *main, + const struct dml2_implicit_svp_meta *meta) +{ + memcpy(phantom, main, sizeof(struct dml2_stream_parameters)); + + phantom->timing.v_total = meta->v_total; + phantom->timing.v_active = meta->v_active; + phantom->timing.v_front_porch = meta->v_front_porch; + phantom->timing.v_blank_end = phantom->timing.v_total - phantom->timing.v_front_porch - phantom->timing.v_active; + phantom->timing.vblank_nom = phantom->timing.v_total - phantom->timing.v_active; + phantom->timing.drr_config.enabled = false; +} + +static void create_phantom_plane_from_main_plane(struct dml2_plane_parameters *phantom, const struct dml2_plane_parameters *main, + const struct dml2_stream_parameters *phantom_stream, int phantom_stream_index, const struct dml2_stream_parameters *main_stream) +{ + memcpy(phantom, main, sizeof(struct dml2_plane_parameters)); + + phantom->stream_index = phantom_stream_index; + phantom->overrides.refresh_from_mall = dml2_refresh_from_mall_mode_override_force_disable; + phantom->overrides.legacy_svp_config = dml2_svp_mode_override_phantom_pipe_no_data_return; + phantom->composition.viewport.plane0.height = (long int unsigned) math_min2(math_ceil2( + (double)main->composition.scaler_info.plane0.v_ratio * (double)phantom_stream->timing.v_active, 16.0), + (double)main->composition.viewport.plane0.height); + phantom->composition.viewport.plane1.height = (long int unsigned) math_min2(math_ceil2( + (double)main->composition.scaler_info.plane1.v_ratio * (double)phantom_stream->timing.v_active, 16.0), + (double)main->composition.viewport.plane1.height); + phantom->immediate_flip = false; + phantom->dynamic_meta_data.enable = false; + phantom->cursor.num_cursors = 0; + phantom->cursor.cursor_width = 0; + phantom->tdlut.setup_for_tdlut = false; +} + +static void expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, + struct dml2_core_scratch *scratch) +{ + unsigned int stream_index, plane_index; + const struct dml2_plane_parameters *main_plane; + const struct dml2_stream_parameters *main_stream; + const struct dml2_stream_parameters *phantom_stream; + + memcpy(svp_expanded_display_cfg, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); + memset(scratch->main_stream_index_from_svp_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); + memset(scratch->svp_stream_index_from_main_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); + memset(scratch->main_plane_index_to_phantom_plane_index, 0, sizeof(int) * DML2_MAX_PLANES); + + if (!display_cfg->display_config.overrides.enable_subvp_implicit_pmo) + return; + + /* disable unbounded requesting for all planes until stage 3 has been performed */ + if (!display_cfg->stage3.performed) { + svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.enable = true; + svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.value = false; + } + // Create the phantom streams + for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { + main_stream = &display_cfg->display_config.stream_descriptors[stream_index]; + scratch->main_stream_index_from_svp_stream_index[stream_index] = stream_index; + scratch->svp_stream_index_from_main_stream_index[stream_index] = stream_index; + + if (display_cfg->stage3.stream_svp_meta[stream_index].valid) { + // Create the phantom stream + create_phantom_stream_from_main_stream(&svp_expanded_display_cfg->stream_descriptors[svp_expanded_display_cfg->num_streams], + main_stream, &display_cfg->stage3.stream_svp_meta[stream_index]); + + // Associate this phantom stream to the main stream + scratch->main_stream_index_from_svp_stream_index[svp_expanded_display_cfg->num_streams] = stream_index; + scratch->svp_stream_index_from_main_stream_index[stream_index] = svp_expanded_display_cfg->num_streams; + + // Increment num streams + svp_expanded_display_cfg->num_streams++; + } + } + + // Create the phantom planes + for (plane_index = 0; plane_index < display_cfg->display_config.num_planes; plane_index++) { + main_plane = &display_cfg->display_config.plane_descriptors[plane_index]; + + if (display_cfg->stage3.stream_svp_meta[main_plane->stream_index].valid) { + main_stream = &display_cfg->display_config.stream_descriptors[main_plane->stream_index]; + phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index]]; + create_phantom_plane_from_main_plane(&svp_expanded_display_cfg->plane_descriptors[svp_expanded_display_cfg->num_planes], + main_plane, phantom_stream, scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index], main_stream); + + // Associate this phantom plane to the main plane + scratch->phantom_plane_index_to_main_plane_index[svp_expanded_display_cfg->num_planes] = plane_index; + scratch->main_plane_index_to_phantom_plane_index[plane_index] = svp_expanded_display_cfg->num_planes; + + // Increment num planes + svp_expanded_display_cfg->num_planes++; + + // Adjust the main plane settings + svp_expanded_display_cfg->plane_descriptors[plane_index].overrides.legacy_svp_config = dml2_svp_mode_override_main_pipe; + } + } +} + +static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_instance *core, const struct display_configuation_with_meta *display_cfg, + const struct dml2_display_cfg *svp_expanded_display_cfg, struct dml2_display_cfg_programming *programming, struct dml2_core_scratch *scratch) +{ + unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index, mcache_index; + unsigned int total_main_mcaches_required = 0; + int total_pipe_regs_copied = 0; + int dml_internal_pipe_index = 0; + const struct dml2_plane_parameters *main_plane; + const struct dml2_plane_parameters *phantom_plane; + const struct dml2_stream_parameters *main_stream; + const struct dml2_stream_parameters *phantom_stream; + + // Copy the unexpanded display config to output + memcpy(&programming->display_config, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); + + // Set the global register values + dml2_core_calcs_get_arb_params(&display_cfg->display_config, &core->clean_me_up.mode_lib, &programming->global_regs.arb_regs); + // Get watermarks uses display config for ref clock override, so it doesn't matter whether we pass the pre or post expansion + // display config + dml2_core_calcs_get_watermarks(&display_cfg->display_config, &core->clean_me_up.mode_lib, &programming->global_regs.wm_regs[0]); + + // Check if FAMS2 is required + if (display_cfg->stage3.performed && display_cfg->stage3.success) { + programming->fams2_required = display_cfg->stage3.fams2_required; + + dml2_core_calcs_get_global_fams2_programming(&core->clean_me_up.mode_lib, display_cfg, &programming->fams2_global_config); + } + + // Only loop over all the main streams (the implicit svp streams will be packed as part of the main stream) + for (stream_index = 0; stream_index < programming->display_config.num_streams; stream_index++) { + main_stream = &svp_expanded_display_cfg->stream_descriptors[stream_index]; + phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[stream_index]]; + + // Set the descriptor + programming->stream_programming[stream_index].stream_descriptor = &programming->display_config.stream_descriptors[stream_index]; + + // Set the odm combine factor + programming->stream_programming[stream_index].num_odms_required = display_cfg->mode_support_result.cfg_support_info.stream_support_info[stream_index].odms_used; + + // Check if the stream has implicit SVP enabled + if (main_stream != phantom_stream) { + // If so, copy the phantom stream descriptor + programming->stream_programming[stream_index].phantom_stream.enabled = true; + memcpy(&programming->stream_programming[stream_index].phantom_stream.descriptor, phantom_stream, sizeof(struct dml2_stream_parameters)); + } else { + programming->stream_programming[stream_index].phantom_stream.enabled = false; + } + + // Due to the way DML indexes data internally, it's easier to populate the rest of the display + // stream programming in the next stage + } + + dml_internal_pipe_index = 0; + total_pipe_regs_copied = 0; + stream_already_populated_mask = 0x0; + + // Loop over all main planes + for (plane_index = 0; plane_index < programming->display_config.num_planes; plane_index++) { + main_plane = &svp_expanded_display_cfg->plane_descriptors[plane_index]; + + // Set the descriptor + programming->plane_programming[plane_index].plane_descriptor = &programming->display_config.plane_descriptors[plane_index]; + + // Set the mpc combine factor + programming->plane_programming[plane_index].num_dpps_required = core->clean_me_up.mode_lib.mp.NoOfDPP[plane_index]; + + // Setup the appropriate p-state strategy + if (display_cfg->stage3.performed && display_cfg->stage3.success) { + programming->plane_programming[plane_index].uclk_pstate_support_method = display_cfg->stage3.pstate_switch_modes[plane_index]; + } else { + programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_na; + } + + dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); + + memcpy(&programming->plane_programming[plane_index].mcache_allocation, + &display_cfg->stage2.mcache_allocations[plane_index], + sizeof(struct dml2_mcache_surface_allocation)); + total_main_mcaches_required += programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane0 + + programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane1 - + (programming->plane_programming[plane_index].mcache_allocation.last_slice_sharing.plane0_plane1 ? 1 : 0); + + for (pipe_offset = 0; pipe_offset < programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { + // Assign storage for this pipe's register values + programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; + memset(programming->plane_programming[plane_index].pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); + total_pipe_regs_copied++; + + // Populate the main plane regs + dml2_core_calcs_get_pipe_regs(svp_expanded_display_cfg, &core->clean_me_up.mode_lib, programming->plane_programming[plane_index].pipe_regs[pipe_offset], dml_internal_pipe_index); + + // Multiple planes can refer to the same stream index, so it's only necessary to populate it once + if (!(stream_already_populated_mask & (0x1 << main_plane->stream_index))) { + dml2_core_calcs_get_stream_programming(&core->clean_me_up.mode_lib, &programming->stream_programming[main_plane->stream_index], dml_internal_pipe_index); + + programming->stream_programming[main_plane->stream_index].uclk_pstate_method = programming->plane_programming[plane_index].uclk_pstate_support_method; + + /* unconditionally populate fams2 params */ + dml2_core_calcs_get_stream_fams2_programming(&core->clean_me_up.mode_lib, + display_cfg, + &programming->stream_programming[main_plane->stream_index].fams2_base_params, + &programming->stream_programming[main_plane->stream_index].fams2_sub_params, + programming->stream_programming[main_plane->stream_index].uclk_pstate_method, + plane_index); + + stream_already_populated_mask |= (0x1 << main_plane->stream_index); + } + dml_internal_pipe_index++; + } + } + + for (plane_index = programming->display_config.num_planes; plane_index < svp_expanded_display_cfg->num_planes; plane_index++) { + phantom_plane = &svp_expanded_display_cfg->plane_descriptors[plane_index]; + main_plane_index = scratch->phantom_plane_index_to_main_plane_index[plane_index]; + main_plane = &svp_expanded_display_cfg->plane_descriptors[main_plane_index]; + + programming->plane_programming[main_plane_index].phantom_plane.valid = true; + memcpy(&programming->plane_programming[main_plane_index].phantom_plane.descriptor, phantom_plane, sizeof(struct dml2_plane_parameters)); + + dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[main_plane_index].svp_size_mall_bytes, dml_internal_pipe_index); + + /* generate mcache allocation, phantoms use identical mcache configuration, but in the MALL set and unique mcache ID's beginning after all main ID's */ + memcpy(&programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation, + &programming->plane_programming[main_plane_index].mcache_allocation, + sizeof(struct dml2_mcache_surface_allocation)); + for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane0; mcache_index++) { + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index] += total_main_mcaches_required; + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane0[mcache_index] = + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index]; + } + for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane1; mcache_index++) { + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index] += total_main_mcaches_required; + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane1[mcache_index] = + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index]; + } + + for (pipe_offset = 0; pipe_offset < programming->plane_programming[main_plane_index].num_dpps_required; pipe_offset++) { + // Assign storage for this pipe's register values + programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; + memset(programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); + total_pipe_regs_copied++; + + // Populate the phantom plane regs + dml2_core_calcs_get_pipe_regs(svp_expanded_display_cfg, &core->clean_me_up.mode_lib, programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset], dml_internal_pipe_index); + // Populate the phantom stream specific programming + if (!(stream_already_populated_mask & (0x1 << phantom_plane->stream_index))) { + dml2_core_calcs_get_global_sync_programming(&core->clean_me_up.mode_lib, &programming->stream_programming[main_plane->stream_index].phantom_stream.global_sync, dml_internal_pipe_index); + + stream_already_populated_mask |= (0x1 << phantom_plane->stream_index); + } + + dml_internal_pipe_index++; + } + } +} + +bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out) +{ + struct dml2_core_instance *core = (struct dml2_core_instance *)in_out->instance; + struct dml2_core_mode_support_locals *l = &core->scratch.mode_support_locals; + + bool result; + unsigned int i, stream_index, stream_bitmask; + int unsigned odm_count, num_odm_output_segments, dpp_count; + + expand_implict_subvp(in_out->display_cfg, &l->svp_expanded_display_cfg, &core->scratch); + + l->mode_support_ex_params.mode_lib = &core->clean_me_up.mode_lib; + l->mode_support_ex_params.in_display_cfg = &l->svp_expanded_display_cfg; + l->mode_support_ex_params.min_clk_table = in_out->min_clk_table; + l->mode_support_ex_params.min_clk_index = in_out->min_clk_index; + l->mode_support_ex_params.out_evaluation_info = &in_out->mode_support_result.cfg_support_info.clean_me_up.support_info; + + result = dml2_core_calcs_mode_support_ex(&l->mode_support_ex_params); + + in_out->mode_support_result.cfg_support_info.is_supported = result; + + if (result) { + in_out->mode_support_result.global.dispclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDISPCLK * 1000); + in_out->mode_support_result.global.dcfclk_deepsleep_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.dcfclk_deepsleep * 1000); + in_out->mode_support_result.global.socclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.SOCCLK * 1000); + + in_out->mode_support_result.global.fclk_pstate_supported = l->mode_support_ex_params.out_evaluation_info->global_fclk_change_supported; + in_out->mode_support_result.global.uclk_pstate_supported = l->mode_support_ex_params.out_evaluation_info->global_dram_clock_change_supported; + + in_out->mode_support_result.global.active.fclk_khz = (unsigned long)(core->clean_me_up.mode_lib.ms.FabricClock * 1000); + in_out->mode_support_result.global.active.dcfclk_khz = (unsigned long)(core->clean_me_up.mode_lib.ms.DCFCLK * 1000); + + + in_out->mode_support_result.global.svp_prefetch.fclk_khz = (unsigned long)core->clean_me_up.mode_lib.ms.FabricClock * 1000; + in_out->mode_support_result.global.svp_prefetch.dcfclk_khz = (unsigned long)core->clean_me_up.mode_lib.ms.DCFCLK * 1000; + + in_out->mode_support_result.global.active.average_bw_sdp_kbps = 0; + in_out->mode_support_result.global.active.urgent_bw_dram_kbps = 0; + in_out->mode_support_result.global.svp_prefetch.average_bw_sdp_kbps = 0; + in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = 0; + + in_out->mode_support_result.global.active.average_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] * 1000), 1.0); + in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] * 1000), 1.0); + in_out->mode_support_result.global.svp_prefetch.average_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] * 1000), 1.0); + in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] * 1000), 1.0); + + in_out->mode_support_result.global.active.average_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] * 1000), 1.0); + in_out->mode_support_result.global.active.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] * 1000), 1.0); + in_out->mode_support_result.global.svp_prefetch.average_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); + in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = (unsigned long)math_ceil2((l->mode_support_ex_params.out_evaluation_info->urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] * 1000), 1.0); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_sdp_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_sdp_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.active.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.active.urgent_bw_dram_kbps); + DML_LOG_VERBOSE("DML::%s: in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps = %ld\n", __func__, in_out->mode_support_result.global.svp_prefetch.urgent_bw_dram_kbps); + + for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) { + in_out->mode_support_result.per_plane[i].dppclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDPPCLK[i] * 1000); + } + + stream_bitmask = 0; + for (i = 0; i < l->svp_expanded_display_cfg.num_planes; i++) { + odm_count = 1; + dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; + num_odm_output_segments = 1; + + switch (l->mode_support_ex_params.out_evaluation_info->ODMMode[i]) { + case dml2_odm_mode_bypass: + odm_count = 1; + dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; + break; + case dml2_odm_mode_combine_2to1: + odm_count = 2; + dpp_count = 2; + break; + case dml2_odm_mode_combine_3to1: + odm_count = 3; + dpp_count = 3; + break; + case dml2_odm_mode_combine_4to1: + odm_count = 4; + dpp_count = 4; + break; + case dml2_odm_mode_split_1to2: + case dml2_odm_mode_mso_1to2: + num_odm_output_segments = 2; + break; + case dml2_odm_mode_mso_1to4: + num_odm_output_segments = 4; + break; + case dml2_odm_mode_auto: + default: + odm_count = 1; + dpp_count = l->mode_support_ex_params.out_evaluation_info->DPPPerSurface[i]; + break; + } + + in_out->mode_support_result.cfg_support_info.plane_support_info[i].dpps_used = dpp_count; + + dml2_core_calcs_get_plane_support_info(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->mode_support_result.cfg_support_info.plane_support_info[i], i); + + stream_index = l->svp_expanded_display_cfg.plane_descriptors[i].stream_index; + + in_out->mode_support_result.per_stream[stream_index].dscclk_khz = (unsigned int)core->clean_me_up.mode_lib.ms.required_dscclk_freq_mhz[i] * 1000; + DML_LOG_VERBOSE("CORE_DCN4::%s: i=%d stream_index=%d, in_out->mode_support_result.per_stream[stream_index].dscclk_khz = %u\n", __func__, i, stream_index, in_out->mode_support_result.per_stream[stream_index].dscclk_khz); + + if (!((stream_bitmask >> stream_index) & 0x1)) { + in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].odms_used = odm_count; + in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].num_odm_output_segments = num_odm_output_segments; + in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].dsc_enable = l->mode_support_ex_params.out_evaluation_info->DSCEnabled[i]; + in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index].num_dsc_slices = l->mode_support_ex_params.out_evaluation_info->NumberOfDSCSlices[i]; + dml2_core_calcs_get_stream_support_info(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->mode_support_result.cfg_support_info.stream_support_info[stream_index], i); + in_out->mode_support_result.per_stream[stream_index].dtbclk_khz = (unsigned int)(core->clean_me_up.mode_lib.ms.RequiredDTBCLK[i] * 1000); + stream_bitmask |= 0x1 << stream_index; + } + } + } + + return result; +} + +static int lookup_uclk_dpm_index_by_freq(unsigned long uclk_freq_khz, struct dml2_soc_bb *soc_bb) +{ + int i; + + for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { + if (uclk_freq_khz == soc_bb->clk_table.uclk.clk_values_khz[i]) + return i; + } + return 0; +} + +bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out) +{ + struct dml2_core_instance *core = (struct dml2_core_instance *)in_out->instance; + struct dml2_core_mode_programming_locals *l = &core->scratch.mode_programming_locals; + + bool result = false; + unsigned int pipe_offset; + int dml_internal_pipe_index; + int total_pipe_regs_copied = 0; + int stream_already_populated_mask = 0; + + int main_stream_index; + unsigned int plane_index; + + expand_implict_subvp(in_out->display_cfg, &l->svp_expanded_display_cfg, &core->scratch); + + l->mode_programming_ex_params.mode_lib = &core->clean_me_up.mode_lib; + l->mode_programming_ex_params.in_display_cfg = &l->svp_expanded_display_cfg; + l->mode_programming_ex_params.min_clk_table = in_out->instance->minimum_clock_table; + l->mode_programming_ex_params.cfg_support_info = in_out->cfg_support_info; + l->mode_programming_ex_params.programming = in_out->programming; + l->mode_programming_ex_params.min_clk_index = lookup_uclk_dpm_index_by_freq(in_out->programming->min_clocks.dcn4x.active.uclk_khz, + &core->clean_me_up.mode_lib.soc); + + result = dml2_core_calcs_mode_programming_ex(&l->mode_programming_ex_params); + + if (result) { + // If the input display configuration contains implict SVP, we need to use a special packer + if (in_out->display_cfg->display_config.overrides.enable_subvp_implicit_pmo) { + pack_mode_programming_params_with_implicit_subvp(core, in_out->display_cfg, &l->svp_expanded_display_cfg, in_out->programming, &core->scratch); + } else { + memcpy(&in_out->programming->display_config, in_out->display_cfg, sizeof(struct dml2_display_cfg)); + + dml2_core_calcs_get_arb_params(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->programming->global_regs.arb_regs); + dml2_core_calcs_get_watermarks(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, &in_out->programming->global_regs.wm_regs[0]); + + dml_internal_pipe_index = 0; + + for (plane_index = 0; plane_index < in_out->programming->display_config.num_planes; plane_index++) { + in_out->programming->plane_programming[plane_index].num_dpps_required = core->clean_me_up.mode_lib.mp.NoOfDPP[plane_index]; + + if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; + else if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe) + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; + else if (in_out->programming->display_config.plane_descriptors[plane_index].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_fw_svp; + else { + if (core->clean_me_up.mode_lib.mp.MaxActiveDRAMClockChangeLatencySupported[plane_index] >= core->clean_me_up.mode_lib.soc.power_management_parameters.dram_clk_change_blackout_us) + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_vactive; + else if (core->clean_me_up.mode_lib.mp.TWait[plane_index] >= core->clean_me_up.mode_lib.soc.power_management_parameters.dram_clk_change_blackout_us) + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_vblank; + else + in_out->programming->plane_programming[plane_index].uclk_pstate_support_method = dml2_pstate_method_na; + } + + dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &in_out->programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); + + memcpy(&in_out->programming->plane_programming[plane_index].mcache_allocation, + &in_out->display_cfg->stage2.mcache_allocations[plane_index], + sizeof(struct dml2_mcache_surface_allocation)); + + for (pipe_offset = 0; pipe_offset < in_out->programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { + in_out->programming->plane_programming[plane_index].plane_descriptor = &in_out->programming->display_config.plane_descriptors[plane_index]; + + // Assign storage for this pipe's register values + in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &in_out->programming->pipe_regs[total_pipe_regs_copied]; + memset(in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset], 0, sizeof(struct dml2_dchub_per_pipe_register_set)); + total_pipe_regs_copied++; + + // Populate + dml2_core_calcs_get_pipe_regs(&l->svp_expanded_display_cfg, &core->clean_me_up.mode_lib, in_out->programming->plane_programming[plane_index].pipe_regs[pipe_offset], dml_internal_pipe_index); + + main_stream_index = in_out->programming->display_config.plane_descriptors[plane_index].stream_index; + + // Multiple planes can refer to the same stream index, so it's only necessary to populate it once + if (!(stream_already_populated_mask & (0x1 << main_stream_index))) { + in_out->programming->stream_programming[main_stream_index].stream_descriptor = &in_out->programming->display_config.stream_descriptors[main_stream_index]; + in_out->programming->stream_programming[main_stream_index].num_odms_required = in_out->cfg_support_info->stream_support_info[main_stream_index].odms_used; + dml2_core_calcs_get_stream_programming(&core->clean_me_up.mode_lib, &in_out->programming->stream_programming[main_stream_index], dml_internal_pipe_index); + + stream_already_populated_mask |= (0x1 << main_stream_index); + } + dml_internal_pipe_index++; + } + } + } + } + + return result; +} + +bool core_dcn4_populate_informative(struct dml2_core_populate_informative_in_out *in_out) +{ + struct dml2_core_internal_display_mode_lib *mode_lib = &in_out->instance->clean_me_up.mode_lib; + + if (in_out->mode_is_supported) + in_out->programming->informative.voltage_level = in_out->instance->scratch.mode_programming_locals.mode_programming_ex_params.min_clk_index; + else + in_out->programming->informative.voltage_level = in_out->instance->scratch.mode_support_locals.mode_support_ex_params.min_clk_index; + + dml2_core_calcs_get_informative(mode_lib, in_out->programming); + return true; +} + +bool core_dcn4_calculate_mcache_allocation(struct dml2_calculate_mcache_allocation_in_out *in_out) +{ + memset(in_out->mcache_allocation, 0, sizeof(struct dml2_mcache_surface_allocation)); + + dml2_core_calcs_get_mcache_allocation(&in_out->instance->clean_me_up.mode_lib, in_out->mcache_allocation, in_out->plane_index); + + if (in_out->mcache_allocation->num_mcaches_plane0 > 0) + in_out->mcache_allocation->mcache_x_offsets_plane0[in_out->mcache_allocation->num_mcaches_plane0 - 1] = in_out->plane_descriptor->surface.plane0.width; + + if (in_out->mcache_allocation->num_mcaches_plane1 > 0) + in_out->mcache_allocation->mcache_x_offsets_plane1[in_out->mcache_allocation->num_mcaches_plane1 - 1] = in_out->plane_descriptor->surface.plane1.width; + + in_out->mcache_allocation->requires_dedicated_mall_mcache = false; + + return true; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h new file mode 100644 index 000000000000..a68bb001a346 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h @@ -0,0 +1,12 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_CORE_DCN4_H__ +#define __DML2_CORE_DCN4_H__ +bool core_dcn4_initialize(struct dml2_core_initialize_in_out *in_out); +bool core_dcn4_mode_support(struct dml2_core_mode_support_in_out *in_out); +bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out); +bool core_dcn4_populate_informative(struct dml2_core_populate_informative_in_out *in_out); +bool core_dcn4_calculate_mcache_allocation(struct dml2_calculate_mcache_allocation_in_out *in_out); +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c new file mode 100644 index 000000000000..f809c4073b43 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -0,0 +1,13341 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + + +#include "dml2_internal_shared_types.h" +#include "dml2_core_dcn4_calcs.h" +#include "dml2_debug.h" +#include "lib_float_math.h" +#include "dml_top_types.h" + +#define DML2_MAX_FMT_420_BUFFER_WIDTH 4096 +#define DML_MAX_NUM_OF_SLICES_PER_DSC 4 +#define DML_MAX_COMPRESSION_RATIO 4 +//#define DML_MODE_SUPPORT_USE_DPM_DRAM_BW +//#define DML_GLOBAL_PREFETCH_CHECK +#define ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE +#define DML_MAX_VSTARTUP_START 1023 + +const char *dml2_core_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type) +{ + switch (bw_type) { + case (dml2_core_internal_bw_sdp): + return("dml2_core_internal_bw_sdp"); + case (dml2_core_internal_bw_dram): + return("dml2_core_internal_bw_dram"); + case (dml2_core_internal_bw_max): + return("dml2_core_internal_bw_max"); + default: + return("dml2_core_internal_bw_unknown"); + } +} + +const char *dml2_core_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type) +{ + switch (dml2_core_internal_soc_state_type) { + case (dml2_core_internal_soc_state_sys_idle): + return("dml2_core_internal_soc_state_sys_idle"); + case (dml2_core_internal_soc_state_sys_active): + return("dml2_core_internal_soc_state_sys_active"); + case (dml2_core_internal_soc_state_svp_prefetch): + return("dml2_core_internal_soc_state_svp_prefetch"); + case dml2_core_internal_soc_state_max: + default: + return("dml2_core_internal_soc_state_unknown"); + } +} + +static double dml2_core_div_rem(double dividend, unsigned int divisor, unsigned int *remainder) +{ + *remainder = ((dividend / divisor) - (int)(dividend / divisor) > 0); + return dividend / divisor; +} + +static void dml2_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) +{ + DML_LOG_VERBOSE("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); + if (!fail_only || support->ScaleRatioAndTapsSupport == 0) + DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); + if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) + DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); + if (!fail_only || support->ViewportSizeSupport == 0) + DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); + if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) + DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); + if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) + DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); + if (!fail_only || support->BPPForMultistreamNotIndicated == 1) + DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); + if (!fail_only || support->MultistreamWithHDMIOreDP == 1) + DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); + if (!fail_only || support->ExceededMultistreamSlots == 1) + DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); + if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) + DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); + if (!fail_only || support->NotEnoughLanesForMSO == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); + if (!fail_only || support->P2IWith420 == 1) + DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); + if (!fail_only || support->DSC422NativeNotSupported == 1) + DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); + if (!fail_only || support->DSCSlicesODMModeSupported == 0) + DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); + if (!fail_only || support->NotEnoughDSCUnits == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); + if (!fail_only || support->NotEnoughDSCSlices == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); + if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) + DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); + if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); + if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) + DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); + if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) + DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); + if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) + DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); + if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); + if (!fail_only || support->ROBSupport == 0) + DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); + if (!fail_only || support->OutstandingRequestsSupport == 0) + DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); + if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) + DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); + if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) + DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); + if (!fail_only || support->TotalAvailablePipesSupport == 0) + DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); + if (!fail_only || support->NumberOfOTGSupport == 0) + DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); + if (!fail_only || support->NumberOfHDMIFRLSupport == 0) + DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); + if (!fail_only || support->NumberOfDP2p0Support == 0) + DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); + if (!fail_only || support->EnoughWritebackUnits == 0) + DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); + if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) + DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); + if (!fail_only || support->WritebackLatencySupport == 0) + DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); + if (!fail_only || support->CursorSupport == 0) + DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); + if (!fail_only || support->PitchSupport == 0) + DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); + if (!fail_only || support->ViewportExceedsSurface == 1) + DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); + if (!fail_only || support->PrefetchSupported == 0) + DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); + if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) + DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); + if (!fail_only || support->AvgBandwidthSupport == 0) + DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); + if (!fail_only || support->DynamicMetadataSupported == 0) + DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); + if (!fail_only || support->VRatioInPrefetchSupported == 0) + DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); + if (!fail_only || support->PTEBufferSizeNotExceeded == 0) + DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); + if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) + DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); + if (!fail_only || support->ExceededMALLSize == 1) + DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); + if (!fail_only || support->g6_temp_read_support == 0) + DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); + if (!fail_only || support->ImmediateFlipSupport == 0) + DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); + if (!fail_only || support->LinkCapacitySupport == 0) + DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); + + if (!fail_only || support->ModeSupport == 0) + DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); + DML_LOG_VERBOSE("DML: ===================================== \n"); +} + +static void get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg) +{ + for (unsigned int k = 0; k < display_cfg->num_planes; k++) { + double bpc = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.bpc; + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_disable) { + switch (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format) { + case dml2_444: + out_bpp[k] = bpc * 3; + break; + case dml2_s422: + out_bpp[k] = bpc * 2; + break; + case dml2_n422: + out_bpp[k] = bpc * 2; + break; + case dml2_420: + default: + out_bpp[k] = bpc * 1.5; + break; + } + } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable) { + out_bpp[k] = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.dsc_compressed_bpp_x16 / 16; + } else { + out_bpp[k] = 0; + } + DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); + DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); + DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); + } +} + +static unsigned int dml_round_to_multiple(unsigned int num, unsigned int multiple, bool up) +{ + unsigned int remainder; + + if (multiple == 0) + return num; + + remainder = num % multiple; + if (remainder == 0) + return num; + + if (up) + return (num + multiple - remainder); + else + return (num - remainder); +} + +static unsigned int dml_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info) +{ + unsigned int num_active_pipes = 0; + + for (unsigned int k = 0; k < num_planes; k++) { + num_active_pipes = num_active_pipes + (unsigned int)cfg_support_info->plane_support_info[k].dpps_used; + } + + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); + return num_active_pipes; +} + +static void dml_calc_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane) +{ + unsigned int pipe_idx = 0; + + for (unsigned int k = 0; k < DML2_MAX_PLANES; ++k) { + pipe_plane[k] = __DML2_CALCS_PIPE_NO_PLANE__; + } + + for (unsigned int plane_idx = 0; plane_idx < DML2_MAX_PLANES; plane_idx++) { + for (int i = 0; i < cfg_support_info->plane_support_info[plane_idx].dpps_used; i++) { + pipe_plane[pipe_idx] = plane_idx; + pipe_idx++; + } + } +} + +static bool dml_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg) +{ + bool is_phantom = false; + + if (plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe || + plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) { + is_phantom = true; + } + + return is_phantom; +} + +static bool dml_get_is_phantom_pipe(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) +{ + unsigned int plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; + + bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_idx]); + DML_LOG_VERBOSE("DML::%s: pipe_idx=%d legacy_svp_config=%0d is_phantom=%d\n", __func__, pipe_idx, display_cfg->plane_descriptors[plane_idx].overrides.legacy_svp_config, is_phantom); + return is_phantom; +} + +#define dml_get_per_pipe_var_func(variable, type, interval_var) static type dml_get_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) \ +{ \ +unsigned int plane_idx; \ +plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; \ +return (type) interval_var[plane_idx]; \ +} + +dml_get_per_pipe_var_func(dpte_group_size_in_bytes, unsigned int, mode_lib->mp.dpte_group_bytes); +dml_get_per_pipe_var_func(vm_group_size_in_bytes, unsigned int, mode_lib->mp.vm_group_bytes); +dml_get_per_pipe_var_func(swath_height_l, unsigned int, mode_lib->mp.SwathHeightY); +dml_get_per_pipe_var_func(swath_height_c, unsigned int, mode_lib->mp.SwathHeightC); +dml_get_per_pipe_var_func(dpte_row_height_linear_l, unsigned int, mode_lib->mp.dpte_row_height_linear); +dml_get_per_pipe_var_func(dpte_row_height_linear_c, unsigned int, mode_lib->mp.dpte_row_height_linear_chroma); + +dml_get_per_pipe_var_func(vstartup_calculated, unsigned int, mode_lib->mp.VStartup); +dml_get_per_pipe_var_func(vupdate_offset, unsigned int, mode_lib->mp.VUpdateOffsetPix); +dml_get_per_pipe_var_func(vupdate_width, unsigned int, mode_lib->mp.VUpdateWidthPix); +dml_get_per_pipe_var_func(vready_offset, unsigned int, mode_lib->mp.VReadyOffsetPix); +dml_get_per_pipe_var_func(pstate_keepout_dst_lines, unsigned int, mode_lib->mp.pstate_keepout_dst_lines); +dml_get_per_pipe_var_func(det_stored_buffer_size_l_bytes, unsigned int, mode_lib->mp.DETBufferSizeY); +dml_get_per_pipe_var_func(det_stored_buffer_size_c_bytes, unsigned int, mode_lib->mp.DETBufferSizeC); +dml_get_per_pipe_var_func(det_buffer_size_kbytes, unsigned int, mode_lib->mp.DETBufferSizeInKByte); +dml_get_per_pipe_var_func(surface_size_in_mall_bytes, unsigned int, mode_lib->mp.SurfaceSizeInTheMALL); + +#define dml_get_per_plane_var_func(variable, type, interval_var) static type dml_get_plane_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int plane_idx) \ +{ \ +return (type) interval_var[plane_idx]; \ +} + +dml_get_per_plane_var_func(num_mcaches_plane0, unsigned int, mode_lib->ms.num_mcaches_l); +dml_get_per_plane_var_func(mcache_row_bytes_plane0, unsigned int, mode_lib->ms.mcache_row_bytes_l); +dml_get_per_plane_var_func(mcache_shift_granularity_plane0, unsigned int, mode_lib->ms.mcache_shift_granularity_l); +dml_get_per_plane_var_func(num_mcaches_plane1, unsigned int, mode_lib->ms.num_mcaches_c); +dml_get_per_plane_var_func(mcache_row_bytes_plane1, unsigned int, mode_lib->ms.mcache_row_bytes_c); +dml_get_per_plane_var_func(mcache_shift_granularity_plane1, unsigned int, mode_lib->ms.mcache_shift_granularity_c); +dml_get_per_plane_var_func(mall_comb_mcache_l, unsigned int, mode_lib->ms.mall_comb_mcache_l); +dml_get_per_plane_var_func(mall_comb_mcache_c, unsigned int, mode_lib->ms.mall_comb_mcache_c); +dml_get_per_plane_var_func(lc_comb_mcache, unsigned int, mode_lib->ms.lc_comb_mcache); +dml_get_per_plane_var_func(subviewport_lines_needed_in_mall, unsigned int, mode_lib->ms.SubViewportLinesNeededInMALL); +dml_get_per_plane_var_func(max_vstartup_lines, unsigned int, mode_lib->ms.MaxVStartupLines); + +#define dml_get_per_plane_array_var_func(variable, type, interval_var) static type dml_get_plane_array_##variable(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int plane_idx, unsigned int array_idx) \ +{ \ +return (type) interval_var[plane_idx][array_idx]; \ +} + +dml_get_per_plane_array_var_func(mcache_offsets_plane0, unsigned int, mode_lib->ms.mcache_offsets_l); +dml_get_per_plane_array_var_func(mcache_offsets_plane1, unsigned int, mode_lib->ms.mcache_offsets_c); + +#define dml_get_var_func(var, type, internal_var) static type dml_get_##var(const struct dml2_core_internal_display_mode_lib *mode_lib) \ +{ \ +return (type) internal_var; \ +} + +dml_get_var_func(wm_urgent, double, mode_lib->mp.Watermark.UrgentWatermark); +dml_get_var_func(wm_stutter_exit, double, mode_lib->mp.Watermark.StutterExitWatermark); +dml_get_var_func(wm_stutter_enter_exit, double, mode_lib->mp.Watermark.StutterEnterPlusExitWatermark); +dml_get_var_func(wm_z8_stutter_exit, double, mode_lib->mp.Watermark.Z8StutterExitWatermark); +dml_get_var_func(wm_z8_stutter_enter_exit, double, mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark); +dml_get_var_func(wm_memory_trip, double, mode_lib->mp.UrgentLatency); +dml_get_var_func(meta_trip_memory_us, double, mode_lib->mp.MetaTripToMemory); + +dml_get_var_func(wm_fclk_change, double, mode_lib->mp.Watermark.FCLKChangeWatermark); +dml_get_var_func(wm_usr_retraining, double, mode_lib->mp.Watermark.USRRetrainingWatermark); +dml_get_var_func(wm_temp_read_or_ppt, double, mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us); +dml_get_var_func(wm_dram_clock_change, double, mode_lib->mp.Watermark.DRAMClockChangeWatermark); +dml_get_var_func(fraction_of_urgent_bandwidth, double, mode_lib->mp.FractionOfUrgentBandwidth); +dml_get_var_func(fraction_of_urgent_bandwidth_imm_flip, double, mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip); +dml_get_var_func(fraction_of_urgent_bandwidth_mall, double, mode_lib->mp.FractionOfUrgentBandwidthMALL); +dml_get_var_func(wm_writeback_dram_clock_change, double, mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); +dml_get_var_func(wm_writeback_fclk_change, double, mode_lib->mp.Watermark.WritebackFCLKChangeWatermark); +dml_get_var_func(stutter_efficiency, double, mode_lib->mp.StutterEfficiency); +dml_get_var_func(stutter_efficiency_no_vblank, double, mode_lib->mp.StutterEfficiencyNotIncludingVBlank); +dml_get_var_func(stutter_num_bursts, double, mode_lib->mp.NumberOfStutterBurstsPerFrame); +dml_get_var_func(stutter_efficiency_z8, double, mode_lib->mp.Z8StutterEfficiency); +dml_get_var_func(stutter_num_bursts_z8, double, mode_lib->mp.Z8NumberOfStutterBurstsPerFrame); +dml_get_var_func(stutter_period, double, mode_lib->mp.StutterPeriod); +dml_get_var_func(stutter_efficiency_z8_bestcase, double, mode_lib->mp.Z8StutterEfficiencyBestCase); +dml_get_var_func(stutter_num_bursts_z8_bestcase, double, mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase); +dml_get_var_func(stutter_period_bestcase, double, mode_lib->mp.StutterPeriodBestCase); +dml_get_var_func(fclk_change_latency, double, mode_lib->mp.MaxActiveFCLKChangeLatencySupported); +dml_get_var_func(global_dppclk_khz, double, mode_lib->mp.GlobalDPPCLK * 1000.0); + +dml_get_var_func(sys_active_avg_bw_required_sdp, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_avg_bw_required_dram, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); + +dml_get_var_func(svp_prefetch_avg_bw_required_sdp, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_avg_bw_required_dram, double, mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(sys_active_avg_bw_available_sdp, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_avg_bw_available_dram, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); + +dml_get_var_func(svp_prefetch_avg_bw_available_sdp, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_avg_bw_available_dram, double, mode_lib->mp.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(sys_active_urg_bw_available_sdp, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_urg_bw_available_dram, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); +dml_get_var_func(sys_active_urg_bw_available_dram_vm_only, double, mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); + +dml_get_var_func(svp_prefetch_urg_bw_available_sdp, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_urg_bw_available_dram, double, mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); +dml_get_var_func(svp_prefetch_urg_bw_available_dram_vm_only, double, mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_svp_prefetch]); + +dml_get_var_func(urgent_latency, double, mode_lib->mp.UrgentLatency); +dml_get_var_func(max_urgent_latency_us, double, mode_lib->ms.support.max_urgent_latency_us); +dml_get_var_func(max_non_urgent_latency_us, double, mode_lib->ms.support.max_non_urgent_latency_us); +dml_get_var_func(avg_non_urgent_latency_us, double, mode_lib->ms.support.avg_non_urgent_latency_us); +dml_get_var_func(avg_urgent_latency_us, double, mode_lib->ms.support.avg_urgent_latency_us); + +dml_get_var_func(sys_active_urg_bw_required_sdp, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_urg_bw_required_dram, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); +dml_get_var_func(svp_prefetch_urg_bw_required_sdp, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_urg_bw_required_dram, double, mode_lib->mp.urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(sys_active_non_urg_required_sdp, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_non_urg_required_dram, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); +dml_get_var_func(svp_prefetch_non_urg_bw_required_sdp, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_non_urg_bw_required_dram, double, mode_lib->mp.non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(sys_active_urg_bw_required_sdp_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_urg_bw_required_dram_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); +dml_get_var_func(svp_prefetch_urg_bw_required_sdp_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_urg_bw_required_dram_flip, double, mode_lib->mp.urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(sys_active_non_urg_required_sdp_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); +dml_get_var_func(sys_active_non_urg_required_dram_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); +dml_get_var_func(svp_prefetch_non_urg_bw_required_sdp_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); +dml_get_var_func(svp_prefetch_non_urg_bw_required_dram_flip, double, mode_lib->mp.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); + +dml_get_var_func(comp_buffer_size_kbytes, unsigned int, mode_lib->mp.CompressedBufferSizeInkByte); + +dml_get_var_func(unbounded_request_enabled, bool, mode_lib->mp.UnboundedRequestEnabled); +dml_get_var_func(wm_writeback_urgent, double, mode_lib->mp.Watermark.WritebackUrgentWatermark); +dml_get_var_func(cstate_max_cap_mode, bool, mode_lib->mp.DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); +dml_get_var_func(compbuf_reserved_space_64b, unsigned int, mode_lib->mp.compbuf_reserved_space_64b); +dml_get_var_func(hw_debug5, bool, mode_lib->mp.hw_debug5); +dml_get_var_func(dcfclk_deep_sleep_hysteresis, unsigned int, mode_lib->mp.dcfclk_deep_sleep_hysteresis); + +static void CalculateMaxDETAndMinCompressedBufferSize( + unsigned int ConfigReturnBufferSizeInKByte, + unsigned int ConfigReturnBufferSegmentSizeInKByte, + unsigned int ROBBufferSizeInKByte, + unsigned int MaxNumDPP, + unsigned int nomDETInKByteOverrideEnable, // VBA_DELTA, allow DV to override default DET size + unsigned int nomDETInKByteOverrideValue, // VBA_DELTA + bool is_mrq_present, + + // Output + unsigned int *MaxTotalDETInKByte, + unsigned int *nomDETInKByte, + unsigned int *MinCompressedBufferSizeInKByte) +{ + if (is_mrq_present) + *MaxTotalDETInKByte = (unsigned int) math_ceil2((double)(ConfigReturnBufferSizeInKByte + ROBBufferSizeInKByte)*4/5, 64); + else + *MaxTotalDETInKByte = ConfigReturnBufferSizeInKByte - ConfigReturnBufferSegmentSizeInKByte; + + *nomDETInKByte = (unsigned int)(math_floor2((double)*MaxTotalDETInKByte / (double)MaxNumDPP, ConfigReturnBufferSegmentSizeInKByte)); + *MinCompressedBufferSizeInKByte = ConfigReturnBufferSizeInKByte - *MaxTotalDETInKByte; + + DML_LOG_VERBOSE("DML::%s: is_mrq_present = %u\n", __func__, is_mrq_present); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u\n", __func__, ROBBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MaxNumDPP = %u\n", __func__, MaxNumDPP); + DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, *MaxTotalDETInKByte); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, *nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, *MinCompressedBufferSizeInKByte); + + if (nomDETInKByteOverrideEnable) { + *nomDETInKByte = nomDETInKByteOverrideValue; + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u (overrided)\n", __func__, *nomDETInKByte); + } +} + +static void PixelClockAdjustmentForProgressiveToInterlaceUnit(const struct dml2_display_cfg *display_cfg, bool ptoi_supported, double *PixelClockBackEnd) +{ + //unsigned int num_active_planes = display_cfg->num_planes; + + //Progressive To Interlace Unit Effect + for (unsigned int k = 0; k < display_cfg->num_planes; ++k) { + PixelClockBackEnd[k] = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced == 1 && ptoi_supported == true) { + // FIXME_STAGE2... can sw pass the pixel rate for interlaced directly + //display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz = 2 * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz; + } + } +} + +static bool dml_is_420(enum dml2_source_format_class source_format) +{ + bool val = false; + + switch (source_format) { + case dml2_444_8: + val = 0; + break; + case dml2_444_16: + val = 0; + break; + case dml2_444_32: + val = 0; + break; + case dml2_444_64: + val = 0; + break; + case dml2_420_8: + val = 1; + break; + case dml2_420_10: + val = 1; + break; + case dml2_420_12: + val = 1; + break; + case dml2_422_planar_8: + val = 0; + break; + case dml2_422_planar_10: + val = 0; + break; + case dml2_422_planar_12: + val = 0; + break; + case dml2_422_packed_8: + val = 0; + break; + case dml2_422_packed_10: + val = 0; + break; + case dml2_422_packed_12: + val = 0; + break; + case dml2_rgbe_alpha: + val = 0; + break; + case dml2_rgbe: + val = 0; + break; + case dml2_mono_8: + val = 0; + break; + case dml2_mono_16: + val = 0; + break; + default: + DML_ASSERT(0); + break; + } + return val; +} + +static unsigned int dml_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode) +{ + if (sw_mode == dml2_sw_linear) + return 256; + else if (sw_mode == dml2_sw_256b_2d) + return 256; + else if (sw_mode == dml2_sw_4kb_2d) + return 4096; + else if (sw_mode == dml2_sw_64kb_2d) + return 65536; + else if (sw_mode == dml2_sw_256kb_2d) + return 262144; + else if (sw_mode == dml2_gfx11_sw_linear) + return 256; + else if (sw_mode == dml2_gfx11_sw_64kb_d) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_d_t) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_d_x) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_r_x) + return 65536; + else if (sw_mode == dml2_gfx11_sw_256kb_d_x) + return 262144; + else if (sw_mode == dml2_gfx11_sw_256kb_r_x) + return 262144; + else { + DML_ASSERT(0); + return 256; + } +} + +static bool dml_is_vertical_rotation(enum dml2_rotation_angle Scan) +{ + bool is_vert = false; + if (Scan == dml2_rotation_90 || Scan == dml2_rotation_270) { + is_vert = true; + } else { + is_vert = false; + } + return is_vert; +} + +static int unsigned dml_get_gfx_version(enum dml2_swizzle_mode sw_mode) +{ + int unsigned version = 0; + + if (sw_mode == dml2_sw_linear || + sw_mode == dml2_sw_256b_2d || + sw_mode == dml2_sw_4kb_2d || + sw_mode == dml2_sw_64kb_2d || + sw_mode == dml2_sw_256kb_2d) { + version = 12; + } else if (sw_mode == dml2_gfx11_sw_linear || + sw_mode == dml2_gfx11_sw_64kb_d || + sw_mode == dml2_gfx11_sw_64kb_d_t || + sw_mode == dml2_gfx11_sw_64kb_d_x || + sw_mode == dml2_gfx11_sw_64kb_r_x || + sw_mode == dml2_gfx11_sw_256kb_d_x || + sw_mode == dml2_gfx11_sw_256kb_r_x) { + version = 11; + } else { + DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); + DML_ASSERT(0); + } + + return version; +} + +static void CalculateBytePerPixelAndBlockSizes( + enum dml2_source_format_class SourcePixelFormat, + enum dml2_swizzle_mode SurfaceTiling, + unsigned int pitch_y, + unsigned int pitch_c, + + // Output + unsigned int *BytePerPixelY, + unsigned int *BytePerPixelC, + double *BytePerPixelDETY, + double *BytePerPixelDETC, + unsigned int *BlockHeight256BytesY, + unsigned int *BlockHeight256BytesC, + unsigned int *BlockWidth256BytesY, + unsigned int *BlockWidth256BytesC, + unsigned int *MacroTileHeightY, + unsigned int *MacroTileHeightC, + unsigned int *MacroTileWidthY, + unsigned int *MacroTileWidthC, + bool *surf_linear128_l, + bool *surf_linear128_c) +{ + *BytePerPixelDETY = 0; + *BytePerPixelDETC = 0; + *BytePerPixelY = 1; + *BytePerPixelC = 1; + + if (SourcePixelFormat == dml2_444_64) { + *BytePerPixelDETY = 8; + *BytePerPixelDETC = 0; + *BytePerPixelY = 8; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml2_444_32 || SourcePixelFormat == dml2_rgbe) { + *BytePerPixelDETY = 4; + *BytePerPixelDETC = 0; + *BytePerPixelY = 4; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml2_444_16 || SourcePixelFormat == dml2_mono_16) { + *BytePerPixelDETY = 2; + *BytePerPixelDETC = 0; + *BytePerPixelY = 2; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml2_444_8 || SourcePixelFormat == dml2_mono_8) { + *BytePerPixelDETY = 1; + *BytePerPixelDETC = 0; + *BytePerPixelY = 1; + *BytePerPixelC = 0; + } else if (SourcePixelFormat == dml2_rgbe_alpha) { + *BytePerPixelDETY = 4; + *BytePerPixelDETC = 1; + *BytePerPixelY = 4; + *BytePerPixelC = 1; + } else if (SourcePixelFormat == dml2_420_8) { + *BytePerPixelDETY = 1; + *BytePerPixelDETC = 2; + *BytePerPixelY = 1; + *BytePerPixelC = 2; + } else if (SourcePixelFormat == dml2_420_12) { + *BytePerPixelDETY = 2; + *BytePerPixelDETC = 4; + *BytePerPixelY = 2; + *BytePerPixelC = 4; + } else if (SourcePixelFormat == dml2_420_10) { + *BytePerPixelDETY = (double)(4.0 / 3); + *BytePerPixelDETC = (double)(8.0 / 3); + *BytePerPixelY = 2; + *BytePerPixelC = 4; + } else { + DML_LOG_VERBOSE("ERROR: DML::%s: SourcePixelFormat = %u not supported!\n", __func__, SourcePixelFormat); + DML_ASSERT(0); + } + + DML_LOG_VERBOSE("DML::%s: SourcePixelFormat = %u\n", __func__, SourcePixelFormat); + DML_LOG_VERBOSE("DML::%s: BytePerPixelDETY = %f\n", __func__, *BytePerPixelDETY); + DML_LOG_VERBOSE("DML::%s: BytePerPixelDETC = %f\n", __func__, *BytePerPixelDETC); + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, *BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, *BytePerPixelC); + DML_LOG_VERBOSE("DML::%s: pitch_y = %u\n", __func__, pitch_y); + DML_LOG_VERBOSE("DML::%s: pitch_c = %u\n", __func__, pitch_c); + DML_LOG_VERBOSE("DML::%s: surf_linear128_l = %u\n", __func__, *surf_linear128_l); + DML_LOG_VERBOSE("DML::%s: surf_linear128_c = %u\n", __func__, *surf_linear128_c); + + if (dml_get_gfx_version(SurfaceTiling) == 11) { + *surf_linear128_l = 0; + *surf_linear128_c = 0; + } else { + if (SurfaceTiling == dml2_sw_linear) { + *surf_linear128_l = (((pitch_y * *BytePerPixelY) % 256) != 0); + + if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) + *surf_linear128_c = (((pitch_c * *BytePerPixelC) % 256) != 0); + } + } + + if (!(dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha)) { + if (SurfaceTiling == dml2_sw_linear) { + *BlockHeight256BytesY = 1; + } else if (SourcePixelFormat == dml2_444_64) { + *BlockHeight256BytesY = 4; + } else if (SourcePixelFormat == dml2_444_8) { + *BlockHeight256BytesY = 16; + } else { + *BlockHeight256BytesY = 8; + } + *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; + *BlockHeight256BytesC = 0; + *BlockWidth256BytesC = 0; + } else { // dual plane + if (SurfaceTiling == dml2_sw_linear) { + *BlockHeight256BytesY = 1; + *BlockHeight256BytesC = 1; + } else if (SourcePixelFormat == dml2_rgbe_alpha) { + *BlockHeight256BytesY = 8; + *BlockHeight256BytesC = 16; + } else if (SourcePixelFormat == dml2_420_8) { + *BlockHeight256BytesY = 16; + *BlockHeight256BytesC = 8; + } else { + *BlockHeight256BytesY = 8; + *BlockHeight256BytesC = 8; + } + *BlockWidth256BytesY = 256U / *BytePerPixelY / *BlockHeight256BytesY; + *BlockWidth256BytesC = 256U / *BytePerPixelC / *BlockHeight256BytesC; + } + DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesY = %u\n", __func__, *BlockWidth256BytesY); + DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesY = %u\n", __func__, *BlockHeight256BytesY); + DML_LOG_VERBOSE("DML::%s: BlockWidth256BytesC = %u\n", __func__, *BlockWidth256BytesC); + DML_LOG_VERBOSE("DML::%s: BlockHeight256BytesC = %u\n", __func__, *BlockHeight256BytesC); + + if (dml_get_gfx_version(SurfaceTiling) == 11) { + if (SurfaceTiling == dml2_gfx11_sw_linear) { + *MacroTileHeightY = *BlockHeight256BytesY; + *MacroTileWidthY = 256 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 256 / *BytePerPixelC / *MacroTileHeightC; + } + } else if (SurfaceTiling == dml2_gfx11_sw_64kb_d || SurfaceTiling == dml2_gfx11_sw_64kb_d_t || SurfaceTiling == dml2_gfx11_sw_64kb_d_x || SurfaceTiling == dml2_gfx11_sw_64kb_r_x) { + *MacroTileHeightY = 16 * *BlockHeight256BytesY; + *MacroTileWidthY = 65536 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = 16 * *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 65536 / *BytePerPixelC / *MacroTileHeightC; + } + } else { + *MacroTileHeightY = 32 * *BlockHeight256BytesY; + *MacroTileWidthY = 65536 * 4 / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = 32 * *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = 65536 * 4 / *BytePerPixelC / *MacroTileHeightC; + } + } + } else { + unsigned int macro_tile_size_bytes = dml_get_tile_block_size_bytes(SurfaceTiling); + unsigned int macro_tile_scale = 1; // macro tile to 256B req scaling + + if (SurfaceTiling == dml2_sw_linear) { + macro_tile_scale = 1; + } else if (SurfaceTiling == dml2_sw_4kb_2d) { + macro_tile_scale = 4; + } else if (SurfaceTiling == dml2_sw_64kb_2d) { + macro_tile_scale = 16; + } else if (SurfaceTiling == dml2_sw_256kb_2d) { + macro_tile_scale = 32; + } else { + DML_LOG_VERBOSE("ERROR: Invalid SurfaceTiling setting! val=%u\n", SurfaceTiling); + DML_ASSERT(0); + } + + *MacroTileHeightY = macro_tile_scale * *BlockHeight256BytesY; + *MacroTileWidthY = macro_tile_size_bytes / *BytePerPixelY / *MacroTileHeightY; + *MacroTileHeightC = macro_tile_scale * *BlockHeight256BytesC; + if (*MacroTileHeightC == 0) { + *MacroTileWidthC = 0; + } else { + *MacroTileWidthC = macro_tile_size_bytes / *BytePerPixelC / *MacroTileHeightC; + } + } + + DML_LOG_VERBOSE("DML::%s: MacroTileWidthY = %u\n", __func__, *MacroTileWidthY); + DML_LOG_VERBOSE("DML::%s: MacroTileHeightY = %u\n", __func__, *MacroTileHeightY); + DML_LOG_VERBOSE("DML::%s: MacroTileWidthC = %u\n", __func__, *MacroTileWidthC); + DML_LOG_VERBOSE("DML::%s: MacroTileHeightC = %u\n", __func__, *MacroTileHeightC); +} + +static void CalculateSinglePipeDPPCLKAndSCLThroughput( + double HRatio, + double HRatioChroma, + double VRatio, + double VRatioChroma, + double MaxDCHUBToPSCLThroughput, + double MaxPSCLToLBThroughput, + double PixelClock, + enum dml2_source_format_class SourcePixelFormat, + unsigned int HTaps, + unsigned int HTapsChroma, + unsigned int VTaps, + unsigned int VTapsChroma, + + // Output + double *PSCL_THROUGHPUT, + double *PSCL_THROUGHPUT_CHROMA, + double *DPPCLKUsingSingleDPP) +{ + double DPPCLKUsingSingleDPPLuma; + double DPPCLKUsingSingleDPPChroma; + + if (HRatio > 1) { + *PSCL_THROUGHPUT = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatio / math_ceil2((double)HTaps / 6.0, 1.0)); + } else { + *PSCL_THROUGHPUT = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); + } + + DPPCLKUsingSingleDPPLuma = PixelClock * math_max3(VTaps / 6 * math_min2(1, HRatio), HRatio * VRatio / *PSCL_THROUGHPUT, 1); + + if ((HTaps > 6 || VTaps > 6) && DPPCLKUsingSingleDPPLuma < 2 * PixelClock) + DPPCLKUsingSingleDPPLuma = 2 * PixelClock; + + if (!dml_is_420(SourcePixelFormat) && SourcePixelFormat != dml2_rgbe_alpha) { + *PSCL_THROUGHPUT_CHROMA = 0; + *DPPCLKUsingSingleDPP = DPPCLKUsingSingleDPPLuma; + } else { + if (HRatioChroma > 1) { + *PSCL_THROUGHPUT_CHROMA = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput * HRatioChroma / math_ceil2((double)HTapsChroma / 6.0, 1.0)); + } else { + *PSCL_THROUGHPUT_CHROMA = math_min2(MaxDCHUBToPSCLThroughput, MaxPSCLToLBThroughput); + } + DPPCLKUsingSingleDPPChroma = PixelClock * math_max3(VTapsChroma / 6 * math_min2(1, HRatioChroma), + HRatioChroma * VRatioChroma / *PSCL_THROUGHPUT_CHROMA, 1); + if ((HTapsChroma > 6 || VTapsChroma > 6) && DPPCLKUsingSingleDPPChroma < 2 * PixelClock) + DPPCLKUsingSingleDPPChroma = 2 * PixelClock; + *DPPCLKUsingSingleDPP = math_max2(DPPCLKUsingSingleDPPLuma, DPPCLKUsingSingleDPPChroma); + } +} + +static void CalculateSwathWidth( + const struct dml2_display_cfg *display_cfg, + bool ForceSingleDPP, + unsigned int NumberOfActiveSurfaces, + enum dml2_odm_mode ODMMode[], + unsigned int BytePerPixY[], + unsigned int BytePerPixC[], + unsigned int Read256BytesBlockHeightY[], + unsigned int Read256BytesBlockHeightC[], + unsigned int Read256BytesBlockWidthY[], + unsigned int Read256BytesBlockWidthC[], + bool surf_linear128_l[], + bool surf_linear128_c[], + unsigned int DPPPerSurface[], + + // Output + unsigned int req_per_swath_ub_l[], + unsigned int req_per_swath_ub_c[], + unsigned int SwathWidthSingleDPPY[], // post-rotated plane width + unsigned int SwathWidthSingleDPPC[], + unsigned int SwathWidthY[], // per-pipe + unsigned int SwathWidthC[], // per-pipe + unsigned int MaximumSwathHeightY[], + unsigned int MaximumSwathHeightC[], + unsigned int swath_width_luma_ub[], // per-pipe + unsigned int swath_width_chroma_ub[]) // per-pipe +{ + enum dml2_odm_mode MainSurfaceODMMode; + double odm_hactive_factor = 1.0; + unsigned int req_width_horz_y; + unsigned int req_width_horz_c; + unsigned int surface_width_ub_l; + unsigned int surface_height_ub_l; + unsigned int surface_width_ub_c; + unsigned int surface_height_ub_c; + + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { + SwathWidthSingleDPPY[k] = (unsigned int)display_cfg->plane_descriptors[k].composition.viewport.plane0.width; + } else { + SwathWidthSingleDPPY[k] = (unsigned int)display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + } + + DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight=%lu\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); + + MainSurfaceODMMode = ODMMode[k]; + + if (ForceSingleDPP) { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + } else { + if (MainSurfaceODMMode == dml2_odm_mode_combine_4to1) + odm_hactive_factor = 4.0; + else if (MainSurfaceODMMode == dml2_odm_mode_combine_3to1) + odm_hactive_factor = 3.0; + else if (MainSurfaceODMMode == dml2_odm_mode_combine_2to1) + odm_hactive_factor = 2.0; + + if (MainSurfaceODMMode == dml2_odm_mode_combine_4to1 || MainSurfaceODMMode == dml2_odm_mode_combine_3to1 || MainSurfaceODMMode == dml2_odm_mode_combine_2to1) { + SwathWidthY[k] = (unsigned int)(math_min2((double)SwathWidthSingleDPPY[k], math_round((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active / odm_hactive_factor * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio))); + } else if (DPPPerSurface[k] == 2) { + SwathWidthY[k] = SwathWidthSingleDPPY[k] / 2; + } else { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + } + } + + DML_LOG_VERBOSE("DML::%s: k=%u HActive=%lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active); + DML_LOG_VERBOSE("DML::%s: k=%u HRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u MainSurfaceODMMode=%u\n", __func__, k, MainSurfaceODMMode); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthSingleDPPY=%u\n", __func__, k, SwathWidthSingleDPPY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthY=%u\n", __func__, k, SwathWidthY[k]); + + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { + SwathWidthC[k] = SwathWidthY[k] / 2; + SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k] / 2; + } else { + SwathWidthC[k] = SwathWidthY[k]; + SwathWidthSingleDPPC[k] = SwathWidthSingleDPPY[k]; + } + + if (ForceSingleDPP == true) { + SwathWidthY[k] = SwathWidthSingleDPPY[k]; + SwathWidthC[k] = SwathWidthSingleDPPC[k]; + } + + req_width_horz_y = Read256BytesBlockWidthY[k]; + req_width_horz_c = Read256BytesBlockWidthC[k]; + + if (surf_linear128_l[k]) + req_width_horz_y = req_width_horz_y / 2; + + if (surf_linear128_c[k]) + req_width_horz_c = req_width_horz_c / 2; + + surface_width_ub_l = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane0.width, req_width_horz_y); + surface_height_ub_l = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane0.height, Read256BytesBlockHeightY[k]); + surface_width_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.width, req_width_horz_c); + surface_height_ub_c = (unsigned int)math_ceil2((double)display_cfg->plane_descriptors[k].surface.plane1.height, Read256BytesBlockHeightC[k]); + + DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_l=%u\n", __func__, k, surface_width_ub_l); + DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_l=%u\n", __func__, k, surface_height_ub_l); + DML_LOG_VERBOSE("DML::%s: k=%u surface_width_ub_c=%u\n", __func__, k, surface_width_ub_c); + DML_LOG_VERBOSE("DML::%s: k=%u surface_height_ub_c=%u\n", __func__, k, surface_height_ub_c); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthY=%u\n", __func__, k, Read256BytesBlockWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightY=%u\n", __func__, k, Read256BytesBlockHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockWidthC=%u\n", __func__, k, Read256BytesBlockWidthC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u Read256BytesBlockHeightC=%u\n", __func__, k, Read256BytesBlockHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_y=%u\n", __func__, k, req_width_horz_y); + DML_LOG_VERBOSE("DML::%s: k=%u req_width_horz_c=%u\n", __func__, k, req_width_horz_c); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportStationary=%u\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.stationary); + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface=%u\n", __func__, k, DPPPerSurface[k]); + + req_per_swath_ub_l[k] = 0; + req_per_swath_ub_c[k] = 0; + if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { + MaximumSwathHeightY[k] = Read256BytesBlockHeightY[k]; + MaximumSwathHeightC[k] = Read256BytesBlockHeightC[k]; + if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { + swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_width_ub_l, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start + SwathWidthY[k] + req_width_horz_y - 1, req_width_horz_y) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start, req_width_horz_y))); + } else { + swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_width_ub_l, math_ceil2((double)SwathWidthY[k] - 1, req_width_horz_y) + req_width_horz_y)); + } + req_per_swath_ub_l[k] = swath_width_luma_ub[k] / req_width_horz_y; + + if (BytePerPixC[k] > 0) { + if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { + swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_width_ub_c, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start + SwathWidthC[k] + req_width_horz_c - 1, req_width_horz_c) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start, req_width_horz_c))); + } else { + swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_width_ub_c, math_ceil2((double)SwathWidthC[k] - 1, req_width_horz_c) + req_width_horz_c)); + } + req_per_swath_ub_c[k] = swath_width_chroma_ub[k] / req_width_horz_c; + } else { + swath_width_chroma_ub[k] = 0; + } + } else { + MaximumSwathHeightY[k] = Read256BytesBlockWidthY[k]; + MaximumSwathHeightC[k] = Read256BytesBlockWidthC[k]; + + if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { + swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_height_ub_l, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start + SwathWidthY[k] + Read256BytesBlockHeightY[k] - 1, Read256BytesBlockHeightY[k]) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start, Read256BytesBlockHeightY[k]))); + } else { + swath_width_luma_ub[k] = (unsigned int)(math_min2(surface_height_ub_l, math_ceil2((double)SwathWidthY[k] - 1, Read256BytesBlockHeightY[k]) + Read256BytesBlockHeightY[k])); + } + req_per_swath_ub_l[k] = swath_width_luma_ub[k] / Read256BytesBlockHeightY[k]; + if (BytePerPixC[k] > 0) { + if (display_cfg->plane_descriptors[k].composition.viewport.stationary && DPPPerSurface[k] == 1) { + swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_height_ub_c, math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start + SwathWidthC[k] + Read256BytesBlockHeightC[k] - 1, Read256BytesBlockHeightC[k]) - math_floor2(display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start, Read256BytesBlockHeightC[k]))); + } else { + swath_width_chroma_ub[k] = (unsigned int)(math_min2(surface_height_ub_c, math_ceil2((double)SwathWidthC[k] - 1, Read256BytesBlockHeightC[k]) + Read256BytesBlockHeightC[k])); + } + req_per_swath_ub_c[k] = swath_width_chroma_ub[k] / Read256BytesBlockHeightC[k]; + } else { + swath_width_chroma_ub[k] = 0; + } + } + + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub=%u\n", __func__, k, swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub=%u\n", __func__, k, swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY=%u\n", __func__, k, MaximumSwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC=%u\n", __func__, k, MaximumSwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_l=%u\n", __func__, k, req_per_swath_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u req_per_swath_ub_c=%u\n", __func__, k, req_per_swath_ub_c[k]); + } +} + +static bool UnboundedRequest(bool unb_req_force_en, bool unb_req_force_val, unsigned int TotalNumberOfActiveDPP, bool NoChromaOrLinear) +{ + bool unb_req_ok = false; + bool unb_req_en = false; + + unb_req_ok = (TotalNumberOfActiveDPP == 1 && NoChromaOrLinear); + unb_req_en = unb_req_ok; + + if (unb_req_force_en) { + unb_req_en = unb_req_force_val && unb_req_ok; + } + DML_LOG_VERBOSE("DML::%s: unb_req_force_en = %u\n", __func__, unb_req_force_en); + DML_LOG_VERBOSE("DML::%s: unb_req_force_val = %u\n", __func__, unb_req_force_val); + DML_LOG_VERBOSE("DML::%s: unb_req_ok = %u\n", __func__, unb_req_ok); + DML_LOG_VERBOSE("DML::%s: unb_req_en = %u\n", __func__, unb_req_en); + return unb_req_en; +} + +static void CalculateDETBufferSize( + struct dml2_core_shared_CalculateDETBufferSize_locals *l, + const struct dml2_display_cfg *display_cfg, + bool ForceSingleDPP, + unsigned int NumberOfActiveSurfaces, + bool UnboundedRequestEnabled, + unsigned int nomDETInKByte, + unsigned int MaxTotalDETInKByte, + unsigned int ConfigReturnBufferSizeInKByte, + unsigned int MinCompressedBufferSizeInKByte, + unsigned int ConfigReturnBufferSegmentSizeInkByte, + unsigned int CompressedBufferSegmentSizeInkByte, + double ReadBandwidthLuma[], + double ReadBandwidthChroma[], + unsigned int full_swath_bytes_l[], + unsigned int full_swath_bytes_c[], + unsigned int DPPPerSurface[], + // Output + unsigned int DETBufferSizeInKByte[], + unsigned int *CompressedBufferSizeInkByte) +{ + memset(l, 0, sizeof(struct dml2_core_shared_CalculateDETBufferSize_locals)); + + bool DETPieceAssignedToThisSurfaceAlready[DML2_MAX_PLANES]; + bool NextPotentialSurfaceToAssignDETPieceFound; + bool MinimizeReallocationSuccess = false; + + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, ForceSingleDPP); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: MaxTotalDETInKByte = %u\n", __func__, MaxTotalDETInKByte); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MinCompressedBufferSizeInKByte = %u\n", __func__, MinCompressedBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSegmentSizeInkByte = %u\n", __func__, CompressedBufferSegmentSizeInkByte); + + // Note: Will use default det size if that fits 2 swaths + if (UnboundedRequestEnabled) { + if (display_cfg->plane_descriptors[0].overrides.det_size_override_kb > 0) { + DETBufferSizeInKByte[0] = display_cfg->plane_descriptors[0].overrides.det_size_override_kb; + } else { + DETBufferSizeInKByte[0] = (unsigned int)math_max2(128.0, math_ceil2(2.0 * ((double)full_swath_bytes_l[0] + (double)full_swath_bytes_c[0]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte)); + } + *CompressedBufferSizeInkByte = ConfigReturnBufferSizeInKByte - DETBufferSizeInKByte[0]; + } else { + l->DETBufferSizePoolInKByte = MaxTotalDETInKByte; + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + DETBufferSizeInKByte[k] = 0; + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { + l->max_minDET = nomDETInKByte - ConfigReturnBufferSegmentSizeInkByte; + } else { + l->max_minDET = nomDETInKByte; + } + l->minDET = 128; + l->minDET_pipe = 0; + + // add DET resource until can hold 2 full swaths + while (l->minDET <= l->max_minDET && l->minDET_pipe == 0) { + if (2.0 * ((double)full_swath_bytes_l[k] + (double)full_swath_bytes_c[k]) / 1024.0 <= l->minDET) + l->minDET_pipe = l->minDET; + l->minDET = l->minDET + ConfigReturnBufferSegmentSizeInkByte; + } + + DML_LOG_VERBOSE("DML::%s: k=%u minDET = %u\n", __func__, k, l->minDET); + DML_LOG_VERBOSE("DML::%s: k=%u max_minDET = %u\n", __func__, k, l->max_minDET); + DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u\n", __func__, k, l->minDET_pipe); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, full_swath_bytes_c[k]); + + if (l->minDET_pipe == 0) { + l->minDET_pipe = (unsigned int)(math_max2(128, math_ceil2(((double)full_swath_bytes_l[k] + (double)full_swath_bytes_c[k]) / 1024.0, ConfigReturnBufferSegmentSizeInkByte))); + DML_LOG_VERBOSE("DML::%s: k=%u minDET_pipe = %u (assume each plane take half DET)\n", __func__, k, l->minDET_pipe); + } + + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + DETBufferSizeInKByte[k] = 0; + } else if (display_cfg->plane_descriptors[k].overrides.det_size_override_kb > 0) { + DETBufferSizeInKByte[k] = display_cfg->plane_descriptors[k].overrides.det_size_override_kb; + l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * display_cfg->plane_descriptors[k].overrides.det_size_override_kb; + } else if ((ForceSingleDPP ? 1 : DPPPerSurface[k]) * l->minDET_pipe <= l->DETBufferSizePoolInKByte) { + DETBufferSizeInKByte[k] = l->minDET_pipe; + l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - (ForceSingleDPP ? 1 : DPPPerSurface[k]) * l->minDET_pipe; + } + + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, DPPPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETSizeOverride = %u\n", __func__, k, display_cfg->plane_descriptors[k].overrides.det_size_override_kb); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + DML_LOG_VERBOSE("DML::%s: DETBufferSizePoolInKByte = %u\n", __func__, l->DETBufferSizePoolInKByte); + } + + if (display_cfg->minimize_det_reallocation) { + MinimizeReallocationSuccess = true; + // To minimize det reallocation, we don't distribute based on each surfaces bandwidth proportional to the global + // but rather distribute DET across streams proportionally based on pixel rate, and only distribute based on + // bandwidth between the planes on the same stream. This ensures that large scale re-distribution only on a + // stream count and/or pixel rate change, which is must less likely then general bandwidth changes per plane. + + // Calculate total pixel rate + for (unsigned int k = 0; k < display_cfg->num_streams; ++k) { + l->TotalPixelRate += display_cfg->stream_descriptors[k].timing.pixel_clock_khz; + } + + // Calculate per stream DET budget + for (unsigned int k = 0; k < display_cfg->num_streams; ++k) { + l->DETBudgetPerStream[k] = (unsigned int)((double) display_cfg->stream_descriptors[k].timing.pixel_clock_khz * MaxTotalDETInKByte / l->TotalPixelRate); + l->RemainingDETBudgetPerStream[k] = l->DETBudgetPerStream[k]; + } + + // Calculate the per stream total bandwidth + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + l->TotalBandwidthPerStream[display_cfg->plane_descriptors[k].stream_index] += (unsigned int)(ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); + + // Check the minimum can be satisfied by budget + if (l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] >= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k])) { + l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] -= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k]); + } else { + MinimizeReallocationSuccess = false; + break; + } + } + } + + if (MinimizeReallocationSuccess) { + // Since a fixed budget per stream is sufficient to satisfy the minimums, just re-distribute each streams + // budget proportionally across its planes + l->ResidualDETAfterRounding = MaxTotalDETInKByte; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + l->IdealDETBudget = (unsigned int)(((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / l->TotalBandwidthPerStream[display_cfg->plane_descriptors[k].stream_index]) + * l->DETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]); + + if (l->IdealDETBudget > DETBufferSizeInKByte[k]) { + l->DeltaDETBudget = l->IdealDETBudget - DETBufferSizeInKByte[k]; + if (l->DeltaDETBudget > l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]) + l->DeltaDETBudget = l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index]; + + /* split the additional budgeted DET among the pipes per plane */ + DETBufferSizeInKByte[k] += (unsigned int)((double)l->DeltaDETBudget / (ForceSingleDPP ? 1 : DPPPerSurface[k])); + l->RemainingDETBudgetPerStream[display_cfg->plane_descriptors[k].stream_index] -= l->DeltaDETBudget; + } + + // Round down to segment size + DETBufferSizeInKByte[k] = (DETBufferSizeInKByte[k] / ConfigReturnBufferSegmentSizeInkByte) * ConfigReturnBufferSegmentSizeInkByte; + + l->ResidualDETAfterRounding -= DETBufferSizeInKByte[k] * (ForceSingleDPP ? 1 : DPPPerSurface[k]); + } + } + } + } + + if (!MinimizeReallocationSuccess) { + l->TotalBandwidth = 0; + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + l->TotalBandwidth = l->TotalBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; + } + } + DML_LOG_VERBOSE("DML::%s: --- Before bandwidth adjustment ---\n", __func__); + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, DETBufferSizeInKByte[k]); + } + DML_LOG_VERBOSE("DML::%s: --- DET allocation with bandwidth ---\n", __func__); + DML_LOG_VERBOSE("DML::%s: TotalBandwidth = %f\n", __func__, l->TotalBandwidth); + l->BandwidthOfSurfacesNotAssignedDETPiece = l->TotalBandwidth; + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + DETPieceAssignedToThisSurfaceAlready[k] = true; + } else if (display_cfg->plane_descriptors[k].overrides.det_size_override_kb > 0 || (((double)(ForceSingleDPP ? 1 : DPPPerSurface[k]) * (double)DETBufferSizeInKByte[k] / (double)MaxTotalDETInKByte) >= ((ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) / l->TotalBandwidth))) { + DETPieceAssignedToThisSurfaceAlready[k] = true; + l->BandwidthOfSurfacesNotAssignedDETPiece = l->BandwidthOfSurfacesNotAssignedDETPiece - ReadBandwidthLuma[k] - ReadBandwidthChroma[k]; + } else { + DETPieceAssignedToThisSurfaceAlready[k] = false; + } + DML_LOG_VERBOSE("DML::%s: k=%u DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, k, DETPieceAssignedToThisSurfaceAlready[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, k, l->BandwidthOfSurfacesNotAssignedDETPiece); + } + + for (unsigned int j = 0; j < NumberOfActiveSurfaces; ++j) { + NextPotentialSurfaceToAssignDETPieceFound = false; + l->NextSurfaceToAssignDETPiece = 0; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[k] = %f\n", __func__, j, k, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[k] = %f\n", __func__, j, k, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthLuma[Next] = %f\n", __func__, j, k, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, ReadBandwidthChroma[Next] = %f\n", __func__, j, k, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, k, l->NextSurfaceToAssignDETPiece); + if (!DETPieceAssignedToThisSurfaceAlready[k] && (!NextPotentialSurfaceToAssignDETPieceFound || + ReadBandwidthLuma[k] + ReadBandwidthChroma[k] < ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece])) { + l->NextSurfaceToAssignDETPiece = k; + NextPotentialSurfaceToAssignDETPieceFound = true; + } + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, DETPieceAssignedToThisSurfaceAlready = %u\n", __func__, j, k, DETPieceAssignedToThisSurfaceAlready[k]); + DML_LOG_VERBOSE("DML::%s: j=%u k=%u, NextPotentialSurfaceToAssignDETPieceFound = %u\n", __func__, j, k, NextPotentialSurfaceToAssignDETPieceFound); + } + + if (NextPotentialSurfaceToAssignDETPieceFound) { + l->NextDETBufferPieceInKByte = (unsigned int)(math_min2( + math_round((double)l->DETBufferSizePoolInKByte * (ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]) / l->BandwidthOfSurfacesNotAssignedDETPiece / + ((ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte)) + * (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte, + math_floor2((double)l->DETBufferSizePoolInKByte, (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]) * ConfigReturnBufferSegmentSizeInkByte))); + + DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizePoolInKByte = %u\n", __func__, j, l->DETBufferSizePoolInKByte); + DML_LOG_VERBOSE("DML::%s: j=%u, NextSurfaceToAssignDETPiece = %u\n", __func__, j, l->NextSurfaceToAssignDETPiece); + DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthLuma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u, ReadBandwidthChroma[%u] = %f\n", __func__, j, l->NextSurfaceToAssignDETPiece, ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("DML::%s: j=%u, BandwidthOfSurfacesNotAssignedDETPiece = %f\n", __func__, j, l->BandwidthOfSurfacesNotAssignedDETPiece); + DML_LOG_VERBOSE("DML::%s: j=%u, NextDETBufferPieceInKByte = %u\n", __func__, j, l->NextDETBufferPieceInKByte); + DML_LOG_VERBOSE("DML::%s: j=%u, DETBufferSizeInKByte[%u] increases from %u ", __func__, j, l->NextSurfaceToAssignDETPiece, DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); + + DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] = DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece] + l->NextDETBufferPieceInKByte / (ForceSingleDPP ? 1 : DPPPerSurface[l->NextSurfaceToAssignDETPiece]); + DML_LOG_VERBOSE("to %u\n", DETBufferSizeInKByte[l->NextSurfaceToAssignDETPiece]); + + l->DETBufferSizePoolInKByte = l->DETBufferSizePoolInKByte - l->NextDETBufferPieceInKByte; + DETPieceAssignedToThisSurfaceAlready[l->NextSurfaceToAssignDETPiece] = true; + l->BandwidthOfSurfacesNotAssignedDETPiece = l->BandwidthOfSurfacesNotAssignedDETPiece - (ReadBandwidthLuma[l->NextSurfaceToAssignDETPiece] + ReadBandwidthChroma[l->NextSurfaceToAssignDETPiece]); + } + } + } + *CompressedBufferSizeInkByte = MinCompressedBufferSizeInKByte; + } + *CompressedBufferSizeInkByte = *CompressedBufferSizeInkByte * CompressedBufferSegmentSizeInkByte / ConfigReturnBufferSegmentSizeInkByte; + + DML_LOG_VERBOSE("DML::%s: --- After bandwidth adjustment ---\n", __func__); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *CompressedBufferSizeInkByte); + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u (TotalReadBandWidth=%f)\n", __func__, k, DETBufferSizeInKByte[k], ReadBandwidthLuma[k] + ReadBandwidthChroma[k]); + } +} + +static double CalculateRequiredDispclk( + enum dml2_odm_mode ODMMode, + double PixelClock, + bool isTMDS420) +{ + double DispClk; + + if (ODMMode == dml2_odm_mode_combine_4to1) { + DispClk = PixelClock / 4.0; + } else if (ODMMode == dml2_odm_mode_combine_3to1) { + DispClk = PixelClock / 3.0; + } else if (ODMMode == dml2_odm_mode_combine_2to1) { + DispClk = PixelClock / 2.0; + } else { + DispClk = PixelClock; + } + + if (isTMDS420) { + double TMDS420MinPixClock = PixelClock / 2.0; + DispClk = math_max2(DispClk, TMDS420MinPixClock); + } + + return DispClk; +} + +static double TruncToValidBPP( + struct dml2_core_shared_TruncToValidBPP_locals *l, + double LinkBitRate, + unsigned int Lanes, + unsigned int HTotal, + unsigned int HActive, + double PixelClock, + double DesiredBPP, + bool DSCEnable, + enum dml2_output_encoder_class Output, + enum dml2_output_format_class Format, + unsigned int DSCInputBitPerComponent, + unsigned int DSCSlices, + unsigned int AudioRate, + unsigned int AudioLayout, + enum dml2_odm_mode ODMModeNoDSC, + enum dml2_odm_mode ODMModeDSC, + + // Output + unsigned int *RequiredSlots) +{ + double MaxLinkBPP; + unsigned int MinDSCBPP; + double MaxDSCBPP; + unsigned int NonDSCBPP0; + unsigned int NonDSCBPP1; + unsigned int NonDSCBPP2; + enum dml2_odm_mode ODMMode; + + if (Format == dml2_420) { + NonDSCBPP0 = 12; + NonDSCBPP1 = 15; + NonDSCBPP2 = 18; + MinDSCBPP = 6; + MaxDSCBPP = 16; + } else if (Format == dml2_444) { + NonDSCBPP0 = 24; + NonDSCBPP1 = 30; + NonDSCBPP2 = 36; + MinDSCBPP = 8; + MaxDSCBPP = 16; + } else { + + if (Output == dml2_hdmi || Output == dml2_hdmifrl) { + NonDSCBPP0 = 24; + NonDSCBPP1 = 24; + NonDSCBPP2 = 24; + } else { + NonDSCBPP0 = 16; + NonDSCBPP1 = 20; + NonDSCBPP2 = 24; + } + if (Format == dml2_n422 || Output == dml2_hdmifrl) { + MinDSCBPP = 7; + MaxDSCBPP = 16; + } else { + MinDSCBPP = 8; + MaxDSCBPP = 16; + } + } + + if (Output == dml2_dp2p0) { + MaxLinkBPP = LinkBitRate * Lanes / PixelClock * 128.0 / 132.0 * 383.0 / 384.0 * 65536.0 / 65540.0; + } else if (DSCEnable && Output == dml2_dp) { + MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock * (1 - 2.4 / 100); + } else { + MaxLinkBPP = LinkBitRate / 10.0 * 8.0 * Lanes / PixelClock; + } + + ODMMode = DSCEnable ? ODMModeDSC : ODMModeNoDSC; + + if (ODMMode == dml2_odm_mode_split_1to2) { + MaxLinkBPP = 2 * MaxLinkBPP; + } + + if (DesiredBPP == 0) { + if (DSCEnable) { + if (MaxLinkBPP < MinDSCBPP) { + return __DML2_CALCS_DPP_INVALID__; + } else if (MaxLinkBPP >= MaxDSCBPP) { + return MaxDSCBPP; + } else { + return math_floor2(16.0 * MaxLinkBPP, 1.0) / 16.0; + } + } else { + if (MaxLinkBPP >= NonDSCBPP2) { + return NonDSCBPP2; + } else if (MaxLinkBPP >= NonDSCBPP1) { + return NonDSCBPP1; + } else if (MaxLinkBPP >= NonDSCBPP0) { + return NonDSCBPP0; + } else { + return __DML2_CALCS_DPP_INVALID__; + } + } + } else { + if (!((DSCEnable == false && (DesiredBPP == NonDSCBPP2 || DesiredBPP == NonDSCBPP1 || DesiredBPP == NonDSCBPP0)) || + (DSCEnable && DesiredBPP >= MinDSCBPP && DesiredBPP <= MaxDSCBPP))) { + return __DML2_CALCS_DPP_INVALID__; + } else { + return DesiredBPP; + } + } +} + +// updated for dcn4 +static unsigned int dscceComputeDelay( + unsigned int bpc, + double BPP, + unsigned int sliceWidth, + unsigned int numSlices, + enum dml2_output_format_class pixelFormat, + enum dml2_output_encoder_class Output) +{ + // valid bpc = source bits per component in the set of {8, 10, 12} + // valid bpp = increments of 1/16 of a bit + // min = 6/7/8 in N420/N422/444, respectively + // max = such that compression is 1:1 + //valid sliceWidth = number of pixels per slice line, must be less than or equal to 5184/numSlices (or 4096/numSlices in 420 mode) + //valid numSlices = number of slices in the horiziontal direction per DSC engine in the set of {1, 2, 3, 4} + //valid pixelFormat = pixel/color format in the set of {:N444_RGB, :S422, :N422, :N420} + + // fixed value + unsigned int rcModelSize = 8192; + + // N422/N420 operate at 2 pixels per clock + unsigned int pixelsPerClock, padding_pixels, ssm_group_priming_delay, ssm_pipeline_delay, obsm_pipeline_delay, slice_padded_pixels, ixd_plus_padding, ixd_plus_padding_groups, cycles_per_group, group_delay, pipeline_delay, pixels, additional_group_delay, lines_to_reach_ixd, groups_to_reach_ixd, slice_width_groups, initial_xmit_delay, number_of_lines_to_reach_ixd, slice_width_modified; + + if (pixelFormat == dml2_420) + pixelsPerClock = 2; + // #all other modes operate at 1 pixel per clock + else if (pixelFormat == dml2_444) + pixelsPerClock = 1; + else if (pixelFormat == dml2_n422 || Output == dml2_hdmifrl) + pixelsPerClock = 2; + else + pixelsPerClock = 1; + + //initial transmit delay as per PPS + initial_xmit_delay = (unsigned int)(math_round(rcModelSize / 2.0 / BPP / pixelsPerClock)); + + //slice width as seen by dscc_bcl in pixels or pixels pairs (depending on number of pixels per pixel container based on pixel format) + slice_width_modified = (pixelFormat == dml2_444 || pixelFormat == dml2_420 || Output == dml2_hdmifrl) ? sliceWidth / 2 : sliceWidth; + + padding_pixels = ((slice_width_modified % 3) != 0) ? (3 - (slice_width_modified % 3)) * (initial_xmit_delay / slice_width_modified) : 0; + + if ((3.0 * pixelsPerClock * BPP) >= ((double)((initial_xmit_delay + 2) / 3) * (double)(3 + (pixelFormat == dml2_n422)))) { + if ((initial_xmit_delay + padding_pixels) % 3 == 1) { + initial_xmit_delay++; + } + } + + //sub-stream multiplexer balance fifo priming delay in groups as per dsc standard + if (bpc == 8) + ssm_group_priming_delay = 83; + else if (bpc == 10) + ssm_group_priming_delay = 91; + else if (bpc == 12) + ssm_group_priming_delay = 115; + else if (bpc == 14) + ssm_group_priming_delay = 123; + else + ssm_group_priming_delay = 128; + + //slice width in groups is rounded up to the nearest group as DSC adds padded pixels such that there are an integer number of groups per slice + slice_width_groups = (slice_width_modified + 2) / 3; + + //determine number of padded pixels in the last group of a slice line, computed as + slice_padded_pixels = 3 * slice_width_groups - slice_width_modified; + + //determine integer number of complete slice lines required to reach initial transmit delay without ssm delay considered + number_of_lines_to_reach_ixd = initial_xmit_delay / slice_width_modified; + + //increase initial transmit delay by the number of padded pixels added to a slice line multipled by the integer number of complete lines to reach initial transmit delay + //this step is necessary as each padded pixel added takes up a clock cycle and, therefore, adds to the overall delay + ixd_plus_padding = initial_xmit_delay + slice_padded_pixels * number_of_lines_to_reach_ixd; + + //convert the padded initial transmit delay from pixels to groups by rounding up to the nearest group as DSC processes in groups of pixels + ixd_plus_padding_groups = (ixd_plus_padding + 2) / 3; + + //number of groups required for a slice to reach initial transmit delay is the sum of the padded initial transmit delay plus the ssm group priming delay + groups_to_reach_ixd = ixd_plus_padding_groups + ssm_group_priming_delay; + + //number of lines required to reach padded initial transmit delay in groups in slices to the left of the last horizontal slice + //needs to be rounded up as a complete slice lines are buffered prior to initial transmit delay being reached in the last horizontal slice + lines_to_reach_ixd = (groups_to_reach_ixd + slice_width_groups - 1) / slice_width_groups; //round up lines to reach ixd to next + + //determine if there are non-zero number of pixels reached in the group where initial transmit delay is reached + //an additional group time (i.e., 3 pixel times) is required before the first output if there are no additional pixels beyond initial transmit delay + additional_group_delay = ((initial_xmit_delay - number_of_lines_to_reach_ixd * slice_width_modified) % 3) == 0 ? 1 : 0; + + //number of pipeline delay cycles in the ssm block (can be determined empirically or analytically by inspecting the ssm block) + ssm_pipeline_delay = 2; + + //number of pipe delay cycles in the obsm block (can be determined empirically or analytically by inspecting the obsm block) + obsm_pipeline_delay = 1; + + //a group of pixels is worth 6 pixels in N422/N420 mode or 3 pixels in all other modes + if (pixelFormat == dml2_420 || pixelFormat == dml2_444 || pixelFormat == dml2_n422 || Output == dml2_hdmifrl) + cycles_per_group = 6; + else + cycles_per_group = 3; + //delay of the bit stream contruction layer in pixels is the sum of: + //1. number of pixel containers in a slice line multipled by the number of lines required to reach initial transmit delay multipled by number of slices to the left of the last horizontal slice + //2. number of pixel containers required to reach initial transmit delay (specifically, in the last horizontal slice) + //3. additional group of delay if initial transmit delay is reached exactly in a group + //4. ssm and obsm pipeline delay (i.e., clock cycles of delay) + group_delay = (lines_to_reach_ixd * slice_width_groups * (numSlices - 1)) + groups_to_reach_ixd + additional_group_delay; + pipeline_delay = ssm_pipeline_delay + obsm_pipeline_delay; + + //pixel delay is group_delay (converted to pixels) + pipeline, however, first group is a special case since it is processed as soon as it arrives (i.e., in 3 cycles regardless of pixel format) + pixels = (group_delay - 1) * cycles_per_group + 3 + pipeline_delay; + + DML_LOG_VERBOSE("DML::%s: bpc: %u\n", __func__, bpc); + DML_LOG_VERBOSE("DML::%s: BPP: %f\n", __func__, BPP); + DML_LOG_VERBOSE("DML::%s: sliceWidth: %u\n", __func__, sliceWidth); + DML_LOG_VERBOSE("DML::%s: numSlices: %u\n", __func__, numSlices); + DML_LOG_VERBOSE("DML::%s: pixelFormat: %u\n", __func__, pixelFormat); + DML_LOG_VERBOSE("DML::%s: Output: %u\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: pixels: %u\n", __func__, pixels); + return pixels; +} + +//updated in dcn4 +static unsigned int dscComputeDelay(enum dml2_output_format_class pixelFormat, enum dml2_output_encoder_class Output) +{ + unsigned int Delay = 0; + unsigned int dispclk_per_dscclk = 3; + + // sfr + Delay = Delay + 2; + + if (pixelFormat == dml2_420 || pixelFormat == dml2_n422 || (Output == dml2_hdmifrl && pixelFormat != dml2_444)) { + dispclk_per_dscclk = 3 * 2; + } + + if (pixelFormat == dml2_420) { + //dscc top delay for pixel compression layer + Delay = Delay + 16 * dispclk_per_dscclk; + + // dscc - input deserializer + Delay = Delay + 5; + + // dscc - input cdc fifo + Delay = Delay + 1 + 4 * dispclk_per_dscclk; + + // dscc - output cdc fifo + Delay = Delay + 3 + 1 * dispclk_per_dscclk; + + // dscc - cdc uncertainty + Delay = Delay + 3 + 3 * dispclk_per_dscclk; + } else if (pixelFormat == dml2_n422 || (Output == dml2_hdmifrl && pixelFormat != dml2_444)) { + //dscc top delay for pixel compression layer + Delay = Delay + 16 * dispclk_per_dscclk; + // dsccif + Delay = Delay + 1; + // dscc - input deserializer + Delay = Delay + 5; + // dscc - input cdc fifo + Delay = Delay + 1 + 4 * dispclk_per_dscclk; + + + // dscc - output cdc fifo + Delay = Delay + 3 + 1 * dispclk_per_dscclk; + // dscc - cdc uncertainty + Delay = Delay + 3 + 3 * dispclk_per_dscclk; + } else if (pixelFormat == dml2_s422) { + //dscc top delay for pixel compression layer + Delay = Delay + 17 * dispclk_per_dscclk; + + // dscc - input deserializer + Delay = Delay + 3; + // dscc - input cdc fifo + Delay = Delay + 1 + 4 * dispclk_per_dscclk; + // dscc - output cdc fifo + Delay = Delay + 3 + 1 * dispclk_per_dscclk; + // dscc - cdc uncertainty + Delay = Delay + 3 + 3 * dispclk_per_dscclk; + } else { + //dscc top delay for pixel compression layer + Delay = Delay + 16 * dispclk_per_dscclk; + // dscc - input deserializer + Delay = Delay + 3; + // dscc - input cdc fifo + Delay = Delay + 1 + 4 * dispclk_per_dscclk; + // dscc - output cdc fifo + Delay = Delay + 3 + 1 * dispclk_per_dscclk; + + // dscc - cdc uncertainty + Delay = Delay + 3 + 3 * dispclk_per_dscclk; + } + + // sft + Delay = Delay + 1; + DML_LOG_VERBOSE("DML::%s: pixelFormat = %u\n", __func__, pixelFormat); + DML_LOG_VERBOSE("DML::%s: Delay = %u\n", __func__, Delay); + + return Delay; +} + +static unsigned int CalculateHostVMDynamicLevels( + bool GPUVMEnable, + bool HostVMEnable, + unsigned int HostVMMinPageSize, + unsigned int HostVMMaxNonCachedPageTableLevels) +{ + unsigned int HostVMDynamicLevels = 0; + + if (GPUVMEnable && HostVMEnable) { + if (HostVMMinPageSize < 2048) + HostVMDynamicLevels = HostVMMaxNonCachedPageTableLevels; + else if (HostVMMinPageSize >= 2048 && HostVMMinPageSize < 1048576) + HostVMDynamicLevels = (unsigned int)math_max2(0, (double)HostVMMaxNonCachedPageTableLevels - 1); + else + HostVMDynamicLevels = (unsigned int)math_max2(0, (double)HostVMMaxNonCachedPageTableLevels - 2); + } else { + HostVMDynamicLevels = 0; + } + return HostVMDynamicLevels; +} + +static unsigned int CalculateVMAndRowBytes(struct dml2_core_shared_calculate_vm_and_row_bytes_params *p) +{ + unsigned int extra_dpde_bytes; + unsigned int extra_mpde_bytes; + unsigned int MacroTileSizeBytes; + unsigned int vp_height_dpte_ub; + + unsigned int meta_surface_bytes; + unsigned int vm_bytes; + unsigned int vp_height_meta_ub; + unsigned int PixelPTEReqWidth_linear = 0; // VBA_DELTA. VBA doesn't calculate this + + *p->MetaRequestHeight = 8 * p->BlockHeight256Bytes; + *p->MetaRequestWidth = 8 * p->BlockWidth256Bytes; + if (p->SurfaceTiling == dml2_sw_linear) { + *p->meta_row_height = 32; + *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->MetaRequestWidth - 1, *p->MetaRequestWidth) - math_floor2(p->ViewportXStart, *p->MetaRequestWidth)); + *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestHeight * p->BytePerPixel / 256.0); // FIXME_DCN4SW missing in old code but no dcc for linear anyways? + } else if (!dml_is_vertical_rotation(p->RotationAngle)) { + *p->meta_row_height = *p->MetaRequestHeight; + if (p->ViewportStationary && p->NumberOfDPPs == 1) { + *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->MetaRequestWidth - 1, *p->MetaRequestWidth) - math_floor2(p->ViewportXStart, *p->MetaRequestWidth)); + } else { + *p->meta_row_width = (unsigned int)(math_ceil2(p->SwathWidth - 1, *p->MetaRequestWidth) + *p->MetaRequestWidth); + } + *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestHeight * p->BytePerPixel / 256.0); + } else { + *p->meta_row_height = *p->MetaRequestWidth; + if (p->ViewportStationary && p->NumberOfDPPs == 1) { + *p->meta_row_width = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + *p->MetaRequestHeight - 1, *p->MetaRequestHeight) - math_floor2(p->ViewportYStart, *p->MetaRequestHeight)); + } else { + *p->meta_row_width = (unsigned int)(math_ceil2(p->SwathWidth - 1, *p->MetaRequestHeight) + *p->MetaRequestHeight); + } + *p->meta_row_bytes = (unsigned int)(*p->meta_row_width * *p->MetaRequestWidth * p->BytePerPixel / 256.0); + } + + if (p->ViewportStationary && p->is_phantom && (p->NumberOfDPPs == 1 || !dml_is_vertical_rotation(p->RotationAngle))) { + vp_height_meta_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + 64 * p->BlockHeight256Bytes - 1, 64 * p->BlockHeight256Bytes) - math_floor2(p->ViewportYStart, 64 * p->BlockHeight256Bytes)); + } else if (!dml_is_vertical_rotation(p->RotationAngle)) { + vp_height_meta_ub = (unsigned int)(math_ceil2(p->ViewportHeight - 1, 64 * p->BlockHeight256Bytes) + 64 * p->BlockHeight256Bytes); + } else { + vp_height_meta_ub = (unsigned int)(math_ceil2(p->SwathWidth - 1, 64 * p->BlockHeight256Bytes) + 64 * p->BlockHeight256Bytes); + } + + meta_surface_bytes = (unsigned int)(p->DCCMetaPitch * vp_height_meta_ub * p->BytePerPixel / 256.0); + DML_LOG_VERBOSE("DML::%s: DCCMetaPitch = %u\n", __func__, p->DCCMetaPitch); + DML_LOG_VERBOSE("DML::%s: meta_surface_bytes = %u\n", __func__, meta_surface_bytes); + if (p->GPUVMEnable == true) { + double meta_vmpg_bytes = 4.0 * 1024.0; + *p->meta_pte_bytes_per_frame_ub = (unsigned int)((math_ceil2((double) (meta_surface_bytes - meta_vmpg_bytes) / (8 * meta_vmpg_bytes), 1) + 1) * 64); + extra_mpde_bytes = 128 * (p->GPUVMMaxPageTableLevels - 1); + } else { + *p->meta_pte_bytes_per_frame_ub = 0; + extra_mpde_bytes = 0; + } + + if (!p->DCCEnable || !p->mrq_present) { + *p->meta_pte_bytes_per_frame_ub = 0; + extra_mpde_bytes = 0; + *p->meta_row_bytes = 0; + } + + if (!p->GPUVMEnable) { + *p->PixelPTEBytesPerRow = 0; + *p->PixelPTEBytesPerRowStorage = 0; + *p->dpte_row_width_ub = 0; + *p->dpte_row_height = 0; + *p->dpte_row_height_linear = 0; + *p->PixelPTEBytesPerRow_one_row_per_frame = 0; + *p->dpte_row_width_ub_one_row_per_frame = 0; + *p->dpte_row_height_one_row_per_frame = 0; + *p->vmpg_width = 0; + *p->vmpg_height = 0; + *p->PixelPTEReqWidth = 0; + *p->PixelPTEReqHeight = 0; + *p->PTERequestSize = 0; + *p->dpde0_bytes_per_frame_ub = 0; + return 0; + } + + MacroTileSizeBytes = p->MacroTileWidth * p->BytePerPixel * p->MacroTileHeight; + + if (p->ViewportStationary && p->is_phantom && (p->NumberOfDPPs == 1 || !dml_is_vertical_rotation(p->RotationAngle))) { + vp_height_dpte_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + p->MacroTileHeight - 1, p->MacroTileHeight) - math_floor2(p->ViewportYStart, p->MacroTileHeight)); + } else if (!dml_is_vertical_rotation(p->RotationAngle)) { + vp_height_dpte_ub = (unsigned int)(math_ceil2((double)p->ViewportHeight - 1, p->MacroTileHeight) + p->MacroTileHeight); + } else { + vp_height_dpte_ub = (unsigned int)(math_ceil2((double)p->SwathWidth - 1, p->MacroTileHeight) + p->MacroTileHeight); + } + + if (p->GPUVMEnable == true && p->GPUVMMaxPageTableLevels > 1) { + *p->dpde0_bytes_per_frame_ub = (unsigned int)(64 * (math_ceil2((double)(p->Pitch * vp_height_dpte_ub * p->BytePerPixel - MacroTileSizeBytes) / (double)(8 * 2097152), 1) + 1)); + extra_dpde_bytes = 128 * (p->GPUVMMaxPageTableLevels - 2); + } else { + *p->dpde0_bytes_per_frame_ub = 0; + extra_dpde_bytes = 0; + } + + vm_bytes = *p->meta_pte_bytes_per_frame_ub + extra_mpde_bytes + *p->dpde0_bytes_per_frame_ub + extra_dpde_bytes; + + DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->DCCEnable); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: SwModeLinear = %u\n", __func__, p->SurfaceTiling == dml2_sw_linear); + DML_LOG_VERBOSE("DML::%s: BytePerPixel = %u\n", __func__, p->BytePerPixel); + DML_LOG_VERBOSE("DML::%s: GPUVMMaxPageTableLevels = %u\n", __func__, p->GPUVMMaxPageTableLevels); + DML_LOG_VERBOSE("DML::%s: BlockHeight256Bytes = %u\n", __func__, p->BlockHeight256Bytes); + DML_LOG_VERBOSE("DML::%s: BlockWidth256Bytes = %u\n", __func__, p->BlockWidth256Bytes); + DML_LOG_VERBOSE("DML::%s: MacroTileHeight = %u\n", __func__, p->MacroTileHeight); + DML_LOG_VERBOSE("DML::%s: MacroTileWidth = %u\n", __func__, p->MacroTileWidth); + DML_LOG_VERBOSE("DML::%s: meta_pte_bytes_per_frame_ub = %u\n", __func__, *p->meta_pte_bytes_per_frame_ub); + DML_LOG_VERBOSE("DML::%s: dpde0_bytes_per_frame_ub = %u\n", __func__, *p->dpde0_bytes_per_frame_ub); + DML_LOG_VERBOSE("DML::%s: extra_mpde_bytes = %u\n", __func__, extra_mpde_bytes); + DML_LOG_VERBOSE("DML::%s: extra_dpde_bytes = %u\n", __func__, extra_dpde_bytes); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: ViewportHeight = %u\n", __func__, p->ViewportHeight); + DML_LOG_VERBOSE("DML::%s: SwathWidth = %u\n", __func__, p->SwathWidth); + DML_LOG_VERBOSE("DML::%s: vp_height_dpte_ub = %u\n", __func__, vp_height_dpte_ub); + + if (p->SurfaceTiling == dml2_sw_linear) { + *p->PixelPTEReqHeight = 1; + *p->PixelPTEReqWidth = p->GPUVMMinPageSizeKBytes * 1024 * 8 / p->BytePerPixel; + PixelPTEReqWidth_linear = p->GPUVMMinPageSizeKBytes * 1024 * 8 / p->BytePerPixel; + *p->PTERequestSize = 64; + + *p->vmpg_height = 1; + *p->vmpg_width = p->GPUVMMinPageSizeKBytes * 1024 / p->BytePerPixel; + } else if (p->GPUVMMinPageSizeKBytes * 1024 >= dml_get_tile_block_size_bytes(p->SurfaceTiling)) { // 1 64B 8x1 PTE + *p->PixelPTEReqHeight = p->MacroTileHeight; + *p->PixelPTEReqWidth = 8 * 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); + *p->PTERequestSize = 64; + + *p->vmpg_height = p->MacroTileHeight; + *p->vmpg_width = 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); + + } else if (p->GPUVMMinPageSizeKBytes == 4 && dml_get_tile_block_size_bytes(p->SurfaceTiling) == 65536) { // 2 64B PTE requests to get 16 PTEs to cover the 64K tile + // one 64KB tile, is 16x16x256B req + *p->PixelPTEReqHeight = 16 * p->BlockHeight256Bytes; + *p->PixelPTEReqWidth = 16 * p->BlockWidth256Bytes; + *p->PTERequestSize = 128; + + *p->vmpg_height = *p->PixelPTEReqHeight; + *p->vmpg_width = *p->PixelPTEReqWidth; + } else { + // default for rest of calculation to go through, when vm is disable, the calulated pte related values shouldnt be used anyways + *p->PixelPTEReqHeight = p->MacroTileHeight; + *p->PixelPTEReqWidth = 8 * 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); + *p->PTERequestSize = 64; + + *p->vmpg_height = p->MacroTileHeight; + *p->vmpg_width = 1024 * p->GPUVMMinPageSizeKBytes / (p->MacroTileHeight * p->BytePerPixel); + + if (p->GPUVMEnable == true) { + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes=%u and sw_mode=%u (tile_size=%d) not supported!\n", + __func__, p->GPUVMMinPageSizeKBytes, p->SurfaceTiling, dml_get_tile_block_size_bytes(p->SurfaceTiling)); + DML_ASSERT(0); + } + } + + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqHeight = %u\n", __func__, *p->PixelPTEReqHeight); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth = %u\n", __func__, *p->PixelPTEReqWidth); + DML_LOG_VERBOSE("DML::%s: PixelPTEReqWidth_linear = %u\n", __func__, PixelPTEReqWidth_linear); + DML_LOG_VERBOSE("DML::%s: PTERequestSize = %u\n", __func__, *p->PTERequestSize); + DML_LOG_VERBOSE("DML::%s: Pitch = %u\n", __func__, p->Pitch); + DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, *p->vmpg_width); + DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, *p->vmpg_height); + + *p->dpte_row_height_one_row_per_frame = vp_height_dpte_ub; + *p->dpte_row_width_ub_one_row_per_frame = (unsigned int)((math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height_one_row_per_frame / (double)*p->PixelPTEReqHeight - 1) / (double)*p->PixelPTEReqWidth, 1) + 1) * (double)*p->PixelPTEReqWidth); + *p->PixelPTEBytesPerRow_one_row_per_frame = (unsigned int)((double)*p->dpte_row_width_ub_one_row_per_frame / (double)*p->PixelPTEReqWidth * *p->PTERequestSize); + *p->dpte_row_height_linear = 0; + + if (p->SurfaceTiling == dml2_sw_linear) { + *p->dpte_row_height = (unsigned int)(math_min2(128, (double)(1ULL << (unsigned int)math_floor2(math_log((float)(p->PTEBufferSizeInRequests * *p->PixelPTEReqWidth / p->Pitch), 2.0), 1)))); + *p->dpte_row_width_ub = (unsigned int)(math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height - 1), (double)*p->PixelPTEReqWidth) + *p->PixelPTEReqWidth); + *p->PixelPTEBytesPerRow = (unsigned int)((double)*p->dpte_row_width_ub / (double)*p->PixelPTEReqWidth * *p->PTERequestSize); + + // VBA_DELTA, VBA doesn't have programming value for pte row height linear. + *p->dpte_row_height_linear = (unsigned int)1 << (unsigned int)math_floor2(math_log((float)(p->PTEBufferSizeInRequests * PixelPTEReqWidth_linear / p->Pitch), 2.0), 1); + if (*p->dpte_row_height_linear > 128) + *p->dpte_row_height_linear = 128; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (linear)\n", __func__, *p->dpte_row_width_ub); +#endif + + } else if (!dml_is_vertical_rotation(p->RotationAngle)) { + *p->dpte_row_height = *p->PixelPTEReqHeight; + + if (p->GPUVMMinPageSizeKBytes > 64) { + *p->dpte_row_width_ub = (unsigned int)((math_ceil2(((double)p->Pitch * (double)*p->dpte_row_height / (double)*p->PixelPTEReqHeight - 1) / (double)*p->PixelPTEReqWidth, 1) + 1) * *p->PixelPTEReqWidth); + } else if (p->ViewportStationary && (p->NumberOfDPPs == 1)) { + *p->dpte_row_width_ub = (unsigned int)(math_floor2(p->ViewportXStart + p->SwathWidth + *p->PixelPTEReqWidth - 1, *p->PixelPTEReqWidth) - math_floor2(p->ViewportXStart, *p->PixelPTEReqWidth)); + } else { + *p->dpte_row_width_ub = (unsigned int)((math_ceil2((double)(p->SwathWidth - 1) / (double)*p->PixelPTEReqWidth, 1) + 1.0) * *p->PixelPTEReqWidth); + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled horz)\n", __func__, *p->dpte_row_width_ub); +#endif + + *p->PixelPTEBytesPerRow = *p->dpte_row_width_ub / *p->PixelPTEReqWidth * *p->PTERequestSize; + } else { + *p->dpte_row_height = (unsigned int)(math_min2(*p->PixelPTEReqWidth, p->MacroTileWidth)); + + if (p->ViewportStationary && (p->NumberOfDPPs == 1)) { + *p->dpte_row_width_ub = (unsigned int)(math_floor2(p->ViewportYStart + p->ViewportHeight + *p->PixelPTEReqHeight - 1, *p->PixelPTEReqHeight) - math_floor2(p->ViewportYStart, *p->PixelPTEReqHeight)); + } else { + *p->dpte_row_width_ub = (unsigned int)((math_ceil2((double)(p->SwathWidth - 1) / (double)*p->PixelPTEReqHeight, 1) + 1) * *p->PixelPTEReqHeight); + } + + *p->PixelPTEBytesPerRow = (unsigned int)((double)*p->dpte_row_width_ub / (double)*p->PixelPTEReqHeight * *p->PTERequestSize); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u (tiled vert)\n", __func__, *p->dpte_row_width_ub); +#endif + } + + if (p->GPUVMEnable != true) { + *p->PixelPTEBytesPerRow = 0; + *p->PixelPTEBytesPerRow_one_row_per_frame = 0; + } + + *p->PixelPTEBytesPerRowStorage = *p->PixelPTEBytesPerRow; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: GPUVMMinPageSizeKBytes = %u\n", __func__, p->GPUVMMinPageSizeKBytes); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: meta_row_height = %u\n", __func__, *p->meta_row_height); + DML_LOG_VERBOSE("DML::%s: dpte_row_height = %u\n", __func__, *p->dpte_row_height); + DML_LOG_VERBOSE("DML::%s: dpte_row_height_linear = %u\n", __func__, *p->dpte_row_height_linear); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub = %u\n", __func__, *p->dpte_row_width_ub); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, *p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRowStorage = %u\n", __func__, *p->PixelPTEBytesPerRowStorage); + DML_LOG_VERBOSE("DML::%s: PTEBufferSizeInRequests = %u\n", __func__, p->PTEBufferSizeInRequests); + DML_LOG_VERBOSE("DML::%s: dpte_row_height_one_row_per_frame = %u\n", __func__, *p->dpte_row_height_one_row_per_frame); + DML_LOG_VERBOSE("DML::%s: dpte_row_width_ub_one_row_per_frame = %u\n", __func__, *p->dpte_row_width_ub_one_row_per_frame); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow_one_row_per_frame = %u\n", __func__, *p->PixelPTEBytesPerRow_one_row_per_frame); +#endif + + return vm_bytes; +} // CalculateVMAndRowBytes + +static unsigned int CalculatePrefetchSourceLines( + double VRatio, + unsigned int VTaps, + bool Interlace, + bool ProgressiveToInterlaceUnitInOPP, + unsigned int SwathHeight, + enum dml2_rotation_angle RotationAngle, + bool mirrored, + bool ViewportStationary, + unsigned int SwathWidth, + unsigned int ViewportHeight, + unsigned int ViewportXStart, + unsigned int ViewportYStart, + + // Output + unsigned int *VInitPreFill, + unsigned int *MaxNumSwath) +{ + + unsigned int vp_start_rot = 0; + unsigned int sw0_tmp = 0; + unsigned int MaxPartialSwath = 0; + double numLines = 0; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); + DML_LOG_VERBOSE("DML::%s: VTaps = %u\n", __func__, VTaps); + DML_LOG_VERBOSE("DML::%s: ViewportXStart = %u\n", __func__, ViewportXStart); + DML_LOG_VERBOSE("DML::%s: ViewportYStart = %u\n", __func__, ViewportYStart); + DML_LOG_VERBOSE("DML::%s: ViewportStationary = %u\n", __func__, ViewportStationary); + DML_LOG_VERBOSE("DML::%s: SwathHeight = %u\n", __func__, SwathHeight); +#endif + if (ProgressiveToInterlaceUnitInOPP) + *VInitPreFill = (unsigned int)(math_floor2((VRatio + (double)VTaps + 1) / 2.0, 1)); + else + *VInitPreFill = (unsigned int)(math_floor2((VRatio + (double)VTaps + 1 + (Interlace ? 1 : 0) * 0.5 * VRatio) / 2.0, 1)); + + if (ViewportStationary) { + if (RotationAngle == dml2_rotation_180) { + vp_start_rot = SwathHeight - (((unsigned int)(ViewportYStart + ViewportHeight - 1) % SwathHeight) + 1); + } else if ((RotationAngle == dml2_rotation_270 && !mirrored) || (RotationAngle == dml2_rotation_90 && mirrored)) { + vp_start_rot = ViewportXStart; + } else if ((RotationAngle == dml2_rotation_90 && !mirrored) || (RotationAngle == dml2_rotation_270 && mirrored)) { + vp_start_rot = SwathHeight - (((unsigned int)(ViewportYStart + SwathWidth - 1) % SwathHeight) + 1); + } else { + vp_start_rot = ViewportYStart; + } + sw0_tmp = SwathHeight - (vp_start_rot % SwathHeight); + if (sw0_tmp < *VInitPreFill) { + *MaxNumSwath = (unsigned int)(math_ceil2((*VInitPreFill - sw0_tmp) / (double)SwathHeight, 1) + 1); + } else { + *MaxNumSwath = 1; + } + MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(vp_start_rot + *VInitPreFill - 1) % SwathHeight)); + } else { + *MaxNumSwath = (unsigned int)(math_ceil2((*VInitPreFill - 1.0) / (double)SwathHeight, 1) + 1); + if (*VInitPreFill > 1) { + MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(*VInitPreFill - 2) % SwathHeight)); + } else { + MaxPartialSwath = (unsigned int)(math_max2(1, (unsigned int)(*VInitPreFill + SwathHeight - 2) % SwathHeight)); + } + } + numLines = *MaxNumSwath * SwathHeight + MaxPartialSwath; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: vp_start_rot = %u\n", __func__, vp_start_rot); + DML_LOG_VERBOSE("DML::%s: VInitPreFill = %u\n", __func__, *VInitPreFill); + DML_LOG_VERBOSE("DML::%s: MaxPartialSwath = %u\n", __func__, MaxPartialSwath); + DML_LOG_VERBOSE("DML::%s: MaxNumSwath = %u\n", __func__, *MaxNumSwath); + DML_LOG_VERBOSE("DML::%s: Prefetch source lines = %3.2f\n", __func__, numLines); +#endif + return (unsigned int)(numLines); + +} + +static void CalculateRowBandwidth( + bool GPUVMEnable, + bool use_one_row_for_frame, + enum dml2_source_format_class SourcePixelFormat, + double VRatio, + double VRatioChroma, + bool DCCEnable, + double LineTime, + unsigned int PixelPTEBytesPerRowLuma, + unsigned int PixelPTEBytesPerRowChroma, + unsigned int dpte_row_height_luma, + unsigned int dpte_row_height_chroma, + + bool mrq_present, + unsigned int meta_row_bytes_per_row_ub_l, + unsigned int meta_row_bytes_per_row_ub_c, + unsigned int meta_row_height_luma, + unsigned int meta_row_height_chroma, + + // Output + double *dpte_row_bw, + double *meta_row_bw) +{ + if (!DCCEnable || !mrq_present) { + *meta_row_bw = 0; + } else if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) { + *meta_row_bw = VRatio * meta_row_bytes_per_row_ub_l / (meta_row_height_luma * LineTime) + + VRatioChroma * meta_row_bytes_per_row_ub_c / (meta_row_height_chroma * LineTime); + } else { + *meta_row_bw = VRatio * meta_row_bytes_per_row_ub_l / (meta_row_height_luma * LineTime); + } + + if (GPUVMEnable != true) { + *dpte_row_bw = 0; + } else if (dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha) { + *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime) + + VRatioChroma * PixelPTEBytesPerRowChroma / (dpte_row_height_chroma * LineTime); + } else { + *dpte_row_bw = VRatio * PixelPTEBytesPerRowLuma / (dpte_row_height_luma * LineTime); + } +} + +static void CalculateMALLUseForStaticScreen( + const struct dml2_display_cfg *display_cfg, + unsigned int NumberOfActiveSurfaces, + unsigned int MALLAllocatedForDCN, + unsigned int SurfaceSizeInMALL[], + bool one_row_per_frame_fits_in_buffer[], + + // Output + bool is_using_mall_for_ss[]) +{ + + unsigned int SurfaceToAddToMALL; + bool CanAddAnotherSurfaceToMALL; + unsigned int TotalSurfaceSizeInMALL; + + TotalSurfaceSizeInMALL = 0; + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + is_using_mall_for_ss[k] = (display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable); + if (is_using_mall_for_ss[k]) + TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k]; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, is_using_mall_for_ss[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TotalSurfaceSizeInMALL = %u\n", __func__, k, TotalSurfaceSizeInMALL); +#endif + } + + SurfaceToAddToMALL = 0; + CanAddAnotherSurfaceToMALL = true; + while (CanAddAnotherSurfaceToMALL) { + CanAddAnotherSurfaceToMALL = false; + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (TotalSurfaceSizeInMALL + SurfaceSizeInMALL[k] <= MALLAllocatedForDCN * 1024 * 1024 && + !is_using_mall_for_ss[k] && display_cfg->plane_descriptors[k].overrides.refresh_from_mall != dml2_refresh_from_mall_mode_override_force_disable && one_row_per_frame_fits_in_buffer[k] && + (!CanAddAnotherSurfaceToMALL || SurfaceSizeInMALL[k] < SurfaceSizeInMALL[SurfaceToAddToMALL])) { + CanAddAnotherSurfaceToMALL = true; + SurfaceToAddToMALL = k; + DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForStaticScreen = %u (dis, en, optimize)\n", __func__, k, display_cfg->plane_descriptors[k].overrides.refresh_from_mall); + } + } + if (CanAddAnotherSurfaceToMALL) { + is_using_mall_for_ss[SurfaceToAddToMALL] = true; + TotalSurfaceSizeInMALL = TotalSurfaceSizeInMALL + SurfaceSizeInMALL[SurfaceToAddToMALL]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: SurfaceToAddToMALL = %u\n", __func__, SurfaceToAddToMALL); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALL = %u\n", __func__, TotalSurfaceSizeInMALL); +#endif + } + } +} + +static void CalculateDCCConfiguration( + bool DCCEnabled, + bool DCCProgrammingAssumesScanDirectionUnknown, + enum dml2_source_format_class SourcePixelFormat, + unsigned int SurfaceWidthLuma, + unsigned int SurfaceWidthChroma, + unsigned int SurfaceHeightLuma, + unsigned int SurfaceHeightChroma, + unsigned int nomDETInKByte, + unsigned int RequestHeight256ByteLuma, + unsigned int RequestHeight256ByteChroma, + enum dml2_swizzle_mode TilingFormat, + unsigned int BytePerPixelY, + unsigned int BytePerPixelC, + double BytePerPixelDETY, + double BytePerPixelDETC, + enum dml2_rotation_angle RotationAngle, + + // Output + enum dml2_core_internal_request_type *RequestLuma, + enum dml2_core_internal_request_type *RequestChroma, + unsigned int *MaxUncompressedBlockLuma, + unsigned int *MaxUncompressedBlockChroma, + unsigned int *MaxCompressedBlockLuma, + unsigned int *MaxCompressedBlockChroma, + unsigned int *IndependentBlockLuma, + unsigned int *IndependentBlockChroma) +{ + unsigned int DETBufferSizeForDCC = nomDETInKByte * 1024; + + unsigned int segment_order_horz_contiguous_luma; + unsigned int segment_order_horz_contiguous_chroma; + unsigned int segment_order_vert_contiguous_luma; + unsigned int segment_order_vert_contiguous_chroma; + + unsigned int req128_horz_wc_l; + unsigned int req128_horz_wc_c; + unsigned int req128_vert_wc_l; + unsigned int req128_vert_wc_c; + + unsigned int yuv420; + unsigned int horz_div_l; + unsigned int horz_div_c; + unsigned int vert_div_l; + unsigned int vert_div_c; + + unsigned int swath_buf_size; + double detile_buf_vp_horz_limit; + double detile_buf_vp_vert_limit; + + unsigned int MAS_vp_horz_limit; + unsigned int MAS_vp_vert_limit; + unsigned int max_vp_horz_width; + unsigned int max_vp_vert_height; + unsigned int eff_surf_width_l; + unsigned int eff_surf_width_c; + unsigned int eff_surf_height_l; + unsigned int eff_surf_height_c; + + unsigned int full_swath_bytes_horz_wc_l; + unsigned int full_swath_bytes_horz_wc_c; + unsigned int full_swath_bytes_vert_wc_l; + unsigned int full_swath_bytes_vert_wc_c; + + if (dml_is_420(SourcePixelFormat)) + yuv420 = 1; + else + yuv420 = 0; + horz_div_l = 1; + horz_div_c = 1; + vert_div_l = 1; + vert_div_c = 1; + + if (BytePerPixelY == 1) + vert_div_l = 0; + if (BytePerPixelC == 1) + vert_div_c = 0; + + if (BytePerPixelC == 0) { + swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 256; + detile_buf_vp_horz_limit = (double)swath_buf_size / ((double)RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l)); + detile_buf_vp_vert_limit = (double)swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l)); + } else { + swath_buf_size = DETBufferSizeForDCC / 2 - 2 * 2 * 256; + detile_buf_vp_horz_limit = (double)swath_buf_size / ((double)RequestHeight256ByteLuma * BytePerPixelY / (1 + horz_div_l) + (double)RequestHeight256ByteChroma * BytePerPixelC / (1 + horz_div_c) / (1 + yuv420)); + detile_buf_vp_vert_limit = (double)swath_buf_size / (256.0 / RequestHeight256ByteLuma / (1 + vert_div_l) + 256.0 / RequestHeight256ByteChroma / (1 + vert_div_c) / (1 + yuv420)); + } + + if (SourcePixelFormat == dml2_420_10) { + detile_buf_vp_horz_limit = 1.5 * detile_buf_vp_horz_limit; + detile_buf_vp_vert_limit = 1.5 * detile_buf_vp_vert_limit; + } + + detile_buf_vp_horz_limit = math_floor2(detile_buf_vp_horz_limit - 1, 16); + detile_buf_vp_vert_limit = math_floor2(detile_buf_vp_vert_limit - 1, 16); + + MAS_vp_horz_limit = SourcePixelFormat == dml2_rgbe_alpha ? 3840 : 6144; + MAS_vp_vert_limit = SourcePixelFormat == dml2_rgbe_alpha ? 3840 : (BytePerPixelY == 8 ? 3072 : 6144); + max_vp_horz_width = (unsigned int)(math_min2((double)MAS_vp_horz_limit, detile_buf_vp_horz_limit)); + max_vp_vert_height = (unsigned int)(math_min2((double)MAS_vp_vert_limit, detile_buf_vp_vert_limit)); + eff_surf_width_l = (SurfaceWidthLuma > max_vp_horz_width ? max_vp_horz_width : SurfaceWidthLuma); + eff_surf_width_c = eff_surf_width_l / (1 + yuv420); + eff_surf_height_l = (SurfaceHeightLuma > max_vp_vert_height ? max_vp_vert_height : SurfaceHeightLuma); + eff_surf_height_c = eff_surf_height_l / (1 + yuv420); + + full_swath_bytes_horz_wc_l = eff_surf_width_l * RequestHeight256ByteLuma * BytePerPixelY; + full_swath_bytes_vert_wc_l = eff_surf_height_l * 256 / RequestHeight256ByteLuma; + if (BytePerPixelC > 0) { + full_swath_bytes_horz_wc_c = eff_surf_width_c * RequestHeight256ByteChroma * BytePerPixelC; + full_swath_bytes_vert_wc_c = eff_surf_height_c * 256 / RequestHeight256ByteChroma; + } else { + full_swath_bytes_horz_wc_c = 0; + full_swath_bytes_vert_wc_c = 0; + } + + if (SourcePixelFormat == dml2_420_10) { + full_swath_bytes_horz_wc_l = (unsigned int)(math_ceil2((double)full_swath_bytes_horz_wc_l * 2.0 / 3.0, 256.0)); + full_swath_bytes_horz_wc_c = (unsigned int)(math_ceil2((double)full_swath_bytes_horz_wc_c * 2.0 / 3.0, 256.0)); + full_swath_bytes_vert_wc_l = (unsigned int)(math_ceil2((double)full_swath_bytes_vert_wc_l * 2.0 / 3.0, 256.0)); + full_swath_bytes_vert_wc_c = (unsigned int)(math_ceil2((double)full_swath_bytes_vert_wc_c * 2.0 / 3.0, 256.0)); + } + + if (2 * full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 0; + req128_horz_wc_c = 0; + } else if (full_swath_bytes_horz_wc_l < 1.5 * full_swath_bytes_horz_wc_c && 2 * full_swath_bytes_horz_wc_l + full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 0; + req128_horz_wc_c = 1; + } else if (full_swath_bytes_horz_wc_l >= 1.5 * full_swath_bytes_horz_wc_c && full_swath_bytes_horz_wc_l + 2 * full_swath_bytes_horz_wc_c <= DETBufferSizeForDCC) { + req128_horz_wc_l = 1; + req128_horz_wc_c = 0; + } else { + req128_horz_wc_l = 1; + req128_horz_wc_c = 1; + } + + if (2 * full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 0; + req128_vert_wc_c = 0; + } else if (full_swath_bytes_vert_wc_l < 1.5 * full_swath_bytes_vert_wc_c && 2 * full_swath_bytes_vert_wc_l + full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 0; + req128_vert_wc_c = 1; + } else if (full_swath_bytes_vert_wc_l >= 1.5 * full_swath_bytes_vert_wc_c && full_swath_bytes_vert_wc_l + 2 * full_swath_bytes_vert_wc_c <= DETBufferSizeForDCC) { + req128_vert_wc_l = 1; + req128_vert_wc_c = 0; + } else { + req128_vert_wc_l = 1; + req128_vert_wc_c = 1; + } + + if (BytePerPixelY == 2) { + segment_order_horz_contiguous_luma = 0; + segment_order_vert_contiguous_luma = 1; + } else { + segment_order_horz_contiguous_luma = 1; + segment_order_vert_contiguous_luma = 0; + } + + if (BytePerPixelC == 2) { + segment_order_horz_contiguous_chroma = 0; + segment_order_vert_contiguous_chroma = 1; + } else { + segment_order_horz_contiguous_chroma = 1; + segment_order_vert_contiguous_chroma = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DCCEnabled = %u\n", __func__, DCCEnabled); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeForDCC = %u\n", __func__, DETBufferSizeForDCC); + DML_LOG_VERBOSE("DML::%s: req128_horz_wc_l = %u\n", __func__, req128_horz_wc_l); + DML_LOG_VERBOSE("DML::%s: req128_horz_wc_c = %u\n", __func__, req128_horz_wc_c); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes_horz_wc_l = %u\n", __func__, full_swath_bytes_horz_wc_l); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes_vert_wc_c = %u\n", __func__, full_swath_bytes_vert_wc_c); + DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_luma = %u\n", __func__, segment_order_horz_contiguous_luma); + DML_LOG_VERBOSE("DML::%s: segment_order_horz_contiguous_chroma = %u\n", __func__, segment_order_horz_contiguous_chroma); +#endif + if (DCCProgrammingAssumesScanDirectionUnknown == true) { + if (req128_horz_wc_l == 0 && req128_vert_wc_l == 0) { + *RequestLuma = dml2_core_internal_request_type_256_bytes; + } else if ((req128_horz_wc_l == 1 && segment_order_horz_contiguous_luma == 0) || (req128_vert_wc_l == 1 && segment_order_vert_contiguous_luma == 0)) { + *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; + } + if (req128_horz_wc_c == 0 && req128_vert_wc_c == 0) { + *RequestChroma = dml2_core_internal_request_type_256_bytes; + } else if ((req128_horz_wc_c == 1 && segment_order_horz_contiguous_chroma == 0) || (req128_vert_wc_c == 1 && segment_order_vert_contiguous_chroma == 0)) { + *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; + } + } else if (!dml_is_vertical_rotation(RotationAngle)) { + if (req128_horz_wc_l == 0) { + *RequestLuma = dml2_core_internal_request_type_256_bytes; + } else if (segment_order_horz_contiguous_luma == 0) { + *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; + } + if (req128_horz_wc_c == 0) { + *RequestChroma = dml2_core_internal_request_type_256_bytes; + } else if (segment_order_horz_contiguous_chroma == 0) { + *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; + } + } else { + if (req128_vert_wc_l == 0) { + *RequestLuma = dml2_core_internal_request_type_256_bytes; + } else if (segment_order_vert_contiguous_luma == 0) { + *RequestLuma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestLuma = dml2_core_internal_request_type_128_bytes_contiguous; + } + if (req128_vert_wc_c == 0) { + *RequestChroma = dml2_core_internal_request_type_256_bytes; + } else if (segment_order_vert_contiguous_chroma == 0) { + *RequestChroma = dml2_core_internal_request_type_128_bytes_non_contiguous; + } else { + *RequestChroma = dml2_core_internal_request_type_128_bytes_contiguous; + } + } + + if (*RequestLuma == dml2_core_internal_request_type_256_bytes) { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 256; + *IndependentBlockLuma = 0; + } else if (*RequestLuma == dml2_core_internal_request_type_128_bytes_contiguous) { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 128; + *IndependentBlockLuma = 128; + } else { + *MaxUncompressedBlockLuma = 256; + *MaxCompressedBlockLuma = 64; + *IndependentBlockLuma = 64; + } + + if (*RequestChroma == dml2_core_internal_request_type_256_bytes) { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 256; + *IndependentBlockChroma = 0; + } else if (*RequestChroma == dml2_core_internal_request_type_128_bytes_contiguous) { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 128; + *IndependentBlockChroma = 128; + } else { + *MaxUncompressedBlockChroma = 256; + *MaxCompressedBlockChroma = 64; + *IndependentBlockChroma = 64; + } + + if (DCCEnabled != true || BytePerPixelC == 0) { + *MaxUncompressedBlockChroma = 0; + *MaxCompressedBlockChroma = 0; + *IndependentBlockChroma = 0; + } + + if (DCCEnabled != true) { + *MaxUncompressedBlockLuma = 0; + *MaxCompressedBlockLuma = 0; + *IndependentBlockLuma = 0; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockLuma = %u\n", __func__, *MaxUncompressedBlockLuma); + DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockLuma = %u\n", __func__, *MaxCompressedBlockLuma); + DML_LOG_VERBOSE("DML::%s: IndependentBlockLuma = %u\n", __func__, *IndependentBlockLuma); + DML_LOG_VERBOSE("DML::%s: MaxUncompressedBlockChroma = %u\n", __func__, *MaxUncompressedBlockChroma); + DML_LOG_VERBOSE("DML::%s: MaxCompressedBlockChroma = %u\n", __func__, *MaxCompressedBlockChroma); + DML_LOG_VERBOSE("DML::%s: IndependentBlockChroma = %u\n", __func__, *IndependentBlockChroma); +#endif + +} + +static void calculate_mcache_row_bytes( + struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_calculate_mcache_row_bytes_params *p) +{ + unsigned int vmpg_bytes = 0; + unsigned int blk_bytes = 0; + float meta_per_mvmpg_per_channel = 0; + unsigned int est_blk_per_vmpg = 2; + unsigned int mvmpg_per_row_ub = 0; + unsigned int full_vp_width_mvmpg_aligned = 0; + unsigned int full_vp_height_mvmpg_aligned = 0; + unsigned int meta_per_mvmpg_per_channel_ub = 0; + unsigned int mvmpg_per_mcache; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: num_chans = %u\n", __func__, p->num_chans); + DML_LOG_VERBOSE("DML::%s: mem_word_bytes = %u\n", __func__, p->mem_word_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_line_size_bytes = %u\n", __func__, p->mcache_line_size_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_size_bytes = %u\n", __func__, p->mcache_size_bytes); + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: gpuvm_page_size_kbytes = %u\n", __func__, p->gpuvm_page_size_kbytes); + DML_LOG_VERBOSE("DML::%s: vp_stationary = %u\n", __func__, p->vp_stationary); + DML_LOG_VERBOSE("DML::%s: tiling_mode = %u\n", __func__, p->tiling_mode); + DML_LOG_VERBOSE("DML::%s: vp_start_x = %u\n", __func__, p->vp_start_x); + DML_LOG_VERBOSE("DML::%s: vp_start_y = %u\n", __func__, p->vp_start_y); + DML_LOG_VERBOSE("DML::%s: full_vp_width = %u\n", __func__, p->full_vp_width); + DML_LOG_VERBOSE("DML::%s: full_vp_height = %u\n", __func__, p->full_vp_height); + DML_LOG_VERBOSE("DML::%s: blk_width = %u\n", __func__, p->blk_width); + DML_LOG_VERBOSE("DML::%s: blk_height = %u\n", __func__, p->blk_height); + DML_LOG_VERBOSE("DML::%s: vmpg_width = %u\n", __func__, p->vmpg_width); + DML_LOG_VERBOSE("DML::%s: vmpg_height = %u\n", __func__, p->vmpg_height); + DML_LOG_VERBOSE("DML::%s: full_swath_bytes = %u\n", __func__, p->full_swath_bytes); +#endif + DML_ASSERT(p->mcache_line_size_bytes != 0); + DML_ASSERT(p->mcache_size_bytes != 0); + + *p->mvmpg_width = 0; + *p->mvmpg_height = 0; + + if (p->full_vp_height == 0 && p->full_vp_width == 0) { + *p->num_mcaches = 0; + *p->mcache_row_bytes = 0; + *p->mcache_row_bytes_per_channel = 0; + } else { + blk_bytes = dml_get_tile_block_size_bytes(p->tiling_mode); + + // if gpuvm is not enable, the alignment boundary should be in terms of tiling block size + vmpg_bytes = p->gpuvm_page_size_kbytes * 1024; + + //With vmpg_bytes >= tile blk_bytes, the meta_row_width alignment equations are relative to the vmpg_width/height. + // But for 4KB page with 64KB tile block, we need the meta for all pages in the tile block. + // Therefore, the alignment is relative to the blk_width/height. The factor of 16 vmpg per 64KB tile block is applied at the end. + *p->mvmpg_width = p->blk_width; + *p->mvmpg_height = p->blk_height; + if (p->gpuvm_enable) { + if (vmpg_bytes >= blk_bytes) { + *p->mvmpg_width = p->vmpg_width; + *p->mvmpg_height = p->vmpg_height; + } else if (!((blk_bytes == 65536) && (vmpg_bytes == 4096))) { + DML_LOG_VERBOSE("ERROR: DML::%s: Tiling size and vm page size combination not supported\n", __func__); + DML_ASSERT(0); + } + } + + //For plane0 & 1, first calculate full_vp_width/height_l/c aligned to vmpg_width/height_l/c + full_vp_width_mvmpg_aligned = (unsigned int)(math_floor2((p->vp_start_x + p->full_vp_width) + *p->mvmpg_width - 1, *p->mvmpg_width) - math_floor2(p->vp_start_x, *p->mvmpg_width)); + full_vp_height_mvmpg_aligned = (unsigned int)(math_floor2((p->vp_start_y + p->full_vp_height) + *p->mvmpg_height - 1, *p->mvmpg_height) - math_floor2(p->vp_start_y, *p->mvmpg_height)); + + *p->full_vp_access_width_mvmpg_aligned = p->surf_vert ? full_vp_height_mvmpg_aligned : full_vp_width_mvmpg_aligned; + + //Use the equation for the exact alignment when possible. Note that the exact alignment cannot be used for horizontal access if vmpg_bytes > blk_bytes. + if (!p->surf_vert) { //horizontal access + if (p->vp_stationary == 1 && vmpg_bytes <= blk_bytes) + *p->meta_row_width_ub = full_vp_width_mvmpg_aligned; + else + *p->meta_row_width_ub = (unsigned int)math_ceil2((double)p->full_vp_width - 1, *p->mvmpg_width) + *p->mvmpg_width; + mvmpg_per_row_ub = *p->meta_row_width_ub / *p->mvmpg_width; + } else { //vertical access + if (p->vp_stationary == 1) + *p->meta_row_width_ub = full_vp_height_mvmpg_aligned; + else + *p->meta_row_width_ub = (unsigned int)math_ceil2((double)p->full_vp_height - 1, *p->mvmpg_height) + *p->mvmpg_height; + mvmpg_per_row_ub = *p->meta_row_width_ub / *p->mvmpg_height; + } + + if (p->gpuvm_enable) { + meta_per_mvmpg_per_channel = (float)vmpg_bytes / (float)256 / p->num_chans; + + //but using the est_blk_per_vmpg between 2 and 4, to be not as pessimestic + if (p->surf_vert && vmpg_bytes > blk_bytes) { + meta_per_mvmpg_per_channel = (float)est_blk_per_vmpg * blk_bytes / (float)256 / p->num_chans; + } + + *p->dcc_dram_bw_nom_overhead_factor = 1 + math_max2(1.0 / 256.0, math_ceil2(meta_per_mvmpg_per_channel, p->mem_word_bytes) / (256 * meta_per_mvmpg_per_channel)); // dcc_dr_oh_nom + } else { + meta_per_mvmpg_per_channel = (float) blk_bytes / (float)256 / p->num_chans; + + if (!p->surf_vert) + *p->dcc_dram_bw_nom_overhead_factor = 1 + 1.0 / 256.0; + else + *p->dcc_dram_bw_nom_overhead_factor = 1 + math_max2(1.0 / 256.0, math_ceil2(meta_per_mvmpg_per_channel, p->mem_word_bytes) / (256 * meta_per_mvmpg_per_channel)); + } + + meta_per_mvmpg_per_channel_ub = (unsigned int)math_ceil2((double)meta_per_mvmpg_per_channel, p->mcache_line_size_bytes); + + //but for 4KB vmpg with 64KB tile blk + if (p->gpuvm_enable && (blk_bytes == 65536) && (vmpg_bytes == 4096)) + meta_per_mvmpg_per_channel_ub = 16 * meta_per_mvmpg_per_channel_ub; + + // If this mcache_row_bytes for the full viewport of the surface is less than or equal to mcache_bytes, + // then one mcache can be used for this request stream. If not, it is useful to know the width of the viewport that can be supported in the mcache_bytes. + if (p->gpuvm_enable || p->surf_vert) { + *p->mcache_row_bytes_per_channel = mvmpg_per_row_ub * meta_per_mvmpg_per_channel_ub; + *p->mcache_row_bytes = *p->mcache_row_bytes_per_channel * p->num_chans; + } else { // horizontal and gpuvm disable + *p->mcache_row_bytes = *p->meta_row_width_ub * p->blk_height * p->bytes_per_pixel / 256; + if (p->mcache_line_size_bytes != 0) + *p->mcache_row_bytes_per_channel = (unsigned int)math_ceil2((double)*p->mcache_row_bytes / p->num_chans, p->mcache_line_size_bytes); + } + + *p->dcc_dram_bw_pref_overhead_factor = 1 + math_max2(1.0 / 256.0, *p->mcache_row_bytes / p->full_swath_bytes); // dcc_dr_oh_pref + if (p->mcache_size_bytes != 0) + *p->num_mcaches = (unsigned int)math_ceil2((double)*p->mcache_row_bytes_per_channel / p->mcache_size_bytes, 1); + + mvmpg_per_mcache = p->mcache_size_bytes / meta_per_mvmpg_per_channel_ub; + *p->mvmpg_per_mcache_lb = (unsigned int)math_floor2(mvmpg_per_mcache, 1); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %u\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %u\n", __func__, vmpg_bytes); + DML_LOG_VERBOSE("DML::%s: blk_bytes = %u\n", __func__, blk_bytes); + DML_LOG_VERBOSE("DML::%s: meta_per_mvmpg_per_channel = %f\n", __func__, meta_per_mvmpg_per_channel); + DML_LOG_VERBOSE("DML::%s: mvmpg_per_row_ub = %u\n", __func__, mvmpg_per_row_ub); + DML_LOG_VERBOSE("DML::%s: meta_row_width_ub = %u\n", __func__, *p->meta_row_width_ub); + DML_LOG_VERBOSE("DML::%s: mvmpg_width = %u\n", __func__, *p->mvmpg_width); + DML_LOG_VERBOSE("DML::%s: mvmpg_height = %u\n", __func__, *p->mvmpg_height); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_nom_overhead_factor); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_pref_overhead_factor = %f\n", __func__, *p->dcc_dram_bw_pref_overhead_factor); +#endif + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: mcache_row_bytes = %u\n", __func__, *p->mcache_row_bytes); + DML_LOG_VERBOSE("DML::%s: mcache_row_bytes_per_channel = %u\n", __func__, *p->mcache_row_bytes_per_channel); + DML_LOG_VERBOSE("DML::%s: num_mcaches = %u\n", __func__, *p->num_mcaches); +#endif + DML_ASSERT(*p->num_mcaches > 0); +} + +static void calculate_mcache_setting( + struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_calculate_mcache_setting_params *p) +{ + unsigned int n; + + struct dml2_core_shared_calculate_mcache_setting_locals *l = &scratch->calculate_mcache_setting_locals; + memset(l, 0, sizeof(struct dml2_core_shared_calculate_mcache_setting_locals)); + + *p->num_mcaches_l = 0; + *p->mcache_row_bytes_l = 0; + *p->mcache_row_bytes_per_channel_l = 0; + *p->dcc_dram_bw_nom_overhead_factor_l = 1.0; + *p->dcc_dram_bw_pref_overhead_factor_l = 1.0; + + *p->num_mcaches_c = 0; + *p->mcache_row_bytes_c = 0; + *p->mcache_row_bytes_per_channel_c = 0; + *p->dcc_dram_bw_nom_overhead_factor_c = 1.0; + *p->dcc_dram_bw_pref_overhead_factor_c = 1.0; + + *p->mall_comb_mcache_l = 0; + *p->mall_comb_mcache_c = 0; + *p->lc_comb_mcache = 0; + + if (!p->dcc_enable) + return; + + l->is_dual_plane = dml_is_420(p->source_format) || p->source_format == dml2_rgbe_alpha; + + l->l_p.num_chans = p->num_chans; + l->l_p.mem_word_bytes = p->mem_word_bytes; + l->l_p.mcache_size_bytes = p->mcache_size_bytes; + l->l_p.mcache_line_size_bytes = p->mcache_line_size_bytes; + l->l_p.gpuvm_enable = p->gpuvm_enable; + l->l_p.gpuvm_page_size_kbytes = p->gpuvm_page_size_kbytes; + l->l_p.surf_vert = p->surf_vert; + l->l_p.vp_stationary = p->vp_stationary; + l->l_p.tiling_mode = p->tiling_mode; + l->l_p.vp_start_x = p->vp_start_x_l; + l->l_p.vp_start_y = p->vp_start_y_l; + l->l_p.full_vp_width = p->full_vp_width_l; + l->l_p.full_vp_height = p->full_vp_height_l; + l->l_p.blk_width = p->blk_width_l; + l->l_p.blk_height = p->blk_height_l; + l->l_p.vmpg_width = p->vmpg_width_l; + l->l_p.vmpg_height = p->vmpg_height_l; + l->l_p.full_swath_bytes = p->full_swath_bytes_l; + l->l_p.bytes_per_pixel = p->bytes_per_pixel_l; + + // output + l->l_p.num_mcaches = p->num_mcaches_l; + l->l_p.mcache_row_bytes = p->mcache_row_bytes_l; + l->l_p.mcache_row_bytes_per_channel = p->mcache_row_bytes_per_channel_l; + l->l_p.dcc_dram_bw_nom_overhead_factor = p->dcc_dram_bw_nom_overhead_factor_l; + l->l_p.dcc_dram_bw_pref_overhead_factor = p->dcc_dram_bw_pref_overhead_factor_l; + l->l_p.mvmpg_width = &l->mvmpg_width_l; + l->l_p.mvmpg_height = &l->mvmpg_height_l; + l->l_p.full_vp_access_width_mvmpg_aligned = &l->full_vp_access_width_mvmpg_aligned_l; + l->l_p.meta_row_width_ub = &l->meta_row_width_l; + l->l_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_l; + + calculate_mcache_row_bytes(scratch, &l->l_p); + DML_ASSERT(*p->num_mcaches_l > 0); + + if (l->is_dual_plane) { + l->c_p.num_chans = p->num_chans; + l->c_p.mem_word_bytes = p->mem_word_bytes; + l->c_p.mcache_size_bytes = p->mcache_size_bytes; + l->c_p.mcache_line_size_bytes = p->mcache_line_size_bytes; + l->c_p.gpuvm_enable = p->gpuvm_enable; + l->c_p.gpuvm_page_size_kbytes = p->gpuvm_page_size_kbytes; + l->c_p.surf_vert = p->surf_vert; + l->c_p.vp_stationary = p->vp_stationary; + l->c_p.tiling_mode = p->tiling_mode; + l->c_p.vp_start_x = p->vp_start_x_c; + l->c_p.vp_start_y = p->vp_start_y_c; + l->c_p.full_vp_width = p->full_vp_width_c; + l->c_p.full_vp_height = p->full_vp_height_c; + l->c_p.blk_width = p->blk_width_c; + l->c_p.blk_height = p->blk_height_c; + l->c_p.vmpg_width = p->vmpg_width_c; + l->c_p.vmpg_height = p->vmpg_height_c; + l->c_p.full_swath_bytes = p->full_swath_bytes_c; + l->c_p.bytes_per_pixel = p->bytes_per_pixel_c; + + // output + l->c_p.num_mcaches = p->num_mcaches_c; + l->c_p.mcache_row_bytes = p->mcache_row_bytes_c; + l->c_p.mcache_row_bytes_per_channel = p->mcache_row_bytes_per_channel_c; + l->c_p.dcc_dram_bw_nom_overhead_factor = p->dcc_dram_bw_nom_overhead_factor_c; + l->c_p.dcc_dram_bw_pref_overhead_factor = p->dcc_dram_bw_pref_overhead_factor_c; + l->c_p.mvmpg_width = &l->mvmpg_width_c; + l->c_p.mvmpg_height = &l->mvmpg_height_c; + l->c_p.full_vp_access_width_mvmpg_aligned = &l->full_vp_access_width_mvmpg_aligned_c; + l->c_p.meta_row_width_ub = &l->meta_row_width_c; + l->c_p.mvmpg_per_mcache_lb = &l->mvmpg_per_mcache_lb_c; + + calculate_mcache_row_bytes(scratch, &l->c_p); + DML_ASSERT(*p->num_mcaches_c > 0); + } + + // Sharing for iMALL access + l->mcache_remainder_l = *p->mcache_row_bytes_per_channel_l % p->mcache_size_bytes; + l->mcache_remainder_c = *p->mcache_row_bytes_per_channel_c % p->mcache_size_bytes; + l->mvmpg_access_width_l = p->surf_vert ? l->mvmpg_height_l : l->mvmpg_width_l; + l->mvmpg_access_width_c = p->surf_vert ? l->mvmpg_height_c : l->mvmpg_width_c; + + if (p->imall_enable) { + *p->mall_comb_mcache_l = (2 * l->mcache_remainder_l <= p->mcache_size_bytes); + + if (l->is_dual_plane) + *p->mall_comb_mcache_c = (2 * l->mcache_remainder_c <= p->mcache_size_bytes); + } + + if (!p->surf_vert) // horizonatal access + l->luma_time_factor = (double)l->mvmpg_height_c / l->mvmpg_height_l * 2; + else // vertical access + l->luma_time_factor = (double)l->mvmpg_width_c / l->mvmpg_width_l * 2; + + // The algorithm starts with computing a non-integer, avg_mcache_element_size_l/c: + if (*p->num_mcaches_l) { + l->avg_mcache_element_size_l = l->meta_row_width_l / *p->num_mcaches_l; + } + if (l->is_dual_plane) { + l->avg_mcache_element_size_c = l->meta_row_width_c / *p->num_mcaches_c; + + /* if either remainder is 0, then mcache sharing is not needed or not possible due to full utilization */ + if (l->mcache_remainder_l && l->mcache_remainder_c) { + if (!p->imall_enable || (*p->mall_comb_mcache_l == *p->mall_comb_mcache_c)) { + l->lc_comb_last_mcache_size = (unsigned int)((l->mcache_remainder_l * (*p->mall_comb_mcache_l ? 2 : 1) * l->luma_time_factor) + + (l->mcache_remainder_c * (*p->mall_comb_mcache_c ? 2 : 1))); + } + *p->lc_comb_mcache = (l->lc_comb_last_mcache_size <= p->mcache_size_bytes) && (*p->mall_comb_mcache_l == *p->mall_comb_mcache_c); + } + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: imall_enable = %u\n", __func__, p->imall_enable); + DML_LOG_VERBOSE("DML::%s: is_dual_plane = %u\n", __func__, l->is_dual_plane); + DML_LOG_VERBOSE("DML::%s: surf_vert = %u\n", __func__, p->surf_vert); + DML_LOG_VERBOSE("DML::%s: mvmpg_width_l = %u\n", __func__, l->mvmpg_width_l); + DML_LOG_VERBOSE("DML::%s: mvmpg_height_l = %u\n", __func__, l->mvmpg_height_l); + DML_LOG_VERBOSE("DML::%s: mcache_remainder_l = %f\n", __func__, l->mcache_remainder_l); + DML_LOG_VERBOSE("DML::%s: num_mcaches_l = %u\n", __func__, *p->num_mcaches_l); + DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_l = %u\n", __func__, l->avg_mcache_element_size_l); + DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_l = %u\n", __func__, l->mvmpg_access_width_l); + DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_l = %u\n", __func__, *p->mall_comb_mcache_l); + + if (l->is_dual_plane) { + DML_LOG_VERBOSE("DML::%s: mvmpg_width_c = %u\n", __func__, l->mvmpg_width_c); + DML_LOG_VERBOSE("DML::%s: mvmpg_height_c = %u\n", __func__, l->mvmpg_height_c); + DML_LOG_VERBOSE("DML::%s: mcache_remainder_c = %f\n", __func__, l->mcache_remainder_c); + DML_LOG_VERBOSE("DML::%s: luma_time_factor = %f\n", __func__, l->luma_time_factor); + DML_LOG_VERBOSE("DML::%s: num_mcaches_c = %u\n", __func__, *p->num_mcaches_c); + DML_LOG_VERBOSE("DML::%s: avg_mcache_element_size_c = %u\n", __func__, l->avg_mcache_element_size_c); + DML_LOG_VERBOSE("DML::%s: mvmpg_access_width_c = %u\n", __func__, l->mvmpg_access_width_c); + DML_LOG_VERBOSE("DML::%s: mall_comb_mcache_c = %u\n", __func__, *p->mall_comb_mcache_c); + DML_LOG_VERBOSE("DML::%s: lc_comb_last_mcache_size = %u\n", __func__, l->lc_comb_last_mcache_size); + DML_LOG_VERBOSE("DML::%s: lc_comb_mcache = %u\n", __func__, *p->lc_comb_mcache); + } +#endif + // calculate split_coordinate + l->full_vp_access_width_l = p->surf_vert ? p->full_vp_height_l : p->full_vp_width_l; + l->full_vp_access_width_c = p->surf_vert ? p->full_vp_height_c : p->full_vp_width_c; + + for (n = 0; n < *p->num_mcaches_l - 1; n++) { + p->mcache_offsets_l[n] = (unsigned int)(math_floor2((n + 1) * l->avg_mcache_element_size_l / l->mvmpg_access_width_l, 1)) * l->mvmpg_access_width_l; + } + p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l; + + if (l->is_dual_plane) { + for (n = 0; n < *p->num_mcaches_c - 1; n++) { + p->mcache_offsets_c[n] = (unsigned int)(math_floor2((n + 1) * l->avg_mcache_element_size_c / l->mvmpg_access_width_c, 1)) * l->mvmpg_access_width_c; + } + p->mcache_offsets_c[*p->num_mcaches_c - 1] = l->full_vp_access_width_c; + } +#ifdef __DML_VBA_DEBUG__ + for (n = 0; n < *p->num_mcaches_l; n++) + DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); + + if (l->is_dual_plane) { + for (n = 0; n < *p->num_mcaches_c; n++) + DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); + } +#endif + + // Luma/Chroma combine in the last mcache + // In the case of Luma/Chroma combine-mCache (with lc_comb_mcache==1), all mCaches except the last segment are filled as much as possible, when stay aligned to mvmpg boundary + if (*p->lc_comb_mcache && l->is_dual_plane) { + for (n = 0; n < *p->num_mcaches_l - 1; n++) + p->mcache_offsets_l[n] = (n + 1) * l->mvmpg_per_mcache_lb_l * l->mvmpg_access_width_l; + p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l; + + for (n = 0; n < *p->num_mcaches_c - 1; n++) + p->mcache_offsets_c[n] = (n + 1) * l->mvmpg_per_mcache_lb_c * l->mvmpg_access_width_c; + p->mcache_offsets_c[*p->num_mcaches_c - 1] = l->full_vp_access_width_c; + +#ifdef __DML_VBA_DEBUG__ + for (n = 0; n < *p->num_mcaches_l; n++) + DML_LOG_VERBOSE("DML::%s: mcache_offsets_l[%u] = %u\n", __func__, n, p->mcache_offsets_l[n]); + + for (n = 0; n < *p->num_mcaches_c; n++) + DML_LOG_VERBOSE("DML::%s: mcache_offsets_c[%u] = %u\n", __func__, n, p->mcache_offsets_c[n]); +#endif + } + + *p->mcache_shift_granularity_l = l->mvmpg_access_width_l; + *p->mcache_shift_granularity_c = l->mvmpg_access_width_c; +} + +static void calculate_mall_bw_overhead_factor( + double mall_prefetch_sdp_overhead_factor[], //mall_sdp_oh_nom/pref + double mall_prefetch_dram_overhead_factor[], //mall_dram_oh_nom/pref + + // input + const struct dml2_display_cfg *display_cfg, + unsigned int num_active_planes) +{ + for (unsigned int k = 0; k < num_active_planes; ++k) { + mall_prefetch_sdp_overhead_factor[k] = 1.0; + mall_prefetch_dram_overhead_factor[k] = 1.0; + + // SDP - on the return side + if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) // always no data return + mall_prefetch_sdp_overhead_factor[k] = 1.25; + else if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) + mall_prefetch_sdp_overhead_factor[k] = 0.25; + + // DRAM + if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) + mall_prefetch_dram_overhead_factor[k] = 2.0; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_sdp_overhead_factor = %f\n", __func__, k, mall_prefetch_sdp_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, mall_prefetch_dram_overhead_factor = %f\n", __func__, k, mall_prefetch_dram_overhead_factor[k]); +#endif + } +} + +static double dml_get_return_bandwidth_available( + const struct dml2_soc_bb *soc, + enum dml2_core_internal_soc_state_type state_type, + enum dml2_core_internal_bw_type bw_type, + bool is_avg_bw, + bool is_hvm_en, + bool is_hvm_only, + double dcfclk_mhz, + double fclk_mhz, + double dram_bw_mbps) +{ + double return_bw_mbps = 0.; + double ideal_sdp_bandwidth = (double)soc->return_bus_width_bytes * dcfclk_mhz; + double ideal_fabric_bandwidth = fclk_mhz * (double)soc->fabric_datapath_to_dcn_data_return_bytes; + double ideal_dram_bandwidth = dram_bw_mbps; //dram_speed_mts * soc->clk_table.dram_config.channel_count * soc->clk_table.dram_config.channel_width_bytes; + + double derate_sdp_factor; + double derate_fabric_factor; + double derate_dram_factor; + + double derate_sdp_bandwidth; + double derate_fabric_bandwidth; + double derate_dram_bandwidth; + + if (is_avg_bw) { + if (state_type == dml2_core_internal_soc_state_svp_prefetch) { + derate_sdp_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.dcfclk_derate_percent / 100.0; + derate_fabric_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.fclk_derate_percent / 100.0; + derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_average.dram_derate_percent_pixel / 100.0; + } else { // just assume sys_active + derate_sdp_factor = soc->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100.0; + derate_fabric_factor = soc->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100.0; + derate_dram_factor = soc->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100.0; + } + } else { // urgent bw + if (state_type == dml2_core_internal_soc_state_svp_prefetch) { + derate_sdp_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dcfclk_derate_percent / 100.0; + derate_fabric_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.fclk_derate_percent / 100.0; + derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100.0; + + if (is_hvm_en) { + if (is_hvm_only) + derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_vm / 100.0; + else + derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel_and_vm / 100.0; + } else { + derate_dram_factor = soc->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100.0; + } + } else { // just assume sys_active + derate_sdp_factor = soc->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0; + derate_fabric_factor = soc->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100.0; + + if (is_hvm_en) { + if (is_hvm_only) + derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_vm / 100.0; + else + derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel_and_vm / 100.0; + } else { + derate_dram_factor = soc->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100.0; + } + } + } + + derate_sdp_bandwidth = ideal_sdp_bandwidth * derate_sdp_factor; + derate_fabric_bandwidth = ideal_fabric_bandwidth * derate_fabric_factor; + derate_dram_bandwidth = ideal_dram_bandwidth * derate_dram_factor; + + if (bw_type == dml2_core_internal_bw_sdp) + return_bw_mbps = math_min2(derate_sdp_bandwidth, derate_fabric_bandwidth); + else // dml2_core_internal_bw_dram + return_bw_mbps = derate_dram_bandwidth; + + DML_LOG_VERBOSE("DML::%s: is_avg_bw = %u\n", __func__, is_avg_bw); + DML_LOG_VERBOSE("DML::%s: is_hvm_en = %u\n", __func__, is_hvm_en); + DML_LOG_VERBOSE("DML::%s: is_hvm_only = %u\n", __func__, is_hvm_only); + DML_LOG_VERBOSE("DML::%s: state_type = %s\n", __func__, dml2_core_internal_soc_state_type_str(state_type)); + DML_LOG_VERBOSE("DML::%s: bw_type = %s\n", __func__, dml2_core_internal_bw_type_str(bw_type)); + DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); + DML_LOG_VERBOSE("DML::%s: ideal_sdp_bandwidth = %f\n", __func__, ideal_sdp_bandwidth); + DML_LOG_VERBOSE("DML::%s: ideal_fabric_bandwidth = %f\n", __func__, ideal_fabric_bandwidth); + DML_LOG_VERBOSE("DML::%s: ideal_dram_bandwidth = %f\n", __func__, ideal_dram_bandwidth); + DML_LOG_VERBOSE("DML::%s: derate_sdp_bandwidth = %f (derate %f)\n", __func__, derate_sdp_bandwidth, derate_sdp_factor); + DML_LOG_VERBOSE("DML::%s: derate_fabric_bandwidth = %f (derate %f)\n", __func__, derate_fabric_bandwidth, derate_fabric_factor); + DML_LOG_VERBOSE("DML::%s: derate_dram_bandwidth = %f (derate %f)\n", __func__, derate_dram_bandwidth, derate_dram_factor); + DML_LOG_VERBOSE("DML::%s: return_bw_mbps = %f\n", __func__, return_bw_mbps); + return return_bw_mbps; +} + +static noinline_for_stack void calculate_bandwidth_available( + double avg_bandwidth_available_min[dml2_core_internal_soc_state_max], + double avg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_bandwidth_available_min[dml2_core_internal_soc_state_max], // min between SDP and DRAM + double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_max], + double urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_max], + + const struct dml2_soc_bb *soc, + bool HostVMEnable, + double dcfclk_mhz, + double fclk_mhz, + double dram_bw_mbps) +{ + unsigned int n, m; + + DML_LOG_VERBOSE("DML::%s: dcfclk_mhz = %f\n", __func__, dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: fclk_mhz = %f\n", __func__, fclk_mhz); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, dram_bw_mbps); + + // Calculate all the bandwidth availabe + for (m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (n = 0; n < dml2_core_internal_bw_max; n++) { + avg_bandwidth_available[m][n] = dml_get_return_bandwidth_available(soc, + m, // soc_state + n, // bw_type + 1, // avg_bw + HostVMEnable, + 0, // hvm_only + dcfclk_mhz, + fclk_mhz, + dram_bw_mbps); + + urg_bandwidth_available[m][n] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 0, dcfclk_mhz, fclk_mhz, dram_bw_mbps); + + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), avg_bandwidth_available[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), urg_bandwidth_available[m][n]); +#endif + + // urg_bandwidth_available_vm_only is indexed by soc_state + if (n == dml2_core_internal_bw_dram) { + urg_bandwidth_available_vm_only[m] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 1, dcfclk_mhz, fclk_mhz, dram_bw_mbps); + urg_bandwidth_available_pixel_and_vm[m] = dml_get_return_bandwidth_available(soc, m, n, 0, HostVMEnable, 0, dcfclk_mhz, fclk_mhz, dram_bw_mbps); + } + } + + avg_bandwidth_available_min[m] = math_min2(avg_bandwidth_available[m][dml2_core_internal_bw_dram], avg_bandwidth_available[m][dml2_core_internal_bw_sdp]); + urg_bandwidth_available_min[m] = math_min2(urg_bandwidth_available[m][dml2_core_internal_bw_dram], urg_bandwidth_available[m][dml2_core_internal_bw_sdp]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), avg_bandwidth_available_min[m]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_min[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_min[m]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_vm_only[%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), urg_bandwidth_available_vm_only[n]); +#endif + } +} + +static void calculate_avg_bandwidth_required( + double avg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + + // input + const struct dml2_display_cfg *display_cfg, + unsigned int num_active_planes, + double ReadBandwidthLuma[], + double ReadBandwidthChroma[], + double cursor_bw[], + double dcc_dram_bw_nom_overhead_factor_p0[], + double dcc_dram_bw_nom_overhead_factor_p1[], + double mall_prefetch_dram_overhead_factor[], + double mall_prefetch_sdp_overhead_factor[]) +{ + unsigned int n, m, k; + double sdp_overhead_factor; + double dram_overhead_factor_p0; + double dram_overhead_factor_p1; + + // Average BW support check + for (m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (n = 0; n < dml2_core_internal_bw_max; n++) { // sdp, dram + avg_bandwidth_required[m][n] = 0; + } + } + + // SysActive and SVP Prefetch AVG bandwidth Check + for (k = 0; k < num_active_planes; ++k) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: plane %0d\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: ReadBandwidthLuma=%f\n", __func__, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: ReadBandwidthChroma=%f\n", __func__, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p0=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p0[k]); + DML_LOG_VERBOSE("DML::%s: dcc_dram_bw_nom_overhead_factor_p1=%f\n", __func__, dcc_dram_bw_nom_overhead_factor_p1[k]); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_dram_overhead_factor=%f\n", __func__, mall_prefetch_dram_overhead_factor[k]); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor=%f\n", __func__, mall_prefetch_sdp_overhead_factor[k]); +#endif + + sdp_overhead_factor = mall_prefetch_sdp_overhead_factor[k]; + dram_overhead_factor_p0 = dcc_dram_bw_nom_overhead_factor_p0[k] * mall_prefetch_dram_overhead_factor[k]; + dram_overhead_factor_p1 = dcc_dram_bw_nom_overhead_factor_p1[k] * mall_prefetch_dram_overhead_factor[k]; + + // FIXME_DCN4, was missing cursor_bw in here, but do I actually need that and tdlut bw for average bandwidth calculation? + // active avg bw not include phantom, but svp_prefetch avg bw should include phantom pipes + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] += sdp_overhead_factor * (ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) + cursor_bw[k]; + avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] += dram_overhead_factor_p0 * ReadBandwidthLuma[k] + dram_overhead_factor_p1 * ReadBandwidthChroma[k] + cursor_bw[k]; + } + avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] += sdp_overhead_factor * (ReadBandwidthLuma[k] + ReadBandwidthChroma[k]) + cursor_bw[k]; + avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] += dram_overhead_factor_p0 * ReadBandwidthLuma[k] + dram_overhead_factor_p1 * ReadBandwidthChroma[k] + cursor_bw[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_sys_active), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_sdp), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_required[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(dml2_core_internal_soc_state_svp_prefetch), dml2_core_internal_bw_type_str(dml2_core_internal_bw_dram), avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]); +#endif + } +} + +static void CalculateVMRowAndSwath(struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_CalculateVMRowAndSwath_params *p) +{ + struct dml2_core_calcs_CalculateVMRowAndSwath_locals *s = &scratch->CalculateVMRowAndSwath_locals; + + s->HostVMDynamicLevels = CalculateHostVMDynamicLevels(p->display_cfg->gpuvm_enable, p->display_cfg->hostvm_enable, p->HostVMMinPageSize, p->display_cfg->hostvm_max_non_cached_page_table_levels); + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->display_cfg->gpuvm_enable == true) { + p->vm_group_bytes[k] = 512; + p->dpte_group_bytes[k] = 512; + } else { + p->vm_group_bytes[k] = 0; + p->dpte_group_bytes[k] = 0; + } + + if (dml_is_420(p->myPipe[k].SourcePixelFormat) || p->myPipe[k].SourcePixelFormat == dml2_rgbe_alpha) { + if ((p->myPipe[k].SourcePixelFormat == dml2_420_10 || p->myPipe[k].SourcePixelFormat == dml2_420_12) && !dml_is_vertical_rotation(p->myPipe[k].RotationAngle)) { + s->PTEBufferSizeInRequestsForLuma[k] = (p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma) / 2; + s->PTEBufferSizeInRequestsForChroma[k] = s->PTEBufferSizeInRequestsForLuma[k]; + } else { + s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma; + s->PTEBufferSizeInRequestsForChroma[k] = p->PTEBufferSizeInRequestsChroma; + } + + scratch->calculate_vm_and_row_bytes_params.ViewportStationary = p->myPipe[k].ViewportStationary; + scratch->calculate_vm_and_row_bytes_params.DCCEnable = p->myPipe[k].DCCEnable; + scratch->calculate_vm_and_row_bytes_params.NumberOfDPPs = p->myPipe[k].DPPPerSurface; + scratch->calculate_vm_and_row_bytes_params.BlockHeight256Bytes = p->myPipe[k].BlockHeight256BytesC; + scratch->calculate_vm_and_row_bytes_params.BlockWidth256Bytes = p->myPipe[k].BlockWidth256BytesC; + scratch->calculate_vm_and_row_bytes_params.SourcePixelFormat = p->myPipe[k].SourcePixelFormat; + scratch->calculate_vm_and_row_bytes_params.SurfaceTiling = p->myPipe[k].SurfaceTiling; + scratch->calculate_vm_and_row_bytes_params.BytePerPixel = p->myPipe[k].BytePerPixelC; + scratch->calculate_vm_and_row_bytes_params.RotationAngle = p->myPipe[k].RotationAngle; + scratch->calculate_vm_and_row_bytes_params.SwathWidth = p->SwathWidthC[k]; + scratch->calculate_vm_and_row_bytes_params.ViewportHeight = p->myPipe[k].ViewportHeightC; + scratch->calculate_vm_and_row_bytes_params.ViewportXStart = p->myPipe[k].ViewportXStartC; + scratch->calculate_vm_and_row_bytes_params.ViewportYStart = p->myPipe[k].ViewportYStartC; + scratch->calculate_vm_and_row_bytes_params.GPUVMEnable = p->display_cfg->gpuvm_enable; + scratch->calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = p->display_cfg->gpuvm_max_page_table_levels; + scratch->calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + scratch->calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = s->PTEBufferSizeInRequestsForChroma[k]; + scratch->calculate_vm_and_row_bytes_params.Pitch = p->myPipe[k].PitchC; + scratch->calculate_vm_and_row_bytes_params.MacroTileWidth = p->myPipe[k].BlockWidthC; + scratch->calculate_vm_and_row_bytes_params.MacroTileHeight = p->myPipe[k].BlockHeightC; + scratch->calculate_vm_and_row_bytes_params.is_phantom = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]); + scratch->calculate_vm_and_row_bytes_params.DCCMetaPitch = p->myPipe[k].DCCMetaPitchC; + scratch->calculate_vm_and_row_bytes_params.mrq_present = p->mrq_present; + + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &s->PixelPTEBytesPerRowC[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &s->PixelPTEBytesPerRowStorageC[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub = &p->dpte_row_width_chroma_ub[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height = &p->dpte_row_height_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height_linear = &p->dpte_row_height_linear_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &s->PixelPTEBytesPerRowC_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &s->dpte_row_width_chroma_ub_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &s->dpte_row_height_chroma_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.vmpg_width = &p->vmpg_width_c[k]; + scratch->calculate_vm_and_row_bytes_params.vmpg_height = &p->vmpg_height_c[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &p->PixelPTEReqWidthC[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &p->PixelPTEReqHeightC[k]; + scratch->calculate_vm_and_row_bytes_params.PTERequestSize = &p->PTERequestSizeC[k]; + scratch->calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &p->dpde0_bytes_per_frame_ub_c[k]; + + scratch->calculate_vm_and_row_bytes_params.meta_row_bytes = &s->meta_row_bytes_per_row_ub_c[k]; + scratch->calculate_vm_and_row_bytes_params.MetaRequestWidth = &p->meta_req_width_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.MetaRequestHeight = &p->meta_req_height_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_row_width = &p->meta_row_width_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_row_height = &p->meta_row_height_chroma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &p->meta_pte_bytes_per_frame_ub_c[k]; + + s->vm_bytes_c = CalculateVMAndRowBytes(&scratch->calculate_vm_and_row_bytes_params); + + p->PrefetchSourceLinesC[k] = CalculatePrefetchSourceLines( + p->myPipe[k].VRatioChroma, + p->myPipe[k].VTapsChroma, + p->myPipe[k].InterlaceEnable, + p->myPipe[k].ProgressiveToInterlaceUnitInOPP, + p->myPipe[k].SwathHeightC, + p->myPipe[k].RotationAngle, + p->myPipe[k].mirrored, + p->myPipe[k].ViewportStationary, + p->SwathWidthC[k], + p->myPipe[k].ViewportHeightC, + p->myPipe[k].ViewportXStartC, + p->myPipe[k].ViewportYStartC, + + // Output + &p->VInitPreFillC[k], + &p->MaxNumSwathC[k]); + } else { + s->PTEBufferSizeInRequestsForLuma[k] = p->PTEBufferSizeInRequestsLuma + p->PTEBufferSizeInRequestsChroma; + s->PTEBufferSizeInRequestsForChroma[k] = 0; + s->PixelPTEBytesPerRowC[k] = 0; + s->PixelPTEBytesPerRowStorageC[k] = 0; + s->vm_bytes_c = 0; + p->MaxNumSwathC[k] = 0; + p->PrefetchSourceLinesC[k] = 0; + s->dpte_row_height_chroma_one_row_per_frame[k] = 0; + s->dpte_row_width_chroma_ub_one_row_per_frame[k] = 0; + s->PixelPTEBytesPerRowC_one_row_per_frame[k] = 0; + } + + scratch->calculate_vm_and_row_bytes_params.ViewportStationary = p->myPipe[k].ViewportStationary; + scratch->calculate_vm_and_row_bytes_params.DCCEnable = p->myPipe[k].DCCEnable; + scratch->calculate_vm_and_row_bytes_params.NumberOfDPPs = p->myPipe[k].DPPPerSurface; + scratch->calculate_vm_and_row_bytes_params.BlockHeight256Bytes = p->myPipe[k].BlockHeight256BytesY; + scratch->calculate_vm_and_row_bytes_params.BlockWidth256Bytes = p->myPipe[k].BlockWidth256BytesY; + scratch->calculate_vm_and_row_bytes_params.SourcePixelFormat = p->myPipe[k].SourcePixelFormat; + scratch->calculate_vm_and_row_bytes_params.SurfaceTiling = p->myPipe[k].SurfaceTiling; + scratch->calculate_vm_and_row_bytes_params.BytePerPixel = p->myPipe[k].BytePerPixelY; + scratch->calculate_vm_and_row_bytes_params.RotationAngle = p->myPipe[k].RotationAngle; + scratch->calculate_vm_and_row_bytes_params.SwathWidth = p->SwathWidthY[k]; + scratch->calculate_vm_and_row_bytes_params.ViewportHeight = p->myPipe[k].ViewportHeight; + scratch->calculate_vm_and_row_bytes_params.ViewportXStart = p->myPipe[k].ViewportXStart; + scratch->calculate_vm_and_row_bytes_params.ViewportYStart = p->myPipe[k].ViewportYStart; + scratch->calculate_vm_and_row_bytes_params.GPUVMEnable = p->display_cfg->gpuvm_enable; + scratch->calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = p->display_cfg->gpuvm_max_page_table_levels; + scratch->calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + scratch->calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = s->PTEBufferSizeInRequestsForLuma[k]; + scratch->calculate_vm_and_row_bytes_params.Pitch = p->myPipe[k].PitchY; + scratch->calculate_vm_and_row_bytes_params.MacroTileWidth = p->myPipe[k].BlockWidthY; + scratch->calculate_vm_and_row_bytes_params.MacroTileHeight = p->myPipe[k].BlockHeightY; + scratch->calculate_vm_and_row_bytes_params.is_phantom = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]); + scratch->calculate_vm_and_row_bytes_params.DCCMetaPitch = p->myPipe[k].DCCMetaPitchY; + scratch->calculate_vm_and_row_bytes_params.mrq_present = p->mrq_present; + + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &s->PixelPTEBytesPerRowY[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &s->PixelPTEBytesPerRowStorageY[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub = &p->dpte_row_width_luma_ub[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height = &p->dpte_row_height_luma[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height_linear = &p->dpte_row_height_linear_luma[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &s->PixelPTEBytesPerRowY_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &s->dpte_row_width_luma_ub_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &s->dpte_row_height_luma_one_row_per_frame[k]; + scratch->calculate_vm_and_row_bytes_params.vmpg_width = &p->vmpg_width_y[k]; + scratch->calculate_vm_and_row_bytes_params.vmpg_height = &p->vmpg_height_y[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &p->PixelPTEReqWidthY[k]; + scratch->calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &p->PixelPTEReqHeightY[k]; + scratch->calculate_vm_and_row_bytes_params.PTERequestSize = &p->PTERequestSizeY[k]; + scratch->calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &p->dpde0_bytes_per_frame_ub_l[k]; + + scratch->calculate_vm_and_row_bytes_params.meta_row_bytes = &s->meta_row_bytes_per_row_ub_l[k]; + scratch->calculate_vm_and_row_bytes_params.MetaRequestWidth = &p->meta_req_width_luma[k]; + scratch->calculate_vm_and_row_bytes_params.MetaRequestHeight = &p->meta_req_height_luma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_row_width = &p->meta_row_width_luma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_row_height = &p->meta_row_height_luma[k]; + scratch->calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &p->meta_pte_bytes_per_frame_ub_l[k]; + + s->vm_bytes_l = CalculateVMAndRowBytes(&scratch->calculate_vm_and_row_bytes_params); + + p->PrefetchSourceLinesY[k] = CalculatePrefetchSourceLines( + p->myPipe[k].VRatio, + p->myPipe[k].VTaps, + p->myPipe[k].InterlaceEnable, + p->myPipe[k].ProgressiveToInterlaceUnitInOPP, + p->myPipe[k].SwathHeightY, + p->myPipe[k].RotationAngle, + p->myPipe[k].mirrored, + p->myPipe[k].ViewportStationary, + p->SwathWidthY[k], + p->myPipe[k].ViewportHeight, + p->myPipe[k].ViewportXStart, + p->myPipe[k].ViewportYStart, + + // Output + &p->VInitPreFillY[k], + &p->MaxNumSwathY[k]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_l = %u (before hvm level)\n", __func__, k, s->vm_bytes_l); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes_c = %u (before hvm level)\n", __func__, k, s->vm_bytes_c); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_l = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes_per_row_ub_c = %u\n", __func__, k, s->meta_row_bytes_per_row_ub_c[k]); +#endif + p->vm_bytes[k] = (s->vm_bytes_l + s->vm_bytes_c) * (1 + 8 * s->HostVMDynamicLevels); + p->meta_row_bytes[k] = s->meta_row_bytes_per_row_ub_l[k] + s->meta_row_bytes_per_row_ub_c[k]; + p->meta_row_bytes_per_row_ub_l[k] = s->meta_row_bytes_per_row_ub_l[k]; + p->meta_row_bytes_per_row_ub_c[k] = s->meta_row_bytes_per_row_ub_c[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_bytes = %u\n", __func__, k, p->meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_bytes = %u (after hvm level)\n", __func__, k, p->vm_bytes[k]); +#endif + if (s->PixelPTEBytesPerRowStorageY[k] <= 64 * s->PTEBufferSizeInRequestsForLuma[k] && s->PixelPTEBytesPerRowStorageC[k] <= 64 * s->PTEBufferSizeInRequestsForChroma[k]) { + p->PTEBufferSizeNotExceeded[k] = true; + } else { + p->PTEBufferSizeNotExceeded[k] = false; + } + + s->one_row_per_frame_fits_in_buffer[k] = (s->PixelPTEBytesPerRowY_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForLuma[k] && + s->PixelPTEBytesPerRowC_one_row_per_frame[k] <= 64 * 2 * s->PTEBufferSizeInRequestsForChroma[k]); +#ifdef __DML_VBA_DEBUG__ + if (p->PTEBufferSizeNotExceeded[k] == 0 || s->one_row_per_frame_fits_in_buffer[k] == 0) { + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (before hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageY = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowStorageC = %u\n", __func__, k, s->PixelPTEBytesPerRowStorageC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForLuma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeInRequestsForChroma = %u\n", __func__, k, s->PTEBufferSizeInRequestsForChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded (not one_row_per_frame) = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); + + DML_LOG_VERBOSE("DML::%s: k=%u, HostVMDynamicLevels = %u\n", __func__, k, s->HostVMDynamicLevels); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowY_one_row_per_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC_one_row_per_frame = %u\n", __func__, k, s->PixelPTEBytesPerRowC_one_row_per_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, one_row_per_frame_fits_in_buffer = %u\n", __func__, k, s->one_row_per_frame_fits_in_buffer[k]); + } +#endif + } + + CalculateMALLUseForStaticScreen( + p->display_cfg, + p->NumberOfActiveSurfaces, + p->MALLAllocatedForDCN, + p->SurfaceSizeInMALL, + s->one_row_per_frame_fits_in_buffer, + // Output + p->is_using_mall_for_ss); + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->display_cfg->gpuvm_enable) { + if (p->display_cfg->plane_descriptors[k].overrides.hw.force_pte_buffer_mode.enable == 1) { + p->PTE_BUFFER_MODE[k] = p->display_cfg->plane_descriptors[k].overrides.hw.force_pte_buffer_mode.value; + } + p->PTE_BUFFER_MODE[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->is_using_mall_for_ss[k] || (p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) || + dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) || (p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes > 64); + p->BIGK_FRAGMENT_SIZE[k] = (unsigned int)(math_log((float)p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes * 1024, 2) - 12); + } else { + p->PTE_BUFFER_MODE[k] = 0; + p->BIGK_FRAGMENT_SIZE[k] = 0; + } + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + p->DCCMetaBufferSizeNotExceeded[k] = true; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, SurfaceSizeInMALL = %u\n", __func__, k, p->SurfaceSizeInMALL[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, is_using_mall_for_ss = %u\n", __func__, k, p->is_using_mall_for_ss[k]); +#endif + p->use_one_row_for_frame[k] = p->myPipe[k].FORCE_ONE_ROW_FOR_FRAME || p->is_using_mall_for_ss[k] || (p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) || + (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) || (p->display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes > 64 && dml_is_vertical_rotation(p->myPipe[k].RotationAngle)); + + p->use_one_row_for_frame_flip[k] = p->use_one_row_for_frame[k] && !(p->display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame); + + if (p->use_one_row_for_frame[k]) { + p->dpte_row_height_luma[k] = s->dpte_row_height_luma_one_row_per_frame[k]; + p->dpte_row_width_luma_ub[k] = s->dpte_row_width_luma_ub_one_row_per_frame[k]; + s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY_one_row_per_frame[k]; + p->dpte_row_height_chroma[k] = s->dpte_row_height_chroma_one_row_per_frame[k]; + p->dpte_row_width_chroma_ub[k] = s->dpte_row_width_chroma_ub_one_row_per_frame[k]; + s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC_one_row_per_frame[k]; + p->PTEBufferSizeNotExceeded[k] = s->one_row_per_frame_fits_in_buffer[k]; + } + + if (p->meta_row_bytes[k] <= p->DCCMetaBufferSizeBytes) { + p->DCCMetaBufferSizeNotExceeded[k] = true; + } else { + p->DCCMetaBufferSizeNotExceeded[k] = false; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bytes = %d\n", __func__, k, p->meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeBytes = %d\n", __func__, k, p->DCCMetaBufferSizeBytes); + DML_LOG_VERBOSE("DML::%s: k=%d, DCCMetaBufferSizeNotExceeded = %d\n", __func__, k, p->DCCMetaBufferSizeNotExceeded[k]); +#endif + } + + s->PixelPTEBytesPerRowY[k] = s->PixelPTEBytesPerRowY[k] * (1 + 8 * s->HostVMDynamicLevels); + s->PixelPTEBytesPerRowC[k] = s->PixelPTEBytesPerRowC[k] * (1 + 8 * s->HostVMDynamicLevels); + p->PixelPTEBytesPerRow[k] = s->PixelPTEBytesPerRowY[k] + s->PixelPTEBytesPerRowC[k]; + p->dpte_row_bytes_per_row_l[k] = s->PixelPTEBytesPerRowY[k]; + p->dpte_row_bytes_per_row_c[k] = s->PixelPTEBytesPerRowC[k]; + + // if one row of dPTEs is meant to span the entire frame, then for these calculations, we will pretend like that one big row is fetched in two halfs + if (p->use_one_row_for_frame[k]) + p->PixelPTEBytesPerRow[k] = p->PixelPTEBytesPerRow[k] / 2; + + CalculateRowBandwidth( + p->display_cfg->gpuvm_enable, + p->use_one_row_for_frame[k], + p->myPipe[k].SourcePixelFormat, + p->myPipe[k].VRatio, + p->myPipe[k].VRatioChroma, + p->myPipe[k].DCCEnable, + p->myPipe[k].HTotal / p->myPipe[k].PixelClock, + s->PixelPTEBytesPerRowY[k], + s->PixelPTEBytesPerRowC[k], + p->dpte_row_height_luma[k], + p->dpte_row_height_chroma[k], + + p->mrq_present, + p->meta_row_bytes_per_row_ub_l[k], + p->meta_row_bytes_per_row_ub_c[k], + p->meta_row_height_luma[k], + p->meta_row_height_chroma[k], + + // Output + &p->dpte_row_bw[k], + &p->meta_row_bw[k]); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame_flip = %u\n", __func__, k, p->use_one_row_for_frame_flip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, UseMALLForPStateChange = %u\n", __func__, k, p->display_cfg->plane_descriptors[k].overrides.legacy_svp_config); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_luma = %u\n", __func__, k, p->dpte_row_height_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowY = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_height_chroma = %u\n", __func__, k, p->dpte_row_height_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRowC = %u (after hvm level)\n", __func__, k, s->PixelPTEBytesPerRowC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEBytesPerRow = %u\n", __func__, k, p->PixelPTEBytesPerRow[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, p->PTEBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, gpuvm_enable = %u\n", __func__, k, p->display_cfg->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: k=%u, PTE_BUFFER_MODE = %u\n", __func__, k, p->PTE_BUFFER_MODE[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, BIGK_FRAGMENT_SIZE = %u\n", __func__, k, p->BIGK_FRAGMENT_SIZE[k]); +#endif + } +} + +static double CalculateUrgentLatency( + double UrgentLatencyPixelDataOnly, + double UrgentLatencyPixelMixedWithVMData, + double UrgentLatencyVMDataOnly, + bool DoUrgentLatencyAdjustment, + double UrgentLatencyAdjustmentFabricClockComponent, + double UrgentLatencyAdjustmentFabricClockReference, + double FabricClock, + double uclk_freq_mhz, + enum dml2_qos_param_type qos_type, + unsigned int urgent_ramp_uclk_cycles, + unsigned int df_qos_response_time_fclk_cycles, + unsigned int max_round_trip_to_furthest_cs_fclk_cycles, + unsigned int mall_overhead_fclk_cycles, + double umc_urgent_ramp_latency_margin, + double fabric_max_transport_latency_margin) +{ + double urgent_latency = 0; + if (qos_type == dml2_qos_param_type_dcn4x) { + urgent_latency = (df_qos_response_time_fclk_cycles + mall_overhead_fclk_cycles) / FabricClock + + max_round_trip_to_furthest_cs_fclk_cycles / FabricClock * (1 + fabric_max_transport_latency_margin / 100.0) + + urgent_ramp_uclk_cycles / uclk_freq_mhz * (1 + umc_urgent_ramp_latency_margin / 100.0); + } else { + urgent_latency = math_max3(UrgentLatencyPixelDataOnly, UrgentLatencyPixelMixedWithVMData, UrgentLatencyVMDataOnly); + if (DoUrgentLatencyAdjustment == true) { + urgent_latency = urgent_latency + UrgentLatencyAdjustmentFabricClockComponent * (UrgentLatencyAdjustmentFabricClockReference / FabricClock - 1); + } + } +#ifdef __DML_VBA_DEBUG__ + if (qos_type == dml2_qos_param_type_dcn4x) { + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: urgent_ramp_uclk_cycles = %d\n", __func__, urgent_ramp_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: umc_urgent_ramp_latency_margin = %f\n", __func__, umc_urgent_ramp_latency_margin); + } else { + DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelDataOnly = %f\n", __func__, UrgentLatencyPixelDataOnly); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyPixelMixedWithVMData = %f\n", __func__, UrgentLatencyPixelMixedWithVMData); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyVMDataOnly = %f\n", __func__, UrgentLatencyVMDataOnly); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockComponent = %f\n", __func__, UrgentLatencyAdjustmentFabricClockComponent); + DML_LOG_VERBOSE("DML::%s: UrgentLatencyAdjustmentFabricClockReference = %f\n", __func__, UrgentLatencyAdjustmentFabricClockReference); + } + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, urgent_latency); +#endif + return urgent_latency; +} + +static double CalculateTripToMemory( + double UrgLatency, + double FabricClock, + double uclk_freq_mhz, + enum dml2_qos_param_type qos_type, + unsigned int trip_to_memory_uclk_cycles, + unsigned int max_round_trip_to_furthest_cs_fclk_cycles, + unsigned int mall_overhead_fclk_cycles, + double umc_max_latency_margin, + double fabric_max_transport_latency_margin) +{ + double trip_to_memory_us; + if (qos_type == dml2_qos_param_type_dcn4x) { + trip_to_memory_us = mall_overhead_fclk_cycles / FabricClock + + max_round_trip_to_furthest_cs_fclk_cycles / FabricClock * (1.0 + fabric_max_transport_latency_margin / 100.0) + + trip_to_memory_uclk_cycles / uclk_freq_mhz * (1.0 + umc_max_latency_margin / 100.0); + } else { + trip_to_memory_us = UrgLatency; + } + +#ifdef __DML_VBA_DEBUG__ + if (qos_type == dml2_qos_param_type_dcn4x) { + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: max_round_trip_to_furthest_cs_fclk_cycles = %d\n", __func__, max_round_trip_to_furthest_cs_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: mall_overhead_fclk_cycles = %d\n", __func__, mall_overhead_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: trip_to_memory_uclk_cycles = %d\n", __func__, trip_to_memory_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: fabric_max_transport_latency_margin = %f\n", __func__, fabric_max_transport_latency_margin); + DML_LOG_VERBOSE("DML::%s: umc_max_latency_margin = %f\n", __func__, umc_max_latency_margin); + } else { + DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); + } + DML_LOG_VERBOSE("DML::%s: trip_to_memory_us = %f\n", __func__, trip_to_memory_us); +#endif + + + return trip_to_memory_us; +} + +static double CalculateMetaTripToMemory( + double UrgLatency, + double FabricClock, + double uclk_freq_mhz, + enum dml2_qos_param_type qos_type, + unsigned int meta_trip_to_memory_uclk_cycles, + unsigned int meta_trip_to_memory_fclk_cycles, + double umc_max_latency_margin, + double fabric_max_transport_latency_margin) +{ + double meta_trip_to_memory_us; + if (qos_type == dml2_qos_param_type_dcn4x) { + meta_trip_to_memory_us = meta_trip_to_memory_fclk_cycles / FabricClock * (1.0 + fabric_max_transport_latency_margin / 100.0) + + meta_trip_to_memory_uclk_cycles / uclk_freq_mhz * (1.0 + umc_max_latency_margin / 100.0); + } else { + meta_trip_to_memory_us = UrgLatency; + } + +#ifdef __DML_VBA_DEBUG__ + if (qos_type == dml2_qos_param_type_dcn4x) { + DML_LOG_VERBOSE("DML::%s: qos_type = %d\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_fclk_cycles = %d\n", __func__, meta_trip_to_memory_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_uclk_cycles = %d\n", __func__, meta_trip_to_memory_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, uclk_freq_mhz); + } else { + DML_LOG_VERBOSE("DML::%s: UrgLatency = %f\n", __func__, UrgLatency); + } + DML_LOG_VERBOSE("DML::%s: meta_trip_to_memory_us = %f\n", __func__, meta_trip_to_memory_us); +#endif + + + return meta_trip_to_memory_us; +} + +static void calculate_cursor_req_attributes( + unsigned int cursor_width, + unsigned int cursor_bpp, + + // output + unsigned int *cursor_lines_per_chunk, + unsigned int *cursor_bytes_per_line, + unsigned int *cursor_bytes_per_chunk, + unsigned int *cursor_bytes) +{ + unsigned int cursor_bytes_per_req = 0; + unsigned int cursor_width_bytes = 0; + unsigned int cursor_height = 0; + + //SW determines the cursor pitch to support the maximum cursor_width that will be used but the following restrictions apply. + //- For 2bpp, cursor_pitch = 256 pixels due to min cursor request size of 64B + //- For 32 or 64 bpp, cursor_pitch = 64, 128 or 256 pixels depending on the cursor width + + //The cursor requestor uses a cursor request size of 64B, 128B, or 256B depending on the cursor_width and cursor_bpp as follows. + + cursor_width_bytes = (unsigned int)math_ceil2((double)cursor_width * cursor_bpp / 8, 1); + if (cursor_width_bytes <= 64) + cursor_bytes_per_req = 64; + else if (cursor_width_bytes <= 128) + cursor_bytes_per_req = 128; + else + cursor_bytes_per_req = 256; + + //If cursor_width_bytes is greater than 256B, then multiple 256B requests are issued to fetch the entire cursor line. + *cursor_bytes_per_line = (unsigned int)math_ceil2((double)cursor_width_bytes, cursor_bytes_per_req); + + //Nominally, the cursor chunk is 1KB or 2KB but it is restricted to a power of 2 number of lines with a maximum of 16 lines. + if (cursor_bpp == 2) { + *cursor_lines_per_chunk = 16; + } else if (cursor_bpp == 32) { + if (cursor_width <= 32) + *cursor_lines_per_chunk = 16; + else if (cursor_width <= 64) + *cursor_lines_per_chunk = 8; + else if (cursor_width <= 128) + *cursor_lines_per_chunk = 4; + else + *cursor_lines_per_chunk = 2; + } else if (cursor_bpp == 64) { + if (cursor_width <= 16) + *cursor_lines_per_chunk = 16; + else if (cursor_width <= 32) + *cursor_lines_per_chunk = 8; + else if (cursor_width <= 64) + *cursor_lines_per_chunk = 4; + else if (cursor_width <= 128) + *cursor_lines_per_chunk = 2; + else + *cursor_lines_per_chunk = 1; + } else { + if (cursor_width > 0) { + DML_LOG_VERBOSE("DML::%s: Invalid cursor_bpp = %d\n", __func__, cursor_bpp); + DML_ASSERT(0); + } + } + + *cursor_bytes_per_chunk = *cursor_bytes_per_line * *cursor_lines_per_chunk; + + // For the cursor implementation, all requested data is stored in the return buffer. Given this fact, the cursor_bytes can be directly compared with the CursorBufferSize. + // Only cursor_width is provided for worst case sizing so assume that the cursor is square + cursor_height = cursor_width; + *cursor_bytes = *cursor_bytes_per_line * cursor_height; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: cursor_bpp = %d\n", __func__, cursor_bpp); + DML_LOG_VERBOSE("DML::%s: cursor_width = %d\n", __func__, cursor_width); + DML_LOG_VERBOSE("DML::%s: cursor_width_bytes = %d\n", __func__, cursor_width_bytes); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_req = %d\n", __func__, cursor_bytes_per_req); + DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %d\n", __func__, *cursor_lines_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, *cursor_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, *cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes = %d\n", __func__, *cursor_bytes); + DML_LOG_VERBOSE("DML::%s: cursor_pitch = %d\n", __func__, cursor_bpp == 2 ? 256 : (unsigned int)1 << (unsigned int)math_ceil2(math_log((float)cursor_width, 2), 1)); +#endif +} + +static void calculate_cursor_urgent_burst_factor( + unsigned int CursorBufferSize, + unsigned int CursorWidth, + unsigned int cursor_bytes_per_chunk, + unsigned int cursor_lines_per_chunk, + double LineTime, + double UrgentLatency, + + double *UrgentBurstFactorCursor, + bool *NotEnoughUrgentLatencyHiding) +{ + unsigned int LinesInCursorBuffer = 0; + double CursorBufferSizeInTime = 0; + + if (CursorWidth > 0) { + LinesInCursorBuffer = (unsigned int)math_floor2(CursorBufferSize * 1024.0 / (double)cursor_bytes_per_chunk, 1) * cursor_lines_per_chunk; + + CursorBufferSizeInTime = LinesInCursorBuffer * LineTime; + if (CursorBufferSizeInTime - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorCursor = 1; + } else { + *NotEnoughUrgentLatencyHiding = 0; + *UrgentBurstFactorCursor = CursorBufferSizeInTime / (CursorBufferSizeInTime - UrgentLatency); + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: LinesInCursorBuffer = %u\n", __func__, LinesInCursorBuffer); + DML_LOG_VERBOSE("DML::%s: CursorBufferSizeInTime = %f\n", __func__, CursorBufferSizeInTime); + DML_LOG_VERBOSE("DML::%s: CursorBufferSize = %u (kbytes)\n", __func__, CursorBufferSize); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %u\n", __func__, cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_lines_per_chunk = %u\n", __func__, cursor_lines_per_chunk); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorCursor = %f\n", __func__, *UrgentBurstFactorCursor); + DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); +#endif + + } +} + +static void CalculateUrgentBurstFactor( + const struct dml2_plane_parameters *plane_cfg, + unsigned int swath_width_luma_ub, + unsigned int swath_width_chroma_ub, + unsigned int SwathHeightY, + unsigned int SwathHeightC, + double LineTime, + double UrgentLatency, + double VRatio, + double VRatioC, + double BytePerPixelInDETY, + double BytePerPixelInDETC, + unsigned int DETBufferSizeY, + unsigned int DETBufferSizeC, + // Output + double *UrgentBurstFactorLuma, + double *UrgentBurstFactorChroma, + bool *NotEnoughUrgentLatencyHiding) +{ + double LinesInDETLuma; + double LinesInDETChroma; + double DETBufferSizeInTimeLuma; + double DETBufferSizeInTimeChroma; + + *NotEnoughUrgentLatencyHiding = 0; + *UrgentBurstFactorLuma = 0; + *UrgentBurstFactorChroma = 0; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); + DML_LOG_VERBOSE("DML::%s: VRatioC = %f\n", __func__, VRatioC); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeY = %d\n", __func__, DETBufferSizeY); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeC = %d\n", __func__, DETBufferSizeC); + DML_LOG_VERBOSE("DML::%s: BytePerPixelInDETY = %f\n", __func__, BytePerPixelInDETY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %d\n", __func__, swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); +#endif + DML_ASSERT(VRatio > 0); + + LinesInDETLuma = (dml_is_phantom_pipe(plane_cfg) ? 1024 * 1024 : DETBufferSizeY) / BytePerPixelInDETY / swath_width_luma_ub; + + DETBufferSizeInTimeLuma = math_floor2(LinesInDETLuma, SwathHeightY) * LineTime / VRatio; + if (DETBufferSizeInTimeLuma - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorLuma = 1; + } else { + *UrgentBurstFactorLuma = DETBufferSizeInTimeLuma / (DETBufferSizeInTimeLuma - UrgentLatency); + } + + if (BytePerPixelInDETC > 0) { + LinesInDETChroma = (dml_is_phantom_pipe(plane_cfg) ? 1024 * 1024 : DETBufferSizeC) / BytePerPixelInDETC / swath_width_chroma_ub; + + DETBufferSizeInTimeChroma = math_floor2(LinesInDETChroma, SwathHeightC) * LineTime / VRatioC; + if (DETBufferSizeInTimeChroma - UrgentLatency <= 0) { + *NotEnoughUrgentLatencyHiding = 1; + *UrgentBurstFactorChroma = 1; + } else { + *UrgentBurstFactorChroma = DETBufferSizeInTimeChroma / (DETBufferSizeInTimeChroma - UrgentLatency); + } + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: LinesInDETLuma = %f\n", __func__, LinesInDETLuma); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); + DML_LOG_VERBOSE("DML::%s: DETBufferSizeInTimeLuma = %f\n", __func__, DETBufferSizeInTimeLuma); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorLuma = %f\n", __func__, *UrgentBurstFactorLuma); + DML_LOG_VERBOSE("DML::%s: UrgentBurstFactorChroma = %f\n", __func__, *UrgentBurstFactorChroma); + DML_LOG_VERBOSE("DML::%s: NotEnoughUrgentLatencyHiding = %d\n", __func__, *NotEnoughUrgentLatencyHiding); +#endif +} + +static void CalculateDCFCLKDeepSleepTdlut( + const struct dml2_display_cfg *display_cfg, + unsigned int NumberOfActiveSurfaces, + unsigned int BytePerPixelY[], + unsigned int BytePerPixelC[], + unsigned int SwathWidthY[], + unsigned int SwathWidthC[], + unsigned int DPPPerSurface[], + double PSCL_THROUGHPUT[], + double PSCL_THROUGHPUT_CHROMA[], + double Dppclk[], + double ReadBandwidthLuma[], + double ReadBandwidthChroma[], + unsigned int ReturnBusWidth, + + double dispclk, + unsigned int tdlut_bytes_to_deliver[], + double prefetch_swath_time_us[], + + // Output + double *DCFClkDeepSleep) +{ + double DisplayPipeLineDeliveryTimeLuma; + double DisplayPipeLineDeliveryTimeChroma; + double DCFClkDeepSleepPerSurface[DML2_MAX_PLANES]; + double ReadBandwidth = 0.0; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + double pixel_rate_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + + if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio <= 1) { + DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] * DPPPerSurface[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_rate_mhz; + } else { + DisplayPipeLineDeliveryTimeLuma = SwathWidthY[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChroma = 0; + } else { + if (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio <= 1) { + DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] * DPPPerSurface[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_rate_mhz; + } else { + DisplayPipeLineDeliveryTimeChroma = SwathWidthC[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } + + if (BytePerPixelC[k] > 0) { + DCFClkDeepSleepPerSurface[k] = math_max2(__DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 32.0 / DisplayPipeLineDeliveryTimeLuma, + __DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthC[k] * BytePerPixelC[k] / 32.0 / DisplayPipeLineDeliveryTimeChroma); + } else { + DCFClkDeepSleepPerSurface[k] = __DML2_CALCS_DCFCLK_FACTOR__ * SwathWidthY[k] * BytePerPixelY[k] / 64.0 / DisplayPipeLineDeliveryTimeLuma; + } + DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], pixel_rate_mhz / 16); + + // adjust for 3dlut delivery time + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && tdlut_bytes_to_deliver[k] > 0) { + double tdlut_required_deepsleep_dcfclk = (double) tdlut_bytes_to_deliver[k] / 64.0 / prefetch_swath_time_us[k]; + + DML_LOG_VERBOSE("DML::%s: k=%d, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_bytes_to_deliver = %d\n", __func__, k, tdlut_bytes_to_deliver[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_swath_time_us = %f\n", __func__, k, prefetch_swath_time_us[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, tdlut_required_deepsleep_dcfclk = %f\n", __func__, k, tdlut_required_deepsleep_dcfclk); + + // increase the deepsleep dcfclk to match the original dispclk throughput rate + if (tdlut_required_deepsleep_dcfclk > DCFClkDeepSleepPerSurface[k]) { + DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], tdlut_required_deepsleep_dcfclk); + DCFClkDeepSleepPerSurface[k] = math_max2(DCFClkDeepSleepPerSurface[k], dispclk / 4.0); + } + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, PixelClock = %f\n", __func__, k, pixel_rate_mhz); + DML_LOG_VERBOSE("DML::%s: k=%u, DCFClkDeepSleepPerSurface = %f\n", __func__, k, DCFClkDeepSleepPerSurface[k]); +#endif + } + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + ReadBandwidth = ReadBandwidth + ReadBandwidthLuma[k] + ReadBandwidthChroma[k]; + } + + *DCFClkDeepSleep = math_max2(8.0, __DML2_CALCS_DCFCLK_FACTOR__ * ReadBandwidth / (double)ReturnBusWidth); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: __DML2_CALCS_DCFCLK_FACTOR__ = %f\n", __func__, __DML2_CALCS_DCFCLK_FACTOR__); + DML_LOG_VERBOSE("DML::%s: ReadBandwidth = %f\n", __func__, ReadBandwidth); + DML_LOG_VERBOSE("DML::%s: ReturnBusWidth = %u\n", __func__, ReturnBusWidth); + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, *DCFClkDeepSleep); +#endif + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + *DCFClkDeepSleep = math_max2(*DCFClkDeepSleep, DCFClkDeepSleepPerSurface[k]); + } + + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f (final)\n", __func__, *DCFClkDeepSleep); +} + +static noinline_for_stack void CalculateDCFCLKDeepSleep( + const struct dml2_display_cfg *display_cfg, + unsigned int NumberOfActiveSurfaces, + unsigned int BytePerPixelY[], + unsigned int BytePerPixelC[], + unsigned int SwathWidthY[], + unsigned int SwathWidthC[], + unsigned int DPPPerSurface[], + double PSCL_THROUGHPUT[], + double PSCL_THROUGHPUT_CHROMA[], + double Dppclk[], + double ReadBandwidthLuma[], + double ReadBandwidthChroma[], + unsigned int ReturnBusWidth, + + // Output + double *DCFClkDeepSleep) +{ + double zero_double[DML2_MAX_PLANES]; + unsigned int zero_integer[DML2_MAX_PLANES]; + + memset(zero_double, 0, DML2_MAX_PLANES * sizeof(double)); + memset(zero_integer, 0, DML2_MAX_PLANES * sizeof(unsigned int)); + + CalculateDCFCLKDeepSleepTdlut( + display_cfg, + NumberOfActiveSurfaces, + BytePerPixelY, + BytePerPixelC, + SwathWidthY, + SwathWidthC, + DPPPerSurface, + PSCL_THROUGHPUT, + PSCL_THROUGHPUT_CHROMA, + Dppclk, + ReadBandwidthLuma, + ReadBandwidthChroma, + ReturnBusWidth, + 0, + zero_integer, //tdlut_bytes_to_deliver, + zero_double, //prefetch_swath_time_us, + + // Output + DCFClkDeepSleep); +} + +static double CalculateWriteBackDelay( + enum dml2_source_format_class WritebackPixelFormat, + double WritebackHRatio, + double WritebackVRatio, + unsigned int WritebackVTaps, + unsigned int WritebackDestinationWidth, + unsigned int WritebackDestinationHeight, + unsigned int WritebackSourceHeight, + unsigned int HTotal) +{ + double CalculateWriteBackDelay; + double Line_length; + double Output_lines_last_notclamped; + double WritebackVInit; + + WritebackVInit = (WritebackVRatio + WritebackVTaps + 1) / 2; + Line_length = math_max2((double)WritebackDestinationWidth, math_ceil2((double)WritebackDestinationWidth / 6.0, 1.0) * WritebackVTaps); + Output_lines_last_notclamped = WritebackDestinationHeight - 1 - math_ceil2(((double)WritebackSourceHeight - (double)WritebackVInit) / (double)WritebackVRatio, 1.0); + if (Output_lines_last_notclamped < 0) { + CalculateWriteBackDelay = 0; + } else { + CalculateWriteBackDelay = Output_lines_last_notclamped * Line_length + (HTotal - WritebackDestinationWidth) + 80; + } + return CalculateWriteBackDelay; +} + +static unsigned int CalculateMaxVStartup( + bool ptoi_supported, + unsigned int vblank_nom_default_us, + const struct dml2_timing_cfg *timing, + double write_back_delay_us) +{ + unsigned int vblank_size = 0; + unsigned int max_vstartup_lines = 0; + + double line_time_us = (double)timing->h_total / ((double)timing->pixel_clock_khz / 1000); + unsigned int vblank_actual = timing->v_total - timing->v_active; + unsigned int vblank_nom_default_in_line = (unsigned int)math_floor2((double)vblank_nom_default_us / line_time_us, 1.0); + unsigned int vblank_avail = (timing->vblank_nom == 0) ? vblank_nom_default_in_line : (unsigned int)timing->vblank_nom; + + vblank_size = (unsigned int)math_min2(vblank_actual, vblank_avail); + + if (timing->interlaced && !ptoi_supported) + max_vstartup_lines = (unsigned int)(math_floor2((vblank_size - 1) / 2.0, 1.0)); + else + max_vstartup_lines = vblank_size - (unsigned int)math_max2(1.0, math_ceil2(write_back_delay_us / line_time_us, 1.0)); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VBlankNom = %lu\n", __func__, timing->vblank_nom); + DML_LOG_VERBOSE("DML::%s: vblank_nom_default_us = %u\n", __func__, vblank_nom_default_us); + DML_LOG_VERBOSE("DML::%s: line_time_us = %f\n", __func__, line_time_us); + DML_LOG_VERBOSE("DML::%s: vblank_actual = %u\n", __func__, vblank_actual); + DML_LOG_VERBOSE("DML::%s: vblank_avail = %u\n", __func__, vblank_avail); + DML_LOG_VERBOSE("DML::%s: max_vstartup_lines = %u\n", __func__, max_vstartup_lines); +#endif + max_vstartup_lines = (unsigned int)math_min2(max_vstartup_lines, DML_MAX_VSTARTUP_START); + return max_vstartup_lines; +} + +static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *p) +{ + unsigned int MaximumSwathHeightY[DML2_MAX_PLANES] = { 0 }; + unsigned int MaximumSwathHeightC[DML2_MAX_PLANES] = { 0 }; + unsigned int RoundedUpSwathSizeBytesY[DML2_MAX_PLANES] = { 0 }; + unsigned int RoundedUpSwathSizeBytesC[DML2_MAX_PLANES] = { 0 }; + unsigned int SwathWidthSingleDPP[DML2_MAX_PLANES] = { 0 }; + unsigned int SwathWidthSingleDPPChroma[DML2_MAX_PLANES] = { 0 }; + + unsigned int TotalActiveDPP = 0; + bool NoChromaOrLinear = true; + unsigned int SurfaceDoingUnboundedRequest = 0; + unsigned int DETBufferSizeInKByteForSwathCalculation; + + const long TTUFIFODEPTH = 8; + const long MAXIMUMCOMPRESSION = 4; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: ForceSingleDPP = %u\n", __func__, p->ForceSingleDPP); + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + DML_LOG_VERBOSE("DML::%s: DPPPerSurface[%u] = %u\n", __func__, k, p->DPPPerSurface[k]); + } +#endif + CalculateSwathWidth( + p->display_cfg, + p->ForceSingleDPP, + p->NumberOfActiveSurfaces, + p->ODMMode, + p->BytePerPixY, + p->BytePerPixC, + p->Read256BytesBlockHeightY, + p->Read256BytesBlockHeightC, + p->Read256BytesBlockWidthY, + p->Read256BytesBlockWidthC, + p->surf_linear128_l, + p->surf_linear128_c, + p->DPPPerSurface, + + // Output + p->req_per_swath_ub_l, + p->req_per_swath_ub_c, + SwathWidthSingleDPP, + SwathWidthSingleDPPChroma, + p->SwathWidth, + p->SwathWidthChroma, + MaximumSwathHeightY, + MaximumSwathHeightC, + p->swath_width_luma_ub, + p->swath_width_chroma_ub); + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + p->full_swath_bytes_l[k] = (unsigned int)(p->swath_width_luma_ub[k] * p->BytePerPixDETY[k] * MaximumSwathHeightY[k]); + p->full_swath_bytes_c[k] = (unsigned int)(p->swath_width_chroma_ub[k] * p->BytePerPixDETC[k] * MaximumSwathHeightC[k]); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u DPPPerSurface = %u\n", __func__, k, p->DPPPerSurface[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_luma_ub = %u\n", __func__, k, p->swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETY = %f\n", __func__, k, p->BytePerPixDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightY = %u\n", __func__, k, MaximumSwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u swath_width_chroma_ub = %u\n", __func__, k, p->swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u BytePerPixDETC = %f\n", __func__, k, p->BytePerPixDETC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathHeightC = %u\n", __func__, k, MaximumSwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); +#endif + if (p->display_cfg->plane_descriptors[k].pixel_format == dml2_420_10) { + p->full_swath_bytes_l[k] = (unsigned int)(math_ceil2((double)p->full_swath_bytes_l[k], 256)); + p->full_swath_bytes_c[k] = (unsigned int)(math_ceil2((double)p->full_swath_bytes_c[k], 256)); + } + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + TotalActiveDPP = TotalActiveDPP + (p->ForceSingleDPP ? 1 : p->DPPPerSurface[k]); + if (p->DPPPerSurface[k] > 0) + SurfaceDoingUnboundedRequest = k; + if (dml_is_420(p->display_cfg->plane_descriptors[k].pixel_format) || p->display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha + || p->display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { + NoChromaOrLinear = false; + } + } + + *p->UnboundedRequestEnabled = UnboundedRequest(p->display_cfg->overrides.hw.force_unbounded_requesting.enable, p->display_cfg->overrides.hw.force_unbounded_requesting.value, TotalActiveDPP, NoChromaOrLinear); + + CalculateDETBufferSize( + &scratch->CalculateDETBufferSize_locals, + p->display_cfg, + p->ForceSingleDPP, + p->NumberOfActiveSurfaces, + *p->UnboundedRequestEnabled, + p->nomDETInKByte, + p->MaxTotalDETInKByte, + p->ConfigReturnBufferSizeInKByte, + p->MinCompressedBufferSizeInKByte, + p->ConfigReturnBufferSegmentSizeInkByte, + p->CompressedBufferSegmentSizeInkByte, + p->ReadBandwidthLuma, + p->ReadBandwidthChroma, + p->full_swath_bytes_l, + p->full_swath_bytes_c, + p->DPPPerSurface, + + // Output + p->DETBufferSizeInKByte, // per hubp pipe + p->CompressedBufferSizeInkByte); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: TotalActiveDPP = %u\n", __func__, TotalActiveDPP); + DML_LOG_VERBOSE("DML::%s: nomDETInKByte = %u\n", __func__, p->nomDETInKByte); + DML_LOG_VERBOSE("DML::%s: ConfigReturnBufferSizeInKByte = %u\n", __func__, p->ConfigReturnBufferSizeInKByte); + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, *p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u\n", __func__, *p->CompressedBufferSizeInkByte); +#endif + + *p->ViewportSizeSupport = true; + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + + DETBufferSizeInKByteForSwathCalculation = (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) ? 1024 : p->DETBufferSizeInKByte[k]); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation = %u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); +#endif + if (p->display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { + p->SwathHeightY[k] = MaximumSwathHeightY[k]; + p->SwathHeightC[k] = MaximumSwathHeightC[k]; + RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; + RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; + + if (p->surf_linear128_l[k]) + p->request_size_bytes_luma[k] = 128; + else + p->request_size_bytes_luma[k] = 256; + + if (p->surf_linear128_c[k]) + p->request_size_bytes_chroma[k] = 128; + else + p->request_size_bytes_chroma[k] = 256; + + } else if (p->full_swath_bytes_l[k] + p->full_swath_bytes_c[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k]; + p->SwathHeightC[k] = MaximumSwathHeightC[k]; + RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; + RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; + p->request_size_bytes_luma[k] = 256; + p->request_size_bytes_chroma[k] = 256; + + } else if (p->full_swath_bytes_l[k] >= 1.5 * p->full_swath_bytes_c[k] && p->full_swath_bytes_l[k] / 2 + p->full_swath_bytes_c[k] <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; + p->SwathHeightC[k] = MaximumSwathHeightC[k]; + RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k] / 2; + RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k]; + p->request_size_bytes_luma[k] = ((p->BytePerPixY[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; + p->request_size_bytes_chroma[k] = 256; + + } else if (p->full_swath_bytes_l[k] < 1.5 * p->full_swath_bytes_c[k] && p->full_swath_bytes_l[k] + p->full_swath_bytes_c[k] / 2 <= DETBufferSizeInKByteForSwathCalculation * 1024 / 2) { + p->SwathHeightY[k] = MaximumSwathHeightY[k]; + p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; + RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k]; + RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k] / 2; + p->request_size_bytes_luma[k] = 256; + p->request_size_bytes_chroma[k] = ((p->BytePerPixC[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; + + } else { + p->SwathHeightY[k] = MaximumSwathHeightY[k] / 2; + p->SwathHeightC[k] = MaximumSwathHeightC[k] / 2; + RoundedUpSwathSizeBytesY[k] = p->full_swath_bytes_l[k] / 2; + RoundedUpSwathSizeBytesC[k] = p->full_swath_bytes_c[k] / 2; + p->request_size_bytes_luma[k] = ((p->BytePerPixY[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; + p->request_size_bytes_chroma[k] = ((p->BytePerPixC[k] == 2) == dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) ? 128 : 64; + } + + if (p->SwathHeightC[k] == 0) + p->request_size_bytes_chroma[k] = 0; + + if ((p->full_swath_bytes_l[k] / 2 + p->full_swath_bytes_c[k] / 2 > DETBufferSizeInKByteForSwathCalculation * 1024 / 2) || + p->SwathWidth[k] > p->MaximumSwathWidthLuma[k] || (p->SwathHeightC[k] > 0 && p->SwathWidthChroma[k] > p->MaximumSwathWidthChroma[k])) { + *p->ViewportSizeSupport = false; + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l=%u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c=%u\n", __func__, k, p->full_swath_bytes_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByteForSwathCalculation=%u\n", __func__, k, DETBufferSizeInKByteForSwathCalculation); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidth=%u\n", __func__, k, p->SwathWidth[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, p->MaximumSwathWidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathWidthChroma=%d\n", __func__, k, p->SwathWidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, p->MaximumSwathWidthChroma[k]); + p->ViewportSizeSupportPerSurface[k] = false; + } else { + p->ViewportSizeSupportPerSurface[k] = true; + } + + if (p->SwathHeightC[k] == 0) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, All DET will be used for plane0\n", __func__, k); +#endif + p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024; + p->DETBufferSizeC[k] = 0; + } else if (RoundedUpSwathSizeBytesY[k] <= 1.5 * RoundedUpSwathSizeBytesC[k]) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, Half DET will be used for plane0, and half for plane1\n", __func__, k); +#endif + p->DETBufferSizeY[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; + p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 / 2; + } else { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, 2/3 DET will be used for plane0, and 1/3 for plane1\n", __func__, k); +#endif + p->DETBufferSizeY[k] = (unsigned int)(math_floor2(p->DETBufferSizeInKByte[k] * 1024 * 2 / 3, 1024)); + p->DETBufferSizeC[k] = p->DETBufferSizeInKByte[k] * 1024 - p->DETBufferSizeY[k]; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u SwathHeightC = %u\n", __func__, k, p->SwathHeightC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_l = %u\n", __func__, k, p->full_swath_bytes_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u full_swath_bytes_c = %u\n", __func__, k, p->full_swath_bytes_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesY = %u\n", __func__, k, RoundedUpSwathSizeBytesY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeInKByte = %u\n", __func__, k, p->DETBufferSizeInKByte[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u DETBufferSizeC = %u\n", __func__, k, p->DETBufferSizeC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportSizeSupportPerSurface = %u\n", __func__, k, p->ViewportSizeSupportPerSurface[k]); +#endif + + } + + *p->compbuf_reserved_space_64b = 2 * p->pixel_chunk_size_kbytes * 1024 / 64; + if (*p->UnboundedRequestEnabled) { + *p->compbuf_reserved_space_64b = (unsigned int)math_ceil2(math_max2(*p->compbuf_reserved_space_64b, + (double)(p->rob_buffer_size_kbytes * 1024 / 64) - (double)(RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest] * TTUFIFODEPTH / (p->mrq_present ? MAXIMUMCOMPRESSION : 1) / 64)), 1.0); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: RoundedUpSwathSizeBytesY[%d] = %u\n", __func__, SurfaceDoingUnboundedRequest, RoundedUpSwathSizeBytesY[SurfaceDoingUnboundedRequest]); + DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %u\n", __func__, p->rob_buffer_size_kbytes); +#endif + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_64b = %u\n", __func__, *p->compbuf_reserved_space_64b); +#endif + + *p->hw_debug5 = false; +#ifdef ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE + if (p->NumberOfActiveSurfaces > 1) + *p->hw_debug5 = true; +#else + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (!(p->mrq_present) && (!(*p->UnboundedRequestEnabled)) && (TotalActiveDPP == 1) + && p->display_cfg->plane_descriptors[k].surface.dcc.enable + && ((p->rob_buffer_size_kbytes * 1024 * (p->mrq_present ? MAXIMUMCOMPRESSION : 1) + + *p->CompressedBufferSizeInkByte * MAXIMUMCOMPRESSION * 1024) > TTUFIFODEPTH * (RoundedUpSwathSizeBytesY[k] + RoundedUpSwathSizeBytesC[k]))) + *p->hw_debug5 = true; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u UnboundedRequestEnabled = %u\n", __func__, k, *p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: k=%u MAXIMUMCOMPRESSION = %lu\n", __func__, k, MAXIMUMCOMPRESSION); + DML_LOG_VERBOSE("DML::%s: k=%u TTUFIFODEPTH = %lu\n", __func__, k, TTUFIFODEPTH); + DML_LOG_VERBOSE("DML::%s: k=%u CompressedBufferSizeInkByte = %u\n", __func__, k, *p->CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: k=%u RoundedUpSwathSizeBytesC = %u\n", __func__, k, RoundedUpSwathSizeBytesC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u hw_debug5 = %u\n", __func__, k, *p->hw_debug5); +#endif + } +#endif +} + +static enum dml2_odm_mode DecideODMMode(unsigned int HActive, + double MaxDispclk, + unsigned int MaximumPixelsPerLinePerDSCUnit, + enum dml2_output_format_class OutFormat, + bool UseDSC, + unsigned int NumberOfDSCSlices, + double SurfaceRequiredDISPCLKWithoutODMCombine, + double SurfaceRequiredDISPCLKWithODMCombineTwoToOne, + double SurfaceRequiredDISPCLKWithODMCombineThreeToOne, + double SurfaceRequiredDISPCLKWithODMCombineFourToOne) +{ + enum dml2_odm_mode MinimumRequiredODMModeForMaxDispClock; + enum dml2_odm_mode MinimumRequiredODMModeForMaxDSCHActive; + enum dml2_odm_mode MinimumRequiredODMModeForMax420HActive; + enum dml2_odm_mode ODMMode = dml2_odm_mode_bypass; + + MinimumRequiredODMModeForMaxDispClock = + (SurfaceRequiredDISPCLKWithoutODMCombine <= MaxDispclk) ? dml2_odm_mode_bypass : + (SurfaceRequiredDISPCLKWithODMCombineTwoToOne <= MaxDispclk) ? dml2_odm_mode_combine_2to1 : + (SurfaceRequiredDISPCLKWithODMCombineThreeToOne <= MaxDispclk) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; + if (ODMMode < MinimumRequiredODMModeForMaxDispClock) + ODMMode = MinimumRequiredODMModeForMaxDispClock; + + if (UseDSC) { + MinimumRequiredODMModeForMaxDSCHActive = + (HActive <= 1 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_bypass : + (HActive <= 2 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_combine_2to1 : + (HActive <= 3 * MaximumPixelsPerLinePerDSCUnit) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; + if (ODMMode < MinimumRequiredODMModeForMaxDSCHActive) + ODMMode = MinimumRequiredODMModeForMaxDSCHActive; + } + + if (OutFormat == dml2_420) { + MinimumRequiredODMModeForMax420HActive = + (HActive <= 1 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_bypass : + (HActive <= 2 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_combine_2to1 : + (HActive <= 3 * DML2_MAX_FMT_420_BUFFER_WIDTH) ? dml2_odm_mode_combine_3to1 : dml2_odm_mode_combine_4to1; + if (ODMMode < MinimumRequiredODMModeForMax420HActive) + ODMMode = MinimumRequiredODMModeForMax420HActive; + } + + if (UseDSC) { + if (ODMMode == dml2_odm_mode_bypass && NumberOfDSCSlices > 4) + ODMMode = dml2_odm_mode_combine_2to1; + if (ODMMode == dml2_odm_mode_combine_2to1 && NumberOfDSCSlices > 8) + ODMMode = dml2_odm_mode_combine_3to1; + if (ODMMode == dml2_odm_mode_combine_3to1 && NumberOfDSCSlices != 12) + ODMMode = dml2_odm_mode_combine_4to1; + } + + return ODMMode; +} + +static void CalculateODMConstraints( + enum dml2_odm_mode ODMUse, + double SurfaceRequiredDISPCLKWithoutODMCombine, + double SurfaceRequiredDISPCLKWithODMCombineTwoToOne, + double SurfaceRequiredDISPCLKWithODMCombineThreeToOne, + double SurfaceRequiredDISPCLKWithODMCombineFourToOne, + unsigned int MaximumPixelsPerLinePerDSCUnit, + /* Output */ + double *DISPCLKRequired, + unsigned int *NumberOfDPPRequired, + unsigned int *MaxHActiveForDSC, + unsigned int *MaxDSCSlices, + unsigned int *MaxHActiveFor420) +{ + switch (ODMUse) { + case dml2_odm_mode_combine_2to1: + *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineTwoToOne; + *NumberOfDPPRequired = 2; + break; + case dml2_odm_mode_combine_3to1: + *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineThreeToOne; + *NumberOfDPPRequired = 3; + break; + case dml2_odm_mode_combine_4to1: + *DISPCLKRequired = SurfaceRequiredDISPCLKWithODMCombineFourToOne; + *NumberOfDPPRequired = 4; + break; + case dml2_odm_mode_auto: + case dml2_odm_mode_split_1to2: + case dml2_odm_mode_mso_1to2: + case dml2_odm_mode_mso_1to4: + case dml2_odm_mode_bypass: + default: + *DISPCLKRequired = SurfaceRequiredDISPCLKWithoutODMCombine; + *NumberOfDPPRequired = 1; + break; + } + *MaxHActiveForDSC = *NumberOfDPPRequired * MaximumPixelsPerLinePerDSCUnit; + *MaxDSCSlices = *NumberOfDPPRequired * DML_MAX_NUM_OF_SLICES_PER_DSC; + *MaxHActiveFor420 = *NumberOfDPPRequired * DML2_MAX_FMT_420_BUFFER_WIDTH; +} + +static bool ValidateODMMode(enum dml2_odm_mode ODMMode, + double MaxDispclk, + unsigned int HActive, + enum dml2_output_format_class OutFormat, + bool UseDSC, + unsigned int NumberOfDSCSlices, + unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, + unsigned int MaxNumDPP, + unsigned int MaxNumOPP, + double DISPCLKRequired, + unsigned int NumberOfDPPRequired, + unsigned int MaxHActiveForDSC, + unsigned int MaxDSCSlices, + unsigned int MaxHActiveFor420) +{ + bool are_odm_segments_symmetrical = (ODMMode == dml2_odm_mode_combine_3to1) ? UseDSC : true; + bool is_max_dsc_slice_required = (ODMMode == dml2_odm_mode_combine_3to1); + unsigned int pixels_per_clock_cycle = (OutFormat == dml2_420 || OutFormat == dml2_n422) ? 2 : 1; + unsigned int h_timing_div_mode = + (ODMMode == dml2_odm_mode_combine_4to1 || ODMMode == dml2_odm_mode_combine_3to1) ? 4 : + (ODMMode == dml2_odm_mode_combine_2to1) ? 2 : pixels_per_clock_cycle; + + if (DISPCLKRequired > MaxDispclk) + return false; + if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP || (TotalNumberOfActiveOPP + NumberOfDPPRequired) > MaxNumOPP) + return false; + if (are_odm_segments_symmetrical) { + if (HActive % (NumberOfDPPRequired * pixels_per_clock_cycle)) + return false; + } + if (HActive % h_timing_div_mode) + /* + * TODO - OTG_H_TOTAL, OTG_H_BLANK_START/END and + * OTG_H_SYNC_A_START/END all need to be visible by h timing div + * mode. This logic only checks H active. + */ + return false; + + if (UseDSC) { + if (HActive > MaxHActiveForDSC) + return false; + if (NumberOfDSCSlices > MaxDSCSlices) + return false; + if (HActive % NumberOfDSCSlices) + return false; + if (NumberOfDSCSlices % NumberOfDPPRequired) + return false; + if (is_max_dsc_slice_required) { + if (NumberOfDSCSlices != MaxDSCSlices) + return false; + } + } + + if (OutFormat == dml2_420) { + if (HActive > MaxHActiveFor420) + return false; + } + + return true; +} + +static noinline_for_stack void CalculateODMMode( + unsigned int MaximumPixelsPerLinePerDSCUnit, + unsigned int HActive, + enum dml2_output_format_class OutFormat, + enum dml2_output_encoder_class Output, + enum dml2_odm_mode ODMUse, + double MaxDispclk, + bool DSCEnable, + unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, + unsigned int MaxNumDPP, + unsigned int MaxNumOPP, + double PixelClock, + unsigned int NumberOfDSCSlices, + + // Output + bool *TotalAvailablePipesSupport, + unsigned int *NumberOfDPP, + enum dml2_odm_mode *ODMMode, + double *RequiredDISPCLKPerSurface) +{ + double SurfaceRequiredDISPCLKWithoutODMCombine; + double SurfaceRequiredDISPCLKWithODMCombineTwoToOne; + double SurfaceRequiredDISPCLKWithODMCombineThreeToOne; + double SurfaceRequiredDISPCLKWithODMCombineFourToOne; + double DISPCLKRequired; + unsigned int NumberOfDPPRequired; + unsigned int MaxHActiveForDSC; + unsigned int MaxDSCSlices; + unsigned int MaxHActiveFor420; + bool success; + bool UseDSC = DSCEnable && (NumberOfDSCSlices > 0); + enum dml2_odm_mode DecidedODMMode; + bool isTMDS420 = (OutFormat == dml2_420 && Output == dml2_hdmi); + + SurfaceRequiredDISPCLKWithoutODMCombine = CalculateRequiredDispclk(dml2_odm_mode_bypass, PixelClock, isTMDS420); + SurfaceRequiredDISPCLKWithODMCombineTwoToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_2to1, PixelClock, isTMDS420); + SurfaceRequiredDISPCLKWithODMCombineThreeToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_3to1, PixelClock, isTMDS420); + SurfaceRequiredDISPCLKWithODMCombineFourToOne = CalculateRequiredDispclk(dml2_odm_mode_combine_4to1, PixelClock, isTMDS420); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: ODMUse = %d\n", __func__, ODMUse); + DML_LOG_VERBOSE("DML::%s: Output = %d\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: DSCEnable = %d\n", __func__, DSCEnable); + DML_LOG_VERBOSE("DML::%s: MaxDispclk = %f\n", __func__, MaxDispclk); + DML_LOG_VERBOSE("DML::%s: MaximumPixelsPerLinePerDSCUnit = %d\n", __func__, MaximumPixelsPerLinePerDSCUnit); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithoutODMCombine = %f\n", __func__, SurfaceRequiredDISPCLKWithoutODMCombine); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineTwoToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineTwoToOne); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineThreeToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineThreeToOne); + DML_LOG_VERBOSE("DML::%s: SurfaceRequiredDISPCLKWithODMCombineFourToOne = %f\n", __func__, SurfaceRequiredDISPCLKWithODMCombineFourToOne); +#endif + if (ODMUse == dml2_odm_mode_auto) + DecidedODMMode = DecideODMMode(HActive, + MaxDispclk, + MaximumPixelsPerLinePerDSCUnit, + OutFormat, + UseDSC, + NumberOfDSCSlices, + SurfaceRequiredDISPCLKWithoutODMCombine, + SurfaceRequiredDISPCLKWithODMCombineTwoToOne, + SurfaceRequiredDISPCLKWithODMCombineThreeToOne, + SurfaceRequiredDISPCLKWithODMCombineFourToOne); + else + DecidedODMMode = ODMUse; + CalculateODMConstraints(DecidedODMMode, + SurfaceRequiredDISPCLKWithoutODMCombine, + SurfaceRequiredDISPCLKWithODMCombineTwoToOne, + SurfaceRequiredDISPCLKWithODMCombineThreeToOne, + SurfaceRequiredDISPCLKWithODMCombineFourToOne, + MaximumPixelsPerLinePerDSCUnit, + &DISPCLKRequired, + &NumberOfDPPRequired, + &MaxHActiveForDSC, + &MaxDSCSlices, + &MaxHActiveFor420); + success = ValidateODMMode(DecidedODMMode, + MaxDispclk, + HActive, + OutFormat, + UseDSC, + NumberOfDSCSlices, + TotalNumberOfActiveDPP, + TotalNumberOfActiveOPP, + MaxNumDPP, + MaxNumOPP, + DISPCLKRequired, + NumberOfDPPRequired, + MaxHActiveForDSC, + MaxDSCSlices, + MaxHActiveFor420); + + *ODMMode = DecidedODMMode; + *TotalAvailablePipesSupport = success; + *NumberOfDPP = NumberOfDPPRequired; + *RequiredDISPCLKPerSurface = success ? DISPCLKRequired : 0; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: ODMMode = %d\n", __func__, *ODMMode); + DML_LOG_VERBOSE("DML::%s: NumberOfDPP = %d\n", __func__, *NumberOfDPP); + DML_LOG_VERBOSE("DML::%s: TotalAvailablePipesSupport = %d\n", __func__, *TotalAvailablePipesSupport); + DML_LOG_VERBOSE("DML::%s: RequiredDISPCLKPerSurface = %f\n", __func__, *RequiredDISPCLKPerSurface); +#endif +} + +static noinline_for_stack void CalculateOutputLink( + struct dml2_core_internal_scratch *s, + double PHYCLK, + double PHYCLKD18, + double PHYCLKD32, + double Downspreading, + enum dml2_output_encoder_class Output, + enum dml2_output_format_class OutputFormat, + unsigned int HTotal, + unsigned int HActive, + double PixelClockBackEnd, + double ForcedOutputLinkBPP, + unsigned int DSCInputBitPerComponent, + unsigned int NumberOfDSCSlices, + double AudioSampleRate, + unsigned int AudioSampleLayout, + enum dml2_odm_mode ODMModeNoDSC, + enum dml2_odm_mode ODMModeDSC, + enum dml2_dsc_enable_option DSCEnable, + unsigned int OutputLinkDPLanes, + enum dml2_output_link_dp_rate OutputLinkDPRate, + + // Output + bool *RequiresDSC, + bool *RequiresFEC, + double *OutBpp, + enum dml2_core_internal_output_type *OutputType, + enum dml2_core_internal_output_type_rate *OutputRate, + unsigned int *RequiredSlots) +{ + bool LinkDSCEnable; + unsigned int dummy; + *RequiresDSC = false; + *RequiresFEC = false; + *OutBpp = 0; + + *OutputType = dml2_core_internal_output_type_unknown; + *OutputRate = dml2_core_internal_output_rate_unknown; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DSCEnable = %u (dis, en, en_if_necessary)\n", __func__, DSCEnable); + DML_LOG_VERBOSE("DML::%s: PHYCLK = %f\n", __func__, PHYCLK); + DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); + DML_LOG_VERBOSE("DML::%s: AudioSampleRate = %f\n", __func__, AudioSampleRate); + DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: ODMModeNoDSC = %u\n", __func__, ODMModeNoDSC); + DML_LOG_VERBOSE("DML::%s: ODMModeDSC = %u\n", __func__, ODMModeDSC); + DML_LOG_VERBOSE("DML::%s: ForcedOutputLinkBPP = %f\n", __func__, ForcedOutputLinkBPP); + DML_LOG_VERBOSE("DML::%s: Output (encoder) = %u\n", __func__, Output); + DML_LOG_VERBOSE("DML::%s: OutputLinkDPRate = %u\n", __func__, OutputLinkDPRate); +#endif + { + if (Output == dml2_hdmi) { + *RequiresDSC = false; + *RequiresFEC = false; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, math_min2(600, PHYCLK) * 10, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, false, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = "HDMI"; + *OutputType = dml2_core_internal_output_type_hdmi; + } else if (Output == dml2_dp || Output == dml2_dp2p0 || Output == dml2_edp) { + if (DSCEnable == dml2_dsc_enable) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml2_dp || Output == dml2_dp2p0) { + *RequiresFEC = true; + } else { + *RequiresFEC = false; + } + } else { + *RequiresDSC = false; + LinkDSCEnable = false; + if (Output == dml2_dp2p0) { + *RequiresFEC = true; + } else { + *RequiresFEC = false; + } + } + if (Output == dml2_dp2p0) { + *OutBpp = 0; + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr10) && PHYCLKD32 >= 10000.0 / 32) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && PHYCLKD32 < 13500.0 / 32 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 10000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR10"; + *OutputType = dml2_core_internal_output_type_dp2p0; + *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr10; + } + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr13p5) && *OutBpp == 0 && PHYCLKD32 >= 13500.0 / 32) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && PHYCLKD32 < 20000.0 / 32 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 13500, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR13p5"; + *OutputType = dml2_core_internal_output_type_dp2p0; + *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr13p5; + } + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_uhbr20) && *OutBpp == 0 && PHYCLKD32 >= 20000.0 / 32) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 20000, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " UHBR20"; + *OutputType = dml2_core_internal_output_type_dp2p0; + *OutputRate = dml2_core_internal_output_rate_dp_rate_uhbr20; + } + } else { // output is dp or edp + *OutBpp = 0; + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr) && PHYCLK >= 270) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + if (*OutBpp == 0 && PHYCLK < 540 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml2_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 2700, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR"; + *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; + *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr; + } + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr2) && *OutBpp == 0 && PHYCLK >= 540) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && PHYCLK < 810 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml2_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 5400, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR2"; + *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; + *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr2; + } + if ((OutputLinkDPRate == dml2_dp_rate_na || OutputLinkDPRate == dml2_dp_rate_hbr3) && *OutBpp == 0 && PHYCLK >= 810) { // VBA_ERROR, vba code doesn't have hbr3 check + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + + if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + if (Output == dml2_dp) { + *RequiresFEC = true; + } + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, (1 - Downspreading / 100) * 8100, OutputLinkDPLanes, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, + OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, RequiredSlots); + } + //OutputTypeAndRate = Output & " HBR3"; + *OutputType = (Output == dml2_dp) ? dml2_core_internal_output_type_dp : dml2_core_internal_output_type_edp; + *OutputRate = dml2_core_internal_output_rate_dp_rate_hbr3; + } + } + } else if (Output == dml2_hdmifrl) { + if (DSCEnable == dml2_dsc_enable) { + *RequiresDSC = true; + LinkDSCEnable = true; + *RequiresFEC = true; + } else { + *RequiresDSC = false; + LinkDSCEnable = false; + *RequiresFEC = false; + } + *OutBpp = 0; + if (PHYCLKD18 >= 3000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 3000, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = Output & "3x3"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_3x3; + } + if (*OutBpp == 0 && PHYCLKD18 >= 6000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 6000, 3, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = Output & "6x3"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_6x3; + } + if (*OutBpp == 0 && PHYCLKD18 >= 6000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 6000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = Output & "6x4"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_6x4; + } + if (*OutBpp == 0 && PHYCLKD18 >= 8000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 8000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + //OutputTypeAndRate = Output & "8x4"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_8x4; + } + if (*OutBpp == 0 && PHYCLKD18 >= 10000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 10000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0 && PHYCLKD18 < 12000.0 / 18) { + *RequiresDSC = true; + LinkDSCEnable = true; + *RequiresFEC = true; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 10000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + } + //OutputTypeAndRate = Output & "10x4"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_10x4; + } + if (*OutBpp == 0 && PHYCLKD18 >= 12000.0 / 18) { + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 12000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + if (*OutBpp == 0 && DSCEnable == dml2_dsc_enable_if_necessary && ForcedOutputLinkBPP == 0) { + *RequiresDSC = true; + LinkDSCEnable = true; + *RequiresFEC = true; + *OutBpp = TruncToValidBPP(&s->TruncToValidBPP_locals, 12000, 4, HTotal, HActive, PixelClockBackEnd, ForcedOutputLinkBPP, LinkDSCEnable, Output, OutputFormat, DSCInputBitPerComponent, NumberOfDSCSlices, (unsigned int)AudioSampleRate, AudioSampleLayout, ODMModeNoDSC, ODMModeDSC, &dummy); + } + //OutputTypeAndRate = Output & "12x4"; + *OutputType = dml2_core_internal_output_type_hdmifrl; + *OutputRate = dml2_core_internal_output_rate_hdmi_rate_12x4; + } + } + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: RequiresDSC = %u\n", __func__, *RequiresDSC); + DML_LOG_VERBOSE("DML::%s: RequiresFEC = %u\n", __func__, *RequiresFEC); + DML_LOG_VERBOSE("DML::%s: OutBpp = %f\n", __func__, *OutBpp); +#endif +} + +static double CalculateWriteBackDISPCLK( + enum dml2_source_format_class WritebackPixelFormat, + double PixelClock, + double WritebackHRatio, + double WritebackVRatio, + unsigned int WritebackHTaps, + unsigned int WritebackVTaps, + unsigned int WritebackSourceWidth, + unsigned int WritebackDestinationWidth, + unsigned int HTotal, + unsigned int WritebackLineBufferSize) +{ + double DISPCLK_H, DISPCLK_V, DISPCLK_HB; + + DISPCLK_H = PixelClock * math_ceil2((double)WritebackHTaps / 8.0, 1) / WritebackHRatio; + DISPCLK_V = PixelClock * (WritebackVTaps * math_ceil2((double)WritebackDestinationWidth / 6.0, 1) + 8.0) / (double)HTotal; + DISPCLK_HB = PixelClock * WritebackVTaps * (WritebackDestinationWidth * WritebackVTaps - WritebackLineBufferSize / 57.0) / 6.0 / (double)WritebackSourceWidth; + return math_max3(DISPCLK_H, DISPCLK_V, DISPCLK_HB); +} + +static double RequiredDTBCLK( + bool DSCEnable, + double PixelClock, + enum dml2_output_format_class OutputFormat, + double OutputBpp, + unsigned int DSCSlices, + unsigned int HTotal, + unsigned int HActive, + unsigned int AudioRate, + unsigned int AudioLayout) +{ + if (DSCEnable != true) { + return math_max2(PixelClock / 4.0 * OutputBpp / 24.0, 25.0); + } else { + double PixelWordRate = PixelClock / (OutputFormat == dml2_444 ? 1 : 2); + double HCActive = math_ceil2(DSCSlices * math_ceil2(OutputBpp * math_ceil2(HActive / DSCSlices, 1) / 8.0, 1) / 3.0, 1); + double HCBlank = 64 + 32 * math_ceil2(AudioRate * (AudioLayout == 1 ? 1 : 0.25) * HTotal / (PixelClock * 1000), 1); + double AverageTribyteRate = PixelWordRate * (HCActive + HCBlank) / HTotal; + double HActiveTribyteRate = PixelWordRate * HCActive / HActive; + return math_max4(PixelWordRate / 4.0, AverageTribyteRate / 4.0, HActiveTribyteRate / 4.0, 25.0) * 1.002; + } +} + +static unsigned int DSCDelayRequirement( + bool DSCEnabled, + enum dml2_odm_mode ODMMode, + unsigned int DSCInputBitPerComponent, + double OutputBpp, + unsigned int HActive, + unsigned int HTotal, + unsigned int NumberOfDSCSlices, + enum dml2_output_format_class OutputFormat, + enum dml2_output_encoder_class Output, + double PixelClock, + double PixelClockBackEnd) +{ + unsigned int DSCDelayRequirement_val = 0; + unsigned int NumberOfDSCSlicesFactor = 1; + + if (DSCEnabled == true && OutputBpp != 0) { + + if (ODMMode == dml2_odm_mode_combine_4to1) + NumberOfDSCSlicesFactor = 4; + else if (ODMMode == dml2_odm_mode_combine_3to1) + NumberOfDSCSlicesFactor = 3; + else if (ODMMode == dml2_odm_mode_combine_2to1) + NumberOfDSCSlicesFactor = 2; + + DSCDelayRequirement_val = NumberOfDSCSlicesFactor * (dscceComputeDelay(DSCInputBitPerComponent, OutputBpp, (unsigned int)(math_ceil2((double)HActive / (double)NumberOfDSCSlices, 1.0)), + (NumberOfDSCSlices / NumberOfDSCSlicesFactor), OutputFormat, Output) + dscComputeDelay(OutputFormat, Output)); + + DSCDelayRequirement_val = (unsigned int)(DSCDelayRequirement_val + (HTotal - HActive) * math_ceil2((double)DSCDelayRequirement_val / (double)HActive, 1.0)); + DSCDelayRequirement_val = (unsigned int)(DSCDelayRequirement_val * PixelClock / PixelClockBackEnd); + + } else { + DSCDelayRequirement_val = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DSCEnabled= %u\n", __func__, DSCEnabled); + DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, ODMMode); + DML_LOG_VERBOSE("DML::%s: OutputBpp = %f\n", __func__, OutputBpp); + DML_LOG_VERBOSE("DML::%s: HActive = %u\n", __func__, HActive); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); + DML_LOG_VERBOSE("DML::%s: PixelClockBackEnd = %f\n", __func__, PixelClockBackEnd); + DML_LOG_VERBOSE("DML::%s: OutputFormat = %u\n", __func__, OutputFormat); + DML_LOG_VERBOSE("DML::%s: DSCInputBitPerComponent = %u\n", __func__, DSCInputBitPerComponent); + DML_LOG_VERBOSE("DML::%s: NumberOfDSCSlices = %u\n", __func__, NumberOfDSCSlices); + DML_LOG_VERBOSE("DML::%s: DSCDelayRequirement_val = %u\n", __func__, DSCDelayRequirement_val); +#endif + + return DSCDelayRequirement_val; +} + +static void CalculateSurfaceSizeInMall( + const struct dml2_display_cfg *display_cfg, + unsigned int NumberOfActiveSurfaces, + unsigned int MALLAllocatedForDCN, + unsigned int BytesPerPixelY[], + unsigned int BytesPerPixelC[], + unsigned int Read256BytesBlockWidthY[], + unsigned int Read256BytesBlockWidthC[], + unsigned int Read256BytesBlockHeightY[], + unsigned int Read256BytesBlockHeightC[], + unsigned int ReadBlockWidthY[], + unsigned int ReadBlockWidthC[], + unsigned int ReadBlockHeightY[], + unsigned int ReadBlockHeightC[], + + // Output + unsigned int SurfaceSizeInMALL[], + bool *ExceededMALLSize) +{ + unsigned int TotalSurfaceSizeInMALLForSS = 0; + unsigned int TotalSurfaceSizeInMALLForSubVP = 0; + unsigned int MALLAllocatedForDCNInBytes = MALLAllocatedForDCN * 1024 * 1024; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + const struct dml2_composition_cfg *composition = &display_cfg->plane_descriptors[k].composition; + const struct dml2_surface_cfg *surface = &display_cfg->plane_descriptors[k].surface; + + if (composition->viewport.stationary) { + SurfaceSizeInMALL[k] = (unsigned int)(math_min2(math_ceil2((double)surface->plane0.width, ReadBlockWidthY[k]), + math_floor2(composition->viewport.plane0.x_start + composition->viewport.plane0.width + ReadBlockWidthY[k] - 1, ReadBlockWidthY[k]) - + math_floor2((double)composition->viewport.plane0.x_start, ReadBlockWidthY[k])) * + math_min2(math_ceil2((double)surface->plane0.height, ReadBlockHeightY[k]), + math_floor2((double)composition->viewport.plane0.y_start + composition->viewport.plane0.height + ReadBlockHeightY[k] - 1, ReadBlockHeightY[k]) - + math_floor2((double)composition->viewport.plane0.y_start, ReadBlockHeightY[k])) * BytesPerPixelY[k]); + + if (ReadBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (unsigned int)(SurfaceSizeInMALL[k] + + math_min2(math_ceil2((double)surface->plane1.width, ReadBlockWidthC[k]), + math_floor2((double)composition->viewport.plane1.y_start + composition->viewport.plane1.width + ReadBlockWidthC[k] - 1, ReadBlockWidthC[k]) - + math_floor2((double)composition->viewport.plane1.y_start, ReadBlockWidthC[k])) * + math_min2(math_ceil2((double)surface->plane1.height, ReadBlockHeightC[k]), + math_floor2((double)composition->viewport.plane1.y_start + composition->viewport.plane1.height + ReadBlockHeightC[k] - 1, ReadBlockHeightC[k]) - + math_floor2(composition->viewport.plane1.y_start, ReadBlockHeightC[k])) * BytesPerPixelC[k]); + } + } else { + SurfaceSizeInMALL[k] = (unsigned int)(math_ceil2(math_min2(surface->plane0.width, composition->viewport.plane0.width + ReadBlockWidthY[k] - 1), ReadBlockWidthY[k]) * + math_ceil2(math_min2(surface->plane0.height, composition->viewport.plane0.height + ReadBlockHeightY[k] - 1), ReadBlockHeightY[k]) * BytesPerPixelY[k]); + if (ReadBlockWidthC[k] > 0) { + SurfaceSizeInMALL[k] = (unsigned int)(SurfaceSizeInMALL[k] + + math_ceil2(math_min2(surface->plane1.width, composition->viewport.plane1.width + ReadBlockWidthC[k] - 1), ReadBlockWidthC[k]) * + math_ceil2(math_min2(surface->plane1.height, composition->viewport.plane1.height + ReadBlockHeightC[k] - 1), ReadBlockHeightC[k]) * BytesPerPixelC[k]); + } + } + } + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + /* SS and Subvp counted separate as they are never used at the same time */ + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) + TotalSurfaceSizeInMALLForSubVP += SurfaceSizeInMALL[k]; + else if (display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable) + TotalSurfaceSizeInMALLForSS += SurfaceSizeInMALL[k]; + } + + *ExceededMALLSize = (TotalSurfaceSizeInMALLForSS > MALLAllocatedForDCNInBytes) || + (TotalSurfaceSizeInMALLForSubVP > MALLAllocatedForDCNInBytes); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: MALLAllocatedForDCN = %u\n", __func__, MALLAllocatedForDCN * 1024 * 1024); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSubVP = %u\n", __func__, TotalSurfaceSizeInMALLForSubVP); + DML_LOG_VERBOSE("DML::%s: TotalSurfaceSizeInMALLForSS = %u\n", __func__, TotalSurfaceSizeInMALLForSS); + DML_LOG_VERBOSE("DML::%s: ExceededMALLSize = %u\n", __func__, *ExceededMALLSize); +#endif +} + +static void calculate_tdlut_setting( + struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_calculate_tdlut_setting_params *p) +{ + // locals + unsigned int tdlut_bpe = 8; + unsigned int tdlut_width; + unsigned int tdlut_pitch_bytes; + unsigned int tdlut_footprint_bytes; + unsigned int vmpg_bytes; + unsigned int tdlut_vmpg_per_frame; + unsigned int tdlut_pte_req_per_frame; + unsigned int tdlut_bytes_per_line; + double tdlut_drain_rate; + unsigned int tdlut_mpc_width; + unsigned int tdlut_bytes_per_group_simple; + + if (!p->setup_for_tdlut) { + *p->tdlut_groups_per_2row_ub = 0; + *p->tdlut_opt_time = 0; + *p->tdlut_drain_time = 0; + *p->tdlut_bytes_to_deliver = 0; + *p->tdlut_bytes_per_group = 0; + *p->tdlut_pte_bytes_per_frame = 0; + *p->tdlut_bytes_per_frame = 0; + return; + } + + if (p->tdlut_mpc_width_flag) { + tdlut_mpc_width = 33; + tdlut_bytes_per_group_simple = 39*256; + } else { + tdlut_mpc_width = 17; + tdlut_bytes_per_group_simple = 10*256; + } + + vmpg_bytes = p->gpuvm_page_size_kbytes * 1024; + + if (p->tdlut_addressing_mode == dml2_tdlut_simple_linear) { + if (p->tdlut_width_mode == dml2_tdlut_width_17_cube) + tdlut_width = 4916; + else + tdlut_width = 35940; + } else { + if (p->tdlut_width_mode == dml2_tdlut_width_17_cube) + tdlut_width = 17; + else // dml2_tdlut_width_33_cube + tdlut_width = 33; + } + + if (p->is_gfx11) + tdlut_pitch_bytes = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 256); //256B alignment + else + tdlut_pitch_bytes = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 128); //128B alignment + + if (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) + tdlut_footprint_bytes = tdlut_pitch_bytes * tdlut_width * tdlut_width; + else + tdlut_footprint_bytes = tdlut_pitch_bytes; + + if (!p->gpuvm_enable) { + tdlut_vmpg_per_frame = 0; + tdlut_pte_req_per_frame = 0; + } else { + tdlut_vmpg_per_frame = (unsigned int)math_ceil2(tdlut_footprint_bytes - 1, vmpg_bytes) / vmpg_bytes + 1; + tdlut_pte_req_per_frame = (unsigned int)math_ceil2(tdlut_vmpg_per_frame - 1, 8) / 8 + 1; + } + tdlut_bytes_per_line = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 64); //64b request + *p->tdlut_pte_bytes_per_frame = tdlut_pte_req_per_frame * 64; + + if (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) { + //the tdlut_width is either 17 or 33 but the 33x33x33 is subsampled every other line/slice + *p->tdlut_bytes_per_frame = tdlut_bytes_per_line * tdlut_mpc_width * tdlut_mpc_width; + *p->tdlut_bytes_per_group = tdlut_bytes_per_line * tdlut_mpc_width; + //the delivery cycles is DispClk cycles per line * number of lines * number of slices + //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width; + tdlut_drain_rate = tdlut_bytes_per_line * p->dispclk_mhz / math_ceil2(tdlut_mpc_width/2.0, 1); + } else { + //tdlut_addressing_mode = tdlut_simple_linear, 3dlut width should be 4*1229=4916 elements + *p->tdlut_bytes_per_frame = (unsigned int)math_ceil2(tdlut_width * tdlut_bpe, 256); + *p->tdlut_bytes_per_group = tdlut_bytes_per_group_simple; + //tdlut_delivery_cycles = (unsigned int)math_ceil2(tdlut_width/2.0, 1); + tdlut_drain_rate = 2 * tdlut_bpe * p->dispclk_mhz; + } + + //the tdlut is fetched during the 2 row times of prefetch. + if (p->setup_for_tdlut) { + *p->tdlut_groups_per_2row_ub = (unsigned int)math_ceil2((double) *p->tdlut_bytes_per_frame / *p->tdlut_bytes_per_group, 1); + if (*p->tdlut_bytes_per_frame > p->cursor_buffer_size * 1024) + *p->tdlut_opt_time = (*p->tdlut_bytes_per_frame - p->cursor_buffer_size * 1024) / tdlut_drain_rate; + else + *p->tdlut_opt_time = 0; + *p->tdlut_drain_time = p->cursor_buffer_size * 1024 / tdlut_drain_rate; + *p->tdlut_bytes_to_deliver = (unsigned int) (p->cursor_buffer_size * 1024.0); + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: gpuvm_enable = %d\n", __func__, p->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: vmpg_bytes = %d\n", __func__, vmpg_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_vmpg_per_frame = %d\n", __func__, tdlut_vmpg_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_pte_req_per_frame = %d\n", __func__, tdlut_pte_req_per_frame); + + DML_LOG_VERBOSE("DML::%s: dispclk_mhz = %f\n", __func__, p->dispclk_mhz); + DML_LOG_VERBOSE("DML::%s: tdlut_width = %u\n", __func__, tdlut_width); + DML_LOG_VERBOSE("DML::%s: tdlut_addressing_mode = %s\n", __func__, (p->tdlut_addressing_mode == dml2_tdlut_sw_linear) ? "sw_linear" : "simple_linear"); + DML_LOG_VERBOSE("DML::%s: tdlut_pitch_bytes = %u\n", __func__, tdlut_pitch_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_footprint_bytes = %u\n", __func__, tdlut_footprint_bytes); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_frame = %u\n", __func__, *p->tdlut_bytes_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_line = %u\n", __func__, tdlut_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_per_group = %u\n", __func__, *p->tdlut_bytes_per_group); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_rate = %f\n", __func__, tdlut_drain_rate); + DML_LOG_VERBOSE("DML::%s: tdlut_delivery_cycles = %u\n", __func__, p->tdlut_addressing_mode == dml2_tdlut_sw_linear ? (unsigned int)math_ceil2(tdlut_mpc_width/2.0, 1) * tdlut_mpc_width * tdlut_mpc_width : (unsigned int)math_ceil2(tdlut_width/2.0, 1)); + DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, *p->tdlut_opt_time); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, *p->tdlut_drain_time); + DML_LOG_VERBOSE("DML::%s: tdlut_bytes_to_deliver = %d\n", __func__, *p->tdlut_bytes_to_deliver); + DML_LOG_VERBOSE("DML::%s: tdlut_groups_per_2row_ub = %d\n", __func__, *p->tdlut_groups_per_2row_ub); +#endif +} + +static void CalculateTarb( + const struct dml2_display_cfg *display_cfg, + unsigned int PixelChunkSizeInKByte, + unsigned int NumberOfActiveSurfaces, + unsigned int NumberOfDPP[], + unsigned int dpte_group_bytes[], + unsigned int tdlut_bytes_per_group[], + double HostVMInefficiencyFactor, + double HostVMInefficiencyFactorPrefetch, + unsigned int HostVMMinPageSize, + double ReturnBW, + unsigned int MetaChunkSize, + + // output + double *Tarb, + double *Tarb_prefetch) +{ + double extra_bytes = 0; + double extra_bytes_prefetch = 0; + double HostVMDynamicLevels = CalculateHostVMDynamicLevels(display_cfg->gpuvm_enable, display_cfg->hostvm_enable, HostVMMinPageSize, display_cfg->hostvm_max_non_cached_page_table_levels); + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + extra_bytes = extra_bytes + (NumberOfDPP[k] * PixelChunkSizeInKByte * 1024); + + if (display_cfg->plane_descriptors[k].surface.dcc.enable) + extra_bytes = extra_bytes + (MetaChunkSize * 1024); + + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) + extra_bytes = extra_bytes + tdlut_bytes_per_group[k]; + } + + extra_bytes_prefetch = extra_bytes; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (display_cfg->gpuvm_enable == true) { + extra_bytes = extra_bytes + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactor; + extra_bytes_prefetch = extra_bytes_prefetch + NumberOfDPP[k] * dpte_group_bytes[k] * (1 + 8 * HostVMDynamicLevels) * HostVMInefficiencyFactorPrefetch; + } + } + *Tarb = extra_bytes / ReturnBW; + *Tarb_prefetch = extra_bytes_prefetch / ReturnBW; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %d\n", __func__, PixelChunkSizeInKByte); + DML_LOG_VERBOSE("DML::%s: MetaChunkSize = %d\n", __func__, MetaChunkSize); + DML_LOG_VERBOSE("DML::%s: extra_bytes = %f\n", __func__, extra_bytes); + DML_LOG_VERBOSE("DML::%s: extra_bytes_prefetch = %f\n", __func__, extra_bytes_prefetch); +#endif +} + +static double CalculateTWait( + long reserved_vblank_time_ns, + double UrgentLatency, + double Ttrip, + double g6_temp_read_blackout_us) +{ + double TWait; + double t_urg_trip = math_max2(UrgentLatency, Ttrip); + TWait = math_max2(reserved_vblank_time_ns/1000.0, g6_temp_read_blackout_us) + t_urg_trip; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: reserved_vblank_time_ns = %ld\n", __func__, reserved_vblank_time_ns); + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, UrgentLatency); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, Ttrip); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, TWait); +#endif + return TWait; +} + + +static void CalculateVUpdateAndDynamicMetadataParameters( + unsigned int MaxInterDCNTileRepeaters, + double Dppclk, + double Dispclk, + double DCFClkDeepSleep, + double PixelClock, + unsigned int HTotal, + unsigned int VBlank, + unsigned int DynamicMetadataTransmittedBytes, + unsigned int DynamicMetadataLinesBeforeActiveRequired, + unsigned int InterlaceEnable, + bool ProgressiveToInterlaceUnitInOPP, + + // Output + double *TSetup, + double *Tdmbf, + double *Tdmec, + double *Tdmsks, + unsigned int *VUpdateOffsetPix, + unsigned int *VUpdateWidthPix, + unsigned int *VReadyOffsetPix) +{ + double TotalRepeaterDelayTime; + TotalRepeaterDelayTime = MaxInterDCNTileRepeaters * (2 / Dppclk + 3 / Dispclk); + *VUpdateWidthPix = (unsigned int)(math_ceil2((14.0 / DCFClkDeepSleep + 12.0 / Dppclk + TotalRepeaterDelayTime) * PixelClock, 1.0)); + *VReadyOffsetPix = (unsigned int)(math_ceil2(math_max2(150.0 / Dppclk, TotalRepeaterDelayTime + 20.0 / DCFClkDeepSleep + 10.0 / Dppclk) * PixelClock, 1.0)); + *VUpdateOffsetPix = (unsigned int)(math_ceil2(HTotal / 4.0, 1.0)); + *TSetup = (*VUpdateOffsetPix + *VUpdateWidthPix + *VReadyOffsetPix) / PixelClock; + *Tdmbf = DynamicMetadataTransmittedBytes / 4.0 / Dispclk; + *Tdmec = HTotal / PixelClock; + + if (DynamicMetadataLinesBeforeActiveRequired == 0) { + *Tdmsks = VBlank * HTotal / PixelClock / 2.0; + } else { + *Tdmsks = DynamicMetadataLinesBeforeActiveRequired * HTotal / PixelClock; + } + if (InterlaceEnable == 1 && ProgressiveToInterlaceUnitInOPP == false) { + *Tdmsks = *Tdmsks / 2; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DynamicMetadataLinesBeforeActiveRequired = %u\n", __func__, DynamicMetadataLinesBeforeActiveRequired); + DML_LOG_VERBOSE("DML::%s: VBlank = %u\n", __func__, VBlank); + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, HTotal); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, PixelClock); + DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, Dppclk); + DML_LOG_VERBOSE("DML::%s: DCFClkDeepSleep = %f\n", __func__, DCFClkDeepSleep); + DML_LOG_VERBOSE("DML::%s: MaxInterDCNTileRepeaters = %u\n", __func__, MaxInterDCNTileRepeaters); + DML_LOG_VERBOSE("DML::%s: TotalRepeaterDelayTime = %f\n", __func__, TotalRepeaterDelayTime); + + DML_LOG_VERBOSE("DML::%s: VUpdateWidthPix = %u\n", __func__, *VUpdateWidthPix); + DML_LOG_VERBOSE("DML::%s: VReadyOffsetPix = %u\n", __func__, *VReadyOffsetPix); + DML_LOG_VERBOSE("DML::%s: VUpdateOffsetPix = %u\n", __func__, *VUpdateOffsetPix); + + DML_LOG_VERBOSE("DML::%s: Tdmsks = %f\n", __func__, *Tdmsks); +#endif +} + +static double get_urgent_bandwidth_required( + struct dml2_core_shared_get_urgent_bandwidth_required_locals *l, + const struct dml2_display_cfg *display_cfg, + enum dml2_core_internal_soc_state_type state_type, + enum dml2_core_internal_bw_type bw_type, + bool inc_flip_bw, // including flip bw + bool use_qual_row_bw, + unsigned int NumberOfActiveSurfaces, + unsigned int NumberOfDPP[], + double dcc_dram_bw_nom_overhead_factor_p0[], + double dcc_dram_bw_nom_overhead_factor_p1[], + double dcc_dram_bw_pref_overhead_factor_p0[], + double dcc_dram_bw_pref_overhead_factor_p1[], + double mall_prefetch_sdp_overhead_factor[], + double mall_prefetch_dram_overhead_factor[], + double ReadBandwidthLuma[], + double ReadBandwidthChroma[], + double PrefetchBandwidthLuma[], + double PrefetchBandwidthChroma[], + double PrefetchBandwidthMax[], + double excess_vactive_fill_bw_l[], + double excess_vactive_fill_bw_c[], + double cursor_bw[], + double dpte_row_bw[], + double meta_row_bw[], + double prefetch_cursor_bw[], + double prefetch_vmrow_bw[], + double flip_bw[], + double UrgentBurstFactorLuma[], + double UrgentBurstFactorChroma[], + double UrgentBurstFactorCursor[], + double UrgentBurstFactorLumaPre[], + double UrgentBurstFactorChromaPre[], + double UrgentBurstFactorCursorPre[], + /* outputs */ + double surface_required_bw[], + double surface_peak_required_bw[]) +{ + // set inc_flip_bw = 0 for total_dchub_urgent_read_bw_noflip calculation, 1 for total_dchub_urgent_read_bw as described in the MAS + // set use_qual_row_bw = 1 to calculate using qualified row bandwidth, used for total_flip_bw calculation + + memset(l, 0, sizeof(struct dml2_core_shared_get_urgent_bandwidth_required_locals)); + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + l->mall_svp_prefetch_factor = (state_type == dml2_core_internal_soc_state_svp_prefetch) ? (bw_type == dml2_core_internal_bw_dram ? mall_prefetch_dram_overhead_factor[k] : mall_prefetch_sdp_overhead_factor[k]) : 1.0; + l->tmp_nom_adj_factor_p0 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p0[k] : 1.0) * l->mall_svp_prefetch_factor; + l->tmp_nom_adj_factor_p1 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p1[k] : 1.0) * l->mall_svp_prefetch_factor; + l->tmp_pref_adj_factor_p0 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_pref_overhead_factor_p0[k] : 1.0) * l->mall_svp_prefetch_factor; + l->tmp_pref_adj_factor_p1 = (bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_pref_overhead_factor_p1[k] : 1.0) * l->mall_svp_prefetch_factor; + + l->adj_factor_p0 = UrgentBurstFactorLuma[k] * l->tmp_nom_adj_factor_p0; + l->adj_factor_p1 = UrgentBurstFactorChroma[k] * l->tmp_nom_adj_factor_p1; + l->adj_factor_cur = UrgentBurstFactorCursor[k]; + l->adj_factor_p0_pre = UrgentBurstFactorLumaPre[k] * l->tmp_pref_adj_factor_p0; + l->adj_factor_p1_pre = UrgentBurstFactorChromaPre[k] * l->tmp_pref_adj_factor_p1; + l->adj_factor_cur_pre = UrgentBurstFactorCursorPre[k]; + + bool is_phantom = dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]); + bool exclude_this_plane = false; + + // Exclude phantom pipe in bw calculation for non svp prefetch state + if (state_type != dml2_core_internal_soc_state_svp_prefetch && is_phantom) + exclude_this_plane = true; + + // The qualified row bandwidth, qual_row_bw, accounts for the regular non-flip row bandwidth when there is no possible immediate flip or HostVM invalidation flip. + // The qual_row_bw is zero if HostVM is possible and only non-zero and equal to row_bw(i) if immediate flip is not allowed for that pipe. + if (use_qual_row_bw) { + if (display_cfg->hostvm_enable) + l->per_plane_flip_bw[k] = 0; // qual_row_bw + else if (!display_cfg->plane_descriptors[k].immediate_flip) + l->per_plane_flip_bw[k] = NumberOfDPP[k] * (dpte_row_bw[k] + meta_row_bw[k]); + } else { + // the final_flip_bw includes the regular row_bw when immediate flip is disallowed (and no HostVM) + if ((!display_cfg->plane_descriptors[k].immediate_flip && !display_cfg->hostvm_enable) || !inc_flip_bw) + l->per_plane_flip_bw[k] = NumberOfDPP[k] * (dpte_row_bw[k] + meta_row_bw[k]); + else + l->per_plane_flip_bw[k] = NumberOfDPP[k] * flip_bw[k]; + } + + if (!exclude_this_plane) { + l->vm_row_bw = NumberOfDPP[k] * prefetch_vmrow_bw[k]; + l->flip_and_active_bw = l->per_plane_flip_bw[k] + ReadBandwidthLuma[k] * l->adj_factor_p0 + ReadBandwidthChroma[k] * l->adj_factor_p1 + cursor_bw[k] * l->adj_factor_cur; + l->flip_and_prefetch_bw = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; + l->flip_and_prefetch_bw_max = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthMax[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; + l->active_and_excess_bw = (ReadBandwidthLuma[k] + excess_vactive_fill_bw_l[k]) * l->tmp_nom_adj_factor_p0 + (ReadBandwidthChroma[k] + excess_vactive_fill_bw_c[k]) * l->tmp_nom_adj_factor_p1 + dpte_row_bw[k] + meta_row_bw[k]; + surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_max); + + /* export peak required bandwidth for the surface */ + surface_peak_required_bw[k] = math_max2(surface_required_bw[k], surface_peak_required_bw[k]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, max1: vm_row_bw=%f\n", __func__, k, l->vm_row_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max2: flip_and_active_bw=%f\n", __func__, k, l->flip_and_active_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max3: flip_and_prefetch_bw=%f\n", __func__, k, l->flip_and_prefetch_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, max4: active_and_excess_bw=%f\n", __func__, k, l->active_and_excess_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, surface_required_bw=%f\n", __func__, k, surface_required_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, surface_peak_required_bw=%f\n", __func__, k, surface_peak_required_bw[k]); +#endif + } else { + surface_required_bw[k] = 0.0; + } + + l->required_bandwidth_mbps += surface_required_bw[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, NumberOfDPP=%d\n", __func__, k, NumberOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, use_qual_row_bw=%d\n", __func__, k, use_qual_row_bw); + DML_LOG_VERBOSE("DML::%s: k=%d, immediate_flip=%d\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); + DML_LOG_VERBOSE("DML::%s: k=%d, mall_svp_prefetch_factor=%f\n", __func__, k, l->mall_svp_prefetch_factor); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0=%f\n", __func__, k, l->adj_factor_p0); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1=%f\n", __func__, k, l->adj_factor_p1); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur=%f\n", __func__, k, l->adj_factor_cur); + + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p0_pre=%f\n", __func__, k, l->adj_factor_p0_pre); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_p1_pre=%f\n", __func__, k, l->adj_factor_p1_pre); + DML_LOG_VERBOSE("DML::%s: k=%d, adj_factor_cur_pre=%f\n", __func__, k, l->adj_factor_cur_pre); + + DML_LOG_VERBOSE("DML::%s: k=%d, per_plane_flip_bw=%f\n", __func__, k, l->per_plane_flip_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_vmrow_bw=%f\n", __func__, k, prefetch_vmrow_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthLuma=%f\n", __func__, k, ReadBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, ReadBandwidthChroma=%f\n", __func__, k, ReadBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_l=%f\n", __func__, k, excess_vactive_fill_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, excess_vactive_fill_bw_c=%f\n", __func__, k, excess_vactive_fill_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, cursor_bw=%f\n", __func__, k, cursor_bw[k]); + + DML_LOG_VERBOSE("DML::%s: k=%d, meta_row_bw=%f\n", __func__, k, meta_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, dpte_row_bw=%f\n", __func__, k, dpte_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthLuma=%f\n", __func__, k, PrefetchBandwidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, PrefetchBandwidthChroma=%f\n", __func__, k, PrefetchBandwidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, prefetch_cursor_bw=%f\n", __func__, k, prefetch_cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), soc_state=%s, inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, dml2_core_internal_soc_state_type_str(state_type), inc_flip_bw, is_phantom, exclude_this_plane); + DML_LOG_VERBOSE("DML::%s: k=%d, required_bandwidth_mbps=%f (total), inc_flip_bw=%d, is_phantom=%d exclude_this_plane=%d\n", __func__, k, l->required_bandwidth_mbps, inc_flip_bw, is_phantom, exclude_this_plane); +#endif + } + + return l->required_bandwidth_mbps; +} + +static void CalculateExtraLatency( + const struct dml2_display_cfg *display_cfg, + unsigned int ROBBufferSizeInKByte, + unsigned int RoundTripPingLatencyCycles, + unsigned int ReorderingBytes, + double DCFCLK, + double FabricClock, + unsigned int PixelChunkSizeInKByte, + double ReturnBW, + unsigned int NumberOfActiveSurfaces, + unsigned int NumberOfDPP[], + unsigned int dpte_group_bytes[], + unsigned int tdlut_bytes_per_group[], + double HostVMInefficiencyFactor, + double HostVMInefficiencyFactorPrefetch, + unsigned int HostVMMinPageSize, + enum dml2_qos_param_type qos_type, + bool max_outstanding_when_urgent_expected, + unsigned int max_outstanding_requests, + unsigned int request_size_bytes_luma[], + unsigned int request_size_bytes_chroma[], + unsigned int MetaChunkSize, + unsigned int dchub_arb_to_ret_delay, + double Ttrip, + unsigned int hostvm_mode, + + // output + double *ExtraLatency, // Tex + double *ExtraLatency_sr, // Tex_sr + double *ExtraLatencyPrefetch) + +{ + double Tarb; + double Tarb_prefetch; + double Tex_trips; + unsigned int max_request_size_bytes = 0; + + CalculateTarb( + display_cfg, + PixelChunkSizeInKByte, + NumberOfActiveSurfaces, + NumberOfDPP, + dpte_group_bytes, + tdlut_bytes_per_group, + HostVMInefficiencyFactor, + HostVMInefficiencyFactorPrefetch, + HostVMMinPageSize, + ReturnBW, + MetaChunkSize, + // output + &Tarb, + &Tarb_prefetch); + + Tex_trips = (display_cfg->hostvm_enable && hostvm_mode == 1) ? (2.0 * Ttrip) : 0.0; + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + if (request_size_bytes_luma[k] > max_request_size_bytes) + max_request_size_bytes = request_size_bytes_luma[k]; + if (request_size_bytes_chroma[k] > max_request_size_bytes) + max_request_size_bytes = request_size_bytes_chroma[k]; + } + + if (qos_type == dml2_qos_param_type_dcn4x) { + *ExtraLatency_sr = dchub_arb_to_ret_delay / DCFCLK; + *ExtraLatency = *ExtraLatency_sr; + if (max_outstanding_when_urgent_expected) + *ExtraLatency = *ExtraLatency + (ROBBufferSizeInKByte * 1024 - max_outstanding_requests * max_request_size_bytes) / ReturnBW; + } else { + *ExtraLatency_sr = dchub_arb_to_ret_delay / DCFCLK + RoundTripPingLatencyCycles / FabricClock + ReorderingBytes / ReturnBW; + *ExtraLatency = *ExtraLatency_sr; + } + *ExtraLatency = *ExtraLatency + Tex_trips; + *ExtraLatencyPrefetch = *ExtraLatency + Tarb_prefetch; + *ExtraLatency = *ExtraLatency + Tarb; + *ExtraLatency_sr = *ExtraLatency_sr + Tarb; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: qos_type=%u\n", __func__, qos_type); + DML_LOG_VERBOSE("DML::%s: hostvm_mode=%u\n", __func__, hostvm_mode); + DML_LOG_VERBOSE("DML::%s: Tex_trips=%f\n", __func__, Tex_trips); + DML_LOG_VERBOSE("DML::%s: max_outstanding_when_urgent_expected=%u\n", __func__, max_outstanding_when_urgent_expected); + DML_LOG_VERBOSE("DML::%s: FabricClock=%f\n", __func__, FabricClock); + DML_LOG_VERBOSE("DML::%s: DCFCLK=%f\n", __func__, DCFCLK); + DML_LOG_VERBOSE("DML::%s: ReturnBW=%f\n", __func__, ReturnBW); + DML_LOG_VERBOSE("DML::%s: RoundTripPingLatencyCycles=%u\n", __func__, RoundTripPingLatencyCycles); + DML_LOG_VERBOSE("DML::%s: ReorderingBytes=%u\n", __func__, ReorderingBytes); + DML_LOG_VERBOSE("DML::%s: Tarb=%f\n", __func__, Tarb); + DML_LOG_VERBOSE("DML::%s: ExtraLatency=%f\n", __func__, *ExtraLatency); + DML_LOG_VERBOSE("DML::%s: ExtraLatency_sr=%f\n", __func__, *ExtraLatency_sr); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch=%f\n", __func__, *ExtraLatencyPrefetch); +#endif +} + +static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch, struct dml2_core_calcs_CalculatePrefetchSchedule_params *p) +{ + struct dml2_core_calcs_CalculatePrefetchSchedule_locals *s = &scratch->CalculatePrefetchSchedule_locals; + bool dcc_mrq_enable; + + unsigned int vm_bytes; + unsigned int extra_tdpe_bytes; + unsigned int tdlut_row_bytes; + unsigned int Lo; + + s->NoTimeToPrefetch = false; + s->DPPCycles = 0; + s->DISPCLKCycles = 0; + s->DSTTotalPixelsAfterScaler = 0.0; + s->LineTime = 0.0; + s->dst_y_prefetch_equ = 0.0; + s->prefetch_bw_oto = 0.0; + s->Tvm_oto = 0.0; + s->Tr0_oto = 0.0; + s->Tvm_oto_lines = 0.0; + s->Tr0_oto_lines = 0.0; + s->dst_y_prefetch_oto = 0.0; + s->TimeForFetchingVM = 0.0; + s->TimeForFetchingRowInVBlank = 0.0; + s->LinesToRequestPrefetchPixelData = 0.0; + s->HostVMDynamicLevelsTrips = 0; + s->trip_to_mem = 0.0; + *p->Tvm_trips = 0.0; + *p->Tr0_trips = 0.0; + s->Tvm_trips_rounded = 0.0; + s->Tr0_trips_rounded = 0.0; + s->max_Tsw = 0.0; + s->Lsw_oto = 0.0; + *p->Tpre_rounded = 0.0; + s->prefetch_bw_equ = 0.0; + s->Tvm_equ = 0.0; + s->Tr0_equ = 0.0; + s->Tdmbf = 0.0; + s->Tdmec = 0.0; + s->Tdmsks = 0.0; + *p->prefetch_sw_bytes = 0.0; + s->prefetch_bw_pr = 0.0; + s->bytes_pp = 0.0; + s->dep_bytes = 0.0; + s->min_Lsw_oto = 0.0; + s->min_Lsw_equ = 0.0; + s->Tsw_est1 = 0.0; + s->Tsw_est2 = 0.0; + s->Tsw_est3 = 0.0; + s->cursor_prefetch_bytes = 0; + *p->prefetch_cursor_bw = 0; + *p->RequiredPrefetchBWMax = 0.0; + + dcc_mrq_enable = (p->dcc_enable && p->mrq_present); + + s->TWait_p = p->TWait - p->Ttrip; // TWait includes max(Turg, Ttrip) and Ttrip here is already max(Turg, Ttrip) + + if (p->display_cfg->gpuvm_enable == true && p->display_cfg->hostvm_enable == true) { + s->HostVMDynamicLevelsTrips = p->display_cfg->hostvm_max_non_cached_page_table_levels; + } else { + s->HostVMDynamicLevelsTrips = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: dcc_enable = %u\n", __func__, p->dcc_enable); + DML_LOG_VERBOSE("DML::%s: mrq_present = %u\n", __func__, p->mrq_present); + DML_LOG_VERBOSE("DML::%s: dcc_mrq_enable = %u\n", __func__, dcc_mrq_enable); + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, p->display_cfg->gpuvm_enable); + DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); + DML_LOG_VERBOSE("DML::%s: DCCEnable = %u\n", __func__, p->myPipe->DCCEnable); + DML_LOG_VERBOSE("DML::%s: VStartup = %u\n", __func__, p->VStartup); + DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, p->display_cfg->hostvm_enable); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: myPipe->Dppclk = %f\n", __func__, p->myPipe->Dppclk); + DML_LOG_VERBOSE("DML::%s: myPipe->Dispclk = %f\n", __func__, p->myPipe->Dispclk); +#endif + CalculateVUpdateAndDynamicMetadataParameters( + p->MaxInterDCNTileRepeaters, + p->myPipe->Dppclk, + p->myPipe->Dispclk, + p->myPipe->DCFClkDeepSleep, + p->myPipe->PixelClock, + p->myPipe->HTotal, + p->myPipe->VBlank, + p->DynamicMetadataTransmittedBytes, + p->DynamicMetadataLinesBeforeActiveRequired, + p->myPipe->InterlaceEnable, + p->myPipe->ProgressiveToInterlaceUnitInOPP, + p->TSetup, + + // Output + &s->Tdmbf, + &s->Tdmec, + &s->Tdmsks, + p->VUpdateOffsetPix, + p->VUpdateWidthPix, + p->VReadyOffsetPix); + + s->LineTime = p->myPipe->HTotal / p->myPipe->PixelClock; + s->trip_to_mem = p->Ttrip; + *p->Tvm_trips = p->ExtraLatencyPrefetch + math_max2(s->trip_to_mem * (p->display_cfg->gpuvm_max_page_table_levels * (s->HostVMDynamicLevelsTrips + 1)), p->Turg); + if (dcc_mrq_enable) + *p->Tvm_trips_flip = *p->Tvm_trips; + else + *p->Tvm_trips_flip = *p->Tvm_trips - s->trip_to_mem; + + *p->Tr0_trips_flip = s->trip_to_mem * (s->HostVMDynamicLevelsTrips + 1); + *p->Tr0_trips = math_max2(*p->Tr0_trips_flip, p->tdlut_opt_time / 2); + + if (p->DynamicMetadataVMEnabled == true) { + *p->Tdmdl_vm = s->TWait_p + *p->Tvm_trips; + *p->Tdmdl = *p->Tdmdl_vm + p->Ttrip; + } else { + *p->Tdmdl_vm = 0; + *p->Tdmdl = s->TWait_p + p->ExtraLatencyPrefetch + p->Ttrip; // Tex + } + + if (p->DynamicMetadataEnable == true) { + if (p->VStartup * s->LineTime < *p->TSetup + *p->Tdmdl + s->Tdmbf + s->Tdmec + s->Tdmsks) { + *p->NotEnoughTimeForDynamicMetadata = true; + DML_LOG_VERBOSE("DML::%s: Not Enough Time for Dynamic Meta!\n", __func__); + DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + } else { + *p->NotEnoughTimeForDynamicMetadata = false; + } + } else { + *p->NotEnoughTimeForDynamicMetadata = false; + } + + if (p->myPipe->ScalerEnabled) + s->DPPCycles = (unsigned int)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCL); + else + s->DPPCycles = (unsigned int)(p->DPPCLKDelaySubtotalPlusCNVCFormater + p->DPPCLKDelaySCLLBOnly); + + s->DPPCycles = (unsigned int)(s->DPPCycles + p->myPipe->NumberOfCursors * p->DPPCLKDelayCNVCCursor); + + s->DISPCLKCycles = (unsigned int)p->DISPCLKDelaySubtotal; + + if (p->myPipe->Dppclk == 0.0 || p->myPipe->Dispclk == 0.0) + return true; + + *p->DSTXAfterScaler = (unsigned int)math_round(s->DPPCycles * p->myPipe->PixelClock / p->myPipe->Dppclk + s->DISPCLKCycles * p->myPipe->PixelClock / p->myPipe->Dispclk + p->DSCDelay); + *p->DSTXAfterScaler = (unsigned int)math_round(*p->DSTXAfterScaler + (p->myPipe->ODMMode != dml2_odm_mode_bypass ? 18 : 0) + (p->myPipe->DPPPerSurface - 1) * p->DPP_RECOUT_WIDTH + + ((p->myPipe->ODMMode == dml2_odm_mode_split_1to2 || p->myPipe->ODMMode == dml2_odm_mode_mso_1to2) ? (double)p->myPipe->HActive / 2.0 : 0) + + ((p->myPipe->ODMMode == dml2_odm_mode_mso_1to4) ? (double)p->myPipe->HActive * 3.0 / 4.0 : 0)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DynamicMetadataVMEnabled = %u\n", __func__, p->DynamicMetadataVMEnabled); + DML_LOG_VERBOSE("DML::%s: DPPCycles = %u\n", __func__, s->DPPCycles); + DML_LOG_VERBOSE("DML::%s: PixelClock = %f\n", __func__, p->myPipe->PixelClock); + DML_LOG_VERBOSE("DML::%s: Dppclk = %f\n", __func__, p->myPipe->Dppclk); + DML_LOG_VERBOSE("DML::%s: DISPCLKCycles = %u\n", __func__, s->DISPCLKCycles); + DML_LOG_VERBOSE("DML::%s: DISPCLK = %f\n", __func__, p->myPipe->Dispclk); + DML_LOG_VERBOSE("DML::%s: DSCDelay = %u\n", __func__, p->DSCDelay); + DML_LOG_VERBOSE("DML::%s: ODMMode = %u\n", __func__, p->myPipe->ODMMode); + DML_LOG_VERBOSE("DML::%s: DPP_RECOUT_WIDTH = %u\n", __func__, p->DPP_RECOUT_WIDTH); + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u\n", __func__, *p->DSTXAfterScaler); + + DML_LOG_VERBOSE("DML::%s: setup_for_tdlut = %u\n", __func__, p->setup_for_tdlut); + DML_LOG_VERBOSE("DML::%s: tdlut_opt_time = %f\n", __func__, p->tdlut_opt_time); + DML_LOG_VERBOSE("DML::%s: tdlut_pte_bytes_per_frame = %u\n", __func__, p->tdlut_pte_bytes_per_frame); + DML_LOG_VERBOSE("DML::%s: tdlut_drain_time = %f\n", __func__, p->tdlut_drain_time); +#endif + + if (p->OutputFormat == dml2_420 || (p->myPipe->InterlaceEnable && p->myPipe->ProgressiveToInterlaceUnitInOPP)) + *p->DSTYAfterScaler = 1; + else + *p->DSTYAfterScaler = 0; + + s->DSTTotalPixelsAfterScaler = *p->DSTYAfterScaler * p->myPipe->HTotal + *p->DSTXAfterScaler; + *p->DSTYAfterScaler = (unsigned int)(math_floor2(s->DSTTotalPixelsAfterScaler / p->myPipe->HTotal, 1)); + *p->DSTXAfterScaler = (unsigned int)(s->DSTTotalPixelsAfterScaler - ((double)(*p->DSTYAfterScaler * p->myPipe->HTotal))); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler = %u (final)\n", __func__, *p->DSTXAfterScaler); + DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler = %u (final)\n", __func__, *p->DSTYAfterScaler); +#endif + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: GPUVMPageTableLevels = %u\n", __func__, p->display_cfg->gpuvm_max_page_table_levels); + DML_LOG_VERBOSE("DML::%s: HostVMDynamicLevelsTrips = %u\n", __func__, s->HostVMDynamicLevelsTrips); +#endif + if (p->display_cfg->gpuvm_enable) { + s->Tvm_trips_rounded = math_ceil2(4.0 * *p->Tvm_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; + *p->Tvm_trips_flip_rounded = math_ceil2(4.0 * *p->Tvm_trips_flip / s->LineTime, 1.0) / 4.0 * s->LineTime; + } else { + if (p->DynamicMetadataEnable || dcc_mrq_enable || p->setup_for_tdlut) + s->Tvm_trips_rounded = math_max2(s->LineTime * math_ceil2(4.0*math_max3(p->ExtraLatencyPrefetch, p->Turg, s->trip_to_mem)/s->LineTime, 1)/4, s->LineTime/4.0); + else + s->Tvm_trips_rounded = s->LineTime / 4.0; + *p->Tvm_trips_flip_rounded = s->LineTime / 4.0; + } + + s->Tvm_trips_rounded = math_max2(s->Tvm_trips_rounded, s->LineTime / 4.0); + *p->Tvm_trips_flip_rounded = math_max2(*p->Tvm_trips_flip_rounded, s->LineTime / 4.0); + + if (p->display_cfg->gpuvm_enable == true || p->setup_for_tdlut || dcc_mrq_enable) { + s->Tr0_trips_rounded = math_ceil2(4.0 * *p->Tr0_trips / s->LineTime, 1.0) / 4.0 * s->LineTime; + *p->Tr0_trips_flip_rounded = math_ceil2(4.0 * *p->Tr0_trips_flip / s->LineTime, 1.0) / 4.0 * s->LineTime; + } else { + s->Tr0_trips_rounded = s->LineTime / 4.0; + *p->Tr0_trips_flip_rounded = s->LineTime / 4.0; + } + s->Tr0_trips_rounded = math_max2(s->Tr0_trips_rounded, s->LineTime / 4.0); + *p->Tr0_trips_flip_rounded = math_max2(*p->Tr0_trips_flip_rounded, s->LineTime / 4.0); + + if (p->display_cfg->gpuvm_enable == true) { + if (p->display_cfg->gpuvm_max_page_table_levels >= 3) { + *p->Tno_bw = p->ExtraLatencyPrefetch + s->trip_to_mem * (double)((p->display_cfg->gpuvm_max_page_table_levels - 2) * (s->HostVMDynamicLevelsTrips + 1)); + } else if (p->display_cfg->gpuvm_max_page_table_levels == 1 && !dcc_mrq_enable && !p->setup_for_tdlut) { + *p->Tno_bw = p->ExtraLatencyPrefetch; + } else { + *p->Tno_bw = 0; + } + } else { + *p->Tno_bw = 0; + } + + if (p->mrq_present || p->display_cfg->gpuvm_max_page_table_levels >= 3) + *p->Tno_bw_flip = *p->Tno_bw; + else + *p->Tno_bw_flip = 0; //because there is no 3DLUT for iFlip + + if (dml_is_420(p->myPipe->SourcePixelFormat)) { + s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC / 4.0; + } else { + s->bytes_pp = p->myPipe->BytePerPixelY + p->myPipe->BytePerPixelC; + } + + *p->prefetch_sw_bytes = p->PrefetchSourceLinesY * p->swath_width_luma_ub * p->myPipe->BytePerPixelY + p->PrefetchSourceLinesC * p->swath_width_chroma_ub * p->myPipe->BytePerPixelC; + *p->prefetch_sw_bytes = *p->prefetch_sw_bytes * p->mall_prefetch_sdp_overhead_factor; + + vm_bytes = p->vm_bytes; // vm_bytes is dpde0_bytes_per_frame_ub_l + dpde0_bytes_per_frame_ub_c + 2*extra_dpde_bytes; + extra_tdpe_bytes = (unsigned int)math_max2(0, (p->display_cfg->gpuvm_max_page_table_levels - 1) * 128); + + if (p->setup_for_tdlut) + vm_bytes = vm_bytes + p->tdlut_pte_bytes_per_frame + (p->display_cfg->gpuvm_enable ? extra_tdpe_bytes : 0); + + tdlut_row_bytes = (unsigned long) math_ceil2(p->tdlut_bytes_per_frame/2.0, 1.0); + + s->min_Lsw_oto = math_max2(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML2_CALCS_MAX_VRATIO_PRE_OTO__; + s->min_Lsw_oto = math_max2(s->min_Lsw_oto, p->tdlut_drain_time / s->LineTime); + s->min_Lsw_oto = math_max2(s->min_Lsw_oto, 2.0); + + // use vactive swath bw for prefetch oto and also cap prefetch_bw_oto to max_vratio_oto + // Note: in prefetch calculation, acounting is done mostly per-pipe. + // vactive swath bw represents the per-surface (aka per dml plane) bw to move vratio_l/c lines of bytes_l/c per line time + s->per_pipe_vactive_sw_bw = p->vactive_sw_bw_l / (double)p->myPipe->DPPPerSurface; + + // one-to-one prefetch bw as one line of bytes per line time (as per vratio_pre_l/c = 1) + s->prefetch_bw_oto = (p->swath_width_luma_ub * p->myPipe->BytePerPixelY) / s->LineTime; + + if (p->myPipe->BytePerPixelC > 0) { + s->per_pipe_vactive_sw_bw += p->vactive_sw_bw_c / (double)p->myPipe->DPPPerSurface; + s->prefetch_bw_oto += (p->swath_width_chroma_ub * p->myPipe->BytePerPixelC) / s->LineTime; + } + + /* oto prefetch bw should be always be less than total vactive bw */ + //DML_ASSERT(s->prefetch_bw_oto < s->per_pipe_vactive_sw_bw * p->myPipe->DPPPerSurface); + + s->prefetch_bw_oto = math_max2(s->per_pipe_vactive_sw_bw, s->prefetch_bw_oto) * p->mall_prefetch_sdp_overhead_factor; + + s->prefetch_bw_oto = math_min2(s->prefetch_bw_oto, *p->prefetch_sw_bytes/(s->min_Lsw_oto*s->LineTime)); + + s->Lsw_oto = math_ceil2(4.0 * *p->prefetch_sw_bytes / s->prefetch_bw_oto / s->LineTime, 1.0) / 4.0; + + s->prefetch_bw_oto = math_max3(s->prefetch_bw_oto, + p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, + (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); + + /* oto bw needs to be outputted even if the oto schedule isn't being used to avoid ms/mp mismatch. + * mp will fail if ms decides to use equ schedule and mp decides to use oto schedule + * and the required bandwidth increases when going from ms to mp + */ + *p->RequiredPrefetchBWMax = s->prefetch_bw_oto; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c = %f\n", __func__, p->vactive_sw_bw_c); + DML_LOG_VERBOSE("DML::%s: per_pipe_vactive_sw_bw = %f\n", __func__, s->per_pipe_vactive_sw_bw); +#endif + + if (p->display_cfg->gpuvm_enable == true) { + s->Tvm_oto = math_max3( + *p->Tvm_trips, + *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto, + s->LineTime / 4.0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Tvm_oto max0 = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_oto max1 = %f\n", __func__, *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto max2 = %f\n", __func__, s->LineTime / 4.0); +#endif + } else { + s->Tvm_oto = s->Tvm_trips_rounded; + } + + if ((p->display_cfg->gpuvm_enable == true || p->setup_for_tdlut || dcc_mrq_enable)) { + s->Tr0_oto = math_max3( + *p->Tr0_trips, + (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto, + s->LineTime / 4.0); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Tr0_oto max0 = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tr0_oto max1 = %f\n", __func__, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tr0_oto max2 = %f\n", __func__, s->LineTime / 4); +#endif + } else + s->Tr0_oto = s->LineTime / 4.0; + + s->Tvm_oto_lines = math_ceil2(4.0 * s->Tvm_oto / s->LineTime, 1) / 4.0; + s->Tr0_oto_lines = math_ceil2(4.0 * s->Tr0_oto / s->LineTime, 1) / 4.0; + s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto; + +#ifdef DML_GLOBAL_PREFETCH_CHECK + DML_LOG_VERBOSE("DML::%s: impacted_Tpre = %f\n", __func__, p->impacted_dst_y_pre); + if (p->impacted_dst_y_pre > 0) { + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); + s->dst_y_prefetch_oto = math_max2(s->dst_y_prefetch_oto, p->impacted_dst_y_pre); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f (impacted)\n", __func__, s->dst_y_prefetch_oto); + } +#endif + *p->Tpre_oto = s->dst_y_prefetch_oto * s->LineTime; + + //To (time for delay after scaler) in line time + Lo = (unsigned int)(*p->DSTYAfterScaler + (double)*p->DSTXAfterScaler / (double)p->myPipe->HTotal); + + s->min_Lsw_equ = math_max2(p->PrefetchSourceLinesY, p->PrefetchSourceLinesC) / __DML2_CALCS_MAX_VRATIO_PRE_EQU__; + s->min_Lsw_equ = math_max2(s->min_Lsw_equ, p->tdlut_drain_time / s->LineTime); + s->min_Lsw_equ = math_max2(s->min_Lsw_equ, 2.0); + //Tpre_equ in line time + if (p->DynamicMetadataVMEnabled && p->DynamicMetadataEnable) + s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + math_max2(p->TCalc, *p->Tvm_trips) + s->TWait_p) / s->LineTime - Lo; + else + s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + math_max2(p->TCalc, p->ExtraLatencyPrefetch) + s->TWait_p) / s->LineTime - Lo; + +#ifdef DML_GLOBAL_PREFETCH_CHECK + s->dst_y_prefetch_equ_impacted = math_max2(p->impacted_dst_y_pre, s->dst_y_prefetch_equ); + + s->dst_y_prefetch_equ_impacted = math_min2(s->dst_y_prefetch_equ_impacted, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH + + if (s->dst_y_prefetch_equ_impacted > s->dst_y_prefetch_equ) + s->dst_y_prefetch_equ -= s->dst_y_prefetch_equ_impacted - s->dst_y_prefetch_equ; +#endif + + s->dst_y_prefetch_equ = math_min2(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal); + DML_LOG_VERBOSE("DML::%s: min_Lsw_oto = %f\n", __func__, s->min_Lsw_oto); + DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); + DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, *p->Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: ExtraLatencyPrefetch = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: trip_to_mem = %f\n", __func__, s->trip_to_mem); + DML_LOG_VERBOSE("DML::%s: mall_prefetch_sdp_overhead_factor = %f\n", __func__, p->mall_prefetch_sdp_overhead_factor); + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: BytePerPixelC = %u\n", __func__, p->myPipe->BytePerPixelC); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + DML_LOG_VERBOSE("DML::%s: swath_width_chroma_ub = %u\n", __func__, p->swath_width_chroma_ub); + DML_LOG_VERBOSE("DML::%s: prefetch_sw_bytes = %f\n", __func__, *p->prefetch_sw_bytes); + DML_LOG_VERBOSE("DML::%s: max_Tsw = %f\n", __func__, s->max_Tsw); + DML_LOG_VERBOSE("DML::%s: bytes_pp = %f\n", __func__, s->bytes_pp); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: Tvm_trips = %f\n", __func__, *p->Tvm_trips); + DML_LOG_VERBOSE("DML::%s: Tr0_trips = %f\n", __func__, *p->Tr0_trips); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, *p->Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, *p->Tr0_trips_flip); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_pr = %f\n", __func__, s->prefetch_bw_pr); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_oto = %f\n", __func__, s->prefetch_bw_oto); + DML_LOG_VERBOSE("DML::%s: Tr0_oto = %f\n", __func__, s->Tr0_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto = %f\n", __func__, s->Tvm_oto); + DML_LOG_VERBOSE("DML::%s: Tvm_oto_lines = %f\n", __func__, s->Tvm_oto_lines); + DML_LOG_VERBOSE("DML::%s: Tr0_oto_lines = %f\n", __func__, s->Tr0_oto_lines); + DML_LOG_VERBOSE("DML::%s: Lsw_oto = %f\n", __func__, s->Lsw_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_oto = %f\n", __func__, s->dst_y_prefetch_oto); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ = %f\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: tdlut_row_bytes = %d\n", __func__, tdlut_row_bytes); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, p->meta_row_bytes); +#endif + s->dst_y_prefetch_equ = math_floor2(4.0 * (s->dst_y_prefetch_equ + 0.125), 1) / 4.0; + *p->Tpre_rounded = s->dst_y_prefetch_equ * s->LineTime; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch_equ: %f (after round)\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: LineTime: %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: VStartup: %u\n", __func__, p->VStartup); + DML_LOG_VERBOSE("DML::%s: Tvstartup: %fus - time between vstartup and first pixel of active\n", __func__, p->VStartup * s->LineTime); + DML_LOG_VERBOSE("DML::%s: TSetup: %fus - time from vstartup to vready\n", __func__, *p->TSetup); + DML_LOG_VERBOSE("DML::%s: TCalc: %fus - time for calculations in dchub starting at vready\n", __func__, p->TCalc); + DML_LOG_VERBOSE("DML::%s: TWait: %fus - time for fabric to become ready max(pstate exit,cstate enter/exit, urgent latency) after TCalc\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: Tdmbf: %fus - time for dmd transfer from dchub to dio output buffer\n", __func__, s->Tdmbf); + DML_LOG_VERBOSE("DML::%s: Tdmec: %fus - time dio takes to transfer dmd\n", __func__, s->Tdmec); + DML_LOG_VERBOSE("DML::%s: Tdmsks: %fus - time before active dmd must complete transmission at dio\n", __func__, s->Tdmsks); + DML_LOG_VERBOSE("DML::%s: TWait = %f\n", __func__, p->TWait); + DML_LOG_VERBOSE("DML::%s: TWait_p = %f\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip = %f\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: Tex = %f\n", __func__, p->ExtraLatencyPrefetch); + DML_LOG_VERBOSE("DML::%s: Tdmdl_vm: %fus - time for vm stages of dmd \n", __func__, *p->Tdmdl_vm); + DML_LOG_VERBOSE("DML::%s: Tdmdl: %fus - time for fabric to become ready and fetch dmd \n", __func__, *p->Tdmdl); + DML_LOG_VERBOSE("DML::%s: TWait_p: %fus\n", __func__, s->TWait_p); + DML_LOG_VERBOSE("DML::%s: Ttrip: %fus\n", __func__, p->Ttrip); + DML_LOG_VERBOSE("DML::%s: DSTXAfterScaler: %u pixels - number of pixel clocks pipeline and buffer delay after scaler \n", __func__, *p->DSTXAfterScaler); + DML_LOG_VERBOSE("DML::%s: DSTYAfterScaler: %u lines - number of lines of pipeline and buffer delay after scaler \n", __func__, *p->DSTYAfterScaler); + DML_LOG_VERBOSE("DML::%s: vm_bytes: %f (hvm inefficiency scaled)\n", __func__, vm_bytes*p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: row_bytes: %f (hvm inefficiency scaled, 1 row)\n", __func__, p->PixelPTEBytesPerRow*p->HostVMInefficiencyFactor+p->meta_row_bytes+tdlut_row_bytes); + DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, (s->dst_y_prefetch_equ * s->LineTime), *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); + DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); +#endif + + *p->dst_y_per_vm_vblank = 0; + *p->dst_y_per_row_vblank = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixelDataBWLuma = 0; + + // Derive bandwidth by finding how much data to move within the time constraint + // Tpre_rounded is Tpre rounding to 2-bit fraction + // Tvm_trips_rounded is Tvm_trips ceiling to 1/4 line time + // Tr0_trips_rounded is Tr0_trips ceiling to 1/4 line time + // So that means prefetch bw calculated can be higher since the total time available for prefetch is less + bool min_Lsw_equ_ok = *p->Tpre_rounded >= s->Tvm_trips_rounded + 2.0*s->Tr0_trips_rounded + s->min_Lsw_equ*s->LineTime; + bool tpre_gt_req_latency = true; +#if 0 + // Check that Tpre_rounded is big enough if all of the stages of the prefetch are time constrained. + // The terms Tvm_trips_rounded and Tr0_trips_rounded represent the min time constraints for the VM and row stages. + // Normally, these terms cover the overall time constraint for Tpre >= (Tex + max{Ttrip, Turg}), but if these terms are at their minimum, an explicit check is necessary. + tpre_gt_req_latency = *p->Tpre_rounded > (math_max2(p->Turg, s->trip_to_mem) + p->ExtraLatencyPrefetch); +#endif + + if (s->dst_y_prefetch_equ > 1 && min_Lsw_equ_ok && tpre_gt_req_latency) { + s->prefetch_bw1 = 0.; + s->prefetch_bw2 = 0.; + s->prefetch_bw3 = 0.; + s->prefetch_bw4 = 0.; + + // prefetch_bw1: VM + 2*R0 + SW + if (*p->Tpre_rounded - *p->Tno_bw > 0) { + s->prefetch_bw1 = (vm_bytes * p->HostVMInefficiencyFactor + + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) + + *p->prefetch_sw_bytes) + / (*p->Tpre_rounded - *p->Tno_bw); + s->Tsw_est1 = *p->prefetch_sw_bytes / s->prefetch_bw1; + } else + s->prefetch_bw1 = 0; + + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f\n", __func__, s->prefetch_bw1); + if ((s->Tsw_est1 < s->min_Lsw_equ * s->LineTime) && (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw > 0)) { + s->prefetch_bw1 = (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / + (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: vm and 2 rows bytes = %f\n", __func__, (vm_bytes * p->HostVMInefficiencyFactor + 2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes))); + DML_LOG_VERBOSE("DML::%s: Tpre_rounded = %f\n", __func__, *p->Tpre_rounded); + DML_LOG_VERBOSE("DML::%s: minus term = %f\n", __func__, s->min_Lsw_equ * s->LineTime + 0.75 * s->LineTime + *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: min_Lsw_equ = %f\n", __func__, s->min_Lsw_equ); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: Tno_bw = %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Time to fetch vm and 2 rows = %f\n", __func__, (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.75 * s->LineTime - *p->Tno_bw)); + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (updated)\n", __func__, s->prefetch_bw1); +#endif + } + + // prefetch_bw2: VM + SW + if (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded > 0) { + s->prefetch_bw2 = (vm_bytes * p->HostVMInefficiencyFactor + *p->prefetch_sw_bytes) / + (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded); + s->Tsw_est2 = *p->prefetch_sw_bytes / s->prefetch_bw2; + } else + s->prefetch_bw2 = 0; + + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f\n", __func__, s->prefetch_bw2); + if ((s->Tsw_est2 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime) > 0)) { + s->prefetch_bw2 = vm_bytes * p->HostVMInefficiencyFactor / (*p->Tpre_rounded - *p->Tno_bw - 2.0 * s->Tr0_trips_rounded - s->min_Lsw_equ * s->LineTime - 0.25 * s->LineTime); + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (updated)\n", __func__, s->prefetch_bw2); + } + + // prefetch_bw3: 2*R0 + SW + if (*p->Tpre_rounded - s->Tvm_trips_rounded > 0) { + s->prefetch_bw3 = (2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) + *p->prefetch_sw_bytes) / + (*p->Tpre_rounded - s->Tvm_trips_rounded); + s->Tsw_est3 = *p->prefetch_sw_bytes / s->prefetch_bw3; + } else + s->prefetch_bw3 = 0; + + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f\n", __func__, s->prefetch_bw3); + if ((s->Tsw_est3 < s->min_Lsw_equ * s->LineTime) && ((*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded) > 0)) { + s->prefetch_bw3 = (2 * (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes)) / (*p->Tpre_rounded - s->min_Lsw_equ * s->LineTime - 0.5 * s->LineTime - s->Tvm_trips_rounded); + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (updated)\n", __func__, s->prefetch_bw3); + } + + // prefetch_bw4: SW + if (*p->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded > 0) + s->prefetch_bw4 = *p->prefetch_sw_bytes / (*p->Tpre_rounded - s->Tvm_trips_rounded - 2 * s->Tr0_trips_rounded); + else + s->prefetch_bw4 = 0; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Tno_bw: %f\n", __func__, *p->Tno_bw); + DML_LOG_VERBOSE("DML::%s: Tpre=%f Tpre_rounded: %f, delta=%f\n", __func__, s->dst_y_prefetch_equ * s->LineTime, *p->Tpre_rounded, (*p->Tpre_rounded - (s->dst_y_prefetch_equ * s->LineTime))); + DML_LOG_VERBOSE("DML::%s: Tvm_trips=%f Tvm_trips_rounded: %f, delta=%f\n", __func__, *p->Tvm_trips, s->Tvm_trips_rounded, (s->Tvm_trips_rounded - *p->Tvm_trips)); + DML_LOG_VERBOSE("DML::%s: Tr0_trips=%f Tr0_trips_rounded: %f, delta=%f\n", __func__, *p->Tr0_trips, s->Tr0_trips_rounded, (s->Tr0_trips_rounded - *p->Tr0_trips)); + DML_LOG_VERBOSE("DML::%s: Tsw_est1: %f\n", __func__, s->Tsw_est1); + DML_LOG_VERBOSE("DML::%s: Tsw_est2: %f\n", __func__, s->Tsw_est2); + DML_LOG_VERBOSE("DML::%s: Tsw_est3: %f\n", __func__, s->Tsw_est3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw1: %f (final)\n", __func__, s->prefetch_bw1); + DML_LOG_VERBOSE("DML::%s: prefetch_bw2: %f (final)\n", __func__, s->prefetch_bw2); + DML_LOG_VERBOSE("DML::%s: prefetch_bw3: %f (final)\n", __func__, s->prefetch_bw3); + DML_LOG_VERBOSE("DML::%s: prefetch_bw4: %f (final)\n", __func__, s->prefetch_bw4); +#endif + { + bool Case1OK = false; + bool Case2OK = false; + bool Case3OK = false; + + // get "equalized" bw among all stages (vm, r0, sw), so based is all 3 stages are just above the latency-based requirement + // so it is not too dis-portionally favor a particular stage, next is either r0 more agressive and next is vm more agressive, the worst is all are agressive + // vs the latency based number + + // prefetch_bw1: VM + 2*R0 + SW + // so prefetch_bw1 will have enough bw to transfer the necessary data within Tpre_rounded - Tno_bw (Tpre is the the worst-case latency based time to fetch the data) + // here is to make sure equ bw wont be more agressive than the latency-based requirement. + // check vm time >= vm_trips + // check r0 time >= r0_trips + + double total_row_bytes = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes); + + DML_LOG_VERBOSE("DML::%s: Tvm_trips_rounded = %f\n", __func__, s->Tvm_trips_rounded); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_rounded = %f\n", __func__, s->Tr0_trips_rounded); + + if (s->prefetch_bw1 > 0) { + double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw1; + double row_transfer_time = total_row_bytes / s->prefetch_bw1; + DML_LOG_VERBOSE("DML::%s: Case1: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case1: row_transfer_time = %f\n", __func__, row_transfer_time); + if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { + Case1OK = true; + } + } + + // prefetch_bw2: VM + SW + // prefetch_bw2 will be enough bw to transfer VM and SW data within (Tpre_rounded - Tr0_trips_rounded - Tno_bw) + // check vm time >= vm_trips + // check r0 time < r0_trips + if (s->prefetch_bw2 > 0) { + double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw2; + double row_transfer_time = total_row_bytes / s->prefetch_bw2; + DML_LOG_VERBOSE("DML::%s: Case2: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case2: row_transfer_time = %f\n", __func__, row_transfer_time); + if (vm_transfer_time >= s->Tvm_trips_rounded && row_transfer_time < s->Tr0_trips_rounded) { + Case2OK = true; + } + } + + // prefetch_bw3: VM + 2*R0 + // check vm time < vm_trips + // check r0 time >= r0_trips + if (s->prefetch_bw3 > 0) { + double vm_transfer_time = *p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw3; + double row_transfer_time = total_row_bytes / s->prefetch_bw3; + DML_LOG_VERBOSE("DML::%s: Case3: vm_transfer_time = %f\n", __func__, vm_transfer_time); + DML_LOG_VERBOSE("DML::%s: Case3: row_transfer_time = %f\n", __func__, row_transfer_time); + if (vm_transfer_time < s->Tvm_trips_rounded && row_transfer_time >= s->Tr0_trips_rounded) { + Case3OK = true; + } + } + + if (Case1OK) { + s->prefetch_bw_equ = s->prefetch_bw1; + } else if (Case2OK) { + s->prefetch_bw_equ = s->prefetch_bw2; + } else if (Case3OK) { + s->prefetch_bw_equ = s->prefetch_bw3; + } else { + s->prefetch_bw_equ = s->prefetch_bw4; + } + + s->prefetch_bw_equ = math_max3(s->prefetch_bw_equ, + p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, + (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Case1OK: %u\n", __func__, Case1OK); + DML_LOG_VERBOSE("DML::%s: Case2OK: %u\n", __func__, Case2OK); + DML_LOG_VERBOSE("DML::%s: Case3OK: %u\n", __func__, Case3OK); + DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ: %f\n", __func__, s->prefetch_bw_equ); +#endif + + if (s->prefetch_bw_equ > 0) { + if (p->display_cfg->gpuvm_enable == true) { + s->Tvm_equ = math_max3(*p->Tno_bw + vm_bytes * p->HostVMInefficiencyFactor / s->prefetch_bw_equ, *p->Tvm_trips, s->LineTime / 4); + } else { + s->Tvm_equ = s->LineTime / 4; + } + + if (p->display_cfg->gpuvm_enable == true || dcc_mrq_enable || p->setup_for_tdlut) { + s->Tr0_equ = math_max3((p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / s->prefetch_bw_equ, // PixelPTEBytesPerRow is dpte_row_bytes + *p->Tr0_trips, + s->LineTime / 4); + } else { + s->Tr0_equ = s->LineTime / 4; + } + } else { + s->Tvm_equ = 0; + s->Tr0_equ = 0; + DML_LOG_VERBOSE("DML::%s: prefetch_bw_equ equals 0!\n", __func__); + } + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Tvm_equ = %f\n", __func__, s->Tvm_equ); + DML_LOG_VERBOSE("DML::%s: Tr0_equ = %f\n", __func__, s->Tr0_equ); +#endif + // Use the more stressful prefetch schedule + if (s->dst_y_prefetch_oto < s->dst_y_prefetch_equ) { + *p->dst_y_prefetch = s->dst_y_prefetch_oto; + s->TimeForFetchingVM = s->Tvm_oto; + s->TimeForFetchingRowInVBlank = s->Tr0_oto; + + *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Using oto scheduling for prefetch\n", __func__); +#endif + + } else { + *p->dst_y_prefetch = s->dst_y_prefetch_equ; + + if (s->dst_y_prefetch_equ < s->dst_y_prefetch_equ_impacted) + *p->dst_y_prefetch = s->dst_y_prefetch_equ_impacted; + + s->TimeForFetchingVM = s->Tvm_equ; + s->TimeForFetchingRowInVBlank = s->Tr0_equ; + + *p->dst_y_per_vm_vblank = math_ceil2(4.0 * s->TimeForFetchingVM / s->LineTime, 1.0) / 4.0; + *p->dst_y_per_row_vblank = math_ceil2(4.0 * s->TimeForFetchingRowInVBlank / s->LineTime, 1.0) / 4.0; + + /* equ bw should be propagated so a ceiling of the equ bw is accounted for prior to mode programming. + * Overall bandwidth may be lower when going from mode support to mode programming but final pixel data + * bandwidth may end up higher than what was calculated in mode support. + */ + *p->RequiredPrefetchBWMax = math_max2(s->prefetch_bw_equ, *p->RequiredPrefetchBWMax); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Using equ bw scheduling for prefetch\n", __func__); +#endif + } + + // Lsw = dst_y_prefetch - (dst_y_per_vm_vblank + 2*dst_y_per_row_vblank) + s->LinesToRequestPrefetchPixelData = *p->dst_y_prefetch - *p->dst_y_per_vm_vblank - 2 * *p->dst_y_per_row_vblank; // Lsw + + s->cursor_prefetch_bytes = (unsigned int)math_max2(p->cursor_bytes_per_chunk, 4 * p->cursor_bytes_per_line); + *p->prefetch_cursor_bw = p->num_cursors * s->cursor_prefetch_bytes / (s->LinesToRequestPrefetchPixelData * s->LineTime); + *p->prefetch_swath_time_us = (s->LinesToRequestPrefetchPixelData * s->LineTime); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: TimeForFetchingVM = %f\n", __func__, s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML::%s: TimeForFetchingRowInVBlank = %f\n", __func__, s->TimeForFetchingRowInVBlank); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: dst_y_prefetch = %f\n", __func__, *p->dst_y_prefetch); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: LinesToRequestPrefetchPixelData = %f\n", __func__, s->LinesToRequestPrefetchPixelData); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: prefetch_swath_time_us = %f\n", __func__, *p->prefetch_swath_time_us); + + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_chunk = %d\n", __func__, p->cursor_bytes_per_chunk); + DML_LOG_VERBOSE("DML::%s: cursor_bytes_per_line = %d\n", __func__, p->cursor_bytes_per_line); + DML_LOG_VERBOSE("DML::%s: cursor_prefetch_bytes = %d\n", __func__, s->cursor_prefetch_bytes); + DML_LOG_VERBOSE("DML::%s: prefetch_cursor_bw = %f\n", __func__, *p->prefetch_cursor_bw); +#endif + DML_ASSERT(*p->dst_y_prefetch < 64); + + unsigned int min_lsw_required = (unsigned int)math_max2(2, p->tdlut_drain_time / s->LineTime); + if (s->LinesToRequestPrefetchPixelData >= min_lsw_required && s->prefetch_bw_equ > 0) { + *p->VRatioPrefetchY = (double)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData; + *p->VRatioPrefetchY = math_max2(*p->VRatioPrefetchY, 1.0); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + DML_LOG_VERBOSE("DML::%s: SwathHeightY = %u\n", __func__, p->SwathHeightY); + DML_LOG_VERBOSE("DML::%s: VInitPreFillY = %u\n", __func__, p->VInitPreFillY); +#endif + if ((p->SwathHeightY > 4) && (p->VInitPreFillY > 3)) { + if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillY - 3.0) / 2.0) { + *p->VRatioPrefetchY = math_max2(*p->VRatioPrefetchY, + (double)p->MaxNumSwathY * p->SwathHeightY / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillY - 3.0) / 2.0)); + } else { + s->NoTimeToPrefetch = true; + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VinitPreFillY=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillY); + *p->VRatioPrefetchY = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchY = %f\n", __func__, *p->VRatioPrefetchY); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesY = %f\n", __func__, p->PrefetchSourceLinesY); + DML_LOG_VERBOSE("DML::%s: MaxNumSwathY = %u\n", __func__, p->MaxNumSwathY); +#endif + } + + *p->VRatioPrefetchC = (double)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData; + *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, 1.0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + DML_LOG_VERBOSE("DML::%s: SwathHeightC = %u\n", __func__, p->SwathHeightC); + DML_LOG_VERBOSE("DML::%s: VInitPreFillC = %u\n", __func__, p->VInitPreFillC); +#endif + if ((p->SwathHeightC > 4) && (p->VInitPreFillC > 3)) { + if (s->LinesToRequestPrefetchPixelData > (p->VInitPreFillC - 3.0) / 2.0) { + *p->VRatioPrefetchC = math_max2(*p->VRatioPrefetchC, (double)p->MaxNumSwathC * p->SwathHeightC / (s->LinesToRequestPrefetchPixelData - (p->VInitPreFillC - 3.0) / 2.0)); + } else { + s->NoTimeToPrefetch = true; + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. LinesToRequestPrefetchPixelData=%f VInitPreFillC=%u\n", __func__, s->LinesToRequestPrefetchPixelData, p->VInitPreFillC); + *p->VRatioPrefetchC = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchC = %f\n", __func__, *p->VRatioPrefetchC); + DML_LOG_VERBOSE("DML::%s: PrefetchSourceLinesC = %f\n", __func__, p->PrefetchSourceLinesC); + DML_LOG_VERBOSE("DML::%s: MaxNumSwathC = %u\n", __func__, p->MaxNumSwathC); +#endif + } + + *p->RequiredPrefetchPixelDataBWLuma = (double)p->PrefetchSourceLinesY / s->LinesToRequestPrefetchPixelData * p->myPipe->BytePerPixelY * p->swath_width_luma_ub / s->LineTime; + *p->RequiredPrefetchPixelDataBWChroma = (double)p->PrefetchSourceLinesC / s->LinesToRequestPrefetchPixelData * p->myPipe->BytePerPixelC * p->swath_width_chroma_ub / s->LineTime; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: BytePerPixelY = %u\n", __func__, p->myPipe->BytePerPixelY); + DML_LOG_VERBOSE("DML::%s: swath_width_luma_ub = %u\n", __func__, p->swath_width_luma_ub); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); +#endif + } else { + s->NoTimeToPrefetch = true; + DML_LOG_VERBOSE("DML::%s: No time to prefetch!, LinesToRequestPrefetchPixelData: %f, should be >= %d\n", __func__, s->LinesToRequestPrefetchPixelData, min_lsw_required); + DML_LOG_VERBOSE("DML::%s: No time to prefetch!, prefetch_bw_equ: %f, should be > 0\n", __func__, s->prefetch_bw_equ); + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixelDataBWLuma = 0; + *p->RequiredPrefetchPixelDataBWChroma = 0; + } + DML_LOG_VERBOSE("DML: Tpre: %fus - sum of time to request 2 x data pte, swaths\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime + 2.0 * s->TimeForFetchingRowInVBlank + s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML: Tvm: %fus - time to fetch vm\n", s->TimeForFetchingVM); + DML_LOG_VERBOSE("DML: Tr0: %fus - time to fetch first row of data pagetables\n", s->TimeForFetchingRowInVBlank); + DML_LOG_VERBOSE("DML: Tsw: %fus = time to fetch enough pixel data and cursor data to feed the scalers init position and detile\n", (double)s->LinesToRequestPrefetchPixelData * s->LineTime); + DML_LOG_VERBOSE("DML: To: %fus - time for propagation from scaler to optc\n", (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime); + DML_LOG_VERBOSE("DML: Tvstartup - TSetup - Tcalc - TWait - Tpre - To > 0\n"); + DML_LOG_VERBOSE("DML: Tslack(pre): %fus - time left over in schedule\n", p->VStartup * s->LineTime - s->TimeForFetchingVM - 2 * s->TimeForFetchingRowInVBlank - (*p->DSTYAfterScaler + ((double)(*p->DSTXAfterScaler) / (double)p->myPipe->HTotal)) * s->LineTime - p->TWait - p->TCalc - *p->TSetup); + DML_LOG_VERBOSE("DML: row_bytes = dpte_row_bytes (per_pipe) = PixelPTEBytesPerRow = : %u\n", p->PixelPTEBytesPerRow); + + } else { + DML_LOG_VERBOSE("DML::%s: No time to prefetch! dst_y_prefetch_equ = %f (should be > 1)\n", __func__, s->dst_y_prefetch_equ); + DML_LOG_VERBOSE("DML::%s: No time to prefetch! min_Lsw_equ_ok = %d, Tpre_rounded (%f) should be >= Tvm_trips_rounded (%f) + 2.0*Tr0_trips_rounded (%f) + min_Tsw_equ (%f)\n", + __func__, min_Lsw_equ_ok, *p->Tpre_rounded, s->Tvm_trips_rounded, 2.0*s->Tr0_trips_rounded, s->min_Lsw_equ*s->LineTime); + s->NoTimeToPrefetch = true; + s->TimeForFetchingVM = 0; + s->TimeForFetchingRowInVBlank = 0; + *p->dst_y_per_vm_vblank = 0; + *p->dst_y_per_row_vblank = 0; + s->LinesToRequestPrefetchPixelData = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixelDataBWLuma = 0; + *p->RequiredPrefetchPixelDataBWChroma = 0; + } + + { + double prefetch_vm_bw; + double prefetch_row_bw; + + if (vm_bytes == 0) { + prefetch_vm_bw = 0; + } else if (*p->dst_y_per_vm_vblank > 0) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, p->HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, s->LineTime); +#endif + prefetch_vm_bw = vm_bytes * p->HostVMInefficiencyFactor / (*p->dst_y_per_vm_vblank * s->LineTime); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: prefetch_vm_bw = %f\n", __func__, prefetch_vm_bw); +#endif + } else { + prefetch_vm_bw = 0; + s->NoTimeToPrefetch = true; + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_vm_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_vm_vblank); + } + + if (p->PixelPTEBytesPerRow == 0 && tdlut_row_bytes == 0) { + prefetch_row_bw = 0; + } else if (*p->dst_y_per_row_vblank > 0) { + prefetch_row_bw = (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + tdlut_row_bytes) / (*p->dst_y_per_row_vblank * s->LineTime); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, p->PixelPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: prefetch_row_bw = %f\n", __func__, prefetch_row_bw); +#endif + } else { + prefetch_row_bw = 0; + s->NoTimeToPrefetch = true; + DML_LOG_VERBOSE("DML::%s: No time to prefetch!. dst_y_per_row_vblank=%f (should be > 0)\n", __func__, *p->dst_y_per_row_vblank); + } + + *p->prefetch_vmrow_bw = math_max2(prefetch_vm_bw, prefetch_row_bw); + } + + if (s->NoTimeToPrefetch) { + s->TimeForFetchingVM = 0; + s->TimeForFetchingRowInVBlank = 0; + *p->dst_y_per_vm_vblank = 0; + *p->dst_y_per_row_vblank = 0; + *p->dst_y_prefetch = 0; + s->LinesToRequestPrefetchPixelData = 0; + *p->VRatioPrefetchY = 0; + *p->VRatioPrefetchC = 0; + *p->RequiredPrefetchPixelDataBWLuma = 0; + *p->RequiredPrefetchPixelDataBWChroma = 0; + *p->prefetch_vmrow_bw = 0; + } + + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_vblank = %f (final)\n", __func__, *p->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_vblank = %f (final)\n", __func__, *p->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML::%s: prefetch_vmrow_bw = %f (final)\n", __func__, *p->prefetch_vmrow_bw); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWLuma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWLuma); + DML_LOG_VERBOSE("DML::%s: RequiredPrefetchPixelDataBWChroma = %f (final)\n", __func__, *p->RequiredPrefetchPixelDataBWChroma); + DML_LOG_VERBOSE("DML::%s: NoTimeToPrefetch=%d\n", __func__, s->NoTimeToPrefetch); + + return s->NoTimeToPrefetch; +} + +static unsigned int get_num_lb_source_lines(unsigned int max_line_buffer_lines, + unsigned int line_buffer_size_bits, + unsigned int num_pipes, + unsigned int vp_width, + unsigned int vp_height, + double h_ratio, + enum dml2_rotation_angle rotation_angle) +{ + unsigned int num_lb_source_lines = 0; + double lb_bit_per_pixel = 57.0; + unsigned recin_width = vp_width/num_pipes; + + if (dml_is_vertical_rotation(rotation_angle)) + recin_width = vp_height/num_pipes; + + num_lb_source_lines = (unsigned int) math_min2((double) max_line_buffer_lines, + math_floor2(line_buffer_size_bits / lb_bit_per_pixel / (recin_width / math_max2(h_ratio, 1.0)), 1.0)); + + return num_lb_source_lines; +} + +static unsigned int find_max_impact_plane(unsigned int this_plane_idx, unsigned int num_planes, unsigned int Trpd_dcfclk_cycles[]) +{ + int max_value = -1; + int max_idx = -1; + for (unsigned int i = 0; i < num_planes; i++) { + if (i != this_plane_idx && (int) Trpd_dcfclk_cycles[i] > max_value) { + max_value = Trpd_dcfclk_cycles[i]; + max_idx = i; + } + } + if (max_idx <= 0) { + DML_ASSERT(max_idx >= 0); + max_idx = this_plane_idx; + } + + return max_idx; +} + +static double calculate_impacted_Tsw(unsigned int exclude_plane_idx, unsigned int num_planes, double *prefetch_swath_bytes, double bw_mbps) +{ + double sum = 0.; + for (unsigned int i = 0; i < num_planes; i++) { + if (i != exclude_plane_idx) { + sum += prefetch_swath_bytes[i]; + } + } + return sum / bw_mbps; +} + +// a global check against the aggregate effect of the per plane prefetch schedule +static noinline_for_stack bool CheckGlobalPrefetchAdmissibility(struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *p) +{ + struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_locals *s = &scratch->CheckGlobalPrefetchAdmissibility_locals; + unsigned int i, k; + + memset(s, 0, sizeof(struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_locals)); + + *p->recalc_prefetch_schedule = 0; + s->prefetch_global_check_passed = 1; + // worst case if the rob and cdb is fully hogged + s->max_Trpd_dcfclk_cycles = (unsigned int) math_ceil2((p->rob_buffer_size_kbytes*1024 + p->compressed_buffer_size_kbytes*DML_MAX_COMPRESSION_RATIO*1024)/64.0, 1.0); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: num_active_planes = %d\n", __func__, p->num_active_planes); + DML_LOG_VERBOSE("DML::%s: rob_buffer_size_kbytes = %d\n", __func__, p->rob_buffer_size_kbytes); + DML_LOG_VERBOSE("DML::%s: compressed_buffer_size_kbytes = %d\n", __func__, p->compressed_buffer_size_kbytes); + DML_LOG_VERBOSE("DML::%s: estimated_urg_bandwidth_required_mbps = %f\n", __func__, p->estimated_urg_bandwidth_required_mbps); + DML_LOG_VERBOSE("DML::%s: estimated_dcfclk_mhz = %f\n", __func__, p->estimated_dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: max_Trpd_dcfclk_cycles = %u\n", __func__, s->max_Trpd_dcfclk_cycles); +#endif + + // calculate the return impact from each plane, request is 256B per dcfclk + for (i = 0; i < p->num_active_planes; i++) { + s->src_detile_buf_size_bytes_l[i] = p->detile_buffer_size_bytes_l[i]; + s->src_detile_buf_size_bytes_c[i] = p->detile_buffer_size_bytes_c[i]; + s->src_swath_bytes_l[i] = p->full_swath_bytes_l[i]; + s->src_swath_bytes_c[i] = p->full_swath_bytes_c[i]; + + if (p->pixel_format[i] == dml2_420_10) { + s->src_detile_buf_size_bytes_l[i] = (unsigned int) (s->src_detile_buf_size_bytes_l[i] * 1.5); + s->src_detile_buf_size_bytes_c[i] = (unsigned int) (s->src_detile_buf_size_bytes_c[i] * 1.5); + s->src_swath_bytes_l[i] = (unsigned int) (s->src_swath_bytes_l[i] * 1.5); + s->src_swath_bytes_c[i] = (unsigned int) (s->src_swath_bytes_c[i] * 1.5); + } + + s->burst_bytes_to_fill_det = (unsigned int) (math_floor2(s->src_detile_buf_size_bytes_l[i] / p->chunk_bytes_l, 1) * p->chunk_bytes_l); + s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(p->lb_source_lines_l[i] / p->swath_height_l[i], 1) * s->src_swath_bytes_l[i]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: i=%u pixel_format = %d\n", __func__, i, p->pixel_format[i]); + DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_l = %d\n", __func__, i, p->chunk_bytes_l); + DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_l = %d\n", __func__, i, p->lb_source_lines_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_l=%d\n", __func__, i, s->src_detile_buf_size_bytes_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_l=%d\n", __func__, i, s->src_swath_bytes_l[i]); + DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d (luma)\n", __func__, i, s->burst_bytes_to_fill_det); +#endif + + if (s->src_swath_bytes_c[i] > 0) { // dual_plane + s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(s->src_detile_buf_size_bytes_c[i] / p->chunk_bytes_c, 1) * p->chunk_bytes_c); + + if (p->pixel_format[i] == dml2_422_planar_8 || p->pixel_format[i] == dml2_422_planar_10 || p->pixel_format[i] == dml2_422_planar_12) { + s->burst_bytes_to_fill_det += (unsigned int) (math_floor2(p->lb_source_lines_c[i] / p->swath_height_c[i], 1) * s->src_swath_bytes_c[i]); + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: i=%u chunk_bytes_c = %d\n", __func__, i, p->chunk_bytes_c); + DML_LOG_VERBOSE("DML::%s: i=%u lb_source_lines_c = %d\n", __func__, i, p->lb_source_lines_c[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_detile_buf_size_bytes_c=%d\n", __func__, i, s->src_detile_buf_size_bytes_c[i]); + DML_LOG_VERBOSE("DML::%s: i=%u src_swath_bytes_c=%d\n", __func__, i, s->src_swath_bytes_c[i]); +#endif + } + + s->time_to_fill_det_us = (double) s->burst_bytes_to_fill_det / (256 * p->estimated_dcfclk_mhz); // fill time assume full burst at request rate + s->accumulated_return_path_dcfclk_cycles[i] = (unsigned int) math_ceil2(((DML_MAX_COMPRESSION_RATIO-1) * 64 * p->estimated_dcfclk_mhz) * s->time_to_fill_det_us / 64.0, 1.0); //for 64B per DCFClk + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: i=%u burst_bytes_to_fill_det=%d\n", __func__, i, s->burst_bytes_to_fill_det); + DML_LOG_VERBOSE("DML::%s: i=%u time_to_fill_det_us=%f\n", __func__, i, s->time_to_fill_det_us); + DML_LOG_VERBOSE("DML::%s: i=%u accumulated_return_path_dcfclk_cycles=%u\n", __func__, i, s->accumulated_return_path_dcfclk_cycles[i]); +#endif + // clamping to worst case delay which is one which occupy the full rob+cdb + if (s->accumulated_return_path_dcfclk_cycles[i] > s->max_Trpd_dcfclk_cycles) + s->accumulated_return_path_dcfclk_cycles[i] = s->max_Trpd_dcfclk_cycles; + } + + // Figure out the impacted prefetch time for each plane + // if impacted_Tre is > equ bw Tpre, we need to fail the prefetch schedule as we need a higher state to support the bw + for (i = 0; i < p->num_active_planes; i++) { + k = find_max_impact_plane(i, p->num_active_planes, s->accumulated_return_path_dcfclk_cycles); // plane k causes most impact to plane i + // the rest of planes (except for k) complete for bw + p->impacted_dst_y_pre[i] = s->accumulated_return_path_dcfclk_cycles[k]/p->estimated_dcfclk_mhz; + p->impacted_dst_y_pre[i] += calculate_impacted_Tsw(k, p->num_active_planes, p->prefetch_sw_bytes, p->estimated_urg_bandwidth_required_mbps); + p->impacted_dst_y_pre[i] = math_ceil2(p->impacted_dst_y_pre[i] / p->line_time[i], 0.25); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: i=%u impacted_Tpre=%f (k=%u)\n", __func__, i, p->impacted_dst_y_pre[i], k); +#endif + } + + if (p->Tpre_rounded != NULL && p->Tpre_oto != NULL) { + for (i = 0; i < p->num_active_planes; i++) { + if (p->impacted_dst_y_pre[i] > p->dst_y_prefetch[i]) { + s->prefetch_global_check_passed = 0; + *p->recalc_prefetch_schedule = 1; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: i=%u Tpre_rounded=%f\n", __func__, i, p->Tpre_rounded[i]); + DML_LOG_VERBOSE("DML::%s: i=%u Tpre_oto=%f\n", __func__, i, p->Tpre_oto[i]); +#endif + } + } else { + // likely a mode programming calls, assume support, and no recalc - not used anyways + s->prefetch_global_check_passed = 1; + *p->recalc_prefetch_schedule = 0; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: prefetch_global_check_passed=%u\n", __func__, s->prefetch_global_check_passed); + DML_LOG_VERBOSE("DML::%s: recalc_prefetch_schedule=%u\n", __func__, *p->recalc_prefetch_schedule); +#endif + + return s->prefetch_global_check_passed; +} + +static void calculate_peak_bandwidth_required( + struct dml2_core_internal_scratch *s, + struct dml2_core_calcs_calculate_peak_bandwidth_required_params *p) +{ + unsigned int n; + unsigned int m; + + struct dml2_core_shared_calculate_peak_bandwidth_required_locals *l = &s->calculate_peak_bandwidth_required_locals; + + memset(l, 0, sizeof(struct dml2_core_shared_calculate_peak_bandwidth_required_locals)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: inc_flip_bw = %d\n", __func__, p->inc_flip_bw); + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %d\n", __func__, p->num_active_planes); +#endif + + for (unsigned int k = 0; k < p->num_active_planes; ++k) { + l->unity_array[k] = 1.0; + l->zero_array[k] = 0.0; + } + + for (m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (n = 0; n < dml2_core_internal_bw_max; n++) { + get_urgent_bandwidth_required( + &s->get_urgent_bandwidth_required_locals, + p->display_cfg, + m, + n, + 0, //inc_flip_bw, + 0, //use_qual_row_bw + p->num_active_planes, + p->num_of_dpp, + p->dcc_dram_bw_nom_overhead_factor_p0, + p->dcc_dram_bw_nom_overhead_factor_p1, + p->dcc_dram_bw_pref_overhead_factor_p0, + p->dcc_dram_bw_pref_overhead_factor_p1, + p->mall_prefetch_sdp_overhead_factor, + p->mall_prefetch_dram_overhead_factor, + p->surface_read_bandwidth_l, + p->surface_read_bandwidth_c, + l->zero_array, //PrefetchBandwidthLuma, + l->zero_array, //PrefetchBandwidthChroma, + l->zero_array, //PrefetchBWMax + l->zero_array, + l->zero_array, + l->zero_array, + p->dpte_row_bw, + p->meta_row_bw, + l->zero_array, //prefetch_cursor_bw, + l->zero_array, //prefetch_vmrow_bw, + l->zero_array, //flip_bw, + l->zero_array, + l->zero_array, + l->zero_array, + l->zero_array, + l->zero_array, + l->zero_array, + p->surface_avg_vactive_required_bw[m][n], + p->surface_peak_required_bw[m][n]); + + p->urg_vactive_bandwidth_required[m][n] = get_urgent_bandwidth_required( + &s->get_urgent_bandwidth_required_locals, + p->display_cfg, + m, + n, + 0, //inc_flip_bw, + 0, //use_qual_row_bw + p->num_active_planes, + p->num_of_dpp, + p->dcc_dram_bw_nom_overhead_factor_p0, + p->dcc_dram_bw_nom_overhead_factor_p1, + p->dcc_dram_bw_pref_overhead_factor_p0, + p->dcc_dram_bw_pref_overhead_factor_p1, + p->mall_prefetch_sdp_overhead_factor, + p->mall_prefetch_dram_overhead_factor, + p->surface_read_bandwidth_l, + p->surface_read_bandwidth_c, + l->zero_array, //PrefetchBandwidthLuma, + l->zero_array, //PrefetchBandwidthChroma, + l->zero_array, //PrefetchBWMax + p->excess_vactive_fill_bw_l, + p->excess_vactive_fill_bw_c, + p->cursor_bw, + p->dpte_row_bw, + p->meta_row_bw, + l->zero_array, //prefetch_cursor_bw, + l->zero_array, //prefetch_vmrow_bw, + l->zero_array, //flip_bw, + p->urgent_burst_factor_l, + p->urgent_burst_factor_c, + p->urgent_burst_factor_cursor, + p->urgent_burst_factor_prefetch_l, + p->urgent_burst_factor_prefetch_c, + p->urgent_burst_factor_prefetch_cursor, + l->surface_dummy_bw, + p->surface_peak_required_bw[m][n]); + + p->urg_bandwidth_required[m][n] = get_urgent_bandwidth_required( + &s->get_urgent_bandwidth_required_locals, + p->display_cfg, + m, + n, + p->inc_flip_bw, + 0, //use_qual_row_bw + p->num_active_planes, + p->num_of_dpp, + p->dcc_dram_bw_nom_overhead_factor_p0, + p->dcc_dram_bw_nom_overhead_factor_p1, + p->dcc_dram_bw_pref_overhead_factor_p0, + p->dcc_dram_bw_pref_overhead_factor_p1, + p->mall_prefetch_sdp_overhead_factor, + p->mall_prefetch_dram_overhead_factor, + p->surface_read_bandwidth_l, + p->surface_read_bandwidth_c, + p->prefetch_bandwidth_l, + p->prefetch_bandwidth_c, + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw + p->excess_vactive_fill_bw_l, + p->excess_vactive_fill_bw_c, + p->cursor_bw, + p->dpte_row_bw, + p->meta_row_bw, + p->prefetch_cursor_bw, + p->prefetch_vmrow_bw, + p->flip_bw, + p->urgent_burst_factor_l, + p->urgent_burst_factor_c, + p->urgent_burst_factor_cursor, + p->urgent_burst_factor_prefetch_l, + p->urgent_burst_factor_prefetch_c, + p->urgent_burst_factor_prefetch_cursor, + l->surface_dummy_bw, + p->surface_peak_required_bw[m][n]); + + p->urg_bandwidth_required_qual[m][n] = get_urgent_bandwidth_required( + &s->get_urgent_bandwidth_required_locals, + p->display_cfg, + m, + n, + 0, //inc_flip_bw + 1, //use_qual_row_bw + p->num_active_planes, + p->num_of_dpp, + p->dcc_dram_bw_nom_overhead_factor_p0, + p->dcc_dram_bw_nom_overhead_factor_p1, + p->dcc_dram_bw_pref_overhead_factor_p0, + p->dcc_dram_bw_pref_overhead_factor_p1, + p->mall_prefetch_sdp_overhead_factor, + p->mall_prefetch_dram_overhead_factor, + p->surface_read_bandwidth_l, + p->surface_read_bandwidth_c, + p->prefetch_bandwidth_l, + p->prefetch_bandwidth_c, + p->prefetch_bandwidth_max, // to prevent ms/mp mismatch where mp prefetch bw > ms prefetch bw + p->excess_vactive_fill_bw_l, + p->excess_vactive_fill_bw_c, + p->cursor_bw, + p->dpte_row_bw, + p->meta_row_bw, + p->prefetch_cursor_bw, + p->prefetch_vmrow_bw, + p->flip_bw, + p->urgent_burst_factor_l, + p->urgent_burst_factor_c, + p->urgent_burst_factor_cursor, + p->urgent_burst_factor_prefetch_l, + p->urgent_burst_factor_prefetch_c, + p->urgent_burst_factor_prefetch_cursor, + l->surface_dummy_bw, + p->surface_peak_required_bw[m][n]); + + p->non_urg_bandwidth_required[m][n] = get_urgent_bandwidth_required( + &s->get_urgent_bandwidth_required_locals, + p->display_cfg, + m, + n, + p->inc_flip_bw, + 0, //use_qual_row_bw + p->num_active_planes, + p->num_of_dpp, + p->dcc_dram_bw_nom_overhead_factor_p0, + p->dcc_dram_bw_nom_overhead_factor_p1, + p->dcc_dram_bw_pref_overhead_factor_p0, + p->dcc_dram_bw_pref_overhead_factor_p1, + p->mall_prefetch_sdp_overhead_factor, + p->mall_prefetch_dram_overhead_factor, + p->surface_read_bandwidth_l, + p->surface_read_bandwidth_c, + p->prefetch_bandwidth_l, + p->prefetch_bandwidth_c, + p->prefetch_bandwidth_max, // to prevent ms/mp mismatches where mp prefetch bw > ms prefetch bw + p->excess_vactive_fill_bw_l, + p->excess_vactive_fill_bw_c, + p->cursor_bw, + p->dpte_row_bw, + p->meta_row_bw, + p->prefetch_cursor_bw, + p->prefetch_vmrow_bw, + p->flip_bw, + l->unity_array, + l->unity_array, + l->unity_array, + l->unity_array, + l->unity_array, + l->unity_array, + l->surface_dummy_bw, + p->surface_peak_required_bw[m][n]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: urg_vactive_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_vactive_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_qual[%s][%s]=%f\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->urg_bandwidth_required[m][n]); + DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required%s[%s][%s]=%f\n", __func__, (p->inc_flip_bw ? "_flip" : ""), dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), p->non_urg_bandwidth_required[m][n]); +#endif + DML_ASSERT(p->urg_bandwidth_required[m][n] >= p->non_urg_bandwidth_required[m][n]); + } + } +} + +static void check_urgent_bandwidth_support( + double *frac_urg_bandwidth_nom, + double *frac_urg_bandwidth_mall, + bool *vactive_bandwidth_support_ok, // vactive ok + bool *bandwidth_support_ok,// max of vm, prefetch, vactive all ok + + unsigned int mall_allocated_for_dcn_mbytes, + double non_urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_vactive_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) +{ + double frac_urg_bandwidth_nom_sdp = non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] / urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + double frac_urg_bandwidth_nom_dram = non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] / urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; + double frac_urg_bandwidth_mall_sdp; + double frac_urg_bandwidth_mall_dram; + if (urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] > 0) + frac_urg_bandwidth_mall_sdp = non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] / urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; + else + frac_urg_bandwidth_mall_sdp = 0.0; + if (urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] > 0) + frac_urg_bandwidth_mall_dram = non_urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] / urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; + else + frac_urg_bandwidth_mall_dram = 0.0; + + *bandwidth_support_ok = 1; + *vactive_bandwidth_support_ok = 1; + + // Check urgent bandwidth required at sdp vs urgent bandwidth avail at sdp -> FractionOfUrgentBandwidth + // Check urgent bandwidth required at dram vs urgent bandwidth avail at dram + // Check urgent bandwidth required at sdp vs urgent bandwidth avail at sdp, svp_prefetch -> FractionOfUrgentBandwidthMALL + // Check urgent bandwidth required at dram vs urgent bandwidth avail at dram, svp_prefetch + + *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; + + if (mall_allocated_for_dcn_mbytes > 0) { + *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; + *bandwidth_support_ok &= urg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; + } + + *frac_urg_bandwidth_nom = math_max2(frac_urg_bandwidth_nom_sdp, frac_urg_bandwidth_nom_dram); + *frac_urg_bandwidth_mall = math_max2(frac_urg_bandwidth_mall_sdp, frac_urg_bandwidth_mall_dram); + + *bandwidth_support_ok &= (*frac_urg_bandwidth_nom <= 1.0); + + if (mall_allocated_for_dcn_mbytes > 0) + *bandwidth_support_ok &= (*frac_urg_bandwidth_mall <= 1.0); + + *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_dram]; + if (mall_allocated_for_dcn_mbytes > 0) { + *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_sdp]; + *vactive_bandwidth_support_ok &= urg_vactive_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram] <= urg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][dml2_core_internal_bw_dram]; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_sdp = %f\n", __func__, frac_urg_bandwidth_nom_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom_dram = %f\n", __func__, frac_urg_bandwidth_nom_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_nom = %f\n", __func__, *frac_urg_bandwidth_nom); + + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_sdp = %f\n", __func__, frac_urg_bandwidth_mall_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall_dram = %f\n", __func__, frac_urg_bandwidth_mall_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_mall = %f\n", __func__, *frac_urg_bandwidth_mall); + DML_LOG_VERBOSE("DML::%s: bandwidth_support_ok = %d\n", __func__, *bandwidth_support_ok); + + for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { + DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", + __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), + urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required[m][n]) ? "<" : ">=", urg_bandwidth_required[m][n]); + } + } +#endif +} + +static double get_bandwidth_available_for_immediate_flip(enum dml2_core_internal_soc_state_type eval_state, + double urg_bandwidth_required[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], // no flip + double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) +{ + double flip_bw_available_mbps; + double flip_bw_available_sdp_mbps; + double flip_bw_available_dram_mbps; + + flip_bw_available_sdp_mbps = urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp] - urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]; + flip_bw_available_dram_mbps = urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram] - urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]; + flip_bw_available_mbps = flip_bw_available_sdp_mbps < flip_bw_available_dram_mbps ? flip_bw_available_sdp_mbps : flip_bw_available_dram_mbps; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_sdp_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available_dram_mbps = %f\n", __func__, urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_sdp_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_sdp]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_dram_mbps = %f\n", __func__, urg_bandwidth_required[eval_state][dml2_core_internal_bw_dram]); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_sdp_mbps = %f\n", __func__, flip_bw_available_sdp_mbps); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_dram_mbps = %f\n", __func__, flip_bw_available_dram_mbps); + DML_LOG_VERBOSE("DML::%s: flip_bw_available_mbps = %f\n", __func__, flip_bw_available_mbps); +#endif + + return flip_bw_available_mbps; +} + +static void calculate_immediate_flip_bandwidth_support( + // Output + double *frac_urg_bandwidth_flip, + bool *flip_bandwidth_support_ok, + + // Input + enum dml2_core_internal_soc_state_type eval_state, + double urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max], + double urg_bandwidth_available[dml2_core_internal_soc_state_max][dml2_core_internal_bw_max]) +{ + double frac_urg_bw_flip_sdp = non_urg_bandwidth_required_flip[eval_state][dml2_core_internal_bw_sdp] / urg_bandwidth_available[eval_state][dml2_core_internal_bw_sdp]; + double frac_urg_bw_flip_dram = non_urg_bandwidth_required_flip[eval_state][dml2_core_internal_bw_dram] / urg_bandwidth_available[eval_state][dml2_core_internal_bw_dram]; + + *flip_bandwidth_support_ok = true; + for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { // check sdp and dram + *flip_bandwidth_support_ok &= urg_bandwidth_available[eval_state][n] >= urg_bandwidth_required_flip[eval_state][n]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: n = %s\n", __func__, dml2_core_internal_bw_type_str(n)); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_available = %f\n", __func__, urg_bandwidth_available[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: non_urg_bandwidth_required_flip = %f\n", __func__, non_urg_bandwidth_required_flip[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_required_flip = %f\n", __func__, urg_bandwidth_required_flip[eval_state][n]); + DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); +#endif + DML_ASSERT(urg_bandwidth_required_flip[eval_state][n] >= non_urg_bandwidth_required_flip[eval_state][n]); + } + + *frac_urg_bandwidth_flip = (frac_urg_bw_flip_sdp > frac_urg_bw_flip_dram) ? frac_urg_bw_flip_sdp : frac_urg_bw_flip_dram; + *flip_bandwidth_support_ok &= (*frac_urg_bandwidth_flip <= 1.0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: eval_state = %s\n", __func__, dml2_core_internal_soc_state_type_str(eval_state)); + DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_sdp = %f\n", __func__, frac_urg_bw_flip_sdp); + DML_LOG_VERBOSE("DML::%s: frac_urg_bw_flip_dram = %f\n", __func__, frac_urg_bw_flip_dram); + DML_LOG_VERBOSE("DML::%s: frac_urg_bandwidth_flip = %f\n", __func__, *frac_urg_bandwidth_flip); + DML_LOG_VERBOSE("DML::%s: flip_bandwidth_support_ok = %d\n", __func__, *flip_bandwidth_support_ok); + + for (unsigned int m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (unsigned int n = 0; n < dml2_core_internal_bw_max; n++) { + DML_LOG_VERBOSE("DML::%s: state:%s bw_type:%s, urg_bandwidth_available=%f %s urg_bandwidth_required=%f\n", + __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n), + urg_bandwidth_available[m][n], (urg_bandwidth_available[m][n] < urg_bandwidth_required_flip[m][n]) ? "<" : ">=", urg_bandwidth_required_flip[m][n]); + } + } +#endif +} + +static void CalculateFlipSchedule( + struct dml2_core_internal_scratch *s, + bool iflip_enable, + bool use_lb_flip_bw, + double HostVMInefficiencyFactor, + double Tvm_trips_flip, + double Tr0_trips_flip, + double Tvm_trips_flip_rounded, + double Tr0_trips_flip_rounded, + bool GPUVMEnable, + double vm_bytes, // vm_bytes + double DPTEBytesPerRow, // dpte_row_bytes + double BandwidthAvailableForImmediateFlip, + unsigned int TotImmediateFlipBytes, + enum dml2_source_format_class SourcePixelFormat, + double LineTime, + double VRatio, + double VRatioChroma, + double Tno_bw_flip, + unsigned int dpte_row_height, + unsigned int dpte_row_height_chroma, + bool use_one_row_for_frame_flip, + unsigned int max_flip_time_us, + unsigned int max_flip_time_lines, + unsigned int per_pipe_flip_bytes, + unsigned int meta_row_bytes, + unsigned int meta_row_height, + unsigned int meta_row_height_chroma, + bool dcc_mrq_enable, + + // Output + double *dst_y_per_vm_flip, + double *dst_y_per_row_flip, + double *final_flip_bw, + bool *ImmediateFlipSupportedForPipe) +{ + struct dml2_core_shared_CalculateFlipSchedule_locals *l = &s->CalculateFlipSchedule_locals; + + l->dual_plane = dml_is_420(SourcePixelFormat) || SourcePixelFormat == dml2_rgbe_alpha; + l->dpte_row_bytes = DPTEBytesPerRow; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: GPUVMEnable = %u\n", __func__, GPUVMEnable); + DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_us = %d\n", __func__, max_flip_time_us); + DML_LOG_VERBOSE("DML::%s: ip.max_flip_time_lines = %d\n", __func__, max_flip_time_lines); + DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); + DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, TotImmediateFlipBytes); + DML_LOG_VERBOSE("DML::%s: use_lb_flip_bw = %u\n", __func__, use_lb_flip_bw); + DML_LOG_VERBOSE("DML::%s: iflip_enable = %u\n", __func__, iflip_enable); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: LineTime = %f\n", __func__, LineTime); + DML_LOG_VERBOSE("DML::%s: Tno_bw_flip = %f\n", __func__, Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip = %f\n", __func__, Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip = %f\n", __func__, Tr0_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded = %f\n", __func__, Tvm_trips_flip_rounded); + DML_LOG_VERBOSE("DML::%s: Tr0_trips_flip_rounded = %f\n", __func__, Tr0_trips_flip_rounded); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %f\n", __func__, vm_bytes); + DML_LOG_VERBOSE("DML::%s: DPTEBytesPerRow = %f\n", __func__, DPTEBytesPerRow); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %d\n", __func__, meta_row_bytes); + DML_LOG_VERBOSE("DML::%s: dpte_row_bytes = %f\n", __func__, l->dpte_row_bytes); + DML_LOG_VERBOSE("DML::%s: dpte_row_height = %d\n", __func__, dpte_row_height); + DML_LOG_VERBOSE("DML::%s: meta_row_height = %d\n", __func__, meta_row_height); + DML_LOG_VERBOSE("DML::%s: VRatio = %f\n", __func__, VRatio); +#endif + + if (TotImmediateFlipBytes > 0 && (GPUVMEnable || dcc_mrq_enable)) { + if (l->dual_plane) { + if (dcc_mrq_enable & GPUVMEnable) { + l->min_row_height = math_min2(dpte_row_height, meta_row_height); + l->min_row_height_chroma = math_min2(dpte_row_height_chroma, meta_row_height_chroma); + } else if (GPUVMEnable) { + l->min_row_height = dpte_row_height; + l->min_row_height_chroma = dpte_row_height_chroma; + } else { + l->min_row_height = meta_row_height; + l->min_row_height_chroma = meta_row_height_chroma; + } + l->min_row_time = math_min2(l->min_row_height * LineTime / VRatio, l->min_row_height_chroma * LineTime / VRatioChroma); + } else { + if (dcc_mrq_enable & GPUVMEnable) + l->min_row_height = math_min2(dpte_row_height, meta_row_height); + else if (GPUVMEnable) + l->min_row_height = dpte_row_height; + else + l->min_row_height = meta_row_height; + + l->min_row_time = l->min_row_height * LineTime / VRatio; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: min_row_time = %f\n", __func__, l->min_row_time); +#endif + DML_ASSERT(l->min_row_time > 0); + + if (use_lb_flip_bw) { + // For mode check, calculation the flip bw requirement with worst case flip time + l->max_flip_time = math_min2(math_min2(l->min_row_time, (double)max_flip_time_lines * LineTime / VRatio), + math_max2(Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded, (double)max_flip_time_us)); + + //The lower bound on flip bandwidth + // Note: The get_urgent_bandwidth_required already consider dpte_row_bw and meta_row_bw in bandwidth calculation, so leave final_flip_bw = 0 if iflip not required + l->lb_flip_bw = 0; + + if (iflip_enable) { + l->hvm_scaled_vm_bytes = vm_bytes * HostVMInefficiencyFactor; + l->num_rows = 2; + l->hvm_scaled_row_bytes = (l->num_rows * l->dpte_row_bytes * HostVMInefficiencyFactor + l->num_rows * meta_row_bytes); + l->hvm_scaled_vm_row_bytes = l->hvm_scaled_vm_bytes + l->hvm_scaled_row_bytes; + l->lb_flip_bw = math_max3( + l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip), + l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded), + l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: max_flip_time = %f\n", __func__, l->max_flip_time); + DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_bytes); + DML_LOG_VERBOSE("DML::%s: total row bytes (%f row, hvm ineff scaled) = %f\n", __func__, l->num_rows, l->hvm_scaled_row_bytes); + DML_LOG_VERBOSE("DML::%s: total vm+row bytes (hvm ineff scaled) = %f\n", __func__, l->hvm_scaled_vm_row_bytes); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm and row = %f\n", __func__, l->hvm_scaled_vm_row_bytes / (l->max_flip_time - Tno_bw_flip)); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm = %f\n", __func__, l->hvm_scaled_vm_bytes / (l->max_flip_time - Tno_bw_flip - 2 * Tr0_trips_flip_rounded)); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row = %f\n", __func__, l->hvm_scaled_row_bytes / (l->max_flip_time - Tvm_trips_flip_rounded)); + + if (l->lb_flip_bw > 0) { + DML_LOG_VERBOSE("DML::%s: mode_support est Tvm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw); + DML_LOG_VERBOSE("DML::%s: mode_support est Tr0_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / l->num_rows); + DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_vm_flip = %f (bw-based)\n", __func__, Tno_bw_flip + l->hvm_scaled_vm_bytes / l->lb_flip_bw / LineTime); + DML_LOG_VERBOSE("DML::%s: mode_support est dst_y_per_row_flip = %f (bw-based)\n", __func__, l->hvm_scaled_row_bytes / l->lb_flip_bw / LineTime / l->num_rows); + DML_LOG_VERBOSE("DML::%s: Tvm_trips_flip_rounded + 2*Tr0_trips_flip_rounded = %f\n", __func__, (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded)); + } +#endif + l->lb_flip_bw = math_max3(l->lb_flip_bw, + l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip, + (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for vm reg limit = %f\n", __func__, l->hvm_scaled_vm_bytes / (31 * LineTime) - Tno_bw_flip); + DML_LOG_VERBOSE("DML::%s: lb_flip_bw for row reg limit = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (15 * LineTime)); +#endif + } + + *final_flip_bw = l->lb_flip_bw; + + *dst_y_per_vm_flip = 1; // not used + *dst_y_per_row_flip = 1; // not used + *ImmediateFlipSupportedForPipe = l->min_row_time >= (Tvm_trips_flip_rounded + 2 * Tr0_trips_flip_rounded); + } else { + if (iflip_enable) { + l->ImmediateFlipBW = (double)per_pipe_flip_bytes * BandwidthAvailableForImmediateFlip / (double)TotImmediateFlipBytes; // flip_bw(i) + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: per_pipe_flip_bytes = %d\n", __func__, per_pipe_flip_bytes); + DML_LOG_VERBOSE("DML::%s: BandwidthAvailableForImmediateFlip = %f\n", __func__, BandwidthAvailableForImmediateFlip); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipBW = %f\n", __func__, l->ImmediateFlipBW); + DML_LOG_VERBOSE("DML::%s: portion of flip bw = %f\n", __func__, (double)per_pipe_flip_bytes / (double)TotImmediateFlipBytes); +#endif + if (l->ImmediateFlipBW == 0) { + l->Tvm_flip = 0; + l->Tr0_flip = 0; + } else { + l->Tvm_flip = math_max3(Tvm_trips_flip, + Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, + LineTime / 4.0); + + l->Tr0_flip = math_max3(Tr0_trips_flip, + (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, + LineTime / 4.0); + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: total vm bytes (hvm ineff scaled) = %f\n", __func__, vm_bytes * HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: total row bytes (hvm ineff scaled, one row) = %f\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes)); + + DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (bw-based), Tvm_trips_flip = %f (latency-based)\n", __func__, Tno_bw_flip + vm_bytes * HostVMInefficiencyFactor / l->ImmediateFlipBW, Tvm_trips_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (bw-based), Tr0_trips_flip = %f (latency-based)\n", __func__, (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / l->ImmediateFlipBW, Tr0_trips_flip); +#endif + *dst_y_per_vm_flip = math_ceil2(4.0 * (l->Tvm_flip / LineTime), 1.0) / 4.0; + *dst_y_per_row_flip = math_ceil2(4.0 * (l->Tr0_flip / LineTime), 1.0) / 4.0; + + *final_flip_bw = math_max2(vm_bytes * HostVMInefficiencyFactor / (*dst_y_per_vm_flip * LineTime), + (l->dpte_row_bytes * HostVMInefficiencyFactor + meta_row_bytes) / (*dst_y_per_row_flip * LineTime)); + + if (*dst_y_per_vm_flip >= 32 || *dst_y_per_row_flip >= 16 || l->Tvm_flip + 2 * l->Tr0_flip > l->min_row_time) { + *ImmediateFlipSupportedForPipe = false; + } else { + *ImmediateFlipSupportedForPipe = iflip_enable; + } + } else { + l->Tvm_flip = 0; + l->Tr0_flip = 0; + *dst_y_per_vm_flip = 0; + *dst_y_per_row_flip = 0; + *final_flip_bw = 0; + *ImmediateFlipSupportedForPipe = iflip_enable; + } + } + } else { + l->Tvm_flip = 0; + l->Tr0_flip = 0; + *dst_y_per_vm_flip = 0; + *dst_y_per_row_flip = 0; + *final_flip_bw = 0; + *ImmediateFlipSupportedForPipe = iflip_enable; + } + +#ifdef __DML_VBA_DEBUG__ + if (!use_lb_flip_bw) { + DML_LOG_VERBOSE("DML::%s: dst_y_per_vm_flip = %f (should be < 32)\n", __func__, *dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML::%s: dst_y_per_row_flip = %f (should be < 16)\n", __func__, *dst_y_per_row_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_flip = %f (final)\n", __func__, l->Tvm_flip); + DML_LOG_VERBOSE("DML::%s: Tr0_flip = %f (final)\n", __func__, l->Tr0_flip); + DML_LOG_VERBOSE("DML::%s: Tvm_flip + 2*Tr0_flip = %f (should be <= min_row_time=%f)\n", __func__, l->Tvm_flip + 2 * l->Tr0_flip, l->min_row_time); + } + DML_LOG_VERBOSE("DML::%s: final_flip_bw = %f\n", __func__, *final_flip_bw); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupportedForPipe = %u\n", __func__, *ImmediateFlipSupportedForPipe); +#endif +} + +static void CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( + struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *p) +{ + struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals *s = &scratch->CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_locals; + + enum dml2_uclk_pstate_change_strategy uclk_pstate_change_strategy; + double reserved_vblank_time_us; + bool FoundCriticalSurface = false; + + s->TotalActiveWriteback = 0; + p->Watermark->UrgentWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); +#endif + + p->Watermark->USRRetrainingWatermark = p->mmSOCParameters.UrgentLatency + p->mmSOCParameters.ExtraLatency + p->mmSOCParameters.USRRetrainingLatency + p->mmSOCParameters.SMNLatency; + p->Watermark->DRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->Watermark->UrgentWatermark; + p->Watermark->FCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->Watermark->UrgentWatermark; + p->Watermark->StutterExitWatermark = p->mmSOCParameters.SRExitTime + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; + p->Watermark->StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitTime + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; + p->Watermark->Z8StutterExitWatermark = p->mmSOCParameters.SRExitZ8Time + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; + p->Watermark->Z8StutterEnterPlusExitWatermark = p->mmSOCParameters.SREnterPlusExitZ8Time + p->mmSOCParameters.ExtraLatency_sr + 10 / p->DCFClkDeepSleep; + if (p->mmSOCParameters.qos_type == dml2_qos_param_type_dcn4x) { + p->Watermark->StutterExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; + p->Watermark->StutterEnterPlusExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; + p->Watermark->Z8StutterExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; + p->Watermark->Z8StutterEnterPlusExitWatermark += p->mmSOCParameters.max_urgent_latency_us + p->mmSOCParameters.df_response_time_us; + } + p->Watermark->temp_read_or_ppt_watermark_us = p->mmSOCParameters.g6_temp_read_blackout_us + p->Watermark->UrgentWatermark; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: UrgentLatency = %f\n", __func__, p->mmSOCParameters.UrgentLatency); + DML_LOG_VERBOSE("DML::%s: ExtraLatency = %f\n", __func__, p->mmSOCParameters.ExtraLatency); + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeLatency = %f\n", __func__, p->mmSOCParameters.DRAMClockChangeLatency); + DML_LOG_VERBOSE("DML::%s: SREnterPlusExitZ8Time = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitZ8Time); + DML_LOG_VERBOSE("DML::%s: SREnterPlusExitTime = %f\n", __func__, p->mmSOCParameters.SREnterPlusExitTime); + DML_LOG_VERBOSE("DML::%s: UrgentWatermark = %f\n", __func__, p->Watermark->UrgentWatermark); + DML_LOG_VERBOSE("DML::%s: USRRetrainingWatermark = %f\n", __func__, p->Watermark->USRRetrainingWatermark); + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeWatermark = %f\n", __func__, p->Watermark->DRAMClockChangeWatermark); + DML_LOG_VERBOSE("DML::%s: FCLKChangeWatermark = %f\n", __func__, p->Watermark->FCLKChangeWatermark); + DML_LOG_VERBOSE("DML::%s: StutterExitWatermark = %f\n", __func__, p->Watermark->StutterExitWatermark); + DML_LOG_VERBOSE("DML::%s: StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: Z8StutterExitWatermark = %f\n", __func__, p->Watermark->Z8StutterExitWatermark); + DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Watermark->Z8StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: temp_read_or_ppt_watermark_us = %f\n", __func__, p->Watermark->temp_read_or_ppt_watermark_us); +#endif + + s->TotalActiveWriteback = 0; + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + s->TotalActiveWriteback = s->TotalActiveWriteback + 1; + } + } + + if (s->TotalActiveWriteback <= 1) { + p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency; + } else { + p->Watermark->WritebackUrgentWatermark = p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; + } + if (p->USRRetrainingRequired) + p->Watermark->WritebackUrgentWatermark = p->Watermark->WritebackUrgentWatermark + p->mmSOCParameters.USRRetrainingLatency; + + if (s->TotalActiveWriteback <= 1) { + p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency; + p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency; + } else { + p->Watermark->WritebackDRAMClockChangeWatermark = p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024.0 / 32.0 / p->SOCCLK; + p->Watermark->WritebackFCLKChangeWatermark = p->mmSOCParameters.FCLKChangeLatency + p->mmSOCParameters.WritebackLatency + p->WritebackChunkSize * 1024 / 32 / p->SOCCLK; + } + + if (p->USRRetrainingRequired) + p->Watermark->WritebackDRAMClockChangeWatermark = p->Watermark->WritebackDRAMClockChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; + + if (p->USRRetrainingRequired) + p->Watermark->WritebackFCLKChangeWatermark = p->Watermark->WritebackFCLKChangeWatermark + p->mmSOCParameters.USRRetrainingLatency; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: WritebackDRAMClockChangeWatermark = %f\n", __func__, p->Watermark->WritebackDRAMClockChangeWatermark); + DML_LOG_VERBOSE("DML::%s: WritebackFCLKChangeWatermark = %f\n", __func__, p->Watermark->WritebackFCLKChangeWatermark); + DML_LOG_VERBOSE("DML::%s: WritebackUrgentWatermark = %f\n", __func__, p->Watermark->WritebackUrgentWatermark); + DML_LOG_VERBOSE("DML::%s: USRRetrainingRequired = %u\n", __func__, p->USRRetrainingRequired); + DML_LOG_VERBOSE("DML::%s: USRRetrainingLatency = %f\n", __func__, p->mmSOCParameters.USRRetrainingLatency); +#endif + + s->TotalPixelBW = 0.0; + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + double h_total = (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; + double pixel_clock_mhz = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000.0; + double v_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + double v_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + s->TotalPixelBW = s->TotalPixelBW + p->DPPPerSurface[k] + * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * v_ratio + p->SwathWidthC[k] * p->BytePerPixelDETC[k] * v_ratio_c) / (h_total / pixel_clock_mhz); + } + + *p->global_fclk_change_supported = true; + *p->global_dram_clock_change_supported = true; + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + double h_total = (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; + double pixel_clock_mhz = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000.0; + double v_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + double v_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + double v_taps = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + double v_taps_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + double h_ratio = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio; + double h_ratio_c = p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio; + double LBBitPerPixel = 57; + + s->LBLatencyHidingSourceLinesY[k] = (unsigned int)(math_min2((double)p->MaxLineBufferLines, math_floor2((double)p->LineBufferSize / LBBitPerPixel / ((double)p->SwathWidthY[k] / math_max2(h_ratio, 1.0)), 1)) - (v_taps - 1)); + s->LBLatencyHidingSourceLinesC[k] = (unsigned int)(math_min2((double)p->MaxLineBufferLines, math_floor2((double)p->LineBufferSize / LBBitPerPixel / ((double)p->SwathWidthC[k] / math_max2(h_ratio_c, 1.0)), 1)) - (v_taps_c - 1)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, MaxLineBufferLines = %u\n", __func__, k, p->MaxLineBufferLines); + DML_LOG_VERBOSE("DML::%s: k=%u, LineBufferSize = %u\n", __func__, k, p->LineBufferSize); + DML_LOG_VERBOSE("DML::%s: k=%u, LBBitPerPixel = %f\n", __func__, k, LBBitPerPixel); + DML_LOG_VERBOSE("DML::%s: k=%u, HRatio = %f\n", __func__, k, h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, VTaps = %f\n", __func__, k, v_taps); +#endif + + s->EffectiveLBLatencyHidingY = s->LBLatencyHidingSourceLinesY[k] / v_ratio * (h_total / pixel_clock_mhz); + s->EffectiveLBLatencyHidingC = s->LBLatencyHidingSourceLinesC[k] / v_ratio_c * (h_total / pixel_clock_mhz); + + s->EffectiveDETBufferSizeY = p->DETBufferSizeY[k]; + if (p->UnboundedRequestEnabled) { + s->EffectiveDETBufferSizeY = s->EffectiveDETBufferSizeY + p->CompressedBufferSizeInkByte * 1024 * (p->SwathWidthY[k] * p->BytePerPixelDETY[k] * v_ratio) / (h_total / pixel_clock_mhz) / s->TotalPixelBW; + } + + s->LinesInDETY[k] = (double)s->EffectiveDETBufferSizeY / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; + s->LinesInDETYRoundedDownToSwath[k] = (unsigned int)(math_floor2(s->LinesInDETY[k], p->SwathHeightY[k])); + s->FullDETBufferingTimeY = s->LinesInDETYRoundedDownToSwath[k] * (h_total / pixel_clock_mhz) / v_ratio; + + s->ActiveClockChangeLatencyHidingY = s->EffectiveLBLatencyHidingY + s->FullDETBufferingTimeY - ((double)p->DSTXAfterScaler[k] / h_total + (double)p->DSTYAfterScaler[k]) * h_total / pixel_clock_mhz; + + if (p->NumberOfActiveSurfaces > 1) { + s->ActiveClockChangeLatencyHidingY = s->ActiveClockChangeLatencyHidingY - (1.0 - 1.0 / (double)p->NumberOfActiveSurfaces) * (double)p->SwathHeightY[k] * (double)h_total / pixel_clock_mhz / v_ratio; + } + + if (p->BytePerPixelDETC[k] > 0) { + s->LinesInDETC[k] = p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k]; + s->LinesInDETCRoundedDownToSwath[k] = (unsigned int)(math_floor2(s->LinesInDETC[k], p->SwathHeightC[k])); + s->FullDETBufferingTimeC = s->LinesInDETCRoundedDownToSwath[k] * (h_total / pixel_clock_mhz) / v_ratio_c; + s->ActiveClockChangeLatencyHidingC = s->EffectiveLBLatencyHidingC + s->FullDETBufferingTimeC - ((double)p->DSTXAfterScaler[k] / (double)h_total + (double)p->DSTYAfterScaler[k]) * (double)h_total / pixel_clock_mhz; + if (p->NumberOfActiveSurfaces > 1) { + s->ActiveClockChangeLatencyHidingC = s->ActiveClockChangeLatencyHidingC - (1.0 - 1.0 / (double)p->NumberOfActiveSurfaces) * (double)p->SwathHeightC[k] * (double)h_total / pixel_clock_mhz / v_ratio_c; + } + s->ActiveClockChangeLatencyHiding = math_min2(s->ActiveClockChangeLatencyHidingY, s->ActiveClockChangeLatencyHidingC); + } else { + s->ActiveClockChangeLatencyHiding = s->ActiveClockChangeLatencyHidingY; + } + + s->ActiveDRAMClockChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->DRAMClockChangeWatermark; + s->ActiveFCLKChangeLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->FCLKChangeWatermark; + s->USRRetrainingLatencyMargin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->USRRetrainingWatermark; + s->g6_temp_read_latency_margin[k] = s->ActiveClockChangeLatencyHiding - p->Watermark->temp_read_or_ppt_watermark_us; + + if (p->VActiveLatencyHidingMargin) + p->VActiveLatencyHidingMargin[k] = s->ActiveDRAMClockChangeLatencyMargin[k]; + + if (p->VActiveLatencyHidingUs) + p->VActiveLatencyHidingUs[k] = s->ActiveClockChangeLatencyHiding; + + if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + s->WritebackLatencyHiding = (double)p->WritebackInterfaceBufferSize * 1024.0 + / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height + * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width + / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height * (double)h_total / pixel_clock_mhz) * 4.0); + if (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format == dml2_444_64) { + s->WritebackLatencyHiding = s->WritebackLatencyHiding / 2; + } + s->WritebackDRAMClockChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackDRAMClockChangeWatermark; + + s->WritebackFCLKChangeLatencyMargin = s->WritebackLatencyHiding - p->Watermark->WritebackFCLKChangeWatermark; + + s->ActiveDRAMClockChangeLatencyMargin[k] = math_min2(s->ActiveDRAMClockChangeLatencyMargin[k], s->WritebackDRAMClockChangeLatencyMargin); + s->ActiveFCLKChangeLatencyMargin[k] = math_min2(s->ActiveFCLKChangeLatencyMargin[k], s->WritebackFCLKChangeLatencyMargin); + } + p->MaxActiveDRAMClockChangeLatencySupported[k] = dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k]) ? 0 : (s->ActiveDRAMClockChangeLatencyMargin[k] + p->mmSOCParameters.DRAMClockChangeLatency); + + uclk_pstate_change_strategy = p->display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy; + reserved_vblank_time_us = (double)p->display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns / 1000; + + p->FCLKChangeSupport[k] = dml2_pstate_change_unsupported; + if (s->ActiveFCLKChangeLatencyMargin[k] > 0) + p->FCLKChangeSupport[k] = dml2_pstate_change_vactive; + else if (reserved_vblank_time_us >= p->mmSOCParameters.FCLKChangeLatency) + p->FCLKChangeSupport[k] = dml2_pstate_change_vblank; + + if (p->FCLKChangeSupport[k] == dml2_pstate_change_unsupported) + *p->global_fclk_change_supported = false; + + p->DRAMClockChangeSupport[k] = dml2_pstate_change_unsupported; + if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_auto) { + if (p->display_cfg->overrides.all_streams_blanked || + (s->ActiveDRAMClockChangeLatencyMargin[k] > 0 && reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency)) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank_and_vactive; + else if (s->ActiveDRAMClockChangeLatencyMargin[k] > 0) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_vactive; + else if (reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank; + } else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_vactive && s->ActiveDRAMClockChangeLatencyMargin[k] > 0) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_vactive; + else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_vblank && reserved_vblank_time_us >= p->mmSOCParameters.DRAMClockChangeLatency) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_vblank; + else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_drr) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_drr; + else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_svp) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_mall_svp; + else if (uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame) + p->DRAMClockChangeSupport[k] = dml2_pstate_change_mall_full_frame; + + if (p->DRAMClockChangeSupport[k] == dml2_pstate_change_unsupported) + *p->global_dram_clock_change_supported = false; + + s->dst_y_pstate = (unsigned int)(math_ceil2((p->mmSOCParameters.DRAMClockChangeLatency + p->mmSOCParameters.UrgentLatency) / (h_total / pixel_clock_mhz), 1)); + s->src_y_pstate_l = (unsigned int)(math_ceil2(s->dst_y_pstate * v_ratio, p->SwathHeightY[k])); + s->src_y_ahead_l = (unsigned int)(math_floor2(p->DETBufferSizeY[k] / p->BytePerPixelDETY[k] / p->SwathWidthY[k], p->SwathHeightY[k]) + s->LBLatencyHidingSourceLinesY[k]); + s->sub_vp_lines_l = s->src_y_pstate_l + s->src_y_ahead_l + p->meta_row_height_l[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u\n", __func__, k, p->DETBufferSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightY = %u\n", __func__, k, p->SwathHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, LBLatencyHidingSourceLinesY = %u\n", __func__, k, s->LBLatencyHidingSourceLinesY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_pstate = %u\n", __func__, k, s->dst_y_pstate); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_l = %u\n", __func__, k, s->src_y_pstate_l); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_l = %u\n", __func__, k, s->src_y_ahead_l); + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_l = %u\n", __func__, k, p->meta_row_height_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_l = %u\n", __func__, k, s->sub_vp_lines_l); +#endif + p->SubViewportLinesNeededInMALL[k] = s->sub_vp_lines_l; + + if (p->BytePerPixelDETC[k] > 0) { + s->src_y_pstate_c = (unsigned int)(math_ceil2(s->dst_y_pstate * v_ratio_c, p->SwathHeightC[k])); + s->src_y_ahead_c = (unsigned int)(math_floor2(p->DETBufferSizeC[k] / p->BytePerPixelDETC[k] / p->SwathWidthC[k], p->SwathHeightC[k]) + s->LBLatencyHidingSourceLinesC[k]); + s->sub_vp_lines_c = s->src_y_pstate_c + s->src_y_ahead_c + p->meta_row_height_c[k]; + + if (dml_is_420(p->display_cfg->plane_descriptors[k].pixel_format)) + p->SubViewportLinesNeededInMALL[k] = (unsigned int)(math_max2(s->sub_vp_lines_l, 2 * s->sub_vp_lines_c)); + else + p->SubViewportLinesNeededInMALL[k] = (unsigned int)(math_max2(s->sub_vp_lines_l, s->sub_vp_lines_c)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, meta_row_height_c = %u\n", __func__, k, p->meta_row_height_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_pstate_c = %u\n", __func__, k, s->src_y_pstate_c); + DML_LOG_VERBOSE("DML::%s: k=%u, src_y_ahead_c = %u\n", __func__, k, s->src_y_ahead_c); + DML_LOG_VERBOSE("DML::%s: k=%u, sub_vp_lines_c = %u\n", __func__, k, s->sub_vp_lines_c); +#endif + } + } + + *p->g6_temp_read_support = true; + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if ((!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) && + (s->g6_temp_read_latency_margin[k] < 0)) { + *p->g6_temp_read_support = false; + } + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if ((!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) && ((!FoundCriticalSurface) + || ((s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency) < *p->MaxActiveFCLKChangeLatencySupported))) { + FoundCriticalSurface = true; + *p->MaxActiveFCLKChangeLatencySupported = s->ActiveFCLKChangeLatencyMargin[k] + p->mmSOCParameters.FCLKChangeLatency; + } + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DRAMClockChangeSupport = %u\n", __func__, *p->global_dram_clock_change_supported); + DML_LOG_VERBOSE("DML::%s: FCLKChangeSupport = %u\n", __func__, *p->global_fclk_change_supported); + DML_LOG_VERBOSE("DML::%s: MaxActiveFCLKChangeLatencySupported = %f\n", __func__, *p->MaxActiveFCLKChangeLatencySupported); + DML_LOG_VERBOSE("DML::%s: USRRetrainingSupport = %u\n", __func__, *p->USRRetrainingSupport); +#endif +} + +static void calculate_bytes_to_fetch_required_to_hide_latency( + struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *p) +{ + unsigned int dst_lines_to_hide; + unsigned int src_lines_to_hide_l; + unsigned int src_lines_to_hide_c; + unsigned int plane_index; + unsigned int stream_index; + + for (plane_index = 0; plane_index < p->num_active_planes; plane_index++) { + if (dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[plane_index])) + continue; + + stream_index = p->display_cfg->plane_descriptors[plane_index].stream_index; + + dst_lines_to_hide = (unsigned int)math_ceil(p->latency_to_hide_us / + ((double)p->display_cfg->stream_descriptors[stream_index].timing.h_total / + (double)p->display_cfg->stream_descriptors[stream_index].timing.pixel_clock_khz * 1000.0)); + + src_lines_to_hide_l = (unsigned int)math_ceil2(p->display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane0.v_ratio * dst_lines_to_hide, + p->swath_height_l[plane_index]); + p->bytes_required_l[plane_index] = src_lines_to_hide_l * p->num_of_dpp[plane_index] * p->swath_width_l[plane_index] * p->byte_per_pix_l[plane_index]; + + src_lines_to_hide_c = (unsigned int)math_ceil2(p->display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane1.v_ratio * dst_lines_to_hide, + p->swath_height_c[plane_index]); + p->bytes_required_c[plane_index] = src_lines_to_hide_c * p->num_of_dpp[plane_index] * p->swath_width_c[plane_index] * p->byte_per_pix_c[plane_index]; + + if (p->display_cfg->plane_descriptors[plane_index].surface.dcc.enable && p->mrq_present) { + p->bytes_required_l[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_l / p->meta_row_height_l[plane_index]) * p->meta_row_bytes_per_row_ub_l[plane_index]; + if (p->meta_row_height_c[plane_index]) { + p->bytes_required_c[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_c / p->meta_row_height_c[plane_index]) * p->meta_row_bytes_per_row_ub_c[plane_index]; + } + } + + if (p->display_cfg->gpuvm_enable == true) { + p->bytes_required_l[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_l / p->dpte_row_height_l[plane_index]) * p->dpte_bytes_per_row_l[plane_index]; + if (p->dpte_row_height_c[plane_index]) { + p->bytes_required_c[plane_index] += (unsigned int)math_ceil((double)src_lines_to_hide_c / p->dpte_row_height_c[plane_index]) * p->dpte_bytes_per_row_c[plane_index]; + } + } + } +} + +static noinline_for_stack void calculate_vactive_det_fill_latency( + const struct dml2_display_cfg *display_cfg, + unsigned int num_active_planes, + unsigned int bytes_required_l[], + unsigned int bytes_required_c[], + double dcc_dram_bw_nom_overhead_factor_p0[], + double dcc_dram_bw_nom_overhead_factor_p1[], + double surface_read_bw_l[], + double surface_read_bw_c[], + double (*surface_avg_vactive_required_bw)[dml2_core_internal_bw_max][DML2_MAX_PLANES], + double (*surface_peak_required_bw)[dml2_core_internal_bw_max][DML2_MAX_PLANES], + /* output */ + double vactive_det_fill_delay_us[]) +{ + double effective_excess_bandwidth; + double effective_excess_bandwidth_l; + double effective_excess_bandwidth_c; + double adj_factor; + unsigned int plane_index; + unsigned int soc_state; + unsigned int bw_type; + + for (plane_index = 0; plane_index < num_active_planes; plane_index++) { + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_index])) + continue; + + vactive_det_fill_delay_us[plane_index] = 0.0; + for (soc_state = 0; soc_state < dml2_core_internal_soc_state_max; soc_state++) { + for (bw_type = 0; bw_type < dml2_core_internal_bw_max; bw_type++) { + effective_excess_bandwidth = (surface_peak_required_bw[soc_state][bw_type][plane_index] - surface_avg_vactive_required_bw[soc_state][bw_type][plane_index]); + + /* luma */ + adj_factor = bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p0[plane_index] : 1.0; + + effective_excess_bandwidth_l = effective_excess_bandwidth * surface_read_bw_l[plane_index] / (surface_read_bw_l[plane_index] + surface_read_bw_c[plane_index]) / adj_factor; + if (effective_excess_bandwidth_l > 0.0) { + vactive_det_fill_delay_us[plane_index] = math_max2(vactive_det_fill_delay_us[plane_index], bytes_required_l[plane_index] / effective_excess_bandwidth_l); + } + + /* chroma */ + adj_factor = bw_type == dml2_core_internal_bw_dram ? dcc_dram_bw_nom_overhead_factor_p1[plane_index] : 1.0; + + effective_excess_bandwidth_c = effective_excess_bandwidth * surface_read_bw_c[plane_index] / (surface_read_bw_l[plane_index] + surface_read_bw_c[plane_index]) / adj_factor; + if (effective_excess_bandwidth_c > 0.0) { + vactive_det_fill_delay_us[plane_index] = math_max2(vactive_det_fill_delay_us[plane_index], bytes_required_c[plane_index] / effective_excess_bandwidth_c); + } + } + } + } +} + +static void calculate_excess_vactive_bandwidth_required( + const struct dml2_display_cfg *display_cfg, + unsigned int num_active_planes, + unsigned int bytes_required_l[], + unsigned int bytes_required_c[], + /* outputs */ + double excess_vactive_fill_bw_l[], + double excess_vactive_fill_bw_c[]) +{ + unsigned int plane_index; + + for (plane_index = 0; plane_index < num_active_planes; plane_index++) { + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[plane_index])) + continue; + + excess_vactive_fill_bw_l[plane_index] = 0.0; + excess_vactive_fill_bw_c[plane_index] = 0.0; + + if (display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us > 0) { + excess_vactive_fill_bw_l[plane_index] = (double)bytes_required_l[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; + excess_vactive_fill_bw_c[plane_index] = (double)bytes_required_c[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; + } + } +} + +static double uclk_khz_to_dram_bw_mbps(unsigned long uclk_khz, const struct dml2_dram_params *dram_config) +{ + double bw_mbps = 0; + bw_mbps = ((double)uclk_khz * dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock) / 1000.0; + + return bw_mbps; +} + +static double dram_bw_kbps_to_uclk_mhz(unsigned long long bw_kbps, const struct dml2_dram_params *dram_config) +{ + double uclk_mhz = 0; + + uclk_mhz = (double)bw_kbps / (dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock) / 1000.0; + + return uclk_mhz; +} + +static unsigned int get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params) +{ + unsigned int i; + unsigned int index = 0; + + for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { + DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); + + if (i == 0) + index = 0; + else + index = i - 1; + + if (uclk_freq_khz < per_uclk_dpm_params[i].minimum_uclk_khz || + per_uclk_dpm_params[i].minimum_uclk_khz == 0) { + break; + } + } + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); + return index; +} + +static unsigned int get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) +{ + unsigned int i; + bool clk_entry_found = false; + + for (i = 0; i < clk_table->uclk.num_clk_values; i++) { + DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); + + if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { + clk_entry_found = true; + break; + } + } + + if (!clk_entry_found) + DML_ASSERT(clk_entry_found); +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); +#endif + return i; +} + +static unsigned int get_pipe_flip_bytes( + double hostvm_inefficiency_factor, + unsigned int vm_bytes, + unsigned int dpte_row_bytes, + unsigned int meta_row_bytes) +{ + unsigned int flip_bytes = 0; + + flip_bytes += (unsigned int) ((vm_bytes * hostvm_inefficiency_factor) + 2*meta_row_bytes); + flip_bytes += (unsigned int) (2*dpte_row_bytes * hostvm_inefficiency_factor); + + return flip_bytes; +} + +static void calculate_hostvm_inefficiency_factor( + double *HostVMInefficiencyFactor, + double *HostVMInefficiencyFactorPrefetch, + + bool gpuvm_enable, + bool hostvm_enable, + unsigned int remote_iommu_outstanding_translations, + unsigned int max_outstanding_reqs, + double urg_bandwidth_avail_active_pixel_and_vm, + double urg_bandwidth_avail_active_vm_only) +{ + *HostVMInefficiencyFactor = 1; + *HostVMInefficiencyFactorPrefetch = 1; + + if (gpuvm_enable && hostvm_enable) { + *HostVMInefficiencyFactor = urg_bandwidth_avail_active_pixel_and_vm / urg_bandwidth_avail_active_vm_only; + *HostVMInefficiencyFactorPrefetch = *HostVMInefficiencyFactor; + + if ((*HostVMInefficiencyFactorPrefetch < 4) && (remote_iommu_outstanding_translations < max_outstanding_reqs)) + *HostVMInefficiencyFactorPrefetch = 4; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_pixel_and_vm = %f\n", __func__, urg_bandwidth_avail_active_pixel_and_vm); + DML_LOG_VERBOSE("DML::%s: urg_bandwidth_avail_active_vm_only = %f\n", __func__, urg_bandwidth_avail_active_vm_only); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactor = %f\n", __func__, *HostVMInefficiencyFactor); + DML_LOG_VERBOSE("DML::%s: HostVMInefficiencyFactorPrefetch = %f\n", __func__, *HostVMInefficiencyFactorPrefetch); +#endif + } +} + +struct dml2_core_internal_g6_temp_read_blackouts_table { + struct { + unsigned int uclk_khz; + unsigned int blackout_us; + } entries[DML_MAX_CLK_TABLE_SIZE]; +}; + +struct dml2_core_internal_g6_temp_read_blackouts_table core_dcn4_g6_temp_read_blackout_table = { + .entries = { + { + .uclk_khz = 96000, + .blackout_us = 23, + }, + { + .uclk_khz = 435000, + .blackout_us = 10, + }, + { + .uclk_khz = 521000, + .blackout_us = 10, + }, + { + .uclk_khz = 731000, + .blackout_us = 8, + }, + { + .uclk_khz = 822000, + .blackout_us = 8, + }, + { + .uclk_khz = 962000, + .blackout_us = 5, + }, + { + .uclk_khz = 1069000, + .blackout_us = 5, + }, + { + .uclk_khz = 1187000, + .blackout_us = 5, + }, + }, +}; + +static double get_g6_temp_read_blackout_us( + struct dml2_soc_bb *soc, + unsigned int uclk_freq_khz, + unsigned int min_clk_index) +{ + unsigned int i; + unsigned int blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[0].blackout_us; + + if (soc->power_management_parameters.g6_temp_read_blackout_us[0] > 0.0) { + /* overrides are present in the SoC BB */ + return soc->power_management_parameters.g6_temp_read_blackout_us[min_clk_index]; + } + + /* use internal table */ + blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[0].blackout_us; + + for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { + if (uclk_freq_khz < core_dcn4_g6_temp_read_blackout_table.entries[i].uclk_khz || + core_dcn4_g6_temp_read_blackout_table.entries[i].uclk_khz == 0) { + break; + } + + blackout_us = core_dcn4_g6_temp_read_blackout_table.entries[i].blackout_us; + } + + return (double)blackout_us; +} + +static double get_max_urgent_latency_us( + struct dml2_dcn4x_soc_qos_params *dcn4x, + double uclk_freq_mhz, + double FabricClock, + unsigned int min_clk_index) +{ + double latency; + latency = dcn4x->per_uclk_dpm_params[min_clk_index].maximum_latency_when_urgent_uclk_cycles / uclk_freq_mhz + * (1 + dcn4x->umc_max_latency_margin / 100.0) + + dcn4x->mall_overhead_fclk_cycles / FabricClock + + dcn4x->max_round_trip_to_furthest_cs_fclk_cycles / FabricClock + * (1 + dcn4x->fabric_max_transport_latency_margin / 100.0); + return latency; +} + +static void calculate_pstate_keepout_dst_lines( + const struct dml2_display_cfg *display_cfg, + const struct dml2_core_internal_watermarks *watermarks, + unsigned int pstate_keepout_dst_lines[]) +{ + const struct dml2_stream_parameters *stream_descriptor; + unsigned int i; + + for (i = 0; i < display_cfg->num_planes; i++) { + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[i])) { + stream_descriptor = &display_cfg->stream_descriptors[display_cfg->plane_descriptors[i].stream_index]; + + pstate_keepout_dst_lines[i] = + (unsigned int)math_ceil(watermarks->DRAMClockChangeWatermark / ((double)stream_descriptor->timing.h_total * 1000.0 / (double)stream_descriptor->timing.pixel_clock_khz)); + + if (pstate_keepout_dst_lines[i] > stream_descriptor->timing.v_total - 1) { + pstate_keepout_dst_lines[i] = stream_descriptor->timing.v_total - 1; + } + } + } +} + +static noinline_for_stack void dml_core_ms_prefetch_check(struct dml2_core_internal_display_mode_lib *mode_lib, + const struct dml2_display_cfg *display_cfg) +{ + struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; + struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; + struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; + struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; +#ifdef DML_GLOBAL_PREFETCH_CHECK + struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; +#endif + struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; + + double min_return_bw_for_latency; + unsigned int k; + + mode_lib->ms.TimeCalc = 24 / mode_lib->ms.dcfclk_deepsleep; + + calculate_hostvm_inefficiency_factor( + &s->HostVMInefficiencyFactor, + &s->HostVMInefficiencyFactorPrefetch, + + display_cfg->gpuvm_enable, + display_cfg->hostvm_enable, + mode_lib->ip.remote_iommu_outstanding_translations, + mode_lib->soc.max_outstanding_reqs, + mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], + mode_lib->ms.support.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); + + mode_lib->ms.Total3dlutActive = 0; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) + mode_lib->ms.Total3dlutActive = mode_lib->ms.Total3dlutActive + 1; + + // Calculate tdlut schedule related terms + calculate_tdlut_setting_params->dispclk_mhz = mode_lib->ms.RequiredDISPCLK; + calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; + calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; + calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; + calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; + calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + calculate_tdlut_setting_params->tdlut_mpc_width_flag = display_cfg->plane_descriptors[k].tdlut.tdlut_mpc_width_flag; + calculate_tdlut_setting_params->is_gfx11 = dml_get_gfx_version(display_cfg->plane_descriptors[k].surface.tiling); + + // output + calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; + calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; + calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; + calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; + + calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); + } + + min_return_bw_for_latency = mode_lib->ms.support.urg_bandwidth_available_min_latency[dml2_core_internal_soc_state_sys_active]; + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) + s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); + + CalculateExtraLatency( + display_cfg, + mode_lib->ip.rob_buffer_size_kbytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, + s->ReorderingBytes, + mode_lib->ms.DCFCLK, + mode_lib->ms.FabricClock, + mode_lib->ip.pixel_chunk_size_kbytes, + min_return_bw_for_latency, + mode_lib->ms.num_active_planes, + mode_lib->ms.NoOfDPP, + mode_lib->ms.dpte_group_bytes, + s->tdlut_bytes_per_group, + s->HostVMInefficiencyFactor, + s->HostVMInefficiencyFactorPrefetch, + mode_lib->soc.hostvm_min_page_size_kbytes, + mode_lib->soc.qos_parameters.qos_type, + !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), + mode_lib->soc.max_outstanding_reqs, + mode_lib->ms.support.request_size_bytes_luma, + mode_lib->ms.support.request_size_bytes_chroma, + mode_lib->ip.meta_chunk_size_kbytes, + mode_lib->ip.dchub_arb_to_ret_delay, + mode_lib->ms.TripToMemory, + mode_lib->ip.hostvm_mode, + + // output + &mode_lib->ms.ExtraLatency, + &mode_lib->ms.ExtraLatency_sr, + &mode_lib->ms.ExtraLatencyPrefetch); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + s->impacted_dst_y_pre[k] = 0; + + s->recalc_prefetch_schedule = 0; + s->recalc_prefetch_done = 0; + do { + mode_lib->ms.support.PrefetchSupported = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; + + s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->ms.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane0.width, + display_cfg->plane_descriptors[k].composition.viewport.plane0.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->ms.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane1.width, + display_cfg->plane_descriptors[k].composition.viewport.plane1.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; + + mode_lib->ms.TWait[k] = CalculateTWait( + display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, + mode_lib->ms.UrgLatency, + mode_lib->ms.TripToMemory, + !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? + get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx) : 0.0); + + myPipe->Dppclk = mode_lib->ms.RequiredDPPCLK[k]; + myPipe->Dispclk = mode_lib->ms.RequiredDISPCLK; + myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + myPipe->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; + myPipe->DPPPerSurface = mode_lib->ms.NoOfDPP[k]; + myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; + myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; + myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; + myPipe->BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; + myPipe->BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; + myPipe->BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; + myPipe->BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; + myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; + myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; + myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; + myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; + myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; + myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; + myPipe->ODMMode = mode_lib->ms.ODMMode[k]; + myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; + myPipe->BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; + myPipe->BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; + myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: MaximumVStartup = %u\n", __func__, s->MaximumVStartup[k]); +#endif + CalculatePrefetchSchedule_params->display_cfg = display_cfg; + CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; + CalculatePrefetchSchedule_params->myPipe = myPipe; + CalculatePrefetchSchedule_params->DSCDelay = mode_lib->ms.DSCDelay[k]; + CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; + CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; + CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; + CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; + CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; + CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->ms.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; + CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; + CalculatePrefetchSchedule_params->VStartup = s->MaximumVStartup[k]; + CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; + CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; + CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; + CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; + CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; + CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->ms.UrgLatency; + CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->ms.ExtraLatencyPrefetch; + CalculatePrefetchSchedule_params->TCalc = mode_lib->ms.TimeCalc; + CalculatePrefetchSchedule_params->vm_bytes = mode_lib->ms.vm_bytes[k]; + CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY[k]; + CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->ms.PrefillY[k]; + CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC[k]; + CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->ms.PrefillC[k]; + CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC[k]; + CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub[k]; + CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub[k]; + CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->ms.SwathHeightY[k]; + CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->ms.SwathHeightC[k]; + CalculatePrefetchSchedule_params->TWait = mode_lib->ms.TWait[k]; + CalculatePrefetchSchedule_params->Ttrip = mode_lib->ms.TripToMemory; + CalculatePrefetchSchedule_params->Turg = mode_lib->ms.UrgLatency; + CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; + CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; + CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); + CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; + CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; + CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; + CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; + CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->ms.meta_row_bytes[k]; + CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor[k]; + CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->ms.vactive_sw_bw_l[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->ms.vactive_sw_bw_c[k]; + + // output + CalculatePrefetchSchedule_params->DSTXAfterScaler = &s->DSTXAfterScaler[k]; + CalculatePrefetchSchedule_params->DSTYAfterScaler = &s->DSTYAfterScaler[k]; + CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->ms.dst_y_prefetch[k]; + CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->ms.LinesForVM[k]; + CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->ms.LinesForDPTERow[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->ms.VRatioPreY[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &mode_lib->ms.RequiredPrefetchBWMax[k]; + CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; + CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; + CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; + CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->ms.prefetch_vmrow_bw[k]; + CalculatePrefetchSchedule_params->Tdmdl_vm = &s->dummy_single[0]; + CalculatePrefetchSchedule_params->Tdmdl = &s->dummy_single[1]; + CalculatePrefetchSchedule_params->TSetup = &s->dummy_single[2]; + CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; + CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->VUpdateOffsetPix = &s->dummy_integer[0]; + CalculatePrefetchSchedule_params->VUpdateWidthPix = &s->dummy_integer[1]; + CalculatePrefetchSchedule_params->VReadyOffsetPix = &s->dummy_integer[2]; + CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->ms.prefetch_cursor_bw[k]; + CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; + CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; + CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; + CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->prefetch_swath_time_us[k]; + + mode_lib->ms.NoTimeForPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); + + mode_lib->ms.support.PrefetchSupported &= !mode_lib->ms.NoTimeForPrefetch[k]; + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_vm_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_per_row_vblank = %f\n", __func__, k, *CalculatePrefetchSchedule_params->dst_y_per_row_vblank); + } // for k num_planes + + CalculateDCFCLKDeepSleepTdlut( + display_cfg, + mode_lib->ms.num_active_planes, + mode_lib->ms.BytePerPixelY, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.SwathWidthY, + mode_lib->ms.SwathWidthC, + mode_lib->ms.NoOfDPP, + mode_lib->ms.PSCL_FACTOR, + mode_lib->ms.PSCL_FACTOR_CHROMA, + mode_lib->ms.RequiredDPPCLK, + mode_lib->ms.vactive_sw_bw_l, + mode_lib->ms.vactive_sw_bw_c, + mode_lib->soc.return_bus_width_bytes, + mode_lib->ms.RequiredDISPCLK, + s->tdlut_bytes_to_deliver, + s->prefetch_swath_time_us, + + /* Output */ + &mode_lib->ms.dcfclk_deepsleep); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.dst_y_prefetch[k] < 2.0 + || mode_lib->ms.LinesForVM[k] >= 32.0 + || mode_lib->ms.LinesForDPTERow[k] >= 16.0 + || mode_lib->ms.NoTimeForPrefetch[k] == true + || s->DSTYAfterScaler[k] > 8) { + mode_lib->ms.support.PrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d, dst_y_prefetch=%f (should not be < 2)\n", __func__, k, mode_lib->ms.dst_y_prefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, LinesForVM=%f (should not be >= 32)\n", __func__, k, mode_lib->ms.LinesForVM[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, LinesForDPTERow=%f (should not be >= 16)\n", __func__, k, mode_lib->ms.LinesForDPTERow[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DSTYAfterScaler=%d (should be <= 8)\n", __func__, k, s->DSTYAfterScaler[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, NoTimeForPrefetch=%d\n", __func__, k, mode_lib->ms.NoTimeForPrefetch[k]); + } + } + + mode_lib->ms.support.DynamicMetadataSupported = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.NoTimeForDynamicMetadata[k] == true) { + mode_lib->ms.support.DynamicMetadataSupported = false; + } + } + + mode_lib->ms.support.VRatioInPrefetchSupported = true; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.VRatioPreY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || + mode_lib->ms.VRatioPreC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { + mode_lib->ms.support.VRatioInPrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreY = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreY[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: k=%d VRatioPreC = %f (should be <= %f)\n", __func__, k, mode_lib->ms.VRatioPreC[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: VRatioInPrefetchSupported = %u\n", __func__, mode_lib->ms.support.VRatioInPrefetchSupported); + } + } + + mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.VRatioInPrefetchSupported; + + // By default, do not recalc prefetch schedule + s->recalc_prefetch_schedule = 0; + + // Only do urg vs prefetch bandwidth check, flip schedule check, power saving feature support check IF the Prefetch Schedule Check is ok + if (mode_lib->ms.support.PrefetchSupported) { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + // Calculate Urgent burst factor for prefetch +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor (for prefetch)\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreY=%f\n", __func__, k, mode_lib->ms.VRatioPreY[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPreC=%f\n", __func__, k, mode_lib->ms.VRatioPreC[k]); +#endif + CalculateUrgentBurstFactor( + &display_cfg->plane_descriptors[k], + mode_lib->ms.swath_width_luma_ub[k], + mode_lib->ms.swath_width_chroma_ub[k], + mode_lib->ms.SwathHeightY[k], + mode_lib->ms.SwathHeightC[k], + s->line_times[k], + mode_lib->ms.UrgLatency, + mode_lib->ms.VRatioPreY[k], + mode_lib->ms.VRatioPreC[k], + mode_lib->ms.BytePerPixelInDETY[k], + mode_lib->ms.BytePerPixelInDETC[k], + mode_lib->ms.DETBufferSizeY[k], + mode_lib->ms.DETBufferSizeC[k], + /* Output */ + &mode_lib->ms.UrgentBurstFactorLumaPre[k], + &mode_lib->ms.UrgentBurstFactorChromaPre[k], + &mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); + } + + // Calculate urgent bandwidth required, both urg and non urg peak bandwidth + // assume flip bw is 0 at this point + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + mode_lib->ms.final_flip_bw[k] = 0; + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->ms.support.urg_vactive_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->ms.support.urg_bandwidth_required_qual; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = mode_lib->ms.surface_avg_vactive_required_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 0; + calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + // Check urg peak bandwidth against available urg bw + // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) + check_urgent_bandwidth_support( + &s->dummy_single[0], // double* frac_urg_bandwidth + &s->dummy_single[1], // double* frac_urg_bandwidth_mall + &mode_lib->ms.support.UrgVactiveBandwidthSupport, + &mode_lib->ms.support.PrefetchBandwidthSupported, + + mode_lib->soc.mall_allocated_for_dcn_mbytes, + mode_lib->ms.support.non_urg_bandwidth_required, + mode_lib->ms.support.urg_vactive_bandwidth_required, + mode_lib->ms.support.urg_bandwidth_required, + mode_lib->ms.support.urg_bandwidth_available); + + mode_lib->ms.support.PrefetchSupported &= mode_lib->ms.support.PrefetchBandwidthSupported; + DML_LOG_VERBOSE("DML::%s: PrefetchBandwidthSupported=%0d\n", __func__, mode_lib->ms.support.PrefetchBandwidthSupported); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]) { + mode_lib->ms.support.PrefetchSupported = false; + DML_LOG_VERBOSE("DML::%s: k=%d, NotEnoughUrgentLatencyHidingPre=%d\n", __func__, k, mode_lib->ms.NotEnoughUrgentLatencyHidingPre[k]); + } + } + +#ifdef DML_GLOBAL_PREFETCH_CHECK + if (mode_lib->ms.support.PrefetchSupported && mode_lib->ms.num_active_planes > 1 && s->recalc_prefetch_done == 0) { + CheckGlobalPrefetchAdmissibility_params->num_active_planes = mode_lib->ms.num_active_planes; + CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; + CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->ms.SwathHeightY; + CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->ms.SwathHeightC; + CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->ms.CompressedBufferSizeInkByte; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->ms.DETBufferSizeY; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->ms.DETBufferSizeC; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; + CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; + CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = s->Tpre_rounded; + CheckGlobalPrefetchAdmissibility_params->Tpre_oto = s->Tpre_oto; + CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->ms.support.urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; + CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->ms.dst_y_prefetch; + if (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps < 10 * 1024) + CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = 10 * 1024; + + CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = (CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps / (double) mode_lib->soc.return_bus_width_bytes) / + ((double)mode_lib->soc.qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100.0); + + // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible + CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->recalc_prefetch_schedule; + CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; + mode_lib->ms.support.PrefetchSupported = CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); + s->recalc_prefetch_done = 1; + s->recalc_prefetch_schedule = 1; + } +#endif + } // prefetch schedule ok, do urg bw and flip schedule + } while (s->recalc_prefetch_schedule); + + // Flip Schedule + // Both prefetch schedule and BW okay + if (mode_lib->ms.support.PrefetchSupported == true) { + mode_lib->ms.BandwidthAvailableForImmediateFlip = + get_bandwidth_available_for_immediate_flip( + dml2_core_internal_soc_state_sys_active, + mode_lib->ms.support.urg_bandwidth_required_qual, // no flip + mode_lib->ms.support.urg_bandwidth_available); + + mode_lib->ms.TotImmediateFlipBytes = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->plane_descriptors[k].immediate_flip) { + s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes( + s->HostVMInefficiencyFactor, + mode_lib->ms.vm_bytes[k], + mode_lib->ms.DPTEBytesPerRow[k], + mode_lib->ms.meta_row_bytes[k]); + } else { + s->per_pipe_flip_bytes[k] = 0; + } + mode_lib->ms.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->ms.NoOfDPP[k]; + + } + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + CalculateFlipSchedule( + &mode_lib->scratch, + display_cfg->plane_descriptors[k].immediate_flip, + 1, // use_lb_flip_bw + s->HostVMInefficiencyFactor, + s->Tvm_trips_flip[k], + s->Tr0_trips_flip[k], + s->Tvm_trips_flip_rounded[k], + s->Tr0_trips_flip_rounded[k], + display_cfg->gpuvm_enable, + mode_lib->ms.vm_bytes[k], + mode_lib->ms.DPTEBytesPerRow[k], + mode_lib->ms.BandwidthAvailableForImmediateFlip, + mode_lib->ms.TotImmediateFlipBytes, + display_cfg->plane_descriptors[k].pixel_format, + (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->ms.Tno_bw_flip[k], + mode_lib->ms.dpte_row_height[k], + mode_lib->ms.dpte_row_height_chroma[k], + mode_lib->ms.use_one_row_for_frame_flip[k], + mode_lib->ip.max_flip_time_us, + mode_lib->ip.max_flip_time_lines, + s->per_pipe_flip_bytes[k], + mode_lib->ms.meta_row_bytes[k], + s->meta_row_height_luma[k], + s->meta_row_height_chroma[k], + mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, + + /* Output */ + &mode_lib->ms.dst_y_per_vm_flip[k], + &mode_lib->ms.dst_y_per_row_flip[k], + &mode_lib->ms.final_flip_bw[k], + &mode_lib->ms.ImmediateFlipSupportedForPipe[k]); + } + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->ms.support.urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->ms.support.non_urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = mode_lib->ms.surface_peak_required_bw; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 1; + calculate_peak_bandwidth_params->num_active_planes = mode_lib->ms.num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->ms.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->ms.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->ms.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->ms.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = mode_lib->ms.RequiredPrefetchBWMax; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->ms.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->ms.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->ms.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->flip_bw = mode_lib->ms.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->ms.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->ms.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->ms.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->ms.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->ms.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->ms.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + calculate_immediate_flip_bandwidth_support( + &s->dummy_single[0], // double* frac_urg_bandwidth_flip + &mode_lib->ms.support.ImmediateFlipSupport, + + dml2_core_internal_soc_state_sys_active, + mode_lib->ms.support.urg_bandwidth_required_flip, + mode_lib->ms.support.non_urg_bandwidth_required_flip, + mode_lib->ms.support.urg_bandwidth_available); + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].immediate_flip == true && mode_lib->ms.ImmediateFlipSupportedForPipe[k] == false) + mode_lib->ms.support.ImmediateFlipSupport = false; + } + + } else { // if prefetch not support, assume iflip is not supported too + mode_lib->ms.support.ImmediateFlipSupport = false; + } + + s->mSOCParameters.UrgentLatency = mode_lib->ms.UrgLatency; + s->mSOCParameters.ExtraLatency = mode_lib->ms.ExtraLatency; + s->mSOCParameters.ExtraLatency_sr = mode_lib->ms.ExtraLatency_sr; + s->mSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; + s->mSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + s->mSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; + s->mSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; + s->mSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; + s->mSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; + s->mSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; + s->mSOCParameters.USRRetrainingLatency = 0; + s->mSOCParameters.SMNLatency = 0; + s->mSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->ms.uclk_freq_mhz * 1000), mode_lib->ms.state_idx); + s->mSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->ms.uclk_freq_mhz, mode_lib->ms.FabricClock, mode_lib->ms.state_idx); + s->mSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->ms.FabricClock; + s->mSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; + + CalculateWatermarks_params->display_cfg = display_cfg; + CalculateWatermarks_params->USRRetrainingRequired = false; + CalculateWatermarks_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; + CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; + CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; + CalculateWatermarks_params->DCFCLK = mode_lib->ms.DCFCLK; + CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; + CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; + CalculateWatermarks_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateWatermarks_params->mmSOCParameters = s->mSOCParameters; + CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; + CalculateWatermarks_params->SOCCLK = mode_lib->ms.SOCCLK; + CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->ms.dcfclk_deepsleep; + CalculateWatermarks_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; + CalculateWatermarks_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; + CalculateWatermarks_params->SwathHeightY = mode_lib->ms.SwathHeightY; + CalculateWatermarks_params->SwathHeightC = mode_lib->ms.SwathHeightC; + CalculateWatermarks_params->SwathWidthY = mode_lib->ms.SwathWidthY; + CalculateWatermarks_params->SwathWidthC = mode_lib->ms.SwathWidthC; + CalculateWatermarks_params->DPPPerSurface = mode_lib->ms.NoOfDPP; + CalculateWatermarks_params->BytePerPixelDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateWatermarks_params->BytePerPixelDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateWatermarks_params->DSTXAfterScaler = s->DSTXAfterScaler; + CalculateWatermarks_params->DSTYAfterScaler = s->DSTYAfterScaler; + CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->ms.UnboundedRequestEnabled; + CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->ms.CompressedBufferSizeInkByte; + CalculateWatermarks_params->meta_row_height_l = s->meta_row_height_luma; + CalculateWatermarks_params->meta_row_height_c = s->meta_row_height_chroma; + + // Output + CalculateWatermarks_params->Watermark = &mode_lib->ms.support.watermarks; // Watermarks *Watermark + CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->ms.support.DRAMClockChangeSupport; + CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->ms.support.global_dram_clock_change_supported; + CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = &s->dummy_single_array[0]; // double *MaxActiveDRAMClockChangeLatencySupported[] + CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->ms.SubViewportLinesNeededInMALL; // unsigned int SubViewportLinesNeededInMALL[] + CalculateWatermarks_params->FCLKChangeSupport = mode_lib->ms.support.FCLKChangeSupport; + CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->ms.support.global_fclk_change_supported; + CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &s->dummy_single[0]; // double *MaxActiveFCLKChangeLatencySupported + CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->ms.support.USRRetrainingSupport; + CalculateWatermarks_params->g6_temp_read_support = &mode_lib->ms.support.g6_temp_read_support; + CalculateWatermarks_params->VActiveLatencyHidingMargin = mode_lib->ms.VActiveLatencyHidingMargin; + CalculateWatermarks_params->VActiveLatencyHidingUs = mode_lib->ms.VActiveLatencyHidingUs; + + CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); + + calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->ms.support.watermarks, s->dummy_integer_array[0]); + DML_LOG_VERBOSE("DML::%s: Done prefetch calculation\n", __func__); + +} + + +static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params) +{ + struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; + const struct dml2_display_cfg *display_cfg = in_out_params->in_display_cfg; + const struct dml2_mcg_min_clock_table *min_clk_table = in_out_params->min_clk_table; + + double outstanding_latency_us = 0; + + struct dml2_core_calcs_mode_support_locals *s = &mode_lib->scratch.dml_core_mode_support_locals; + struct dml2_core_calcs_CalculateVMRowAndSwath_params *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; + struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; + struct dml2_core_calcs_calculate_mcache_setting_params *calculate_mcache_setting_params = &mode_lib->scratch.calculate_mcache_setting_params; + struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *calculate_bytes_to_fetch_required_to_hide_latency_params = &mode_lib->scratch.calculate_bytes_to_fetch_required_to_hide_latency_params; + unsigned int k, m, n; + + memset(&mode_lib->scratch, 0, sizeof(struct dml2_core_internal_scratch)); + memset(&mode_lib->ms, 0, sizeof(struct dml2_core_internal_mode_support)); + + mode_lib->ms.num_active_planes = display_cfg->num_planes; + get_stream_output_bpp(s->OutputBpp, display_cfg); + + mode_lib->ms.state_idx = in_out_params->min_clk_index; + mode_lib->ms.SOCCLK = ((double)mode_lib->soc.clk_table.socclk.clk_values_khz[0] / 1000); + mode_lib->ms.DCFCLK = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_dcfclk_khz / 1000); + mode_lib->ms.FabricClock = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz / 1000); + mode_lib->ms.MaxDCFCLK = (double)min_clk_table->max_clocks_khz.dcfclk / 1000; + mode_lib->ms.MaxFabricClock = (double)min_clk_table->max_clocks_khz.fclk / 1000; + mode_lib->ms.max_dispclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dispclk / 1000; + mode_lib->ms.max_dscclk_freq_mhz = (double)min_clk_table->max_clocks_khz.dscclk / 1000; + mode_lib->ms.max_dppclk_freq_mhz = (double)min_clk_table->max_ss_clocks_khz.dppclk / 1000; + mode_lib->ms.uclk_freq_mhz = dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config); + mode_lib->ms.dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps / 1000); + mode_lib->ms.max_dram_bw_mbps = ((double)min_clk_table->dram_bw_table.entries[min_clk_table->dram_bw_table.num_entries - 1].pre_derate_dram_bw_kbps / 1000); + mode_lib->ms.qos_param_index = get_qos_param_index((unsigned int) (mode_lib->ms.uclk_freq_mhz * 1000.0), mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params); + mode_lib->ms.active_min_uclk_dpm_index = get_active_min_uclk_dpm_index((unsigned int) (mode_lib->ms.uclk_freq_mhz * 1000.0), &mode_lib->soc.clk_table); + +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); + DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, mode_lib->ms.num_active_planes); + DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: qos_param_index = %0d\n", __func__, mode_lib->ms.qos_param_index); + DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, mode_lib->ms.SOCCLK); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->ms.dram_bw_mbps); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); + DML_LOG_VERBOSE("DML::%s: MaxDCFCLK = %f\n", __func__, mode_lib->ms.MaxDCFCLK); + DML_LOG_VERBOSE("DML::%s: max_dispclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dispclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: max_dscclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dscclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: max_dppclk_freq_mhz = %f\n", __func__, mode_lib->ms.max_dppclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: MaxFabricClock = %f\n", __func__, mode_lib->ms.MaxFabricClock); + DML_LOG_VERBOSE("DML::%s: ip.compressed_buffer_segment_size_in_kbytes = %u\n", __func__, mode_lib->ip.compressed_buffer_segment_size_in_kbytes); + DML_LOG_VERBOSE("DML::%s: ip.dcn_mrq_present = %u\n", __func__, mode_lib->ip.dcn_mrq_present); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); +#endif + + CalculateMaxDETAndMinCompressedBufferSize( + mode_lib->ip.config_return_buffer_size_in_kbytes, + mode_lib->ip.config_return_buffer_segment_size_in_kbytes, + mode_lib->ip.rob_buffer_size_kbytes, + mode_lib->ip.max_num_dpp, + display_cfg->overrides.hw.force_nom_det_size_kbytes.enable, + display_cfg->overrides.hw.force_nom_det_size_kbytes.value, + mode_lib->ip.dcn_mrq_present, + + /* Output */ + &mode_lib->ms.MaxTotalDETInKByte, + &mode_lib->ms.NomDETInKByte, + &mode_lib->ms.MinCompressedBufferSizeInKByte); + + PixelClockAdjustmentForProgressiveToInterlaceUnit(display_cfg, mode_lib->ip.ptoi_supported, s->PixelClockBackEnd); + + /*MODE SUPPORT, VOLTAGE STATE AND SOC CONFIGURATION*/ + + /*Scale Ratio, taps Support Check*/ + mode_lib->ms.support.ScaleRatioAndTapsSupport = true; + // Many core tests are still setting scaling parameters "incorrectly" + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].composition.scaler_info.enabled == false + && (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio != 1.0 + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps != 1.0 + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio != 1.0 + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps != 1.0)) { + mode_lib->ms.support.ScaleRatioAndTapsSupport = false; + } else if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps < 1.0 || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps > 8.0 + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps < 1.0 || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps > 8.0 + || (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps > 1.0 && (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps % 2) == 1) + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio > mode_lib->ip.max_hscl_ratio + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio > mode_lib->ip.max_vscl_ratio + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps + || display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps + || (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) + && (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps < 1 || display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps > 8 || + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps < 1 || display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps > 8 || + (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps > 1 && display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps % 2 == 1) || + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio > mode_lib->ip.max_hscl_ratio || + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio > mode_lib->ip.max_vscl_ratio || + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps || + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio > display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps))) { + mode_lib->ms.support.ScaleRatioAndTapsSupport = false; + } + } + + /*Source Format, Pixel Format and Scan Support Check*/ + mode_lib->ms.support.SourceFormatPixelAndScanSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear && dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { + mode_lib->ms.support.SourceFormatPixelAndScanSupport = false; + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateBytePerPixelAndBlockSizes( + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->plane_descriptors[k].surface.tiling, + display_cfg->plane_descriptors[k].surface.plane0.pitch, + display_cfg->plane_descriptors[k].surface.plane1.pitch, + + /* Output */ + &mode_lib->ms.BytePerPixelY[k], + &mode_lib->ms.BytePerPixelC[k], + &mode_lib->ms.BytePerPixelInDETY[k], + &mode_lib->ms.BytePerPixelInDETC[k], + &mode_lib->ms.Read256BlockHeightY[k], + &mode_lib->ms.Read256BlockHeightC[k], + &mode_lib->ms.Read256BlockWidthY[k], + &mode_lib->ms.Read256BlockWidthC[k], + &mode_lib->ms.MacroTileHeightY[k], + &mode_lib->ms.MacroTileHeightC[k], + &mode_lib->ms.MacroTileWidthY[k], + &mode_lib->ms.MacroTileWidthC[k], + &mode_lib->ms.surf_linear128_l[k], + &mode_lib->ms.surf_linear128_c[k]); + } + + /*Bandwidth Support Check*/ + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle)) { + mode_lib->ms.SwathWidthYSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; + mode_lib->ms.SwathWidthCSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; + } else { + mode_lib->ms.SwathWidthYSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + mode_lib->ms.SwathWidthCSingleDPP[k] = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + mode_lib->ms.vactive_sw_bw_l[k] = mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + mode_lib->ms.vactive_sw_bw_c[k] = mode_lib->ms.SwathWidthCSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + + mode_lib->ms.cursor_bw[k] = display_cfg->plane_descriptors[k].cursor.num_cursors * display_cfg->plane_descriptors[k].cursor.cursor_width * + display_cfg->plane_descriptors[k].cursor.cursor_bpp / 8.0 / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthLuma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] * math_ceil2(mode_lib->ms.BytePerPixelInDETY[k], 1.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, old_ReadBandwidthChroma = %f\n", __func__, k, mode_lib->ms.SwathWidthYSingleDPP[k] / 2 * math_ceil2(mode_lib->ms.BytePerPixelInDETC[k], 2.0) / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio / 2.0); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->ms.vactive_sw_bw_c[k]); +#endif + } + + // Writeback bandwidth + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format == dml2_444_64) { + mode_lib->ms.WriteBandwidth[k][0] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height + * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width + / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height + * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total + / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * 8.0; + } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + mode_lib->ms.WriteBandwidth[k][0] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height + * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width + / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height + * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total + / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * 4.0; + } else { + mode_lib->ms.WriteBandwidth[k][0] = 0.0; + } + } + + /*Writeback Latency support check*/ + mode_lib->ms.support.WritebackLatencySupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0 && + (mode_lib->ms.WriteBandwidth[k][0] > mode_lib->ip.writeback_interface_buffer_size_kbytes * 1024 / ((double)mode_lib->soc.qos_parameters.writeback.base_latency_us))) { + mode_lib->ms.support.WritebackLatencySupport = false; + } + } + + + /* Writeback Scale Ratio and Taps Support Check */ + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = true; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio > mode_lib->ip.writeback_max_hscl_ratio + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio > mode_lib->ip.writeback_max_vscl_ratio + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio < mode_lib->ip.writeback_min_hscl_ratio + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio < mode_lib->ip.writeback_min_vscl_ratio + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps > (unsigned int) mode_lib->ip.writeback_max_hscl_taps + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps > (unsigned int) mode_lib->ip.writeback_max_vscl_taps + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio > (unsigned int)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps + || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio > (unsigned int)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps + || (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps > 2.0 && ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps % 2) == 1))) { + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; + } + if (2.0 * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height * (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps - 1) * 57 > mode_lib->ip.writeback_line_buffer_buffer_size) { + mode_lib->ms.support.WritebackScaleRatioAndTapsSupport = false; + } + } + } + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + CalculateSinglePipeDPPCLKAndSCLThroughput( + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->ip.max_dchub_pscl_bw_pix_per_clk, + mode_lib->ip.max_pscl_lb_bw_pix_per_clk, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps, + /* Output */ + &mode_lib->ms.PSCL_FACTOR[k], + &mode_lib->ms.PSCL_FACTOR_CHROMA[k], + &mode_lib->ms.MinDPPCLKUsingSingleDPP[k]); + } + + // Max Viewport Size support + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->plane_descriptors[k].surface.tiling == dml2_sw_linear) { + s->MaximumSwathWidthSupportLuma = 15360; + } else if (!dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelC[k] > 0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { // horz video + s->MaximumSwathWidthSupportLuma = 7680 + 16; + } else if (dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelC[k] > 0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { // vert video + s->MaximumSwathWidthSupportLuma = 4320 + 16; + } else if (display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { // rgbe + alpha + s->MaximumSwathWidthSupportLuma = 5120 + 16; + } else if (dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle) && mode_lib->ms.BytePerPixelY[k] == 8 && display_cfg->plane_descriptors[k].surface.dcc.enable == true) { // vert 64bpp + s->MaximumSwathWidthSupportLuma = 3072 + 16; + } else { + s->MaximumSwathWidthSupportLuma = 6144 + 16; + } + + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format)) { + s->MaximumSwathWidthSupportChroma = (unsigned int)(s->MaximumSwathWidthSupportLuma / 2.0); + } else { + s->MaximumSwathWidthSupportChroma = s->MaximumSwathWidthSupportLuma; + } + + unsigned lb_buffer_size_bits_luma = mode_lib->ip.line_buffer_size_bits; + unsigned lb_buffer_size_bits_chroma = mode_lib->ip.line_buffer_size_bits; + +/* +#if defined(DV_BUILD) + // Assume a memory config setting of 3 in 420 mode or get a new ip parameter that reflects the programming. + if (mode_lib->ms.BytePerPixelC[k] != 0.0 && display_cfg->plane_descriptors[k].pixel_format != dml2_rgbe_alpha) { + lb_buffer_size_bits_luma = 34620 * 57; + lb_buffer_size_bits_chroma = 13560 * 57; + } +#endif +*/ + mode_lib->ms.MaximumSwathWidthInLineBufferLuma = lb_buffer_size_bits_luma * math_max2(display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, 1.0) / 57 / + (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps + math_max2(math_ceil2(display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, 1.0) - 2, 0.0)); + if (mode_lib->ms.BytePerPixelC[k] == 0.0) { + mode_lib->ms.MaximumSwathWidthInLineBufferChroma = 0; + } else { + mode_lib->ms.MaximumSwathWidthInLineBufferChroma = lb_buffer_size_bits_chroma * math_max2(display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, 1.0) / 57 / + (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps + math_max2(math_ceil2(display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, 1.0) - 2, 0.0)); + } + + mode_lib->ms.MaximumSwathWidthLuma[k] = math_min2(s->MaximumSwathWidthSupportLuma, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); + mode_lib->ms.MaximumSwathWidthChroma[k] = math_min2(s->MaximumSwathWidthSupportChroma, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); + + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportLuma=%u\n", __func__, k, s->MaximumSwathWidthSupportLuma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferLuma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferLuma); + + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthSupportChroma=%u\n", __func__, k, s->MaximumSwathWidthSupportChroma); + DML_LOG_VERBOSE("DML::%s: k=%u MaximumSwathWidthInLineBufferChroma=%f\n", __func__, k, mode_lib->ms.MaximumSwathWidthInLineBufferChroma); + } + + /* Cursor Support Check */ + mode_lib->ms.support.CursorSupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { + if (display_cfg->plane_descriptors[k].cursor.cursor_bpp == 64 && mode_lib->ip.cursor_64bpp_support == false) + mode_lib->ms.support.CursorSupport = false; + } + } + + /* Valid Pitch Check */ + mode_lib->ms.support.PitchSupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + + // data pitch + unsigned int alignment_l = mode_lib->ms.MacroTileWidthY[k]; + + if (mode_lib->ms.surf_linear128_l[k]) + alignment_l = alignment_l / 2; + + mode_lib->ms.support.AlignedYPitch[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.plane0.pitch, display_cfg->plane_descriptors[k].surface.plane0.width), alignment_l); + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { + unsigned int alignment_c = mode_lib->ms.MacroTileWidthC[k]; + + if (mode_lib->ms.surf_linear128_c[k]) + alignment_c = alignment_c / 2; + mode_lib->ms.support.AlignedCPitch[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.plane1.pitch, display_cfg->plane_descriptors[k].surface.plane1.width), alignment_c); + } else { + mode_lib->ms.support.AlignedCPitch[k] = display_cfg->plane_descriptors[k].surface.plane1.pitch; + } + + if (mode_lib->ms.support.AlignedYPitch[k] > display_cfg->plane_descriptors[k].surface.plane0.pitch || + mode_lib->ms.support.AlignedCPitch[k] > display_cfg->plane_descriptors[k].surface.plane1.pitch) { + mode_lib->ms.support.PitchSupport = false; +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: k=%u AlignedYPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedYPitch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u PitchY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.pitch); + DML_LOG_VERBOSE("DML::%s: k=%u AlignedCPitch = %d\n", __func__, k, mode_lib->ms.support.AlignedCPitch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u PitchC = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane1.pitch); + DML_LOG_VERBOSE("DML::%s: k=%u PitchSupport = %d\n", __func__, k, mode_lib->ms.support.PitchSupport); +#endif + } + + // meta pitch + if (mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable) { + mode_lib->ms.support.AlignedDCCMetaPitchY[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch, + display_cfg->plane_descriptors[k].surface.plane0.width), 64.0 * mode_lib->ms.Read256BlockWidthY[k]); + + if (mode_lib->ms.support.AlignedDCCMetaPitchY[k] > display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch) + mode_lib->ms.support.PitchSupport = false; + + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { + mode_lib->ms.support.AlignedDCCMetaPitchC[k] = (unsigned int)math_ceil2(math_max2(display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch, + display_cfg->plane_descriptors[k].surface.plane1.width), 64.0 * mode_lib->ms.Read256BlockWidthC[k]); + + if (mode_lib->ms.support.AlignedDCCMetaPitchC[k] > display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch) + mode_lib->ms.support.PitchSupport = false; + } + } else { + mode_lib->ms.support.AlignedDCCMetaPitchY[k] = 0; + mode_lib->ms.support.AlignedDCCMetaPitchC[k] = 0; + } + } + + mode_lib->ms.support.ViewportExceedsSurface = false; + if (!display_cfg->overrides.hw.surface_viewport_size_check_disable) { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + if (display_cfg->plane_descriptors[k].composition.viewport.plane0.width > display_cfg->plane_descriptors[k].surface.plane0.width || + display_cfg->plane_descriptors[k].composition.viewport.plane0.height > display_cfg->plane_descriptors[k].surface.plane0.height) { + mode_lib->ms.support.ViewportExceedsSurface = true; +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: k=%u ViewportWidth = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u SurfaceWidthY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.width); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportHeight = %ld\n", __func__, k, display_cfg->plane_descriptors[k].composition.viewport.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u SurfaceHeightY = %ld\n", __func__, k, display_cfg->plane_descriptors[k].surface.plane0.height); + DML_LOG_VERBOSE("DML::%s: k=%u ViewportExceedsSurface = %d\n", __func__, k, mode_lib->ms.support.ViewportExceedsSurface); +#endif + } + if (dml_is_420(display_cfg->plane_descriptors[k].pixel_format) || display_cfg->plane_descriptors[k].pixel_format == dml2_rgbe_alpha) { + if (display_cfg->plane_descriptors[k].composition.viewport.plane1.width > display_cfg->plane_descriptors[k].surface.plane1.width || + display_cfg->plane_descriptors[k].composition.viewport.plane1.height > display_cfg->plane_descriptors[k].surface.plane1.height) { + mode_lib->ms.support.ViewportExceedsSurface = true; + } + } + } + } + + CalculateSwathAndDETConfiguration_params->display_cfg = display_cfg; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ip.config_return_buffer_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = mode_lib->ms.MaxTotalDETInKByte; + CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = mode_lib->ms.MinCompressedBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = 1; + CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateSwathAndDETConfiguration_params->nomDETInKByte = mode_lib->ms.NomDETInKByte; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ip.config_return_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByte = mode_lib->ip.compressed_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->ms.vactive_sw_bw_l; + CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->ms.vactive_sw_bw_c; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = mode_lib->ms.MaximumSwathWidthLuma; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = mode_lib->ms.MaximumSwathWidthChroma; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->ms.Read256BlockHeightY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->ms.Read256BlockHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->ms.Read256BlockWidthY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->ms.Read256BlockWidthC; + CalculateSwathAndDETConfiguration_params->surf_linear128_l = mode_lib->ms.surf_linear128_l; + CalculateSwathAndDETConfiguration_params->surf_linear128_c = mode_lib->ms.surf_linear128_c; + CalculateSwathAndDETConfiguration_params->ODMMode = s->dummy_odm_mode; + CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->ms.BytePerPixelY; + CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->ms.BytePerPixelC; + CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->ms.BytePerPixelInDETY; + CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->ms.BytePerPixelInDETC; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = s->dummy_integer_array[2]; + CalculateSwathAndDETConfiguration_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + // output + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = s->dummy_integer_array[0]; + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = s->dummy_integer_array[1]; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_integer_array[3]; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_integer_array[4]; + CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_integer_array[5]; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_integer_array[6]; + CalculateSwathAndDETConfiguration_params->SwathHeightY = s->dummy_integer_array[7]; + CalculateSwathAndDETConfiguration_params->SwathHeightC = s->dummy_integer_array[8]; + CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = s->dummy_integer_array[26]; + CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = s->dummy_integer_array[27]; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = s->dummy_integer_array[9]; + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = s->dummy_integer_array[10]; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = s->dummy_integer_array[11]; + CalculateSwathAndDETConfiguration_params->full_swath_bytes_l = s->full_swath_bytes_l; + CalculateSwathAndDETConfiguration_params->full_swath_bytes_c = s->full_swath_bytes_c; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &s->dummy_boolean[0]; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &s->dummy_integer[1]; + CalculateSwathAndDETConfiguration_params->hw_debug5 = &s->dummy_boolean[2]; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &s->dummy_integer[0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = mode_lib->ms.SingleDPPViewportSizeSupportPerSurface; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[1]; + + // This calls is just to find out if there is enough DET space to support full vp in 1 pipe. + CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); + + mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; + mode_lib->ms.support.TotalAvailablePipesSupport = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + /*Number Of DSC Slices*/ + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable_if_necessary) { + + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices != 0) + mode_lib->ms.support.NumberOfDSCSlices[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices; + else { + if (s->PixelClockBackEnd[k] > 4800) { + mode_lib->ms.support.NumberOfDSCSlices[k] = (unsigned int)(math_ceil2(s->PixelClockBackEnd[k] / 600, 4)); + } else if (s->PixelClockBackEnd[k] > 2400) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 8; + } else if (s->PixelClockBackEnd[k] > 1200) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 4; + } else if (s->PixelClockBackEnd[k] > 340) { + mode_lib->ms.support.NumberOfDSCSlices[k] = 2; + } else { + mode_lib->ms.support.NumberOfDSCSlices[k] = 1; + } + } + } else { + mode_lib->ms.support.NumberOfDSCSlices[k] = 0; + } + + CalculateODMMode( + mode_lib->ip.maximum_pixels_per_line_per_dsc_unit, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode, + mode_lib->ms.max_dispclk_freq_mhz, + false, // DSCEnable + mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, + mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + mode_lib->ms.support.NumberOfDSCSlices[k], + + /* Output */ + &s->TotalAvailablePipesSupportNoDSC, + &s->NumberOfDPPNoDSC, + &s->ODMModeNoDSC, + &s->RequiredDISPCLKPerSurfaceNoDSC); + + CalculateODMMode( + mode_lib->ip.maximum_pixels_per_line_per_dsc_unit, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode, + mode_lib->ms.max_dispclk_freq_mhz, + true, // DSCEnable + mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, + mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + mode_lib->ms.support.NumberOfDSCSlices[k], + + /* Output */ + &s->TotalAvailablePipesSupportDSC, + &s->NumberOfDPPDSC, + &s->ODMModeDSC, + &s->RequiredDISPCLKPerSurfaceDSC); + + CalculateOutputLink( + &mode_lib->scratch, + ((double)mode_lib->soc.clk_table.phyclk.clk_values_khz[0] / 1000), + ((double)mode_lib->soc.clk_table.phyclk_d18.clk_values_khz[0] / 1000), + ((double)mode_lib->soc.clk_table.phyclk_d32.clk_values_khz[0] / 1000), + mode_lib->soc.phy_downspread_percent, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + s->PixelClockBackEnd[k], + s->OutputBpp[k], + mode_lib->ip.maximum_dsc_bits_per_component, + mode_lib->ms.support.NumberOfDSCSlices[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_rate, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_layout, + s->ODMModeNoDSC, + s->ODMModeDSC, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate, + + /* Output */ + &mode_lib->ms.RequiresDSC[k], + &mode_lib->ms.RequiresFEC[k], + &mode_lib->ms.OutputBpp[k], + &mode_lib->ms.OutputType[k], + &mode_lib->ms.OutputRate[k], + &mode_lib->ms.RequiredSlots[k]); + + if (s->OutputBpp[k] == 0.0) { + s->OutputBpp[k] = mode_lib->ms.OutputBpp[k]; + } + + if (mode_lib->ms.RequiresDSC[k] == false) { + mode_lib->ms.ODMMode[k] = s->ODMModeNoDSC; + mode_lib->ms.RequiredDISPCLKPerSurface[k] = s->RequiredDISPCLKPerSurfaceNoDSC; + if (!s->TotalAvailablePipesSupportNoDSC) + mode_lib->ms.support.TotalAvailablePipesSupport = false; + mode_lib->ms.TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP + s->NumberOfDPPNoDSC; + } else { + mode_lib->ms.ODMMode[k] = s->ODMModeDSC; + mode_lib->ms.RequiredDISPCLKPerSurface[k] = s->RequiredDISPCLKPerSurfaceDSC; + if (!s->TotalAvailablePipesSupportDSC) + mode_lib->ms.support.TotalAvailablePipesSupport = false; + mode_lib->ms.TotalNumberOfActiveDPP = mode_lib->ms.TotalNumberOfActiveDPP + s->NumberOfDPPDSC; + } +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: k=%d RequiresDSC = %d\n", __func__, k, mode_lib->ms.RequiresDSC[k]); + DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); +#endif + + // ensure the number dsc slices is integer multiple based on ODM mode + mode_lib->ms.support.DSCSlicesODMModeSupported = true; + if (mode_lib->ms.RequiresDSC[k]) { + // fail a ms check if the override num_slices doesn't align with odm mode setting + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices != 0) { + if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) + mode_lib->ms.support.DSCSlicesODMModeSupported = ((mode_lib->ms.support.NumberOfDSCSlices[k] % 2) == 0); + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) + mode_lib->ms.support.DSCSlicesODMModeSupported = (mode_lib->ms.support.NumberOfDSCSlices[k] == 12); + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) + mode_lib->ms.support.DSCSlicesODMModeSupported = ((mode_lib->ms.support.NumberOfDSCSlices[k] % 4) == 0); +#if defined(__DML_VBA_DEBUG__) + if (!mode_lib->ms.support.DSCSlicesODMModeSupported) { + DML_LOG_VERBOSE("DML::%s: k=%d Invalid dsc num_slices and ODM mode setting\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d num_slices = %d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.overrides.num_slices); + DML_LOG_VERBOSE("DML::%s: k=%d ODMMode = %d\n", __func__, k, mode_lib->ms.ODMMode[k]); + } +#endif + } else { + // safe guard to ensure the dml derived dsc slices and odm setting are compatible + if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) + mode_lib->ms.support.NumberOfDSCSlices[k] = 2 * (unsigned int)math_ceil2(mode_lib->ms.support.NumberOfDSCSlices[k] / 2.0, 1.0); + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) + mode_lib->ms.support.NumberOfDSCSlices[k] = 12; + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) + mode_lib->ms.support.NumberOfDSCSlices[k] = 4 * (unsigned int)math_ceil2(mode_lib->ms.support.NumberOfDSCSlices[k] / 4.0, 1.0); + } + + } else { + mode_lib->ms.support.NumberOfDSCSlices[k] = 0; + } + } + + mode_lib->ms.support.incorrect_imall_usage = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall) + mode_lib->ms.support.incorrect_imall_usage = 1; + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.MPCCombine[k] = false; + mode_lib->ms.NoOfDPP[k] = 1; + mode_lib->ms.NoOfOPP[k] = 1; + + if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) { + mode_lib->ms.MPCCombine[k] = false; + mode_lib->ms.NoOfDPP[k] = 4; + mode_lib->ms.NoOfOPP[k] = 4; + } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) { + mode_lib->ms.MPCCombine[k] = false; + mode_lib->ms.NoOfDPP[k] = 3; + mode_lib->ms.NoOfOPP[k] = 3; + } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) { + mode_lib->ms.MPCCombine[k] = false; + mode_lib->ms.NoOfDPP[k] = 2; + mode_lib->ms.NoOfOPP[k] = 2; + } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 2) { + mode_lib->ms.MPCCombine[k] = true; + mode_lib->ms.NoOfDPP[k] = 2; + } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 1) { + mode_lib->ms.MPCCombine[k] = false; + mode_lib->ms.NoOfDPP[k] = 1; + if (!mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { + DML_LOG_VERBOSE("WARNING: DML::%s: MPCC is override to disable but viewport is too large to be supported with single pipe!\n", __func__); + } + } else { + if ((mode_lib->ms.MinDPPCLKUsingSingleDPP[k] > mode_lib->ms.max_dppclk_freq_mhz) || !mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { + mode_lib->ms.MPCCombine[k] = true; + mode_lib->ms.NoOfDPP[k] = 2; + } + } +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: k=%d, NoOfDPP = %d\n", __func__, k, mode_lib->ms.NoOfDPP[k]); +#endif + } + + mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.TotalNumberOfActiveDPP += mode_lib->ms.NoOfDPP[k]; + mode_lib->ms.TotalNumberOfActiveOPP += mode_lib->ms.NoOfOPP[k]; + } + if (mode_lib->ms.TotalNumberOfActiveDPP > (unsigned int)mode_lib->ip.max_num_dpp) + mode_lib->ms.support.TotalAvailablePipesSupport = false; + if (mode_lib->ms.TotalNumberOfActiveOPP > (unsigned int)mode_lib->ip.max_num_opp) + mode_lib->ms.support.TotalAvailablePipesSupport = false; + + + mode_lib->ms.TotalNumberOfSingleDPPSurfaces = 0; + for (k = 0; k < (unsigned int)mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.NoOfDPP[k] == 1) + mode_lib->ms.TotalNumberOfSingleDPPSurfaces = mode_lib->ms.TotalNumberOfSingleDPPSurfaces + 1; + } + + //DISPCLK/DPPCLK + mode_lib->ms.WritebackRequiredDISPCLK = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + mode_lib->ms.WritebackRequiredDISPCLK = math_max2(mode_lib->ms.WritebackRequiredDISPCLK, + CalculateWriteBackDISPCLK(display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_taps, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_width, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, + mode_lib->ip.writeback_line_buffer_buffer_size)); + } + } + + mode_lib->ms.RequiredDISPCLK = mode_lib->ms.WritebackRequiredDISPCLK; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.RequiredDISPCLK = math_max2(mode_lib->ms.RequiredDISPCLK, mode_lib->ms.RequiredDISPCLKPerSurface[k]); + } + + mode_lib->ms.GlobalDPPCLK = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.RequiredDPPCLK[k] = mode_lib->ms.MinDPPCLKUsingSingleDPP[k] / mode_lib->ms.NoOfDPP[k]; + mode_lib->ms.GlobalDPPCLK = math_max2(mode_lib->ms.GlobalDPPCLK, mode_lib->ms.RequiredDPPCLK[k]); + } + + mode_lib->ms.support.DISPCLK_DPPCLK_Support = !((mode_lib->ms.RequiredDISPCLK > mode_lib->ms.max_dispclk_freq_mhz) || (mode_lib->ms.GlobalDPPCLK > mode_lib->ms.max_dppclk_freq_mhz)); + + /* Total Available OTG, Writeback, HDMIFRL, DP Support Check */ + s->TotalNumberOfActiveOTG = 0; + s->TotalNumberOfActiveHDMIFRL = 0; + s->TotalNumberOfActiveDP2p0 = 0; + s->TotalNumberOfActiveDP2p0Outputs = 0; + s->TotalNumberOfActiveWriteback = 0; + memset(s->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (!dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + if (!s->stream_visited[display_cfg->plane_descriptors[k].stream_index]) { + s->stream_visited[display_cfg->plane_descriptors[k].stream_index] = 1; + + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) + s->TotalNumberOfActiveWriteback = s->TotalNumberOfActiveWriteback + 1; + + s->TotalNumberOfActiveOTG = s->TotalNumberOfActiveOTG + 1; + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) + s->TotalNumberOfActiveHDMIFRL = s->TotalNumberOfActiveHDMIFRL + 1; + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0) { + s->TotalNumberOfActiveDP2p0 = s->TotalNumberOfActiveDP2p0 + 1; + // FIXME_STAGE2: SW not using backend related stuff, need mapping for mst setup + //if (display_cfg->output.OutputMultistreamId[k] == k || display_cfg->output.OutputMultistreamEn[k] == false) { + s->TotalNumberOfActiveDP2p0Outputs = s->TotalNumberOfActiveDP2p0Outputs + 1; + //} + } + } + } + } + + /* Writeback Mode Support Check */ + mode_lib->ms.support.EnoughWritebackUnits = 1; + if (s->TotalNumberOfActiveWriteback > (unsigned int)mode_lib->ip.max_num_wb) { + mode_lib->ms.support.EnoughWritebackUnits = false; + } + mode_lib->ms.support.NumberOfOTGSupport = (s->TotalNumberOfActiveOTG <= (unsigned int)mode_lib->ip.max_num_otg); + mode_lib->ms.support.NumberOfHDMIFRLSupport = (s->TotalNumberOfActiveHDMIFRL <= (unsigned int)mode_lib->ip.max_num_hdmi_frl_outputs); + mode_lib->ms.support.NumberOfDP2p0Support = (s->TotalNumberOfActiveDP2p0 <= (unsigned int)mode_lib->ip.max_num_dp2p0_streams && s->TotalNumberOfActiveDP2p0Outputs <= (unsigned int)mode_lib->ip.max_num_dp2p0_outputs); + + + mode_lib->ms.support.ExceededMultistreamSlots = false; + mode_lib->ms.support.LinkCapacitySupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_disabled == false && + (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) && mode_lib->ms.OutputBpp[k] == 0) { + mode_lib->ms.support.LinkCapacitySupport = false; + } + } + + mode_lib->ms.support.P2IWith420 = false; + mode_lib->ms.support.DSCOnlyIfNecessaryWithBPP = false; + mode_lib->ms.support.DSC422NativeNotSupported = false; + mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = false; + mode_lib->ms.support.LinkRateForMultistreamNotIndicated = false; + mode_lib->ms.support.BPPForMultistreamNotIndicated = false; + mode_lib->ms.support.MultistreamWithHDMIOreDP = false; + mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = false; + mode_lib->ms.support.NotEnoughLanesForMSO = false; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_420 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced == 1 && mode_lib->ip.ptoi_supported == true) + mode_lib->ms.support.P2IWith420 = true; + + if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable_if_necessary) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_n422 && !mode_lib->ip.dsc422_native_support) + mode_lib->ms.support.DSC422NativeNotSupported = true; + + if (((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr2 || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_hbr3) && + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_edp) || + ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr10 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr13p5 || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_uhbr20) && + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp2p0)) + mode_lib->ms.support.LinkRateDoesNotMatchDPVersion = true; + + // FIXME_STAGE2 + //if (display_cfg->output.OutputMultistreamEn[k] == 1) { + // if (display_cfg->output.OutputMultistreamId[k] == k && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_link_rate == dml2_dp_rate_na) + // mode_lib->ms.support.LinkRateForMultistreamNotIndicated = true; + // if (display_cfg->output.OutputMultistreamId[k] == k && s->OutputBpp[k] == 0) + // mode_lib->ms.support.BPPForMultistreamNotIndicated = true; + // for (n = 0; n < mode_lib->ms.num_active_planes; ++n) { + // if (display_cfg->output.OutputMultistreamId[k] == n && s->OutputBpp[k] == 0) + // mode_lib->ms.support.BPPForMultistreamNotIndicated = true; + // } + //} + + if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmi || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl)) { + // FIXME_STAGE2 + //if (display_cfg->output.OutputMultistreamEn[k] == 1 && display_cfg->output.OutputMultistreamId[k] == k) + // mode_lib->ms.support.MultistreamWithHDMIOreDP = true; + //for (n = 0; n < mode_lib->ms.num_active_planes; ++n) { + // if (display_cfg->output.OutputMultistreamEn[k] == 1 && display_cfg->output.OutputMultistreamId[k] == n) + // mode_lib->ms.support.MultistreamWithHDMIOreDP = true; + //} + } + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder != dml2_dp && (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_split_1to2 || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to2 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to4)) + mode_lib->ms.support.MSOOrODMSplitWithNonDPLink = true; + + if ((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to2 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count < 2) || + (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].overrides.odm_mode == dml2_odm_mode_mso_1to4 && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_dp_lane_count < 4)) + mode_lib->ms.support.NotEnoughLanesForMSO = true; + } + } + + mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl && + !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) { + mode_lib->ms.RequiredDTBCLK[k] = RequiredDTBCLK( + mode_lib->ms.RequiresDSC[k], + s->PixelClockBackEnd[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + mode_lib->ms.OutputBpp[k], + mode_lib->ms.support.NumberOfDSCSlices[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_rate, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.audio_sample_layout); + + if (mode_lib->ms.RequiredDTBCLK[k] > ((double)min_clk_table->max_ss_clocks_khz.dtbclk / 1000)) { + mode_lib->ms.support.DTBCLKRequiredMoreThanSupported = true; + } + } else { + /* Phantom DTBCLK can be calculated different from main because phantom has no DSC and thus + * will have a different output BPP. Ignore phantom DTBCLK requirement and only consider + * non-phantom DTBCLK requirements. In map_mode_to_soc_dpm we choose the highest DTBCLK + * required - by setting phantom dtbclk to 0 we ignore it. + */ + mode_lib->ms.RequiredDTBCLK[k] = 0; + } + } + + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = false; + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_dp2p0 || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_edp || + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_420) { + s->DSCFormatFactor = 2; + } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_444) { + s->DSCFormatFactor = 1; + } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format == dml2_n422 || display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder == dml2_hdmifrl) { + s->DSCFormatFactor = 2; + } else { + s->DSCFormatFactor = 1; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, RequiresDSC = %u\n", __func__, k, mode_lib->ms.RequiresDSC[k]); +#endif + if (mode_lib->ms.RequiresDSC[k] == true) { + s->PixelClockBackEndFactor = 3.0; + + if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) + s->PixelClockBackEndFactor = 12.0; + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) + s->PixelClockBackEndFactor = 9.0; + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) + s->PixelClockBackEndFactor = 6.0; + + mode_lib->ms.required_dscclk_freq_mhz[k] = s->PixelClockBackEnd[k] / s->PixelClockBackEndFactor / (double)s->DSCFormatFactor; + if (mode_lib->ms.required_dscclk_freq_mhz[k] > mode_lib->ms.max_dscclk_freq_mhz) { + mode_lib->ms.support.DSCCLKRequiredMoreThanSupported = true; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, PixelClockBackEnd = %f\n", __func__, k, s->PixelClockBackEnd[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, required_dscclk_freq_mhz = %f\n", __func__, k, mode_lib->ms.required_dscclk_freq_mhz[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DSCFormatFactor = %u\n", __func__, k, s->DSCFormatFactor); + DML_LOG_VERBOSE("DML::%s: k=%u, DSCCLKRequiredMoreThanSupported = %u\n", __func__, k, mode_lib->ms.support.DSCCLKRequiredMoreThanSupported); +#endif + } + } + } + + /* Check DSC Unit and Slices Support */ + mode_lib->ms.support.NotEnoughDSCSlices = false; + s->TotalDSCUnitsRequired = 0; + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = true; + memset(s->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.RequiresDSC[k] == true && !s->stream_visited[display_cfg->plane_descriptors[k].stream_index]) { + s->NumDSCUnitRequired = 1; + + if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) + s->NumDSCUnitRequired = 4; + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) + s->NumDSCUnitRequired = 3; + else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) + s->NumDSCUnitRequired = 2; + + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active > s->NumDSCUnitRequired * (unsigned int)mode_lib->ip.maximum_pixels_per_line_per_dsc_unit) + mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport = false; + s->TotalDSCUnitsRequired = s->TotalDSCUnitsRequired + s->NumDSCUnitRequired; + + if (mode_lib->ms.support.NumberOfDSCSlices[k] > 4 * s->NumDSCUnitRequired) + mode_lib->ms.support.NotEnoughDSCSlices = true; + } + s->stream_visited[display_cfg->plane_descriptors[k].stream_index] = 1; + } + + mode_lib->ms.support.NotEnoughDSCUnits = false; + if (s->TotalDSCUnitsRequired > (unsigned int)mode_lib->ip.num_dsc) { + mode_lib->ms.support.NotEnoughDSCUnits = true; + } + + /*DSC Delay per state*/ + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.DSCDelay[k] = DSCDelayRequirement(mode_lib->ms.RequiresDSC[k], + mode_lib->ms.ODMMode[k], + mode_lib->ip.maximum_dsc_bits_per_component, + s->OutputBpp[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, + mode_lib->ms.support.NumberOfDSCSlices[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + s->PixelClockBackEnd[k]); + } + + // Figure out the swath and DET configuration after the num dpp per plane is figured out + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; + CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->ms.ODMMode; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->ms.NoOfDPP; + + // output + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = s->dummy_integer_array[0]; + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = s->dummy_integer_array[1]; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = mode_lib->ms.swath_width_luma_ub; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = mode_lib->ms.swath_width_chroma_ub; + CalculateSwathAndDETConfiguration_params->SwathWidth = mode_lib->ms.SwathWidthY; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = mode_lib->ms.SwathWidthC; + CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->ms.SwathHeightY; + CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->ms.SwathHeightC; + CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = mode_lib->ms.support.request_size_bytes_luma; + CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = mode_lib->ms.support.request_size_bytes_chroma; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->ms.DETBufferSizeInKByte; // FIXME: This is per pipe but the pipes in plane will use that + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->ms.DETBufferSizeY; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->ms.DETBufferSizeC; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->ms.UnboundedRequestEnabled; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = s->dummy_integer_array[3]; + CalculateSwathAndDETConfiguration_params->hw_debug5 = s->dummy_boolean_array[1]; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->ms.CompressedBufferSizeInkByte; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = s->dummy_boolean_array[0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &mode_lib->ms.support.ViewportSizeSupport; + + CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); + + if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0) { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) + mode_lib->ms.SurfaceSizeInMALL[k] = 0; + mode_lib->ms.support.ExceededMALLSize = 0; + } else { + CalculateSurfaceSizeInMall( + display_cfg, + mode_lib->ms.num_active_planes, + mode_lib->soc.mall_allocated_for_dcn_mbytes, + + mode_lib->ms.BytePerPixelY, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.Read256BlockWidthY, + mode_lib->ms.Read256BlockWidthC, + mode_lib->ms.Read256BlockHeightY, + mode_lib->ms.Read256BlockHeightC, + mode_lib->ms.MacroTileWidthY, + mode_lib->ms.MacroTileWidthC, + mode_lib->ms.MacroTileHeightY, + mode_lib->ms.MacroTileHeightC, + + /* Output */ + mode_lib->ms.SurfaceSizeInMALL, + &mode_lib->ms.support.ExceededMALLSize); + } + + mode_lib->ms.TotalNumberOfDCCActiveDPP = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->plane_descriptors[k].surface.dcc.enable == true) { + mode_lib->ms.TotalNumberOfDCCActiveDPP = mode_lib->ms.TotalNumberOfDCCActiveDPP + mode_lib->ms.NoOfDPP[k]; + } + } + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->SurfParameters[k].PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + s->SurfParameters[k].DPPPerSurface = mode_lib->ms.NoOfDPP[k]; + s->SurfParameters[k].RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; + s->SurfParameters[k].ViewportHeight = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + s->SurfParameters[k].ViewportHeightC = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; + s->SurfParameters[k].BlockWidth256BytesY = mode_lib->ms.Read256BlockWidthY[k]; + s->SurfParameters[k].BlockHeight256BytesY = mode_lib->ms.Read256BlockHeightY[k]; + s->SurfParameters[k].BlockWidth256BytesC = mode_lib->ms.Read256BlockWidthC[k]; + s->SurfParameters[k].BlockHeight256BytesC = mode_lib->ms.Read256BlockHeightC[k]; + s->SurfParameters[k].BlockWidthY = mode_lib->ms.MacroTileWidthY[k]; + s->SurfParameters[k].BlockHeightY = mode_lib->ms.MacroTileHeightY[k]; + s->SurfParameters[k].BlockWidthC = mode_lib->ms.MacroTileWidthC[k]; + s->SurfParameters[k].BlockHeightC = mode_lib->ms.MacroTileHeightC[k]; + s->SurfParameters[k].InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; + s->SurfParameters[k].HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; + s->SurfParameters[k].DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; + s->SurfParameters[k].SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; + s->SurfParameters[k].SurfaceTiling = display_cfg->plane_descriptors[k].surface.tiling; + s->SurfParameters[k].BytePerPixelY = mode_lib->ms.BytePerPixelY[k]; + s->SurfParameters[k].BytePerPixelC = mode_lib->ms.BytePerPixelC[k]; + s->SurfParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + s->SurfParameters[k].VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + s->SurfParameters[k].VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + s->SurfParameters[k].VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + s->SurfParameters[k].VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + s->SurfParameters[k].PitchY = display_cfg->plane_descriptors[k].surface.plane0.pitch; + s->SurfParameters[k].PitchC = display_cfg->plane_descriptors[k].surface.plane1.pitch; + s->SurfParameters[k].ViewportStationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; + s->SurfParameters[k].ViewportXStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; + s->SurfParameters[k].ViewportYStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; + s->SurfParameters[k].ViewportXStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + s->SurfParameters[k].ViewportYStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + s->SurfParameters[k].FORCE_ONE_ROW_FOR_FRAME = display_cfg->plane_descriptors[k].overrides.hw.force_one_row_for_frame; + s->SurfParameters[k].SwathHeightY = mode_lib->ms.SwathHeightY[k]; + s->SurfParameters[k].SwathHeightC = mode_lib->ms.SwathHeightC[k]; + + s->SurfParameters[k].DCCMetaPitchY = display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch; + s->SurfParameters[k].DCCMetaPitchC = display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch; + } + + CalculateVMRowAndSwath_params->display_cfg = display_cfg; + CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = mode_lib->ms.num_active_planes; + CalculateVMRowAndSwath_params->myPipe = s->SurfParameters; + CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->ms.SurfaceSizeInMALL; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma; + CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->soc.mall_allocated_for_dcn_mbytes; + CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->ms.SwathWidthY; + CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->ms.SwathWidthC; + CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ip.dcc_meta_buffer_size_bytes; + CalculateVMRowAndSwath_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + // output + CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = mode_lib->ms.PTEBufferSizeNotExceeded; + CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = s->dummy_integer_array[12]; + CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = s->dummy_integer_array[13]; + CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->ms.dpte_row_height; + CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->ms.dpte_row_height_chroma; + CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = s->dummy_integer_array[14]; // VBA_DELTA + CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = s->dummy_integer_array[15]; // VBA_DELTA + CalculateVMRowAndSwath_params->vm_group_bytes = s->dummy_integer_array[16]; + CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->ms.dpte_group_bytes; + CalculateVMRowAndSwath_params->PixelPTEReqWidthY = s->dummy_integer_array[17]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightY = s->dummy_integer_array[18]; + CalculateVMRowAndSwath_params->PTERequestSizeY = s->dummy_integer_array[19]; + CalculateVMRowAndSwath_params->PixelPTEReqWidthC = s->dummy_integer_array[20]; + CalculateVMRowAndSwath_params->PixelPTEReqHeightC = s->dummy_integer_array[21]; + CalculateVMRowAndSwath_params->PTERequestSizeC = s->dummy_integer_array[22]; + CalculateVMRowAndSwath_params->vmpg_width_y = s->vmpg_width_y; + CalculateVMRowAndSwath_params->vmpg_height_y = s->vmpg_height_y; + CalculateVMRowAndSwath_params->vmpg_width_c = s->vmpg_width_c; + CalculateVMRowAndSwath_params->vmpg_height_c = s->vmpg_height_c; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = s->dummy_integer_array[23]; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = s->dummy_integer_array[24]; + CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->ms.PrefetchLinesY; + CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->ms.PrefetchLinesC; + CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->ms.PrefillY; + CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->ms.PrefillC; + CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->ms.MaxNumSwathY; + CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->ms.MaxNumSwathC; + CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->ms.dpte_row_bw; + CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->ms.DPTEBytesPerRow; + CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_l = s->dpte_row_bytes_per_row_l; + CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_c = s->dpte_row_bytes_per_row_c; + CalculateVMRowAndSwath_params->vm_bytes = mode_lib->ms.vm_bytes; + CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->ms.use_one_row_for_frame; + CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->ms.use_one_row_for_frame_flip; + CalculateVMRowAndSwath_params->is_using_mall_for_ss = s->dummy_boolean_array[0]; + CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = s->dummy_boolean_array[1]; + CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = s->dummy_integer_array[25]; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = mode_lib->ms.DCCMetaBufferSizeNotExceeded; + CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->ms.meta_row_bw; + CalculateVMRowAndSwath_params->meta_row_bytes = mode_lib->ms.meta_row_bytes; + CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; + CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; + CalculateVMRowAndSwath_params->meta_req_width_luma = s->dummy_integer_array[26]; + CalculateVMRowAndSwath_params->meta_req_height_luma = s->dummy_integer_array[27]; + CalculateVMRowAndSwath_params->meta_row_width_luma = s->dummy_integer_array[28]; + CalculateVMRowAndSwath_params->meta_row_height_luma = s->meta_row_height_luma; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = s->dummy_integer_array[29]; + CalculateVMRowAndSwath_params->meta_req_width_chroma = s->dummy_integer_array[30]; + CalculateVMRowAndSwath_params->meta_req_height_chroma = s->dummy_integer_array[31]; + CalculateVMRowAndSwath_params->meta_row_width_chroma = s->dummy_integer_array[32]; + CalculateVMRowAndSwath_params->meta_row_height_chroma = s->meta_row_height_chroma; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = s->dummy_integer_array[33]; + + CalculateVMRowAndSwath(&mode_lib->scratch, CalculateVMRowAndSwath_params); + + mode_lib->ms.support.PTEBufferSizeNotExceeded = true; + mode_lib->ms.support.DCCMetaBufferSizeNotExceeded = true; + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.PTEBufferSizeNotExceeded[k] == false) + mode_lib->ms.support.PTEBufferSizeNotExceeded = false; + + if (mode_lib->ms.DCCMetaBufferSizeNotExceeded[k] == false) + mode_lib->ms.support.DCCMetaBufferSizeNotExceeded = false; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, PTEBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.PTEBufferSizeNotExceeded[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DCCMetaBufferSizeNotExceeded = %u\n", __func__, k, mode_lib->ms.DCCMetaBufferSizeNotExceeded[k]); +#endif + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: PTEBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.PTEBufferSizeNotExceeded); + DML_LOG_VERBOSE("DML::%s: DCCMetaBufferSizeNotExceeded = %u\n", __func__, mode_lib->ms.support.DCCMetaBufferSizeNotExceeded); +#endif + + /* VActive bytes to fetch for UCLK P-State */ + calculate_bytes_to_fetch_required_to_hide_latency_params->display_cfg = display_cfg; + calculate_bytes_to_fetch_required_to_hide_latency_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + calculate_bytes_to_fetch_required_to_hide_latency_params->num_active_planes = mode_lib->ms.num_active_planes; + calculate_bytes_to_fetch_required_to_hide_latency_params->num_of_dpp = mode_lib->ms.NoOfDPP; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_l = s->meta_row_height_luma; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_c = s->meta_row_height_chroma; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_l = mode_lib->ms.dpte_row_height; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_c = mode_lib->ms.dpte_row_height_chroma; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_l = s->dpte_row_bytes_per_row_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_c = s->dpte_row_bytes_per_row_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_l = mode_lib->ms.BytePerPixelY; + calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_c = mode_lib->ms.BytePerPixelC; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_l = mode_lib->ms.SwathWidthY; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->ms.SwathWidthC; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->ms.SwathHeightY; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->ms.SwathHeightC; + calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + + /* outputs */ + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; + + calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); + + /* Excess VActive bandwidth required to fill DET */ + calculate_excess_vactive_bandwidth_required( + display_cfg, + mode_lib->ms.num_active_planes, + s->pstate_bytes_required_l, + s->pstate_bytes_required_c, + /* outputs */ + mode_lib->ms.excess_vactive_fill_bw_l, + mode_lib->ms.excess_vactive_fill_bw_c); + + mode_lib->ms.UrgLatency = CalculateUrgentLatency( + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_pixel_vm_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_vm_us, + mode_lib->soc.do_urgent_latency_adjustment, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_fclk_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_mhz, + mode_lib->ms.FabricClock, + mode_lib->ms.uclk_freq_mhz, + mode_lib->soc.qos_parameters.qos_type, + mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].urgent_ramp_uclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_urgent_ramp_latency_margin, + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); + + mode_lib->ms.TripToMemory = CalculateTripToMemory( + mode_lib->ms.UrgLatency, + mode_lib->ms.FabricClock, + mode_lib->ms.uclk_freq_mhz, + mode_lib->soc.qos_parameters.qos_type, + mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].trip_to_memory_uclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); + + mode_lib->ms.TripToMemory = math_max2(mode_lib->ms.UrgLatency, mode_lib->ms.TripToMemory); + + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + double line_time_us = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + bool cursor_not_enough_urgent_latency_hiding = false; + + if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { + calculate_cursor_req_attributes( + display_cfg->plane_descriptors[k].cursor.cursor_width, + display_cfg->plane_descriptors[k].cursor.cursor_bpp, + + // output + &s->cursor_lines_per_chunk[k], + &s->cursor_bytes_per_line[k], + &s->cursor_bytes_per_chunk[k], + &s->cursor_bytes[k]); + + calculate_cursor_urgent_burst_factor( + mode_lib->ip.cursor_buffer_size, + display_cfg->plane_descriptors[k].cursor.cursor_width, + s->cursor_bytes_per_chunk[k], + s->cursor_lines_per_chunk[k], + line_time_us, + mode_lib->ms.UrgLatency, + + // output + &mode_lib->ms.UrgentBurstFactorCursor[k], + &cursor_not_enough_urgent_latency_hiding); + } + + mode_lib->ms.UrgentBurstFactorCursorPre[k] = mode_lib->ms.UrgentBurstFactorCursor[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, Calling CalculateUrgentBurstFactor\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatio=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioChroma=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); +#endif + + CalculateUrgentBurstFactor( + &display_cfg->plane_descriptors[k], + mode_lib->ms.swath_width_luma_ub[k], + mode_lib->ms.swath_width_chroma_ub[k], + mode_lib->ms.SwathHeightY[k], + mode_lib->ms.SwathHeightC[k], + line_time_us, + mode_lib->ms.UrgLatency, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->ms.BytePerPixelInDETY[k], + mode_lib->ms.BytePerPixelInDETC[k], + mode_lib->ms.DETBufferSizeY[k], + mode_lib->ms.DETBufferSizeC[k], + + // Output + &mode_lib->ms.UrgentBurstFactorLuma[k], + &mode_lib->ms.UrgentBurstFactorChroma[k], + &mode_lib->ms.NotEnoughUrgentLatencyHiding[k]); + + mode_lib->ms.NotEnoughUrgentLatencyHiding[k] = mode_lib->ms.NotEnoughUrgentLatencyHiding[k] || cursor_not_enough_urgent_latency_hiding; + } + + CalculateDCFCLKDeepSleep( + display_cfg, + mode_lib->ms.num_active_planes, + mode_lib->ms.BytePerPixelY, + mode_lib->ms.BytePerPixelC, + mode_lib->ms.SwathWidthY, + mode_lib->ms.SwathWidthC, + mode_lib->ms.NoOfDPP, + mode_lib->ms.PSCL_FACTOR, + mode_lib->ms.PSCL_FACTOR_CHROMA, + mode_lib->ms.RequiredDPPCLK, + mode_lib->ms.vactive_sw_bw_l, + mode_lib->ms.vactive_sw_bw_c, + mode_lib->soc.return_bus_width_bytes, + + /* Output */ + &mode_lib->ms.dcfclk_deepsleep); + + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + mode_lib->ms.WritebackDelayTime[k] = mode_lib->soc.qos_parameters.writeback.base_latency_us + CalculateWriteBackDelay( + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) / mode_lib->ms.RequiredDISPCLK; + } else { + mode_lib->ms.WritebackDelayTime[k] = 0.0; + } + } + + // MaximumVStartup is actually Tvstartup_min in DCN4 programming guide + for (k = 0; k <= mode_lib->ms.num_active_planes - 1; k++) { + bool isInterlaceTiming = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !mode_lib->ip.ptoi_supported); + s->MaximumVStartup[k] = CalculateMaxVStartup( + mode_lib->ip.ptoi_supported, + mode_lib->ip.vblank_nom_default_us, + &display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing, + mode_lib->ms.WritebackDelayTime[k]); + mode_lib->ms.MaxVStartupLines[k] = (isInterlaceTiming ? (2 * s->MaximumVStartup[k]) : s->MaximumVStartup[k]); + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumVStartup = %u\n", __func__, k, s->MaximumVStartup[k]); +#endif + + /* Immediate Flip and MALL parameters */ + s->ImmediateFlipRequired = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + s->ImmediateFlipRequired = s->ImmediateFlipRequired || display_cfg->plane_descriptors[k].immediate_flip; + } + + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = + mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe || + ((display_cfg->hostvm_enable == true || display_cfg->plane_descriptors[k].immediate_flip == true) && + (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame || dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]))); + } + + mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen || + ((display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_enable || display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_auto) && (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]))) || + ((display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_force_disable || display_cfg->plane_descriptors[k].overrides.refresh_from_mall == dml2_refresh_from_mall_mode_override_auto) && (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame)); + } + + s->FullFrameMALLPStateMethod = false; + s->SubViewportMALLPStateMethod = false; + s->PhantomPipeMALLPStateMethod = false; + s->SubViewportMALLRefreshGreaterThan120Hz = false; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (display_cfg->plane_descriptors[k].overrides.uclk_pstate_change_strategy == dml2_uclk_pstate_change_strategy_force_mall_full_frame) + s->FullFrameMALLPStateMethod = true; + if (display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_main_pipe) { + s->SubViewportMALLPStateMethod = true; + if (!display_cfg->overrides.enable_subvp_implicit_pmo) { + // For dv, small frame tests will have very high refresh rate + unsigned long long refresh_rate = (unsigned long long) ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz * 1000 / + (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / + (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); + if (refresh_rate > 120) + s->SubViewportMALLRefreshGreaterThan120Hz = true; + } + } + if (dml_is_phantom_pipe(&display_cfg->plane_descriptors[k])) + s->PhantomPipeMALLPStateMethod = true; + } + mode_lib->ms.support.InvalidCombinationOfMALLUseForPState = (s->SubViewportMALLPStateMethod != s->PhantomPipeMALLPStateMethod) || + (s->SubViewportMALLPStateMethod && s->FullFrameMALLPStateMethod) || s->SubViewportMALLRefreshGreaterThan120Hz; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: SubViewportMALLPStateMethod = %u\n", __func__, s->SubViewportMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: PhantomPipeMALLPStateMethod = %u\n", __func__, s->PhantomPipeMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: FullFrameMALLPStateMethod = %u\n", __func__, s->FullFrameMALLPStateMethod); + DML_LOG_VERBOSE("DML::%s: SubViewportMALLRefreshGreaterThan120Hz = %u\n", __func__, s->SubViewportMALLRefreshGreaterThan120Hz); + DML_LOG_VERBOSE("DML::%s: InvalidCombinationOfMALLUseForPState = %u\n", __func__, mode_lib->ms.support.InvalidCombinationOfMALLUseForPState); + DML_LOG_VERBOSE("DML::%s: in_out_params->min_clk_index = %u\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.DCFCLK = %f\n", __func__, mode_lib->ms.DCFCLK); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.FabricClock = %f\n", __func__, mode_lib->ms.FabricClock); + DML_LOG_VERBOSE("DML::%s: mode_lib->ms.uclk_freq_mhz = %f\n", __func__, mode_lib->ms.uclk_freq_mhz); +#endif + + mode_lib->ms.support.OutstandingRequestsSupport = true; + mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = true; + + mode_lib->ms.support.avg_urgent_latency_us + = (mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].average_latency_when_urgent_uclk_cycles / mode_lib->ms.uclk_freq_mhz + * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_average_latency_margin / 100.0) + + mode_lib->soc.qos_parameters.qos_params.dcn4x.average_transport_distance_fclk_cycles / mode_lib->ms.FabricClock) + * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_average_transport_latency_margin / 100.0); + + mode_lib->ms.support.avg_non_urgent_latency_us + = (mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].average_latency_when_non_urgent_uclk_cycles / mode_lib->ms.uclk_freq_mhz + * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_average_latency_margin / 100.0) + + mode_lib->soc.qos_parameters.qos_params.dcn4x.average_transport_distance_fclk_cycles / mode_lib->ms.FabricClock) + * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_average_transport_latency_margin / 100.0); + + mode_lib->ms.support.max_non_urgent_latency_us + = mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->ms.qos_param_index].maximum_latency_when_non_urgent_uclk_cycles + / mode_lib->ms.uclk_freq_mhz * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin / 100.0) + + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles / mode_lib->ms.FabricClock + + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles / mode_lib->ms.FabricClock + * (1 + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin / 100.0); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { + outstanding_latency_us = (mode_lib->soc.max_outstanding_reqs * mode_lib->ms.support.request_size_bytes_luma[k] + / (mode_lib->ms.DCFCLK * mode_lib->soc.return_bus_width_bytes)); + + if (outstanding_latency_us < mode_lib->ms.support.avg_urgent_latency_us) { + mode_lib->ms.support.OutstandingRequestsSupport = false; + } + + if (outstanding_latency_us < mode_lib->ms.support.avg_non_urgent_latency_us) { + mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = false; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: avg_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: avg_non_urgent_latency_us = %f\n", __func__, mode_lib->ms.support.avg_non_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_luma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (luma)\n", __func__, k, outstanding_latency_us); +#endif + } + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x && mode_lib->ms.BytePerPixelC[k] > 0) { + outstanding_latency_us = (mode_lib->soc.max_outstanding_reqs * mode_lib->ms.support.request_size_bytes_chroma[k] + / (mode_lib->ms.DCFCLK * mode_lib->soc.return_bus_width_bytes)); + + if (outstanding_latency_us < mode_lib->ms.support.avg_urgent_latency_us) { + mode_lib->ms.support.OutstandingRequestsSupport = false; + } + + if (outstanding_latency_us < mode_lib->ms.support.avg_non_urgent_latency_us) { + mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance = false; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, request_size_bytes_chroma = %d\n", __func__, k, mode_lib->ms.support.request_size_bytes_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, outstanding_latency_us = %f (chroma)\n", __func__, k, outstanding_latency_us); +#endif + } + } + + memset(calculate_mcache_setting_params, 0, sizeof(struct dml2_core_calcs_calculate_mcache_setting_params)); + if (mode_lib->soc.mcache_size_bytes == 0 || mode_lib->ip.dcn_mrq_present) { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + mode_lib->ms.mall_prefetch_sdp_overhead_factor[k] = 1.0; + mode_lib->ms.mall_prefetch_dram_overhead_factor[k] = 1.0; + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0[k] = 1.0; + mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0[k] = 1.0; + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1[k] = 1.0; + mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1[k] = 1.0; + } + } else { + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + calculate_mcache_setting_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; + calculate_mcache_setting_params->num_chans = mode_lib->soc.clk_table.dram_config.channel_count; + calculate_mcache_setting_params->mem_word_bytes = mode_lib->soc.mem_word_bytes; + calculate_mcache_setting_params->mcache_size_bytes = mode_lib->soc.mcache_size_bytes; + calculate_mcache_setting_params->mcache_line_size_bytes = mode_lib->soc.mcache_line_size_bytes; + calculate_mcache_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; + calculate_mcache_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + + calculate_mcache_setting_params->source_format = display_cfg->plane_descriptors[k].pixel_format; + calculate_mcache_setting_params->surf_vert = dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle); + calculate_mcache_setting_params->vp_stationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; + calculate_mcache_setting_params->tiling_mode = display_cfg->plane_descriptors[k].surface.tiling; + calculate_mcache_setting_params->imall_enable = mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall; + + calculate_mcache_setting_params->vp_start_x_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; + calculate_mcache_setting_params->vp_start_y_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; + calculate_mcache_setting_params->full_vp_width_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; + calculate_mcache_setting_params->full_vp_height_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + calculate_mcache_setting_params->blk_width_l = mode_lib->ms.MacroTileWidthY[k]; + calculate_mcache_setting_params->blk_height_l = mode_lib->ms.MacroTileHeightY[k]; + calculate_mcache_setting_params->vmpg_width_l = s->vmpg_width_y[k]; + calculate_mcache_setting_params->vmpg_height_l = s->vmpg_height_y[k]; + calculate_mcache_setting_params->full_swath_bytes_l = s->full_swath_bytes_l[k]; + calculate_mcache_setting_params->bytes_per_pixel_l = mode_lib->ms.BytePerPixelY[k]; + + calculate_mcache_setting_params->vp_start_x_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.x_start; + calculate_mcache_setting_params->vp_start_y_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + calculate_mcache_setting_params->full_vp_width_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; + calculate_mcache_setting_params->full_vp_height_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; + calculate_mcache_setting_params->blk_width_c = mode_lib->ms.MacroTileWidthC[k]; + calculate_mcache_setting_params->blk_height_c = mode_lib->ms.MacroTileHeightC[k]; + calculate_mcache_setting_params->vmpg_width_c = s->vmpg_width_c[k]; + calculate_mcache_setting_params->vmpg_height_c = s->vmpg_height_c[k]; + calculate_mcache_setting_params->full_swath_bytes_c = s->full_swath_bytes_c[k]; + calculate_mcache_setting_params->bytes_per_pixel_c = mode_lib->ms.BytePerPixelC[k]; + + // output + calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_l = &mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0[k]; + calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_l = &mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p0[k]; + calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_c = &mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1[k]; + calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_c = &mode_lib->ms.dcc_dram_bw_pref_overhead_factor_p1[k]; + + calculate_mcache_setting_params->num_mcaches_l = &mode_lib->ms.num_mcaches_l[k]; + calculate_mcache_setting_params->mcache_row_bytes_l = &mode_lib->ms.mcache_row_bytes_l[k]; + calculate_mcache_setting_params->mcache_row_bytes_per_channel_l = &mode_lib->ms.mcache_row_bytes_per_channel_l[k]; + calculate_mcache_setting_params->mcache_offsets_l = mode_lib->ms.mcache_offsets_l[k]; + calculate_mcache_setting_params->mcache_shift_granularity_l = &mode_lib->ms.mcache_shift_granularity_l[k]; + + calculate_mcache_setting_params->num_mcaches_c = &mode_lib->ms.num_mcaches_c[k]; + calculate_mcache_setting_params->mcache_row_bytes_c = &mode_lib->ms.mcache_row_bytes_c[k]; + calculate_mcache_setting_params->mcache_row_bytes_per_channel_c = &mode_lib->ms.mcache_row_bytes_per_channel_c[k]; + calculate_mcache_setting_params->mcache_offsets_c = mode_lib->ms.mcache_offsets_c[k]; + calculate_mcache_setting_params->mcache_shift_granularity_c = &mode_lib->ms.mcache_shift_granularity_c[k]; + + calculate_mcache_setting_params->mall_comb_mcache_l = &mode_lib->ms.mall_comb_mcache_l[k]; + calculate_mcache_setting_params->mall_comb_mcache_c = &mode_lib->ms.mall_comb_mcache_c[k]; + calculate_mcache_setting_params->lc_comb_mcache = &mode_lib->ms.lc_comb_mcache[k]; + + calculate_mcache_setting(&mode_lib->scratch, calculate_mcache_setting_params); + } + + calculate_mall_bw_overhead_factor( + mode_lib->ms.mall_prefetch_sdp_overhead_factor, + mode_lib->ms.mall_prefetch_dram_overhead_factor, + + // input + display_cfg, + mode_lib->ms.num_active_planes); + } + + // Calculate all the bandwidth available + // Need anothe bw for latency evaluation + calculate_bandwidth_available( + mode_lib->ms.support.avg_bandwidth_available_min, // not used + mode_lib->ms.support.avg_bandwidth_available, // not used + mode_lib->ms.support.urg_bandwidth_available_min_latency, + mode_lib->ms.support.urg_bandwidth_available, // not used + mode_lib->ms.support.urg_bandwidth_available_vm_only, // not used + mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm, // not used + + &mode_lib->soc, + display_cfg->hostvm_enable, + mode_lib->ms.DCFCLK, + mode_lib->ms.FabricClock, + mode_lib->ms.dram_bw_mbps); + + calculate_bandwidth_available( + mode_lib->ms.support.avg_bandwidth_available_min, + mode_lib->ms.support.avg_bandwidth_available, + mode_lib->ms.support.urg_bandwidth_available_min, + mode_lib->ms.support.urg_bandwidth_available, + mode_lib->ms.support.urg_bandwidth_available_vm_only, + mode_lib->ms.support.urg_bandwidth_available_pixel_and_vm, + + &mode_lib->soc, + display_cfg->hostvm_enable, + mode_lib->ms.MaxDCFCLK, + mode_lib->ms.MaxFabricClock, +#ifdef DML_MODE_SUPPORT_USE_DPM_DRAM_BW + mode_lib->ms.dram_bw_mbps); +#else + mode_lib->ms.max_dram_bw_mbps); +#endif + + // Average BW support check + calculate_avg_bandwidth_required( + mode_lib->ms.support.avg_bandwidth_required, + // input + display_cfg, + mode_lib->ms.num_active_planes, + mode_lib->ms.vactive_sw_bw_l, + mode_lib->ms.vactive_sw_bw_c, + mode_lib->ms.cursor_bw, + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0, + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1, + mode_lib->ms.mall_prefetch_dram_overhead_factor, + mode_lib->ms.mall_prefetch_sdp_overhead_factor); + + for (m = 0; m < dml2_core_internal_bw_max; m++) { // check sdp and dram + mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_sys_idle][m] = 1; + mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_sys_active][m] = (mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_sys_active][m] <= mode_lib->ms.support.avg_bandwidth_available[dml2_core_internal_soc_state_sys_active][m]); + mode_lib->ms.support.avg_bandwidth_support_ok[dml2_core_internal_soc_state_svp_prefetch][m] = (mode_lib->ms.support.avg_bandwidth_required[dml2_core_internal_soc_state_svp_prefetch][m] <= mode_lib->ms.support.avg_bandwidth_available[dml2_core_internal_soc_state_svp_prefetch][m]); + } + + mode_lib->ms.support.AvgBandwidthSupport = true; + mode_lib->ms.support.EnoughUrgentLatencyHidingSupport = true; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + if (mode_lib->ms.NotEnoughUrgentLatencyHiding[k]) { + mode_lib->ms.support.EnoughUrgentLatencyHidingSupport = false; + DML_LOG_VERBOSE("DML::%s: k=%u NotEnoughUrgentLatencyHiding set\n", __func__, k); + + } + } + for (m = 0; m < dml2_core_internal_soc_state_max; m++) { + for (n = 0; n < dml2_core_internal_bw_max; n++) { // check sdp and dram + if (!mode_lib->ms.support.avg_bandwidth_support_ok[m][n] && (m == dml2_core_internal_soc_state_sys_active || mode_lib->soc.mall_allocated_for_dcn_mbytes > 0)) { + mode_lib->ms.support.AvgBandwidthSupport = false; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: avg_bandwidth_support_ok[%s][%s] not ok\n", __func__, dml2_core_internal_soc_state_type_str(m), dml2_core_internal_bw_type_str(n)); +#endif + } + } + } + + dml_core_ms_prefetch_check(mode_lib, display_cfg); + + mode_lib->ms.support.max_urgent_latency_us = s->mSOCParameters.max_urgent_latency_us; + + //Re-ordering Buffer Support Check + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { + if (((mode_lib->ip.rob_buffer_size_kbytes - mode_lib->ip.pixel_chunk_size_kbytes) * 1024 + / mode_lib->ms.support.non_urg_bandwidth_required_flip[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]) >= s->mSOCParameters.max_urgent_latency_us) { + mode_lib->ms.support.ROBSupport = true; + } else { + mode_lib->ms.support.ROBSupport = false; + } + } else { + if (mode_lib->ip.rob_buffer_size_kbytes * 1024 >= mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles * mode_lib->soc.fabric_datapath_to_dcn_data_return_bytes) { + mode_lib->ms.support.ROBSupport = true; + } else { + mode_lib->ms.support.ROBSupport = false; + } + } + + /* VActive fill time calculations (informative) */ + calculate_vactive_det_fill_latency( + display_cfg, + mode_lib->ms.num_active_planes, + s->pstate_bytes_required_l, + s->pstate_bytes_required_c, + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0, + mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1, + mode_lib->ms.vactive_sw_bw_l, + mode_lib->ms.vactive_sw_bw_c, + mode_lib->ms.surface_avg_vactive_required_bw, + mode_lib->ms.surface_peak_required_bw, + /* outputs */ + mode_lib->ms.dram_change_vactive_det_fill_delay_us); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: max_urgent_latency_us = %f\n", __func__, s->mSOCParameters.max_urgent_latency_us); + DML_LOG_VERBOSE("DML::%s: ROBSupport = %u\n", __func__, mode_lib->ms.support.ROBSupport); +#endif + + /*Mode Support, Voltage State and SOC Configuration*/ + { + if (mode_lib->ms.support.ScaleRatioAndTapsSupport + && mode_lib->ms.support.SourceFormatPixelAndScanSupport + && mode_lib->ms.support.ViewportSizeSupport + && !mode_lib->ms.support.LinkRateDoesNotMatchDPVersion + && !mode_lib->ms.support.LinkRateForMultistreamNotIndicated + && !mode_lib->ms.support.BPPForMultistreamNotIndicated + && !mode_lib->ms.support.MultistreamWithHDMIOreDP + && !mode_lib->ms.support.ExceededMultistreamSlots + && !mode_lib->ms.support.MSOOrODMSplitWithNonDPLink + && !mode_lib->ms.support.NotEnoughLanesForMSO + && !mode_lib->ms.support.P2IWith420 + && !mode_lib->ms.support.DSC422NativeNotSupported + && mode_lib->ms.support.DSCSlicesODMModeSupported + && !mode_lib->ms.support.NotEnoughDSCUnits + && !mode_lib->ms.support.NotEnoughDSCSlices + && !mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe + && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen + && !mode_lib->ms.support.DSCCLKRequiredMoreThanSupported + && mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport + && !mode_lib->ms.support.DTBCLKRequiredMoreThanSupported + && !mode_lib->ms.support.InvalidCombinationOfMALLUseForPState + && mode_lib->ms.support.ROBSupport + && mode_lib->ms.support.OutstandingRequestsSupport + && mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance + && mode_lib->ms.support.DISPCLK_DPPCLK_Support + && mode_lib->ms.support.TotalAvailablePipesSupport + && mode_lib->ms.support.NumberOfOTGSupport + && mode_lib->ms.support.NumberOfHDMIFRLSupport + && mode_lib->ms.support.NumberOfDP2p0Support + && mode_lib->ms.support.EnoughWritebackUnits + && mode_lib->ms.support.WritebackLatencySupport + && mode_lib->ms.support.WritebackScaleRatioAndTapsSupport + && mode_lib->ms.support.CursorSupport + && mode_lib->ms.support.PitchSupport + && !mode_lib->ms.support.ViewportExceedsSurface + && mode_lib->ms.support.PrefetchSupported + && mode_lib->ms.support.EnoughUrgentLatencyHidingSupport + && mode_lib->ms.support.AvgBandwidthSupport + && mode_lib->ms.support.DynamicMetadataSupported + && mode_lib->ms.support.VRatioInPrefetchSupported + && mode_lib->ms.support.PTEBufferSizeNotExceeded + && mode_lib->ms.support.DCCMetaBufferSizeNotExceeded + && !mode_lib->ms.support.ExceededMALLSize + && mode_lib->ms.support.g6_temp_read_support + && ((!display_cfg->hostvm_enable && !s->ImmediateFlipRequired) || mode_lib->ms.support.ImmediateFlipSupport)) { + DML_LOG_VERBOSE("DML::%s: mode is supported\n", __func__); + mode_lib->ms.support.ModeSupport = true; + } else { + DML_LOG_VERBOSE("DML::%s: mode is NOT supported\n", __func__); + mode_lib->ms.support.ModeSupport = false; + } + } + + // Since now the mode_support work on 1 particular power state, so there is only 1 state idx (index 0). + DML_LOG_VERBOSE("DML::%s: ModeSupport = %u\n", __func__, mode_lib->ms.support.ModeSupport); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupport = %u\n", __func__, mode_lib->ms.support.ImmediateFlipSupport); + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + mode_lib->ms.support.MPCCombineEnable[k] = mode_lib->ms.MPCCombine[k]; + mode_lib->ms.support.DPPPerSurface[k] = mode_lib->ms.NoOfDPP[k]; + } + + for (k = 0; k < mode_lib->ms.num_active_planes; k++) { + mode_lib->ms.support.ODMMode[k] = mode_lib->ms.ODMMode[k]; + mode_lib->ms.support.DSCEnabled[k] = mode_lib->ms.RequiresDSC[k]; + mode_lib->ms.support.FECEnabled[k] = mode_lib->ms.RequiresFEC[k]; + mode_lib->ms.support.OutputBpp[k] = mode_lib->ms.OutputBpp[k]; + mode_lib->ms.support.OutputType[k] = mode_lib->ms.OutputType[k]; + mode_lib->ms.support.OutputRate[k] = mode_lib->ms.OutputRate[k]; + +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: k=%d, ODMMode = %u\n", __func__, k, mode_lib->ms.support.ODMMode[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DSCEnabled = %u\n", __func__, k, mode_lib->ms.support.DSCEnabled[k]); +#endif + } + +#if defined(__DML_VBA_DEBUG__) + if (!mode_lib->ms.support.ModeSupport) + dml2_print_mode_support_info(&mode_lib->ms.support, true); + + DML_LOG_VERBOSE("DML::%s: --- DONE --- \n", __func__); +#endif + + return mode_lib->ms.support.ModeSupport; +} + +unsigned int dml2_core_calcs_mode_support_ex(struct dml2_core_calcs_mode_support_ex *in_out_params) +{ + unsigned int result; + + DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); + result = dml_core_mode_support(in_out_params); + + if (result) + *in_out_params->out_evaluation_info = in_out_params->mode_lib->ms.support; + + DML_LOG_VERBOSE("DML::%s: is_mode_support = %u (min_clk_index=%d)\n", __func__, result, in_out_params->min_clk_index); + + for (unsigned int k = 0; k < in_out_params->in_display_cfg->num_planes; k++) + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, in_out_params->in_display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); + + DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); + + return result; +} + +static void CalculatePixelDeliveryTimes( + const struct dml2_display_cfg *display_cfg, + const struct core_display_cfg_support_info *cfg_support_info, + unsigned int NumberOfActiveSurfaces, + double VRatioPrefetchY[], + double VRatioPrefetchC[], + unsigned int swath_width_luma_ub[], + unsigned int swath_width_chroma_ub[], + double PSCL_THROUGHPUT[], + double PSCL_THROUGHPUT_CHROMA[], + double Dppclk[], + unsigned int BytePerPixelC[], + unsigned int req_per_swath_ub_l[], + unsigned int req_per_swath_ub_c[], + + // Output + double DisplayPipeLineDeliveryTimeLuma[], + double DisplayPipeLineDeliveryTimeChroma[], + double DisplayPipeLineDeliveryTimeLumaPrefetch[], + double DisplayPipeLineDeliveryTimeChromaPrefetch[], + double DisplayPipeRequestDeliveryTimeLuma[], + double DisplayPipeRequestDeliveryTimeChroma[], + double DisplayPipeRequestDeliveryTimeLumaPrefetch[], + double DisplayPipeRequestDeliveryTimeChromaPrefetch[]) +{ + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u : HRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatio = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : HRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioChroma = %f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchY = %f\n", __func__, k, VRatioPrefetchY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : VRatioPrefetchC = %f\n", __func__, k, VRatioPrefetchC[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_luma_ub = %u\n", __func__, k, swath_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : swath_width_chroma_ub = %u\n", __func__, k, swath_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT = %f\n", __func__, k, PSCL_THROUGHPUT[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : PSCL_THROUGHPUT_CHROMA = %f\n", __func__, k, PSCL_THROUGHPUT_CHROMA[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DPPPerSurface = %u\n", __func__, k, cfg_support_info->plane_support_info[k].dpps_used); + DML_LOG_VERBOSE("DML::%s: k=%u : pixel_clock_mhz = %f\n", __func__, k, pixel_clock_mhz); + DML_LOG_VERBOSE("DML::%s: k=%u : Dppclk = %f\n", __func__, k, Dppclk[k]); +#endif + if (display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio <= 1) { + DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_clock_mhz; + } else { + DisplayPipeLineDeliveryTimeLuma[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChroma[k] = 0; + } else { + if (display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio <= 1) { + DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_clock_mhz; + } else { + DisplayPipeLineDeliveryTimeChroma[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } + + if (VRatioPrefetchY[k] <= 1) { + DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio / pixel_clock_mhz; + } else { + DisplayPipeLineDeliveryTimeLumaPrefetch[k] = swath_width_luma_ub[k] / PSCL_THROUGHPUT[k] / Dppclk[k]; + } + + if (BytePerPixelC[k] == 0) { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = 0; + } else { + if (VRatioPrefetchC[k] <= 1) { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] * cfg_support_info->plane_support_info[k].dpps_used / display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio / pixel_clock_mhz; + } else { + DisplayPipeLineDeliveryTimeChromaPrefetch[k] = swath_width_chroma_ub[k] / PSCL_THROUGHPUT_CHROMA[k] / Dppclk[k]; + } + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeLumaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeLineDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeLineDeliveryTimeChromaPrefetch[k]); +#endif + } + + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + + DisplayPipeRequestDeliveryTimeLuma[k] = DisplayPipeLineDeliveryTimeLuma[k] / req_per_swath_ub_l[k]; + DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = DisplayPipeLineDeliveryTimeLumaPrefetch[k] / req_per_swath_ub_l[k]; + if (BytePerPixelC[k] == 0) { + DisplayPipeRequestDeliveryTimeChroma[k] = 0; + DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = 0; + } else { + DisplayPipeRequestDeliveryTimeChroma[k] = DisplayPipeLineDeliveryTimeChroma[k] / req_per_swath_ub_c[k]; + DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = DisplayPipeLineDeliveryTimeChromaPrefetch[k] / req_per_swath_ub_c[k]; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLuma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeLumaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeLumaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_l = %d\n", __func__, k, req_per_swath_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChroma = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : DisplayPipeRequestDeliveryTimeChromaPrefetch = %f\n", __func__, k, DisplayPipeRequestDeliveryTimeChromaPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u : req_per_swath_ub_c = %d\n", __func__, k, req_per_swath_ub_c[k]); +#endif + } +} + +static void CalculateMetaAndPTETimes(struct dml2_core_shared_CalculateMetaAndPTETimes_params *p) +{ + unsigned int meta_chunk_width; + unsigned int min_meta_chunk_width; + unsigned int meta_chunk_per_row_int; + unsigned int meta_row_remainder; + unsigned int meta_chunk_threshold; + unsigned int meta_chunks_per_row_ub; + unsigned int meta_chunk_width_chroma; + unsigned int min_meta_chunk_width_chroma; + unsigned int meta_chunk_per_row_int_chroma; + unsigned int meta_row_remainder_chroma; + unsigned int meta_chunk_threshold_chroma; + unsigned int meta_chunks_per_row_ub_chroma; + unsigned int dpte_group_width_luma; + unsigned int dpte_groups_per_row_luma_ub; + unsigned int dpte_group_width_chroma; + unsigned int dpte_groups_per_row_chroma_ub; + double pixel_clock_mhz; + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + p->DST_Y_PER_PTE_ROW_NOM_L[k] = p->dpte_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + if (p->BytePerPixelC[k] == 0) { + p->DST_Y_PER_PTE_ROW_NOM_C[k] = 0; + } else { + p->DST_Y_PER_PTE_ROW_NOM_C[k] = p->dpte_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + } + p->DST_Y_PER_META_ROW_NOM_L[k] = p->meta_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + if (p->BytePerPixelC[k] == 0) { + p->DST_Y_PER_META_ROW_NOM_C[k] = 0; + } else { + p->DST_Y_PER_META_ROW_NOM_C[k] = p->meta_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + } + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (p->display_cfg->plane_descriptors[k].surface.dcc.enable == true && p->mrq_present) { + meta_chunk_width = p->MetaChunkSize * 1024 * 256 / p->BytePerPixelY[k] / p->meta_row_height[k]; + min_meta_chunk_width = p->MinMetaChunkSizeBytes * 256 / p->BytePerPixelY[k] / p->meta_row_height[k]; + meta_chunk_per_row_int = p->meta_row_width[k] / meta_chunk_width; + meta_row_remainder = p->meta_row_width[k] % meta_chunk_width; + if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { + meta_chunk_threshold = 2 * min_meta_chunk_width - p->meta_req_width[k]; + } else { + meta_chunk_threshold = 2 * min_meta_chunk_width - p->meta_req_height[k]; + } + if (meta_row_remainder <= meta_chunk_threshold) { + meta_chunks_per_row_ub = meta_chunk_per_row_int + 1; + } else { + meta_chunks_per_row_ub = meta_chunk_per_row_int + 2; + } + p->TimePerMetaChunkNominal[k] = p->meta_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio * + p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / + (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; + p->TimePerMetaChunkVBlank[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / + (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; + p->TimePerMetaChunkFlip[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / + (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub; + if (p->BytePerPixelC[k] == 0) { + p->TimePerChromaMetaChunkNominal[k] = 0; + p->TimePerChromaMetaChunkVBlank[k] = 0; + p->TimePerChromaMetaChunkFlip[k] = 0; + } else { + meta_chunk_width_chroma = p->MetaChunkSize * 1024 * 256 / p->BytePerPixelC[k] / p->meta_row_height_chroma[k]; + min_meta_chunk_width_chroma = p->MinMetaChunkSizeBytes * 256 / p->BytePerPixelC[k] / p->meta_row_height_chroma[k]; + meta_chunk_per_row_int_chroma = (unsigned int)((double)p->meta_row_width_chroma[k] / meta_chunk_width_chroma); + meta_row_remainder_chroma = p->meta_row_width_chroma[k] % meta_chunk_width_chroma; + if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { + meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - p->meta_req_width_chroma[k]; + } else { + meta_chunk_threshold_chroma = 2 * min_meta_chunk_width_chroma - p->meta_req_height_chroma[k]; + } + if (meta_row_remainder_chroma <= meta_chunk_threshold_chroma) { + meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 1; + } else { + meta_chunks_per_row_ub_chroma = meta_chunk_per_row_int_chroma + 2; + } + p->TimePerChromaMetaChunkNominal[k] = p->meta_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; + p->TimePerChromaMetaChunkVBlank[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; + p->TimePerChromaMetaChunkFlip[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / (p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) / meta_chunks_per_row_ub_chroma; + } + } else { + p->TimePerMetaChunkNominal[k] = 0; + p->TimePerMetaChunkVBlank[k] = 0; + p->TimePerMetaChunkFlip[k] = 0; + p->TimePerChromaMetaChunkNominal[k] = 0; + p->TimePerChromaMetaChunkVBlank[k] = 0; + p->TimePerChromaMetaChunkFlip[k] = 0; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_L[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, DST_Y_PER_META_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_META_ROW_NOM_C[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkNominal = %f\n", __func__, k, p->TimePerMetaChunkNominal[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkVBlank = %f\n", __func__, k, p->TimePerMetaChunkVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerMetaChunkFlip = %f\n", __func__, k, p->TimePerMetaChunkFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkNominal = %f\n", __func__, k, p->TimePerChromaMetaChunkNominal[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkVBlank = %f\n", __func__, k, p->TimePerChromaMetaChunkVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, TimePerChromaMetaChunkFlip = %f\n", __func__, k, p->TimePerChromaMetaChunkFlip[k]); +#endif + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + p->DST_Y_PER_PTE_ROW_NOM_L[k] = p->dpte_row_height[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + if (p->BytePerPixelC[k] == 0) { + p->DST_Y_PER_PTE_ROW_NOM_C[k] = 0; + } else { + p->DST_Y_PER_PTE_ROW_NOM_C[k] = p->dpte_row_height_chroma[k] / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + } + } + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + pixel_clock_mhz = ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + + if (p->display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut) + p->time_per_tdlut_group[k] = 2 * p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / p->tdlut_groups_per_2row_ub[k]; + else + p->time_per_tdlut_group[k] = 0; + + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_tdlut_group = %f\n", __func__, k, p->time_per_tdlut_group[k]); + + if (p->display_cfg->gpuvm_enable == true) { + if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { + dpte_group_width_luma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeY[k] * p->PixelPTEReqWidthY[k]); + } else { + dpte_group_width_luma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeY[k] * p->PixelPTEReqHeightY[k]); + } + if (p->use_one_row_for_frame[k]) { + dpte_groups_per_row_luma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_luma_ub[k] / (double)dpte_group_width_luma / 2.0, 1.0)); + } else { + dpte_groups_per_row_luma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_luma_ub[k] / (double)dpte_group_width_luma, 1.0)); + } + if (dpte_groups_per_row_luma_ub <= 2) { + dpte_groups_per_row_luma_ub = dpte_groups_per_row_luma_ub + 1; + } + DML_LOG_VERBOSE("DML::%s: k=%u, use_one_row_for_frame = %u\n", __func__, k, p->use_one_row_for_frame[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_bytes = %u\n", __func__, k, p->dpte_group_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PTERequestSizeY = %u\n", __func__, k, p->PTERequestSizeY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqWidthY = %u\n", __func__, k, p->PixelPTEReqWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, PixelPTEReqHeightY = %u\n", __func__, k, p->PixelPTEReqHeightY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_luma_ub = %u\n", __func__, k, p->dpte_row_width_luma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_luma = %u\n", __func__, k, dpte_group_width_luma); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_luma_ub = %u\n", __func__, k, dpte_groups_per_row_luma_ub); + + p->time_per_pte_group_nom_luma[k] = p->DST_Y_PER_PTE_ROW_NOM_L[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; + p->time_per_pte_group_vblank_luma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; + p->time_per_pte_group_flip_luma[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_luma_ub; + if (p->BytePerPixelC[k] == 0) { + p->time_per_pte_group_nom_chroma[k] = 0; + p->time_per_pte_group_vblank_chroma[k] = 0; + p->time_per_pte_group_flip_chroma[k] = 0; + } else { + if (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle)) { + dpte_group_width_chroma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeC[k] * p->PixelPTEReqWidthC[k]); + } else { + dpte_group_width_chroma = (unsigned int)((double)p->dpte_group_bytes[k] / (double)p->PTERequestSizeC[k] * p->PixelPTEReqHeightC[k]); + } + + if (p->use_one_row_for_frame[k]) { + dpte_groups_per_row_chroma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_chroma_ub[k] / (double)dpte_group_width_chroma / 2.0, 1.0)); + } else { + dpte_groups_per_row_chroma_ub = (unsigned int)(math_ceil2((double)p->dpte_row_width_chroma_ub[k] / (double)dpte_group_width_chroma, 1.0)); + } + if (dpte_groups_per_row_chroma_ub <= 2) { + dpte_groups_per_row_chroma_ub = dpte_groups_per_row_chroma_ub + 1; + } + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_row_width_chroma_ub = %u\n", __func__, k, p->dpte_row_width_chroma_ub[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_group_width_chroma = %u\n", __func__, k, dpte_group_width_chroma); + DML_LOG_VERBOSE("DML::%s: k=%u, dpte_groups_per_row_chroma_ub = %u\n", __func__, k, dpte_groups_per_row_chroma_ub); + + p->time_per_pte_group_nom_chroma[k] = p->DST_Y_PER_PTE_ROW_NOM_C[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; + p->time_per_pte_group_vblank_chroma[k] = p->dst_y_per_row_vblank[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; + p->time_per_pte_group_flip_chroma[k] = p->dst_y_per_row_flip[k] * p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz / dpte_groups_per_row_chroma_ub; + } + } else { + p->time_per_pte_group_nom_luma[k] = 0; + p->time_per_pte_group_vblank_luma[k] = 0; + p->time_per_pte_group_flip_luma[k] = 0; + p->time_per_pte_group_nom_chroma[k] = 0; + p->time_per_pte_group_vblank_chroma[k] = 0; + p->time_per_pte_group_flip_chroma[k] = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_vblank = %f\n", __func__, k, p->dst_y_per_row_vblank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_row_flip = %f\n", __func__, k, p->dst_y_per_row_flip[k]); + + DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_L = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_L[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DST_Y_PER_PTE_ROW_NOM_C = %f\n", __func__, k, p->DST_Y_PER_PTE_ROW_NOM_C[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_luma = %f\n", __func__, k, p->time_per_pte_group_nom_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_luma = %f\n", __func__, k, p->time_per_pte_group_vblank_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_luma = %f\n", __func__, k, p->time_per_pte_group_flip_luma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_nom_chroma = %f\n", __func__, k, p->time_per_pte_group_nom_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_vblank_chroma = %f\n", __func__, k, p->time_per_pte_group_vblank_chroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, time_per_pte_group_flip_chroma = %f\n", __func__, k, p->time_per_pte_group_flip_chroma[k]); +#endif + } +} // CalculateMetaAndPTETimes + +static void CalculateVMGroupAndRequestTimes( + const struct dml2_display_cfg *display_cfg, + unsigned int NumberOfActiveSurfaces, + unsigned int BytePerPixelC[], + double dst_y_per_vm_vblank[], + double dst_y_per_vm_flip[], + unsigned int dpte_row_width_luma_ub[], + unsigned int dpte_row_width_chroma_ub[], + unsigned int vm_group_bytes[], + unsigned int dpde0_bytes_per_frame_ub_l[], + unsigned int dpde0_bytes_per_frame_ub_c[], + unsigned int tdlut_pte_bytes_per_frame[], + unsigned int meta_pte_bytes_per_frame_ub_l[], + unsigned int meta_pte_bytes_per_frame_ub_c[], + bool mrq_present, + + // Output + double TimePerVMGroupVBlank[], + double TimePerVMGroupFlip[], + double TimePerVMRequestVBlank[], + double TimePerVMRequestFlip[]) +{ + unsigned int num_group_per_lower_vm_stage = 0; + unsigned int num_req_per_lower_vm_stage = 0; + unsigned int num_group_per_lower_vm_stage_flip; + unsigned int num_group_per_lower_vm_stage_pref; + unsigned int num_req_per_lower_vm_stage_flip; + unsigned int num_req_per_lower_vm_stage_pref; + double line_time; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: NumberOfActiveSurfaces = %u\n", __func__, NumberOfActiveSurfaces); +#endif + for (unsigned int k = 0; k < NumberOfActiveSurfaces; ++k) { + double pixel_clock_mhz = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + bool dcc_mrq_enable = display_cfg->plane_descriptors[k].surface.dcc.enable && mrq_present; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, dcc_mrq_enable = %u\n", __func__, k, dcc_mrq_enable); + DML_LOG_VERBOSE("DML::%s: k=%u, vm_group_bytes = %u\n", __func__, k, vm_group_bytes[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_l = %u\n", __func__, k, dpde0_bytes_per_frame_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dpde0_bytes_per_frame_ub_c = %u\n", __func__, k, dpde0_bytes_per_frame_ub_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_l = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%d, meta_pte_bytes_per_frame_ub_c = %d\n", __func__, k, meta_pte_bytes_per_frame_ub_c[k]); +#endif + + if (display_cfg->gpuvm_enable) { + if (display_cfg->gpuvm_max_page_table_levels >= 2) { + num_group_per_lower_vm_stage += (unsigned int) math_ceil2((double) (dpde0_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1); + + if (BytePerPixelC[k] > 0) + num_group_per_lower_vm_stage += (unsigned int) math_ceil2((double) (dpde0_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1); + } + + if (dcc_mrq_enable) { + if (BytePerPixelC[k] > 0) { + num_group_per_lower_vm_stage += (unsigned int)(2.0 /*for each mpde0 group*/ + math_ceil2((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1) + + math_ceil2((double) (meta_pte_bytes_per_frame_ub_c[k]) / (double) (vm_group_bytes[k]), 1)); + } else { + num_group_per_lower_vm_stage += (unsigned int)(1.0 + math_ceil2((double) (meta_pte_bytes_per_frame_ub_l[k]) / (double) (vm_group_bytes[k]), 1)); + } + } + + num_group_per_lower_vm_stage_flip = num_group_per_lower_vm_stage; + num_group_per_lower_vm_stage_pref = num_group_per_lower_vm_stage; + + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && display_cfg->gpuvm_enable) { + num_group_per_lower_vm_stage_pref += (unsigned int) math_ceil2(tdlut_pte_bytes_per_frame[k] / vm_group_bytes[k], 1); + if (display_cfg->gpuvm_max_page_table_levels >= 2) + num_group_per_lower_vm_stage_pref += 1; // tdpe0 group + } + + if (display_cfg->gpuvm_max_page_table_levels >= 2) { + num_req_per_lower_vm_stage += dpde0_bytes_per_frame_ub_l[k] / 64; + if (BytePerPixelC[k] > 0) + num_req_per_lower_vm_stage += dpde0_bytes_per_frame_ub_c[k]; + } + + if (dcc_mrq_enable) { + num_req_per_lower_vm_stage += meta_pte_bytes_per_frame_ub_l[k] / 64; + if (BytePerPixelC[k] > 0) + num_req_per_lower_vm_stage += meta_pte_bytes_per_frame_ub_c[k] / 64; + } + + num_req_per_lower_vm_stage_flip = num_req_per_lower_vm_stage; + num_req_per_lower_vm_stage_pref = num_req_per_lower_vm_stage; + + if (display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut && display_cfg->gpuvm_enable) { + num_req_per_lower_vm_stage_pref += tdlut_pte_bytes_per_frame[k] / 64; + } + + line_time = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / pixel_clock_mhz; + + if (num_group_per_lower_vm_stage_pref > 0) + TimePerVMGroupVBlank[k] = dst_y_per_vm_vblank[k] * line_time / num_group_per_lower_vm_stage_pref; + else + TimePerVMGroupVBlank[k] = 0; + + if (num_group_per_lower_vm_stage_flip > 0) + TimePerVMGroupFlip[k] = dst_y_per_vm_flip[k] * line_time / num_group_per_lower_vm_stage_flip; + else + TimePerVMGroupFlip[k] = 0; + + if (num_req_per_lower_vm_stage_pref > 0) + TimePerVMRequestVBlank[k] = dst_y_per_vm_vblank[k] * line_time / num_req_per_lower_vm_stage_pref; + else + TimePerVMRequestVBlank[k] = 0.0; + if (num_req_per_lower_vm_stage_flip > 0) + TimePerVMRequestFlip[k] = dst_y_per_vm_flip[k] * line_time / num_req_per_lower_vm_stage_flip; + else + TimePerVMRequestFlip[k] = 0.0; + + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_vblank = %f\n", __func__, k, dst_y_per_vm_vblank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, dst_y_per_vm_flip = %f\n", __func__, k, dst_y_per_vm_flip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, line_time = %f\n", __func__, k, line_time); + DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_pref = %d\n", __func__, k, num_group_per_lower_vm_stage_pref); + DML_LOG_VERBOSE("DML::%s: k=%u, num_group_per_lower_vm_stage_flip = %d\n", __func__, k, num_group_per_lower_vm_stage_flip); + DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_pref = %d\n", __func__, k, num_req_per_lower_vm_stage_pref); + DML_LOG_VERBOSE("DML::%s: k=%u, num_req_per_lower_vm_stage_flip = %d\n", __func__, k, num_req_per_lower_vm_stage_flip); + + if (display_cfg->gpuvm_max_page_table_levels > 2) { + TimePerVMGroupVBlank[k] = TimePerVMGroupVBlank[k] / 2; + TimePerVMGroupFlip[k] = TimePerVMGroupFlip[k] / 2; + TimePerVMRequestVBlank[k] = TimePerVMRequestVBlank[k] / 2; + TimePerVMRequestFlip[k] = TimePerVMRequestFlip[k] / 2; + } + + } else { + TimePerVMGroupVBlank[k] = 0; + TimePerVMGroupFlip[k] = 0; + TimePerVMRequestVBlank[k] = 0; + TimePerVMRequestFlip[k] = 0; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupVBlank = %f\n", __func__, k, TimePerVMGroupVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMGroupFlip = %f\n", __func__, k, TimePerVMGroupFlip[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestVBlank = %f\n", __func__, k, TimePerVMRequestVBlank[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TimePerVMRequestFlip = %f\n", __func__, k, TimePerVMRequestFlip[k]); +#endif + } +} + +static void CalculateStutterEfficiency(struct dml2_core_internal_scratch *scratch, + struct dml2_core_calcs_CalculateStutterEfficiency_params *p) +{ + struct dml2_core_calcs_CalculateStutterEfficiency_locals *l = &scratch->CalculateStutterEfficiency_locals; + + unsigned int TotalNumberOfActiveOTG = 0; + double SinglePixelClock = 0; + unsigned int SingleHTotal = 0; + unsigned int SingleVTotal = 0; + bool SameTiming = true; + bool FoundCriticalSurface = false; + + memset(l, 0, sizeof(struct dml2_core_calcs_CalculateStutterEfficiency_locals)); + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { + if (p->display_cfg->plane_descriptors[k].surface.dcc.enable == true) { + if ((dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockWidth256BytesY[k] > p->SwathHeightY[k]) || (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockHeight256BytesY[k] > p->SwathHeightY[k]) || p->DCCYMaxUncompressedBlock[k] < 256) { + l->MaximumEffectiveCompressionLuma = 2; + } else { + l->MaximumEffectiveCompressionLuma = 4; + } + l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0, l->MaximumEffectiveCompressionLuma); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateLuma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane0); + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionLuma = %f\n", __func__, k, l->MaximumEffectiveCompressionLuma); +#endif + l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0; + l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane0 / l->MaximumEffectiveCompressionLuma; + + if (p->ReadBandwidthSurfaceChroma[k] > 0) { + if ((dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockWidth256BytesC[k] > p->SwathHeightC[k]) || (!dml_is_vertical_rotation(p->display_cfg->plane_descriptors[k].composition.rotation_angle) && p->BlockHeight256BytesC[k] > p->SwathHeightC[k]) || p->DCCCMaxUncompressedBlock[k] < 256) { + l->MaximumEffectiveCompressionChroma = 2; + } else { + l->MaximumEffectiveCompressionChroma = 4; + } + l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] / math_min2(p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1, l->MaximumEffectiveCompressionChroma); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceChroma = %f\n", __func__, k, p->ReadBandwidthSurfaceChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NetDCCRateChroma = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].surface.dcc.informative.dcc_rate_plane1); + DML_LOG_VERBOSE("DML::%s: k=%u, MaximumEffectiveCompressionChroma = %f\n", __func__, k, l->MaximumEffectiveCompressionChroma); +#endif + l->TotalZeroSizeRequestReadBandwidth = l->TotalZeroSizeRequestReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1; + l->TotalZeroSizeCompressedReadBandwidth = l->TotalZeroSizeCompressedReadBandwidth + p->ReadBandwidthSurfaceChroma[k] * p->display_cfg->plane_descriptors[k].surface.dcc.informative.fraction_of_zero_size_request_plane1 / l->MaximumEffectiveCompressionChroma; + } + } else { + l->TotalCompressedReadBandwidth = l->TotalCompressedReadBandwidth + p->ReadBandwidthSurfaceLuma[k] + p->ReadBandwidthSurfaceChroma[k]; + } + l->TotalRowReadBandwidth = l->TotalRowReadBandwidth + p->DPPPerSurface[k] * (p->meta_row_bw[k] + p->dpte_row_bw[k]); + } + } + + l->AverageDCCCompressionRate = p->TotalDataReadBandwidth / l->TotalCompressedReadBandwidth; + l->AverageDCCZeroSizeFraction = l->TotalZeroSizeRequestReadBandwidth / p->TotalDataReadBandwidth; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: UnboundedRequestEnabled = %u\n", __func__, p->UnboundedRequestEnabled); + DML_LOG_VERBOSE("DML::%s: TotalCompressedReadBandwidth = %f\n", __func__, l->TotalCompressedReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalZeroSizeRequestReadBandwidth = %f\n", __func__, l->TotalZeroSizeRequestReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalZeroSizeCompressedReadBandwidth = %f\n", __func__, l->TotalZeroSizeCompressedReadBandwidth); + DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionLuma = %f\n", __func__, l->MaximumEffectiveCompressionLuma); + DML_LOG_VERBOSE("DML::%s: MaximumEffectiveCompressionChroma = %f\n", __func__, l->MaximumEffectiveCompressionChroma); + DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: AverageDCCZeroSizeFraction = %f\n", __func__, l->AverageDCCZeroSizeFraction); + + DML_LOG_VERBOSE("DML::%s: CompbufReservedSpace64B = %u (%f kbytes)\n", __func__, p->CompbufReservedSpace64B, p->CompbufReservedSpace64B * 64 / 1024.0); + DML_LOG_VERBOSE("DML::%s: CompbufReservedSpaceZs = %u\n", __func__, p->CompbufReservedSpaceZs); + DML_LOG_VERBOSE("DML::%s: CompressedBufferSizeInkByte = %u kbytes\n", __func__, p->CompressedBufferSizeInkByte); + DML_LOG_VERBOSE("DML::%s: ROBBufferSizeInKByte = %u kbytes\n", __func__, p->ROBBufferSizeInKByte); +#endif + if (l->AverageDCCZeroSizeFraction == 1) { + l->AverageZeroSizeCompressionRate = l->TotalZeroSizeRequestReadBandwidth / l->TotalZeroSizeCompressedReadBandwidth; + l->EffectiveCompressedBufferSize = (double)p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageZeroSizeCompressionRate + ((double)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 * l->AverageZeroSizeCompressionRate; + + + } else if (l->AverageDCCZeroSizeFraction > 0) { + l->AverageZeroSizeCompressionRate = l->TotalZeroSizeRequestReadBandwidth / l->TotalZeroSizeCompressedReadBandwidth; + l->EffectiveCompressedBufferSize = math_min2((double)p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate, + (double)p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)) + + (p->rob_alloc_compressed ? math_min2(((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * l->AverageDCCCompressionRate, + ((double)p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)) + : ((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); + + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate + 1 / l->AverageDCCCompressionRate)); + DML_LOG_VERBOSE("DML::%s: min 3 = %d\n", __func__, (p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64)); + DML_LOG_VERBOSE("DML::%s: min 4 = %f\n", __func__, (p->ZeroSizeBufferEntries - p->CompbufReservedSpaceZs) * 64 / (l->AverageDCCZeroSizeFraction / l->AverageZeroSizeCompressionRate)); +#endif + } else { + l->EffectiveCompressedBufferSize = math_min2((double)p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate, + (double)p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate) + + ((double)p->ROBBufferSizeInKByte * 1024 - p->CompbufReservedSpace64B * 64) * (p->rob_alloc_compressed ? l->AverageDCCCompressionRate : 1.0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: min 1 = %f\n", __func__, p->CompressedBufferSizeInkByte * 1024 * l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: min 2 = %f\n", __func__, p->MetaFIFOSizeInKEntries * 1024 * 64 * l->AverageDCCCompressionRate); +#endif + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: MetaFIFOSizeInKEntries = %u\n", __func__, p->MetaFIFOSizeInKEntries); + DML_LOG_VERBOSE("DML::%s: ZeroSizeBufferEntries = %u\n", __func__, p->ZeroSizeBufferEntries); + DML_LOG_VERBOSE("DML::%s: AverageZeroSizeCompressionRate = %f\n", __func__, l->AverageZeroSizeCompressionRate); + DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); +#endif + + *p->StutterPeriod = 0; + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { + l->LinesInDETY = ((double)p->DETBufferSizeY[k] + (p->UnboundedRequestEnabled == true ? l->EffectiveCompressedBufferSize : 0) * p->ReadBandwidthSurfaceLuma[k] / p->TotalDataReadBandwidth) / p->BytePerPixelDETY[k] / p->SwathWidthY[k]; + l->LinesInDETYRoundedDownToSwath = math_floor2(l->LinesInDETY, p->SwathHeightY[k]); + l->DETBufferingTimeY = l->LinesInDETYRoundedDownToSwath * ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) / p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferSizeY = %u (%u kbytes)\n", __func__, k, p->DETBufferSizeY[k], p->DETBufferSizeY[k] / 1024); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelDETY = %f\n", __func__, k, p->BytePerPixelDETY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthY = %u\n", __func__, k, p->SwathWidthY[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, ReadBandwidthSurfaceLuma = %f\n", __func__, k, p->ReadBandwidthSurfaceLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, p->TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETY = %f\n", __func__, k, l->LinesInDETY); + DML_LOG_VERBOSE("DML::%s: k=%u, LinesInDETYRoundedDownToSwath = %f\n", __func__, k, l->LinesInDETYRoundedDownToSwath); + DML_LOG_VERBOSE("DML::%s: k=%u, VRatio = %f\n", __func__, k, p->display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + DML_LOG_VERBOSE("DML::%s: k=%u, DETBufferingTimeY = %f\n", __func__, k, l->DETBufferingTimeY); +#endif + + if (!FoundCriticalSurface || l->DETBufferingTimeY < *p->StutterPeriod) { + bool isInterlaceTiming = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !p->ProgressiveToInterlaceUnitInOPP; + + FoundCriticalSurface = true; + *p->StutterPeriod = l->DETBufferingTimeY; + l->FrameTimeCriticalSurface = (isInterlaceTiming ? math_floor2((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total / 2.0, 1.0) : p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total) * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + l->VActiveTimeCriticalSurface = (isInterlaceTiming ? math_floor2((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_active / 2.0, 1.0) : p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_active) * (double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + l->BytePerPixelYCriticalSurface = p->BytePerPixelY[k]; + l->SwathWidthYCriticalSurface = p->SwathWidthY[k]; + l->SwathHeightYCriticalSurface = p->SwathHeightY[k]; + l->BlockWidth256BytesYCriticalSurface = p->BlockWidth256BytesY[k]; + l->DETBufferSizeYCriticalSurface = p->DETBufferSizeY[k]; + l->MinTTUVBlankCriticalSurface = p->MinTTUVBlank[k]; + l->SinglePlaneCriticalSurface = (p->ReadBandwidthSurfaceChroma[k] == 0); + l->SinglePipeCriticalSurface = (p->DPPPerSurface[k] == 1); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, FoundCriticalSurface = %u\n", __func__, k, FoundCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, StutterPeriod = %f\n", __func__, k, *p->StutterPeriod); + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlankCriticalSurface = %f\n", __func__, k, l->MinTTUVBlankCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, FrameTimeCriticalSurface= %f\n", __func__, k, l->FrameTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, VActiveTimeCriticalSurface = %f\n", __func__, k, l->VActiveTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, BytePerPixelYCriticalSurface = %u\n", __func__, k, l->BytePerPixelYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathWidthYCriticalSurface = %f\n", __func__, k, l->SwathWidthYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SwathHeightYCriticalSurface = %f\n", __func__, k, l->SwathHeightYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, BlockWidth256BytesYCriticalSurface = %u\n", __func__, k, l->BlockWidth256BytesYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SinglePlaneCriticalSurface = %u\n", __func__, k, l->SinglePlaneCriticalSurface); + DML_LOG_VERBOSE("DML::%s: k=%u, SinglePipeCriticalSurface = %u\n", __func__, k, l->SinglePipeCriticalSurface); +#endif + } + } + } + + // for bounded req, the stutter period is calculated only based on DET size, but during burst there can be some return inside ROB/compressed buffer + // stutter period is calculated only on the det sizing + // if (cdb + rob >= det) the stutter burst will be absorbed by the cdb + rob which is before decompress + // else + // the cdb + rob part will be in compressed rate with urg bw (idea bw) + // the det part will be return at uncompressed rate with 64B/dcfclk + // + // for unbounded req, the stutter period should be calculated as total of CDB+ROB+DET, so the term "PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer" + // should be == EffectiveCompressedBufferSize which will returned a compressed rate, the rest of stutter period is from the DET will be returned at uncompressed rate with 64B/dcfclk + + l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = math_min2(*p->StutterPeriod * p->TotalDataReadBandwidth, l->EffectiveCompressedBufferSize); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: AverageDCCCompressionRate = %f\n", __func__, l->AverageDCCCompressionRate); + DML_LOG_VERBOSE("DML::%s: StutterPeriod*TotalDataReadBandwidth = %f (%f kbytes)\n", __func__, *p->StutterPeriod * p->TotalDataReadBandwidth, (*p->StutterPeriod * p->TotalDataReadBandwidth) / 1024.0); + DML_LOG_VERBOSE("DML::%s: EffectiveCompressedBufferSize = %f (%f kbytes)\n", __func__, l->EffectiveCompressedBufferSize, l->EffectiveCompressedBufferSize / 1024.0); + DML_LOG_VERBOSE("DML::%s: PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer = %f (%f kbytes)\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / 1024); + DML_LOG_VERBOSE("DML::%s: ReturnBW = %f\n", __func__, p->ReturnBW); + DML_LOG_VERBOSE("DML::%s: TotalDataReadBandwidth = %f\n", __func__, p->TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: TotalRowReadBandwidth = %f\n", __func__, l->TotalRowReadBandwidth); + DML_LOG_VERBOSE("DML::%s: DCFCLK = %f\n", __func__, p->DCFCLK); +#endif + + l->StutterBurstTime = l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer + / (p->ReturnBW * (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)) + + (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) + / math_min2(p->DCFCLK * 64, p->ReturnBW * (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)) + + *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Part 1 = %f\n", __func__, l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer / p->ReturnBW / (p->hw_debug5 ? 1 : l->AverageDCCCompressionRate)); + DML_LOG_VERBOSE("DML::%s: Part 2 = %f\n", __func__, (*p->StutterPeriod * p->TotalDataReadBandwidth - l->PartOfUncompressedPixelBurstThatFitsInROBAndCompressedBuffer) / (p->DCFCLK * 64)); + DML_LOG_VERBOSE("DML::%s: Part 3 = %f\n", __func__, *p->StutterPeriod * l->TotalRowReadBandwidth / p->ReturnBW); + DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); +#endif + l->TotalActiveWriteback = 0; + memset(l->stream_visited, 0, DML2_MAX_PLANES * sizeof(bool)); + + for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { + if (!dml_is_phantom_pipe(&p->display_cfg->plane_descriptors[k])) { + if (!l->stream_visited[p->display_cfg->plane_descriptors[k].stream_index]) { + + if (p->display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) + l->TotalActiveWriteback = l->TotalActiveWriteback + 1; + + if (TotalNumberOfActiveOTG == 0) { // first otg + SinglePixelClock = ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + SingleHTotal = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total; + SingleVTotal = p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total; + } else if (SinglePixelClock != ((double)p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) || + SingleHTotal != p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.h_total || + SingleVTotal != p->display_cfg->stream_descriptors[p->display_cfg->plane_descriptors[k].stream_index].timing.v_total) { + SameTiming = false; + } + TotalNumberOfActiveOTG = TotalNumberOfActiveOTG + 1; + l->stream_visited[p->display_cfg->plane_descriptors[k].stream_index] = 1; + } + } + } + + if (l->TotalActiveWriteback == 0) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: SRExitTime = %f\n", __func__, p->SRExitTime); + DML_LOG_VERBOSE("DML::%s: SRExitZ8Time = %f\n", __func__, p->SRExitZ8Time); + DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); +#endif + *p->StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitTime + l->StutterBurstTime) / *p->StutterPeriod) * 100; + *p->Z8StutterEfficiencyNotIncludingVBlank = math_max2(0., 1 - (p->SRExitZ8Time + l->StutterBurstTime) / *p->StutterPeriod) * 100; + *p->NumberOfStutterBurstsPerFrame = (*p->StutterEfficiencyNotIncludingVBlank > 0 ? (unsigned int)(math_ceil2(l->VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); + *p->Z8NumberOfStutterBurstsPerFrame = (*p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? (unsigned int)(math_ceil2(l->VActiveTimeCriticalSurface / *p->StutterPeriod, 1)) : 0); + } else { + *p->StutterEfficiencyNotIncludingVBlank = 0.; + *p->Z8StutterEfficiencyNotIncludingVBlank = 0.; + *p->NumberOfStutterBurstsPerFrame = 0; + *p->Z8NumberOfStutterBurstsPerFrame = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: VActiveTimeCriticalSurface = %f\n", __func__, l->VActiveTimeCriticalSurface); + DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->NumberOfStutterBurstsPerFrame); + DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); +#endif + + if (*p->StutterEfficiencyNotIncludingVBlank > 0) { + if (!((p->SynchronizeTimings || TotalNumberOfActiveOTG == 1) && SameTiming)) { + *p->StutterEfficiency = *p->StutterEfficiencyNotIncludingVBlank; + } else { + *p->StutterEfficiency = (1 - (*p->NumberOfStutterBurstsPerFrame * p->SRExitTime + l->StutterBurstTime * l->VActiveTimeCriticalSurface / *p->StutterPeriod) / l->FrameTimeCriticalSurface) * 100; + } + } else { + *p->StutterEfficiency = 0; + *p->NumberOfStutterBurstsPerFrame = 0; + } + + if (*p->Z8StutterEfficiencyNotIncludingVBlank > 0) { + //LastZ8StutterPeriod = l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod; + if (!((p->SynchronizeTimings || TotalNumberOfActiveOTG == 1) && SameTiming)) { + *p->Z8StutterEfficiency = *p->Z8StutterEfficiencyNotIncludingVBlank; + } else { + *p->Z8StutterEfficiency = (1 - (*p->Z8NumberOfStutterBurstsPerFrame * p->SRExitZ8Time + l->StutterBurstTime * l->VActiveTimeCriticalSurface / *p->StutterPeriod) / l->FrameTimeCriticalSurface) * 100; + } + } else { + *p->Z8StutterEfficiency = 0.; + *p->Z8NumberOfStutterBurstsPerFrame = 0; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: TotalNumberOfActiveOTG = %u\n", __func__, TotalNumberOfActiveOTG); + DML_LOG_VERBOSE("DML::%s: SameTiming = %u\n", __func__, SameTiming); + DML_LOG_VERBOSE("DML::%s: SynchronizeTimings = %u\n", __func__, p->SynchronizeTimings); + DML_LOG_VERBOSE("DML::%s: LastZ8StutterPeriod = %f\n", __func__, *p->Z8StutterEfficiencyNotIncludingVBlank > 0 ? l->VActiveTimeCriticalSurface - (*p->Z8NumberOfStutterBurstsPerFrame - 1) * *p->StutterPeriod : 0); + DML_LOG_VERBOSE("DML::%s: Z8StutterEnterPlusExitWatermark = %f\n", __func__, p->Z8StutterEnterPlusExitWatermark); + DML_LOG_VERBOSE("DML::%s: StutterBurstTime = %f\n", __func__, l->StutterBurstTime); + DML_LOG_VERBOSE("DML::%s: StutterPeriod = %f\n", __func__, *p->StutterPeriod); + DML_LOG_VERBOSE("DML::%s: StutterEfficiency = %f\n", __func__, *p->StutterEfficiency); + DML_LOG_VERBOSE("DML::%s: Z8StutterEfficiency = %f\n", __func__, *p->Z8StutterEfficiency); + DML_LOG_VERBOSE("DML::%s: StutterEfficiencyNotIncludingVBlank = %f\n", __func__, *p->StutterEfficiencyNotIncludingVBlank); + DML_LOG_VERBOSE("DML::%s: Z8NumberOfStutterBurstsPerFrame = %u\n", __func__, *p->Z8NumberOfStutterBurstsPerFrame); +#endif + + *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = !(!p->UnboundedRequestEnabled && (p->NumberOfActiveSurfaces == 1) && l->SinglePlaneCriticalSurface && l->SinglePipeCriticalSurface); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: DETBufferSizeYCriticalSurface = %u\n", __func__, l->DETBufferSizeYCriticalSurface); + DML_LOG_VERBOSE("DML::%s: PixelChunkSizeInKByte = %u\n", __func__, p->PixelChunkSizeInKByte); + DML_LOG_VERBOSE("DML::%s: DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = %u\n", __func__, *p->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE); +#endif +} + +static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex *in_out_params) +{ + const struct dml2_display_cfg *display_cfg = in_out_params->in_display_cfg; + const struct dml2_mcg_min_clock_table *min_clk_table = in_out_params->min_clk_table; + const struct core_display_cfg_support_info *cfg_support_info = in_out_params->cfg_support_info; + struct dml2_core_internal_display_mode_lib *mode_lib = in_out_params->mode_lib; + struct dml2_display_cfg_programming *programming = in_out_params->programming; + + struct dml2_core_calcs_mode_programming_locals *s = &mode_lib->scratch.dml_core_mode_programming_locals; + struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params *CalculateWatermarks_params = &mode_lib->scratch.CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_params; + struct dml2_core_calcs_CalculateVMRowAndSwath_params *CalculateVMRowAndSwath_params = &mode_lib->scratch.CalculateVMRowAndSwath_params; + struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params *CalculateSwathAndDETConfiguration_params = &mode_lib->scratch.CalculateSwathAndDETConfiguration_params; + struct dml2_core_calcs_CalculateStutterEfficiency_params *CalculateStutterEfficiency_params = &mode_lib->scratch.CalculateStutterEfficiency_params; + struct dml2_core_calcs_CalculatePrefetchSchedule_params *CalculatePrefetchSchedule_params = &mode_lib->scratch.CalculatePrefetchSchedule_params; + struct dml2_core_calcs_CheckGlobalPrefetchAdmissibility_params *CheckGlobalPrefetchAdmissibility_params = &mode_lib->scratch.CheckGlobalPrefetchAdmissibility_params; + struct dml2_core_calcs_calculate_mcache_setting_params *calculate_mcache_setting_params = &mode_lib->scratch.calculate_mcache_setting_params; + struct dml2_core_calcs_calculate_tdlut_setting_params *calculate_tdlut_setting_params = &mode_lib->scratch.calculate_tdlut_setting_params; + struct dml2_core_shared_CalculateMetaAndPTETimes_params *CalculateMetaAndPTETimes_params = &mode_lib->scratch.CalculateMetaAndPTETimes_params; + struct dml2_core_calcs_calculate_peak_bandwidth_required_params *calculate_peak_bandwidth_params = &mode_lib->scratch.calculate_peak_bandwidth_params; + struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params *calculate_bytes_to_fetch_required_to_hide_latency_params = &mode_lib->scratch.calculate_bytes_to_fetch_required_to_hide_latency_params; + + unsigned int k; + bool must_support_iflip; + const long min_return_uclk_cycles = 83; + const long min_return_fclk_cycles = 75; + const double max_fclk_mhz = min_clk_table->max_clocks_khz.fclk / 1000.0; + double hard_minimum_dcfclk_mhz = (double)min_clk_table->dram_bw_table.entries[0].min_dcfclk_khz / 1000.0; + double max_uclk_mhz = 0; + double min_return_latency_in_DCFCLK_cycles = 0; + + DML_LOG_VERBOSE("DML::%s: --- START --- \n", __func__); + + memset(&mode_lib->scratch, 0, sizeof(struct dml2_core_internal_scratch)); + memset(&mode_lib->mp, 0, sizeof(struct dml2_core_internal_mode_program)); + + s->num_active_planes = display_cfg->num_planes; + get_stream_output_bpp(s->OutputBpp, display_cfg); + + mode_lib->mp.num_active_pipes = dml_get_num_active_pipes(display_cfg->num_planes, cfg_support_info); + dml_calc_pipe_plane_mapping(cfg_support_info, mode_lib->mp.pipe_plane); + + mode_lib->mp.Dcfclk = programming->min_clocks.dcn4x.active.dcfclk_khz / 1000.0; + mode_lib->mp.FabricClock = programming->min_clocks.dcn4x.active.fclk_khz / 1000.0; + mode_lib->mp.dram_bw_mbps = uclk_khz_to_dram_bw_mbps(programming->min_clocks.dcn4x.active.uclk_khz, &mode_lib->soc.clk_table.dram_config); + mode_lib->mp.uclk_freq_mhz = programming->min_clocks.dcn4x.active.uclk_khz / 1000.0; + mode_lib->mp.GlobalDPPCLK = programming->min_clocks.dcn4x.dpprefclk_khz / 1000.0; + s->SOCCLK = (double)programming->min_clocks.dcn4x.socclk_khz / 1000; + mode_lib->mp.qos_param_index = get_qos_param_index(programming->min_clocks.dcn4x.active.uclk_khz, mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params); + mode_lib->mp.active_min_uclk_dpm_index = get_active_min_uclk_dpm_index(programming->min_clocks.dcn4x.active.uclk_khz, &mode_lib->soc.clk_table); + + for (k = 0; k < s->num_active_planes; ++k) { + unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].odms_used <= 4); + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4 || + cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 2 || + cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); + + if (cfg_support_info->stream_support_info[stream_index].odms_used > 1) + DML_ASSERT(cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 1); + + switch (cfg_support_info->stream_support_info[stream_index].odms_used) { + case (4): + mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_4to1; + break; + case (3): + mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_3to1; + break; + case (2): + mode_lib->mp.ODMMode[k] = dml2_odm_mode_combine_2to1; + break; + default: + if (cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 4) + mode_lib->mp.ODMMode[k] = dml2_odm_mode_mso_1to4; + else if (cfg_support_info->stream_support_info[stream_index].num_odm_output_segments == 2) + mode_lib->mp.ODMMode[k] = dml2_odm_mode_mso_1to2; + else + mode_lib->mp.ODMMode[k] = dml2_odm_mode_bypass; + break; + } + } + + for (k = 0; k < s->num_active_planes; ++k) { + mode_lib->mp.NoOfDPP[k] = cfg_support_info->plane_support_info[k].dpps_used; + mode_lib->mp.Dppclk[k] = programming->plane_programming[k].min_clocks.dcn4x.dppclk_khz / 1000.0; + DML_ASSERT(mode_lib->mp.Dppclk[k] > 0); + } + + for (k = 0; k < s->num_active_planes; ++k) { + unsigned int stream_index = display_cfg->plane_descriptors[k].stream_index; + mode_lib->mp.DSCCLK[k] = programming->stream_programming[stream_index].min_clocks.dcn4x.dscclk_khz / 1000.0; + DML_LOG_VERBOSE("DML::%s: k=%d stream_index=%d, mode_lib->mp.DSCCLK = %f\n", __func__, k, stream_index, mode_lib->mp.DSCCLK[k]); + } + + mode_lib->mp.Dispclk = programming->min_clocks.dcn4x.dispclk_khz / 1000.0; + mode_lib->mp.DCFCLKDeepSleep = programming->min_clocks.dcn4x.deepsleep_dcfclk_khz / 1000.0; + + DML_ASSERT(mode_lib->mp.Dcfclk > 0); + DML_ASSERT(mode_lib->mp.FabricClock > 0); + DML_ASSERT(mode_lib->mp.dram_bw_mbps > 0); + DML_ASSERT(mode_lib->mp.uclk_freq_mhz > 0); + DML_ASSERT(mode_lib->mp.GlobalDPPCLK > 0); + DML_ASSERT(mode_lib->mp.Dispclk > 0); + DML_ASSERT(mode_lib->mp.DCFCLKDeepSleep > 0); + DML_ASSERT(s->SOCCLK > 0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: num_active_planes = %u\n", __func__, s->num_active_planes); + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %u\n", __func__, mode_lib->mp.num_active_pipes); + DML_LOG_VERBOSE("DML::%s: Dcfclk = %f\n", __func__, mode_lib->mp.Dcfclk); + DML_LOG_VERBOSE("DML::%s: FabricClock = %f\n", __func__, mode_lib->mp.FabricClock); + DML_LOG_VERBOSE("DML::%s: dram_bw_mbps = %f\n", __func__, mode_lib->mp.dram_bw_mbps); + DML_LOG_VERBOSE("DML::%s: uclk_freq_mhz = %f\n", __func__, mode_lib->mp.uclk_freq_mhz); + DML_LOG_VERBOSE("DML::%s: Dispclk = %f\n", __func__, mode_lib->mp.Dispclk); + for (k = 0; k < s->num_active_planes; ++k) { + DML_LOG_VERBOSE("DML::%s: Dppclk[%0d] = %f\n", __func__, k, mode_lib->mp.Dppclk[k]); + } + DML_LOG_VERBOSE("DML::%s: GlobalDPPCLK = %f\n", __func__, mode_lib->mp.GlobalDPPCLK); + DML_LOG_VERBOSE("DML::%s: DCFCLKDeepSleep = %f\n", __func__, mode_lib->mp.DCFCLKDeepSleep); + DML_LOG_VERBOSE("DML::%s: SOCCLK = %f\n", __func__, s->SOCCLK); + DML_LOG_VERBOSE("DML::%s: min_clk_index = %0d\n", __func__, in_out_params->min_clk_index); + DML_LOG_VERBOSE("DML::%s: min_clk_table min_fclk_khz = %ld\n", __func__, min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].min_fclk_khz); + DML_LOG_VERBOSE("DML::%s: min_clk_table uclk_mhz = %f\n", __func__, dram_bw_kbps_to_uclk_mhz(min_clk_table->dram_bw_table.entries[in_out_params->min_clk_index].pre_derate_dram_bw_kbps, &mode_lib->soc.clk_table.dram_config)); + for (k = 0; k < mode_lib->mp.num_active_pipes; ++k) { + DML_LOG_VERBOSE("DML::%s: pipe=%d is in plane=%d\n", __func__, k, mode_lib->mp.pipe_plane[k]); + DML_LOG_VERBOSE("DML::%s: Per-plane DPPPerSurface[%0d] = %d\n", __func__, k, mode_lib->mp.NoOfDPP[k]); + } + + for (k = 0; k < s->num_active_planes; k++) + DML_LOG_VERBOSE("DML::%s: plane_%d: reserved_vblank_time_ns = %lu\n", __func__, k, display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns); +#endif + + CalculateMaxDETAndMinCompressedBufferSize( + mode_lib->ip.config_return_buffer_size_in_kbytes, + mode_lib->ip.config_return_buffer_segment_size_in_kbytes, + mode_lib->ip.rob_buffer_size_kbytes, + mode_lib->ip.max_num_dpp, + display_cfg->overrides.hw.force_nom_det_size_kbytes.enable, + display_cfg->overrides.hw.force_nom_det_size_kbytes.value, + mode_lib->ip.dcn_mrq_present, + + /* Output */ + &s->MaxTotalDETInKByte, + &s->NomDETInKByte, + &s->MinCompressedBufferSizeInKByte); + + + PixelClockAdjustmentForProgressiveToInterlaceUnit(display_cfg, mode_lib->ip.ptoi_supported, s->PixelClockBackEnd); + + for (k = 0; k < s->num_active_planes; ++k) { + CalculateSinglePipeDPPCLKAndSCLThroughput( + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->ip.max_dchub_pscl_bw_pix_per_clk, + mode_lib->ip.max_pscl_lb_bw_pix_per_clk, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_taps, + + /* Output */ + &mode_lib->mp.PSCL_THROUGHPUT[k], + &mode_lib->mp.PSCL_THROUGHPUT_CHROMA[k], + &mode_lib->mp.DPPCLKUsingSingleDPP[k]); + } + + for (k = 0; k < s->num_active_planes; ++k) { + CalculateBytePerPixelAndBlockSizes( + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->plane_descriptors[k].surface.tiling, + display_cfg->plane_descriptors[k].surface.plane0.pitch, + display_cfg->plane_descriptors[k].surface.plane1.pitch, + + // Output + &mode_lib->mp.BytePerPixelY[k], + &mode_lib->mp.BytePerPixelC[k], + &mode_lib->mp.BytePerPixelInDETY[k], + &mode_lib->mp.BytePerPixelInDETC[k], + &mode_lib->mp.Read256BlockHeightY[k], + &mode_lib->mp.Read256BlockHeightC[k], + &mode_lib->mp.Read256BlockWidthY[k], + &mode_lib->mp.Read256BlockWidthC[k], + &mode_lib->mp.MacroTileHeightY[k], + &mode_lib->mp.MacroTileHeightC[k], + &mode_lib->mp.MacroTileWidthY[k], + &mode_lib->mp.MacroTileWidthC[k], + &mode_lib->mp.surf_linear128_l[k], + &mode_lib->mp.surf_linear128_c[k]); + } + + CalculateSwathWidth( + display_cfg, + false, // ForceSingleDPP + s->num_active_planes, + mode_lib->mp.ODMMode, + mode_lib->mp.BytePerPixelY, + mode_lib->mp.BytePerPixelC, + mode_lib->mp.Read256BlockHeightY, + mode_lib->mp.Read256BlockHeightC, + mode_lib->mp.Read256BlockWidthY, + mode_lib->mp.Read256BlockWidthC, + mode_lib->mp.surf_linear128_l, + mode_lib->mp.surf_linear128_c, + mode_lib->mp.NoOfDPP, + + /* Output */ + mode_lib->mp.req_per_swath_ub_l, + mode_lib->mp.req_per_swath_ub_c, + mode_lib->mp.SwathWidthSingleDPPY, + mode_lib->mp.SwathWidthSingleDPPC, + mode_lib->mp.SwathWidthY, + mode_lib->mp.SwathWidthC, + s->dummy_integer_array[0], // unsigned int MaximumSwathHeightY[] + s->dummy_integer_array[1], // unsigned int MaximumSwathHeightC[] + mode_lib->mp.swath_width_luma_ub, + mode_lib->mp.swath_width_chroma_ub); + + for (k = 0; k < s->num_active_planes; ++k) { + mode_lib->mp.cursor_bw[k] = display_cfg->plane_descriptors[k].cursor.num_cursors * display_cfg->plane_descriptors[k].cursor.cursor_width * display_cfg->plane_descriptors[k].cursor.cursor_bpp / 8.0 / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)); + mode_lib->mp.vactive_sw_bw_l[k] = mode_lib->mp.SwathWidthSingleDPPY[k] * mode_lib->mp.BytePerPixelY[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + mode_lib->mp.vactive_sw_bw_c[k] = mode_lib->mp.SwathWidthSingleDPPC[k] * mode_lib->mp.BytePerPixelC[k] / (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)) * display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_l[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: vactive_sw_bw_c[%i] = %fBps\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); + } + + CalculateSwathAndDETConfiguration_params->display_cfg = display_cfg; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSizeInKByte = mode_lib->ip.config_return_buffer_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->MaxTotalDETInKByte = s->MaxTotalDETInKByte; + CalculateSwathAndDETConfiguration_params->MinCompressedBufferSizeInKByte = s->MinCompressedBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CalculateSwathAndDETConfiguration_params->pixel_chunk_size_kbytes = mode_lib->ip.pixel_chunk_size_kbytes; + CalculateSwathAndDETConfiguration_params->ForceSingleDPP = false; + CalculateSwathAndDETConfiguration_params->NumberOfActiveSurfaces = s->num_active_planes; + CalculateSwathAndDETConfiguration_params->nomDETInKByte = s->NomDETInKByte; + CalculateSwathAndDETConfiguration_params->ConfigReturnBufferSegmentSizeInkByte = mode_lib->ip.config_return_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->CompressedBufferSegmentSizeInkByte = mode_lib->ip.compressed_buffer_segment_size_in_kbytes; + CalculateSwathAndDETConfiguration_params->ReadBandwidthLuma = mode_lib->mp.vactive_sw_bw_l; + CalculateSwathAndDETConfiguration_params->ReadBandwidthChroma = mode_lib->mp.vactive_sw_bw_c; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthLuma = s->dummy_single_array[0]; + CalculateSwathAndDETConfiguration_params->MaximumSwathWidthChroma = s->dummy_single_array[1]; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightY = mode_lib->mp.Read256BlockHeightY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockHeightC = mode_lib->mp.Read256BlockHeightC; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthY = mode_lib->mp.Read256BlockWidthY; + CalculateSwathAndDETConfiguration_params->Read256BytesBlockWidthC = mode_lib->mp.Read256BlockWidthC; + CalculateSwathAndDETConfiguration_params->surf_linear128_l = mode_lib->mp.surf_linear128_l; + CalculateSwathAndDETConfiguration_params->surf_linear128_c = mode_lib->mp.surf_linear128_c; + CalculateSwathAndDETConfiguration_params->ODMMode = mode_lib->mp.ODMMode; + CalculateSwathAndDETConfiguration_params->DPPPerSurface = mode_lib->mp.NoOfDPP; + CalculateSwathAndDETConfiguration_params->BytePerPixY = mode_lib->mp.BytePerPixelY; + CalculateSwathAndDETConfiguration_params->BytePerPixC = mode_lib->mp.BytePerPixelC; + CalculateSwathAndDETConfiguration_params->BytePerPixDETY = mode_lib->mp.BytePerPixelInDETY; + CalculateSwathAndDETConfiguration_params->BytePerPixDETC = mode_lib->mp.BytePerPixelInDETC; + CalculateSwathAndDETConfiguration_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + // output + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_l = mode_lib->mp.req_per_swath_ub_l; + CalculateSwathAndDETConfiguration_params->req_per_swath_ub_c = mode_lib->mp.req_per_swath_ub_c; + CalculateSwathAndDETConfiguration_params->swath_width_luma_ub = s->dummy_long_array[0]; + CalculateSwathAndDETConfiguration_params->swath_width_chroma_ub = s->dummy_long_array[1]; + CalculateSwathAndDETConfiguration_params->SwathWidth = s->dummy_long_array[2]; + CalculateSwathAndDETConfiguration_params->SwathWidthChroma = s->dummy_long_array[3]; + CalculateSwathAndDETConfiguration_params->SwathHeightY = mode_lib->mp.SwathHeightY; + CalculateSwathAndDETConfiguration_params->SwathHeightC = mode_lib->mp.SwathHeightC; + CalculateSwathAndDETConfiguration_params->request_size_bytes_luma = mode_lib->mp.request_size_bytes_luma; + CalculateSwathAndDETConfiguration_params->request_size_bytes_chroma = mode_lib->mp.request_size_bytes_chroma; + CalculateSwathAndDETConfiguration_params->DETBufferSizeInKByte = mode_lib->mp.DETBufferSizeInKByte; + CalculateSwathAndDETConfiguration_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; + CalculateSwathAndDETConfiguration_params->DETBufferSizeC = mode_lib->mp.DETBufferSizeC; + CalculateSwathAndDETConfiguration_params->full_swath_bytes_l = s->full_swath_bytes_l; + CalculateSwathAndDETConfiguration_params->full_swath_bytes_c = s->full_swath_bytes_c; + CalculateSwathAndDETConfiguration_params->UnboundedRequestEnabled = &mode_lib->mp.UnboundedRequestEnabled; + CalculateSwathAndDETConfiguration_params->compbuf_reserved_space_64b = &mode_lib->mp.compbuf_reserved_space_64b; + CalculateSwathAndDETConfiguration_params->hw_debug5 = &mode_lib->mp.hw_debug5; + CalculateSwathAndDETConfiguration_params->CompressedBufferSizeInkByte = &mode_lib->mp.CompressedBufferSizeInkByte; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupportPerSurface = &s->dummy_boolean_array[0][0]; + CalculateSwathAndDETConfiguration_params->ViewportSizeSupport = &s->dummy_boolean[0]; + + // Calculate DET size, swath height here. + CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); + + // DSC Delay + for (k = 0; k < s->num_active_planes; ++k) { + mode_lib->mp.DSCDelay[k] = DSCDelayRequirement(cfg_support_info->stream_support_info[display_cfg->plane_descriptors[k].stream_index].dsc_enable, + mode_lib->mp.ODMMode[k], + mode_lib->ip.maximum_dsc_bits_per_component, + s->OutputBpp[k], + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total, + cfg_support_info->stream_support_info[display_cfg->plane_descriptors[k].stream_index].num_dsc_slices, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_encoder, + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + s->PixelClockBackEnd[k]); + } + + // Prefetch + if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0) { + for (k = 0; k < s->num_active_planes; ++k) + mode_lib->mp.SurfaceSizeInTheMALL[k] = 0; + } else { + CalculateSurfaceSizeInMall( + display_cfg, + s->num_active_planes, + mode_lib->soc.mall_allocated_for_dcn_mbytes, + mode_lib->mp.BytePerPixelY, + mode_lib->mp.BytePerPixelC, + mode_lib->mp.Read256BlockWidthY, + mode_lib->mp.Read256BlockWidthC, + mode_lib->mp.Read256BlockHeightY, + mode_lib->mp.Read256BlockHeightC, + mode_lib->mp.MacroTileWidthY, + mode_lib->mp.MacroTileWidthC, + mode_lib->mp.MacroTileHeightY, + mode_lib->mp.MacroTileHeightC, + + /* Output */ + mode_lib->mp.SurfaceSizeInTheMALL, + &s->dummy_boolean[0]); /* bool *ExceededMALLSize */ + } + + for (k = 0; k < s->num_active_planes; ++k) { + s->SurfaceParameters[k].PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + s->SurfaceParameters[k].DPPPerSurface = mode_lib->mp.NoOfDPP[k]; + s->SurfaceParameters[k].RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; + s->SurfaceParameters[k].ViewportHeight = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + s->SurfaceParameters[k].ViewportHeightC = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; + s->SurfaceParameters[k].BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY[k]; + s->SurfaceParameters[k].BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY[k]; + s->SurfaceParameters[k].BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC[k]; + s->SurfaceParameters[k].BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC[k]; + s->SurfaceParameters[k].BlockWidthY = mode_lib->mp.MacroTileWidthY[k]; + s->SurfaceParameters[k].BlockHeightY = mode_lib->mp.MacroTileHeightY[k]; + s->SurfaceParameters[k].BlockWidthC = mode_lib->mp.MacroTileWidthC[k]; + s->SurfaceParameters[k].BlockHeightC = mode_lib->mp.MacroTileHeightC[k]; + s->SurfaceParameters[k].InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; + s->SurfaceParameters[k].HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; + s->SurfaceParameters[k].DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; + s->SurfaceParameters[k].SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; + s->SurfaceParameters[k].SurfaceTiling = display_cfg->plane_descriptors[k].surface.tiling; + s->SurfaceParameters[k].BytePerPixelY = mode_lib->mp.BytePerPixelY[k]; + s->SurfaceParameters[k].BytePerPixelC = mode_lib->mp.BytePerPixelC[k]; + s->SurfaceParameters[k].ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + s->SurfaceParameters[k].VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + s->SurfaceParameters[k].VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + s->SurfaceParameters[k].VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + s->SurfaceParameters[k].VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + s->SurfaceParameters[k].PitchY = display_cfg->plane_descriptors[k].surface.plane0.pitch; + s->SurfaceParameters[k].PitchC = display_cfg->plane_descriptors[k].surface.plane1.pitch; + s->SurfaceParameters[k].ViewportStationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; + s->SurfaceParameters[k].ViewportXStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; + s->SurfaceParameters[k].ViewportYStart = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; + s->SurfaceParameters[k].ViewportXStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + s->SurfaceParameters[k].ViewportYStartC = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + s->SurfaceParameters[k].FORCE_ONE_ROW_FOR_FRAME = display_cfg->plane_descriptors[k].overrides.hw.force_one_row_for_frame; + s->SurfaceParameters[k].SwathHeightY = mode_lib->mp.SwathHeightY[k]; + s->SurfaceParameters[k].SwathHeightC = mode_lib->mp.SwathHeightC[k]; + s->SurfaceParameters[k].DCCMetaPitchY = display_cfg->plane_descriptors[k].surface.dcc.plane0.pitch; + s->SurfaceParameters[k].DCCMetaPitchC = display_cfg->plane_descriptors[k].surface.dcc.plane1.pitch; + } + + CalculateVMRowAndSwath_params->display_cfg = display_cfg; + CalculateVMRowAndSwath_params->NumberOfActiveSurfaces = s->num_active_planes; + CalculateVMRowAndSwath_params->myPipe = s->SurfaceParameters; + CalculateVMRowAndSwath_params->SurfaceSizeInMALL = mode_lib->mp.SurfaceSizeInTheMALL; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsLuma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; + CalculateVMRowAndSwath_params->PTEBufferSizeInRequestsChroma = mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma; + CalculateVMRowAndSwath_params->MALLAllocatedForDCN = mode_lib->soc.mall_allocated_for_dcn_mbytes; + CalculateVMRowAndSwath_params->SwathWidthY = mode_lib->mp.SwathWidthY; + CalculateVMRowAndSwath_params->SwathWidthC = mode_lib->mp.SwathWidthC; + CalculateVMRowAndSwath_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeBytes = mode_lib->ip.dcc_meta_buffer_size_bytes; + CalculateVMRowAndSwath_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + // output + CalculateVMRowAndSwath_params->PTEBufferSizeNotExceeded = s->dummy_boolean_array[0]; + CalculateVMRowAndSwath_params->dpte_row_width_luma_ub = mode_lib->mp.dpte_row_width_luma_ub; + CalculateVMRowAndSwath_params->dpte_row_width_chroma_ub = mode_lib->mp.dpte_row_width_chroma_ub; + CalculateVMRowAndSwath_params->dpte_row_height_luma = mode_lib->mp.dpte_row_height; + CalculateVMRowAndSwath_params->dpte_row_height_chroma = mode_lib->mp.dpte_row_height_chroma; + CalculateVMRowAndSwath_params->dpte_row_height_linear_luma = mode_lib->mp.dpte_row_height_linear; + CalculateVMRowAndSwath_params->dpte_row_height_linear_chroma = mode_lib->mp.dpte_row_height_linear_chroma; + CalculateVMRowAndSwath_params->vm_group_bytes = mode_lib->mp.vm_group_bytes; + CalculateVMRowAndSwath_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; + CalculateVMRowAndSwath_params->PixelPTEReqWidthY = mode_lib->mp.PixelPTEReqWidthY; + CalculateVMRowAndSwath_params->PixelPTEReqHeightY = mode_lib->mp.PixelPTEReqHeightY; + CalculateVMRowAndSwath_params->PTERequestSizeY = mode_lib->mp.PTERequestSizeY; + CalculateVMRowAndSwath_params->PixelPTEReqWidthC = mode_lib->mp.PixelPTEReqWidthC; + CalculateVMRowAndSwath_params->PixelPTEReqHeightC = mode_lib->mp.PixelPTEReqHeightC; + CalculateVMRowAndSwath_params->PTERequestSizeC = mode_lib->mp.PTERequestSizeC; + CalculateVMRowAndSwath_params->vmpg_width_y = s->vmpg_width_y; + CalculateVMRowAndSwath_params->vmpg_height_y = s->vmpg_height_y; + CalculateVMRowAndSwath_params->vmpg_width_c = s->vmpg_width_c; + CalculateVMRowAndSwath_params->vmpg_height_c = s->vmpg_height_c; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_l = mode_lib->mp.dpde0_bytes_per_frame_ub_l; + CalculateVMRowAndSwath_params->dpde0_bytes_per_frame_ub_c = mode_lib->mp.dpde0_bytes_per_frame_ub_c; + CalculateVMRowAndSwath_params->PrefetchSourceLinesY = mode_lib->mp.PrefetchSourceLinesY; + CalculateVMRowAndSwath_params->PrefetchSourceLinesC = mode_lib->mp.PrefetchSourceLinesC; + CalculateVMRowAndSwath_params->VInitPreFillY = mode_lib->mp.VInitPreFillY; + CalculateVMRowAndSwath_params->VInitPreFillC = mode_lib->mp.VInitPreFillC; + CalculateVMRowAndSwath_params->MaxNumSwathY = mode_lib->mp.MaxNumSwathY; + CalculateVMRowAndSwath_params->MaxNumSwathC = mode_lib->mp.MaxNumSwathC; + CalculateVMRowAndSwath_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; + CalculateVMRowAndSwath_params->PixelPTEBytesPerRow = mode_lib->mp.PixelPTEBytesPerRow; + CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_l = s->dpte_row_bytes_per_row_l; + CalculateVMRowAndSwath_params->dpte_row_bytes_per_row_c = s->dpte_row_bytes_per_row_c; + CalculateVMRowAndSwath_params->vm_bytes = mode_lib->mp.vm_bytes; + CalculateVMRowAndSwath_params->use_one_row_for_frame = mode_lib->mp.use_one_row_for_frame; + CalculateVMRowAndSwath_params->use_one_row_for_frame_flip = mode_lib->mp.use_one_row_for_frame_flip; + CalculateVMRowAndSwath_params->is_using_mall_for_ss = mode_lib->mp.is_using_mall_for_ss; + CalculateVMRowAndSwath_params->PTE_BUFFER_MODE = mode_lib->mp.PTE_BUFFER_MODE; + CalculateVMRowAndSwath_params->BIGK_FRAGMENT_SIZE = mode_lib->mp.BIGK_FRAGMENT_SIZE; + CalculateVMRowAndSwath_params->DCCMetaBufferSizeNotExceeded = s->dummy_boolean_array[1]; + CalculateVMRowAndSwath_params->meta_row_bw = mode_lib->mp.meta_row_bw; + CalculateVMRowAndSwath_params->meta_row_bytes = mode_lib->mp.meta_row_bytes; + CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; + CalculateVMRowAndSwath_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; + CalculateVMRowAndSwath_params->meta_req_width_luma = mode_lib->mp.meta_req_width; + CalculateVMRowAndSwath_params->meta_req_height_luma = mode_lib->mp.meta_req_height; + CalculateVMRowAndSwath_params->meta_row_width_luma = mode_lib->mp.meta_row_width; + CalculateVMRowAndSwath_params->meta_row_height_luma = mode_lib->mp.meta_row_height; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_l = mode_lib->mp.meta_pte_bytes_per_frame_ub_l; + CalculateVMRowAndSwath_params->meta_req_width_chroma = mode_lib->mp.meta_req_width_chroma; + CalculateVMRowAndSwath_params->meta_row_height_chroma = mode_lib->mp.meta_row_height_chroma; + CalculateVMRowAndSwath_params->meta_row_width_chroma = mode_lib->mp.meta_row_width_chroma; + CalculateVMRowAndSwath_params->meta_req_height_chroma = mode_lib->mp.meta_req_height_chroma; + CalculateVMRowAndSwath_params->meta_pte_bytes_per_frame_ub_c = mode_lib->mp.meta_pte_bytes_per_frame_ub_c; + + CalculateVMRowAndSwath(&mode_lib->scratch, CalculateVMRowAndSwath_params); + + memset(calculate_mcache_setting_params, 0, sizeof(struct dml2_core_calcs_calculate_mcache_setting_params)); + if (mode_lib->soc.mall_allocated_for_dcn_mbytes == 0 || mode_lib->ip.dcn_mrq_present) { + for (k = 0; k < s->num_active_planes; k++) { + mode_lib->mp.mall_prefetch_sdp_overhead_factor[k] = 1.0; + mode_lib->mp.mall_prefetch_dram_overhead_factor[k] = 1.0; + mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0[k] = 1.0; + mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0[k] = 1.0; + mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1[k] = 1.0; + mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1[k] = 1.0; + } + } else { + for (k = 0; k < s->num_active_planes; k++) { + calculate_mcache_setting_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; + calculate_mcache_setting_params->num_chans = mode_lib->soc.clk_table.dram_config.channel_count; + calculate_mcache_setting_params->mem_word_bytes = mode_lib->soc.mem_word_bytes; + calculate_mcache_setting_params->mcache_size_bytes = mode_lib->soc.mcache_size_bytes; + calculate_mcache_setting_params->mcache_line_size_bytes = mode_lib->soc.mcache_line_size_bytes; + calculate_mcache_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; + calculate_mcache_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + + calculate_mcache_setting_params->source_format = display_cfg->plane_descriptors[k].pixel_format; + calculate_mcache_setting_params->surf_vert = dml_is_vertical_rotation(display_cfg->plane_descriptors[k].composition.rotation_angle); + calculate_mcache_setting_params->vp_stationary = display_cfg->plane_descriptors[k].composition.viewport.stationary; + calculate_mcache_setting_params->tiling_mode = display_cfg->plane_descriptors[k].surface.tiling; + calculate_mcache_setting_params->imall_enable = mode_lib->ip.imall_supported && display_cfg->plane_descriptors[k].overrides.legacy_svp_config == dml2_svp_mode_override_imall; + + calculate_mcache_setting_params->vp_start_x_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.x_start; + calculate_mcache_setting_params->vp_start_y_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.y_start; + calculate_mcache_setting_params->full_vp_width_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.width; + calculate_mcache_setting_params->full_vp_height_l = display_cfg->plane_descriptors[k].composition.viewport.plane0.height; + calculate_mcache_setting_params->blk_width_l = mode_lib->mp.MacroTileWidthY[k]; + calculate_mcache_setting_params->blk_height_l = mode_lib->mp.MacroTileHeightY[k]; + calculate_mcache_setting_params->vmpg_width_l = s->vmpg_width_y[k]; + calculate_mcache_setting_params->vmpg_height_l = s->vmpg_height_y[k]; + calculate_mcache_setting_params->full_swath_bytes_l = s->full_swath_bytes_l[k]; + calculate_mcache_setting_params->bytes_per_pixel_l = mode_lib->mp.BytePerPixelY[k]; + + calculate_mcache_setting_params->vp_start_x_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + calculate_mcache_setting_params->vp_start_y_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.y_start; + calculate_mcache_setting_params->full_vp_width_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.width; + calculate_mcache_setting_params->full_vp_height_c = display_cfg->plane_descriptors[k].composition.viewport.plane1.height; + calculate_mcache_setting_params->blk_width_c = mode_lib->mp.MacroTileWidthC[k]; + calculate_mcache_setting_params->blk_height_c = mode_lib->mp.MacroTileHeightC[k]; + calculate_mcache_setting_params->vmpg_width_c = s->vmpg_width_c[k]; + calculate_mcache_setting_params->vmpg_height_c = s->vmpg_height_c[k]; + calculate_mcache_setting_params->full_swath_bytes_c = s->full_swath_bytes_c[k]; + calculate_mcache_setting_params->bytes_per_pixel_c = mode_lib->mp.BytePerPixelC[k]; + + // output + calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_l = &mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0[k]; + calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_l = &mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0[k]; + calculate_mcache_setting_params->dcc_dram_bw_nom_overhead_factor_c = &mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1[k]; + calculate_mcache_setting_params->dcc_dram_bw_pref_overhead_factor_c = &mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1[k]; + + calculate_mcache_setting_params->num_mcaches_l = &mode_lib->mp.num_mcaches_l[k]; + calculate_mcache_setting_params->mcache_row_bytes_l = &mode_lib->mp.mcache_row_bytes_l[k]; + calculate_mcache_setting_params->mcache_row_bytes_per_channel_l = &mode_lib->mp.mcache_row_bytes_per_channel_l[k]; + calculate_mcache_setting_params->mcache_offsets_l = mode_lib->mp.mcache_offsets_l[k]; + calculate_mcache_setting_params->mcache_shift_granularity_l = &mode_lib->mp.mcache_shift_granularity_l[k]; + + calculate_mcache_setting_params->num_mcaches_c = &mode_lib->mp.num_mcaches_c[k]; + calculate_mcache_setting_params->mcache_row_bytes_c = &mode_lib->mp.mcache_row_bytes_c[k]; + calculate_mcache_setting_params->mcache_row_bytes_per_channel_c = &mode_lib->mp.mcache_row_bytes_per_channel_c[k]; + calculate_mcache_setting_params->mcache_offsets_c = mode_lib->mp.mcache_offsets_c[k]; + calculate_mcache_setting_params->mcache_shift_granularity_c = &mode_lib->mp.mcache_shift_granularity_c[k]; + + calculate_mcache_setting_params->mall_comb_mcache_l = &mode_lib->mp.mall_comb_mcache_l[k]; + calculate_mcache_setting_params->mall_comb_mcache_c = &mode_lib->mp.mall_comb_mcache_c[k]; + calculate_mcache_setting_params->lc_comb_mcache = &mode_lib->mp.lc_comb_mcache[k]; + calculate_mcache_setting(&mode_lib->scratch, calculate_mcache_setting_params); + } + + calculate_mall_bw_overhead_factor( + mode_lib->mp.mall_prefetch_sdp_overhead_factor, + mode_lib->mp.mall_prefetch_dram_overhead_factor, + + // input + display_cfg, + s->num_active_planes); + } + + // Calculate all the bandwidth availabe + calculate_bandwidth_available( + mode_lib->mp.avg_bandwidth_available_min, + mode_lib->mp.avg_bandwidth_available, + mode_lib->mp.urg_bandwidth_available_min, + mode_lib->mp.urg_bandwidth_available, + mode_lib->mp.urg_bandwidth_available_vm_only, + mode_lib->mp.urg_bandwidth_available_pixel_and_vm, + + &mode_lib->soc, + display_cfg->hostvm_enable, + mode_lib->mp.Dcfclk, + mode_lib->mp.FabricClock, + mode_lib->mp.dram_bw_mbps); + + + calculate_hostvm_inefficiency_factor( + &s->HostVMInefficiencyFactor, + &s->HostVMInefficiencyFactorPrefetch, + + display_cfg->gpuvm_enable, + display_cfg->hostvm_enable, + mode_lib->ip.remote_iommu_outstanding_translations, + mode_lib->soc.max_outstanding_reqs, + mode_lib->mp.urg_bandwidth_available_pixel_and_vm[dml2_core_internal_soc_state_sys_active], + mode_lib->mp.urg_bandwidth_available_vm_only[dml2_core_internal_soc_state_sys_active]); + + s->TotalDCCActiveDPP = 0; + s->TotalActiveDPP = 0; + for (k = 0; k < s->num_active_planes; ++k) { + s->TotalActiveDPP = s->TotalActiveDPP + mode_lib->mp.NoOfDPP[k]; + if (display_cfg->plane_descriptors[k].surface.dcc.enable) + s->TotalDCCActiveDPP = s->TotalDCCActiveDPP + mode_lib->mp.NoOfDPP[k]; + } + // Calculate tdlut schedule related terms + for (k = 0; k <= s->num_active_planes - 1; k++) { + calculate_tdlut_setting_params->dispclk_mhz = mode_lib->mp.Dispclk; + calculate_tdlut_setting_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + calculate_tdlut_setting_params->tdlut_width_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_width_mode; + calculate_tdlut_setting_params->tdlut_addressing_mode = display_cfg->plane_descriptors[k].tdlut.tdlut_addressing_mode; + calculate_tdlut_setting_params->cursor_buffer_size = mode_lib->ip.cursor_buffer_size; + calculate_tdlut_setting_params->gpuvm_enable = display_cfg->gpuvm_enable; + calculate_tdlut_setting_params->gpuvm_page_size_kbytes = display_cfg->plane_descriptors[k].overrides.gpuvm_min_page_size_kbytes; + + // output + calculate_tdlut_setting_params->tdlut_pte_bytes_per_frame = &s->tdlut_pte_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_frame = &s->tdlut_bytes_per_frame[k]; + calculate_tdlut_setting_params->tdlut_groups_per_2row_ub = &s->tdlut_groups_per_2row_ub[k]; + calculate_tdlut_setting_params->tdlut_opt_time = &s->tdlut_opt_time[k]; + calculate_tdlut_setting_params->tdlut_drain_time = &s->tdlut_drain_time[k]; + calculate_tdlut_setting_params->tdlut_bytes_to_deliver = &s->tdlut_bytes_to_deliver[k]; + calculate_tdlut_setting_params->tdlut_bytes_per_group = &s->tdlut_bytes_per_group[k]; + calculate_tdlut_setting(&mode_lib->scratch, calculate_tdlut_setting_params); + } + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn3) + s->ReorderingBytes = (unsigned int)(mode_lib->soc.clk_table.dram_config.channel_count * math_max3(mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_only_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_out_of_order_return_per_channel_vm_only_bytes)); + + CalculateExtraLatency( + display_cfg, + mode_lib->ip.rob_buffer_size_kbytes, + mode_lib->soc.qos_parameters.qos_params.dcn32x.loaded_round_trip_latency_fclk_cycles, + s->ReorderingBytes, + mode_lib->mp.Dcfclk, + mode_lib->mp.FabricClock, + mode_lib->ip.pixel_chunk_size_kbytes, + mode_lib->mp.urg_bandwidth_available_min[dml2_core_internal_soc_state_sys_active], + s->num_active_planes, + mode_lib->mp.NoOfDPP, + mode_lib->mp.dpte_group_bytes, + s->tdlut_bytes_per_group, + s->HostVMInefficiencyFactor, + s->HostVMInefficiencyFactorPrefetch, + mode_lib->soc.hostvm_min_page_size_kbytes, + mode_lib->soc.qos_parameters.qos_type, + !(display_cfg->overrides.max_outstanding_when_urgent_expected_disable), + mode_lib->soc.max_outstanding_reqs, + mode_lib->mp.request_size_bytes_luma, + mode_lib->mp.request_size_bytes_chroma, + mode_lib->ip.meta_chunk_size_kbytes, + mode_lib->ip.dchub_arb_to_ret_delay, + mode_lib->mp.TripToMemory, + mode_lib->ip.hostvm_mode, + + // output + &mode_lib->mp.ExtraLatency, + &mode_lib->mp.ExtraLatency_sr, + &mode_lib->mp.ExtraLatencyPrefetch); + + mode_lib->mp.TCalc = 24.0 / mode_lib->mp.DCFCLKDeepSleep; + + for (k = 0; k < s->num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + mode_lib->mp.WritebackDelay[k] = + mode_lib->soc.qos_parameters.writeback.base_latency_us + + CalculateWriteBackDelay( + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].pixel_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].h_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_ratio, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].v_taps, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_width, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].output_height, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.writeback_stream[0].input_height, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) / mode_lib->mp.Dispclk; + } else + mode_lib->mp.WritebackDelay[k] = 0; + } + + /* VActive bytes to fetch for UCLK P-State */ + calculate_bytes_to_fetch_required_to_hide_latency_params->display_cfg = display_cfg; + calculate_bytes_to_fetch_required_to_hide_latency_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + calculate_bytes_to_fetch_required_to_hide_latency_params->num_active_planes = s->num_active_planes; + calculate_bytes_to_fetch_required_to_hide_latency_params->num_of_dpp = mode_lib->mp.NoOfDPP; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_l = mode_lib->mp.meta_row_height; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_height_c = mode_lib->mp.meta_row_height_chroma; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_l = s->meta_row_bytes_per_row_ub_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->meta_row_bytes_per_row_ub_c = s->meta_row_bytes_per_row_ub_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_l = mode_lib->mp.dpte_row_height; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_row_height_c = mode_lib->mp.dpte_row_height_chroma; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_l = s->dpte_row_bytes_per_row_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->dpte_bytes_per_row_c = s->dpte_row_bytes_per_row_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_l = mode_lib->mp.BytePerPixelY; + calculate_bytes_to_fetch_required_to_hide_latency_params->byte_per_pix_c = mode_lib->mp.BytePerPixelC; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_l = mode_lib->mp.SwathWidthY; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->mp.SwathWidthC; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->mp.SwathHeightY; + calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->mp.SwathHeightC; + calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + + /* outputs */ + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; + + calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); + + /* Excess VActive bandwidth required to fill DET */ + calculate_excess_vactive_bandwidth_required( + display_cfg, + s->num_active_planes, + s->pstate_bytes_required_l, + s->pstate_bytes_required_c, + /* outputs */ + mode_lib->mp.excess_vactive_fill_bw_l, + mode_lib->mp.excess_vactive_fill_bw_c); + + mode_lib->mp.UrgentLatency = CalculateUrgentLatency( + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_pixel_vm_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.base_latency_vm_us, + mode_lib->soc.do_urgent_latency_adjustment, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_fclk_us, + mode_lib->soc.qos_parameters.qos_params.dcn32x.urgent_latency_us.scaling_factor_mhz, + mode_lib->mp.FabricClock, + mode_lib->mp.uclk_freq_mhz, + mode_lib->soc.qos_parameters.qos_type, + mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].urgent_ramp_uclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_urgent_ramp_latency_margin, + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); + + mode_lib->mp.TripToMemory = CalculateTripToMemory( + mode_lib->mp.UrgentLatency, + mode_lib->mp.FabricClock, + mode_lib->mp.uclk_freq_mhz, + mode_lib->soc.qos_parameters.qos_type, + mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].trip_to_memory_uclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.max_round_trip_to_furthest_cs_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.mall_overhead_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); + + mode_lib->mp.TripToMemory = math_max2(mode_lib->mp.UrgentLatency, mode_lib->mp.TripToMemory); + + mode_lib->mp.MetaTripToMemory = CalculateMetaTripToMemory( + mode_lib->mp.UrgentLatency, + mode_lib->mp.FabricClock, + mode_lib->mp.uclk_freq_mhz, + mode_lib->soc.qos_parameters.qos_type, + mode_lib->soc.qos_parameters.qos_params.dcn4x.per_uclk_dpm_params[mode_lib->mp.qos_param_index].meta_trip_to_memory_uclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.meta_trip_adder_fclk_cycles, + mode_lib->soc.qos_parameters.qos_params.dcn4x.umc_max_latency_margin, + mode_lib->soc.qos_parameters.qos_params.dcn4x.fabric_max_transport_latency_margin); + + for (k = 0; k < s->num_active_planes; ++k) { + bool cursor_not_enough_urgent_latency_hiding = false; + s->line_times[k] = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + + s->pixel_format[k] = display_cfg->plane_descriptors[k].pixel_format; + + s->lb_source_lines_l[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->mp.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane0.width, + display_cfg->plane_descriptors[k].composition.viewport.plane0.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + s->lb_source_lines_c[k] = get_num_lb_source_lines(mode_lib->ip.max_line_buffer_lines, mode_lib->ip.line_buffer_size_bits, + mode_lib->mp.NoOfDPP[k], + display_cfg->plane_descriptors[k].composition.viewport.plane1.width, + display_cfg->plane_descriptors[k].composition.viewport.plane1.height, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.h_ratio, + display_cfg->plane_descriptors[k].composition.rotation_angle); + + if (display_cfg->plane_descriptors[k].cursor.num_cursors > 0) { + calculate_cursor_req_attributes( + display_cfg->plane_descriptors[k].cursor.cursor_width, + display_cfg->plane_descriptors[k].cursor.cursor_bpp, + + // output + &s->cursor_lines_per_chunk[k], + &s->cursor_bytes_per_line[k], + &s->cursor_bytes_per_chunk[k], + &s->cursor_bytes[k]); + + calculate_cursor_urgent_burst_factor( + mode_lib->ip.cursor_buffer_size, + display_cfg->plane_descriptors[k].cursor.cursor_width, + s->cursor_bytes_per_chunk[k], + s->cursor_lines_per_chunk[k], + s->line_times[k], + mode_lib->mp.UrgentLatency, + + // output + &mode_lib->mp.UrgentBurstFactorCursor[k], + &cursor_not_enough_urgent_latency_hiding); + } + mode_lib->mp.UrgentBurstFactorCursorPre[k] = mode_lib->mp.UrgentBurstFactorCursor[k]; + + CalculateUrgentBurstFactor( + &display_cfg->plane_descriptors[k], + mode_lib->mp.swath_width_luma_ub[k], + mode_lib->mp.swath_width_chroma_ub[k], + mode_lib->mp.SwathHeightY[k], + mode_lib->mp.SwathHeightC[k], + s->line_times[k], + mode_lib->mp.UrgentLatency, + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->mp.BytePerPixelInDETY[k], + mode_lib->mp.BytePerPixelInDETC[k], + mode_lib->mp.DETBufferSizeY[k], + mode_lib->mp.DETBufferSizeC[k], + + /* output */ + &mode_lib->mp.UrgentBurstFactorLuma[k], + &mode_lib->mp.UrgentBurstFactorChroma[k], + &mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); + + mode_lib->mp.NotEnoughUrgentLatencyHiding[k] = mode_lib->mp.NotEnoughUrgentLatencyHiding[k] || cursor_not_enough_urgent_latency_hiding; + } + + for (k = 0; k < s->num_active_planes; ++k) { + s->MaxVStartupLines[k] = CalculateMaxVStartup( + mode_lib->ip.ptoi_supported, + mode_lib->ip.vblank_nom_default_us, + &display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing, + mode_lib->mp.WritebackDelay[k]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%u WritebackDelay = %f\n", __func__, k, mode_lib->mp.WritebackDelay[k]); +#endif + } + + s->immediate_flip_required = false; + for (k = 0; k < s->num_active_planes; ++k) { + s->immediate_flip_required = s->immediate_flip_required || display_cfg->plane_descriptors[k].immediate_flip; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: immediate_flip_required = %u\n", __func__, s->immediate_flip_required); +#endif + + if (s->num_active_planes > 1) { + CheckGlobalPrefetchAdmissibility_params->num_active_planes = s->num_active_planes; + CheckGlobalPrefetchAdmissibility_params->pixel_format = s->pixel_format; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_l = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->chunk_bytes_c = mode_lib->ip.pixel_chunk_size_kbytes * 1024; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_l = s->lb_source_lines_l; + CheckGlobalPrefetchAdmissibility_params->lb_source_lines_c = s->lb_source_lines_c; + CheckGlobalPrefetchAdmissibility_params->swath_height_l = mode_lib->mp.SwathHeightY; + CheckGlobalPrefetchAdmissibility_params->swath_height_c = mode_lib->mp.SwathHeightC; + CheckGlobalPrefetchAdmissibility_params->rob_buffer_size_kbytes = mode_lib->ip.rob_buffer_size_kbytes; + CheckGlobalPrefetchAdmissibility_params->compressed_buffer_size_kbytes = mode_lib->mp.CompressedBufferSizeInkByte; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_l = mode_lib->mp.DETBufferSizeY; + CheckGlobalPrefetchAdmissibility_params->detile_buffer_size_bytes_c = mode_lib->mp.DETBufferSizeC; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_l = s->full_swath_bytes_l; + CheckGlobalPrefetchAdmissibility_params->full_swath_bytes_c = s->full_swath_bytes_c; + CheckGlobalPrefetchAdmissibility_params->prefetch_sw_bytes = s->prefetch_sw_bytes; + CheckGlobalPrefetchAdmissibility_params->Tpre_rounded = 0; // don't care + CheckGlobalPrefetchAdmissibility_params->Tpre_oto = 0; // don't care + CheckGlobalPrefetchAdmissibility_params->estimated_urg_bandwidth_required_mbps = mode_lib->mp.urg_bandwidth_available[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]; + CheckGlobalPrefetchAdmissibility_params->estimated_dcfclk_mhz = mode_lib->mp.Dcfclk; + CheckGlobalPrefetchAdmissibility_params->line_time = s->line_times; + CheckGlobalPrefetchAdmissibility_params->dst_y_prefetch = mode_lib->mp.dst_y_prefetch; + + // if recalc_prefetch_schedule is set, recalculate the prefetch schedule with the new impacted_Tpre, prefetch should be possible + CheckGlobalPrefetchAdmissibility_params->recalc_prefetch_schedule = &s->dummy_boolean[0]; + CheckGlobalPrefetchAdmissibility_params->impacted_dst_y_pre = s->impacted_dst_y_pre; + CheckGlobalPrefetchAdmissibility(&mode_lib->scratch, CheckGlobalPrefetchAdmissibility_params); // dont care about the check output for mode programming + } + + { + s->DestinationLineTimesForPrefetchLessThan2 = false; + s->VRatioPrefetchMoreThanMax = false; + + DML_LOG_VERBOSE("DML::%s: Start one iteration of prefetch schedule evaluation\n", __func__); + + for (k = 0; k < s->num_active_planes; ++k) { + struct dml2_core_internal_DmlPipe *myPipe = &s->myPipe; + + DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + mode_lib->mp.TWait[k] = CalculateTWait( + display_cfg->plane_descriptors[k].overrides.reserved_vblank_time_ns, + mode_lib->mp.UrgentLatency, + mode_lib->mp.TripToMemory, + !dml_is_phantom_pipe(&display_cfg->plane_descriptors[k]) && display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.drr_config.enabled ? + get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->mp.uclk_freq_mhz * 1000), in_out_params->min_clk_index) : 0.0); + + myPipe->Dppclk = mode_lib->mp.Dppclk[k]; + myPipe->Dispclk = mode_lib->mp.Dispclk; + myPipe->PixelClock = ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + myPipe->DCFClkDeepSleep = mode_lib->mp.DCFCLKDeepSleep; + myPipe->DPPPerSurface = mode_lib->mp.NoOfDPP[k]; + myPipe->ScalerEnabled = display_cfg->plane_descriptors[k].composition.scaler_info.enabled; + myPipe->VRatio = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio; + myPipe->VRatioChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio; + myPipe->VTaps = display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_taps; + myPipe->VTapsChroma = display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_taps; + myPipe->RotationAngle = display_cfg->plane_descriptors[k].composition.rotation_angle; + myPipe->mirrored = display_cfg->plane_descriptors[k].composition.mirrored; + myPipe->BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY[k]; + myPipe->BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY[k]; + myPipe->BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC[k]; + myPipe->BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC[k]; + myPipe->InterlaceEnable = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced; + myPipe->NumberOfCursors = display_cfg->plane_descriptors[k].cursor.num_cursors; + myPipe->VBlank = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active; + myPipe->HTotal = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total; + myPipe->HActive = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_active; + myPipe->DCCEnable = display_cfg->plane_descriptors[k].surface.dcc.enable; + myPipe->ODMMode = mode_lib->mp.ODMMode[k]; + myPipe->SourcePixelFormat = display_cfg->plane_descriptors[k].pixel_format; + myPipe->BytePerPixelY = mode_lib->mp.BytePerPixelY[k]; + myPipe->BytePerPixelC = mode_lib->mp.BytePerPixelC[k]; + myPipe->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Calling CalculatePrefetchSchedule for k=%u\n", __func__, k); +#endif + CalculatePrefetchSchedule_params->display_cfg = display_cfg; + CalculatePrefetchSchedule_params->HostVMInefficiencyFactor = s->HostVMInefficiencyFactorPrefetch; + CalculatePrefetchSchedule_params->myPipe = myPipe; + CalculatePrefetchSchedule_params->DSCDelay = mode_lib->mp.DSCDelay[k]; + CalculatePrefetchSchedule_params->DPPCLKDelaySubtotalPlusCNVCFormater = mode_lib->ip.dppclk_delay_subtotal + mode_lib->ip.dppclk_delay_cnvc_formatter; + CalculatePrefetchSchedule_params->DPPCLKDelaySCL = mode_lib->ip.dppclk_delay_scl; + CalculatePrefetchSchedule_params->DPPCLKDelaySCLLBOnly = mode_lib->ip.dppclk_delay_scl_lb_only; + CalculatePrefetchSchedule_params->DPPCLKDelayCNVCCursor = mode_lib->ip.dppclk_delay_cnvc_cursor; + CalculatePrefetchSchedule_params->DISPCLKDelaySubtotal = mode_lib->ip.dispclk_delay_subtotal; + CalculatePrefetchSchedule_params->DPP_RECOUT_WIDTH = (unsigned int)(mode_lib->mp.SwathWidthY[k] / display_cfg->plane_descriptors[k].composition.scaler_info.plane0.h_ratio); + CalculatePrefetchSchedule_params->OutputFormat = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format; + CalculatePrefetchSchedule_params->MaxInterDCNTileRepeaters = mode_lib->ip.max_inter_dcn_tile_repeaters; + CalculatePrefetchSchedule_params->VStartup = s->MaxVStartupLines[k]; + CalculatePrefetchSchedule_params->HostVMMinPageSize = mode_lib->soc.hostvm_min_page_size_kbytes; + CalculatePrefetchSchedule_params->DynamicMetadataEnable = display_cfg->plane_descriptors[k].dynamic_meta_data.enable; + CalculatePrefetchSchedule_params->DynamicMetadataVMEnabled = mode_lib->ip.dynamic_metadata_vm_enabled; + CalculatePrefetchSchedule_params->DynamicMetadataLinesBeforeActiveRequired = display_cfg->plane_descriptors[k].dynamic_meta_data.lines_before_active_required; + CalculatePrefetchSchedule_params->DynamicMetadataTransmittedBytes = display_cfg->plane_descriptors[k].dynamic_meta_data.transmitted_bytes; + CalculatePrefetchSchedule_params->UrgentLatency = mode_lib->mp.UrgentLatency; + CalculatePrefetchSchedule_params->ExtraLatencyPrefetch = mode_lib->mp.ExtraLatencyPrefetch; + CalculatePrefetchSchedule_params->TCalc = mode_lib->mp.TCalc; + CalculatePrefetchSchedule_params->vm_bytes = mode_lib->mp.vm_bytes[k]; + CalculatePrefetchSchedule_params->PixelPTEBytesPerRow = mode_lib->mp.PixelPTEBytesPerRow[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesY = mode_lib->mp.PrefetchSourceLinesY[k]; + CalculatePrefetchSchedule_params->VInitPreFillY = mode_lib->mp.VInitPreFillY[k]; + CalculatePrefetchSchedule_params->MaxNumSwathY = mode_lib->mp.MaxNumSwathY[k]; + CalculatePrefetchSchedule_params->PrefetchSourceLinesC = mode_lib->mp.PrefetchSourceLinesC[k]; + CalculatePrefetchSchedule_params->VInitPreFillC = mode_lib->mp.VInitPreFillC[k]; + CalculatePrefetchSchedule_params->MaxNumSwathC = mode_lib->mp.MaxNumSwathC[k]; + CalculatePrefetchSchedule_params->swath_width_luma_ub = mode_lib->mp.swath_width_luma_ub[k]; + CalculatePrefetchSchedule_params->swath_width_chroma_ub = mode_lib->mp.swath_width_chroma_ub[k]; + CalculatePrefetchSchedule_params->SwathHeightY = mode_lib->mp.SwathHeightY[k]; + CalculatePrefetchSchedule_params->SwathHeightC = mode_lib->mp.SwathHeightC[k]; + CalculatePrefetchSchedule_params->TWait = mode_lib->mp.TWait[k]; + CalculatePrefetchSchedule_params->Ttrip = mode_lib->mp.TripToMemory; + CalculatePrefetchSchedule_params->Turg = mode_lib->mp.UrgentLatency; + CalculatePrefetchSchedule_params->setup_for_tdlut = display_cfg->plane_descriptors[k].tdlut.setup_for_tdlut; + CalculatePrefetchSchedule_params->tdlut_pte_bytes_per_frame = s->tdlut_pte_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_bytes_per_frame = s->tdlut_bytes_per_frame[k]; + CalculatePrefetchSchedule_params->tdlut_opt_time = s->tdlut_opt_time[k]; + CalculatePrefetchSchedule_params->tdlut_drain_time = s->tdlut_drain_time[k]; + CalculatePrefetchSchedule_params->num_cursors = (display_cfg->plane_descriptors[k].cursor.cursor_width > 0); + CalculatePrefetchSchedule_params->cursor_bytes_per_chunk = s->cursor_bytes_per_chunk[k]; + CalculatePrefetchSchedule_params->cursor_bytes_per_line = s->cursor_bytes_per_line[k]; + CalculatePrefetchSchedule_params->dcc_enable = display_cfg->plane_descriptors[k].surface.dcc.enable; + CalculatePrefetchSchedule_params->mrq_present = mode_lib->ip.dcn_mrq_present; + CalculatePrefetchSchedule_params->meta_row_bytes = mode_lib->mp.meta_row_bytes[k]; + CalculatePrefetchSchedule_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor[k]; + CalculatePrefetchSchedule_params->impacted_dst_y_pre = s->impacted_dst_y_pre[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_l = mode_lib->mp.vactive_sw_bw_l[k]; + CalculatePrefetchSchedule_params->vactive_sw_bw_c = mode_lib->mp.vactive_sw_bw_c[k]; + + // output + CalculatePrefetchSchedule_params->DSTXAfterScaler = &mode_lib->mp.DSTXAfterScaler[k]; + CalculatePrefetchSchedule_params->DSTYAfterScaler = &mode_lib->mp.DSTYAfterScaler[k]; + CalculatePrefetchSchedule_params->dst_y_prefetch = &mode_lib->mp.dst_y_prefetch[k]; + CalculatePrefetchSchedule_params->dst_y_per_vm_vblank = &mode_lib->mp.dst_y_per_vm_vblank[k]; + CalculatePrefetchSchedule_params->dst_y_per_row_vblank = &mode_lib->mp.dst_y_per_row_vblank[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchY = &mode_lib->mp.VRatioPrefetchY[k]; + CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->mp.VRatioPrefetchC[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchBWMax = &s->dummy_single_array[0][k]; + CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]; + CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->mp.Tno_bw[k]; + CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->mp.Tno_bw_flip[k]; + CalculatePrefetchSchedule_params->prefetch_vmrow_bw = &mode_lib->mp.prefetch_vmrow_bw[k]; + CalculatePrefetchSchedule_params->Tdmdl_vm = &mode_lib->mp.Tdmdl_vm[k]; + CalculatePrefetchSchedule_params->Tdmdl = &mode_lib->mp.Tdmdl[k]; + CalculatePrefetchSchedule_params->TSetup = &mode_lib->mp.TSetup[k]; + CalculatePrefetchSchedule_params->Tvm_trips = &s->Tvm_trips[k]; + CalculatePrefetchSchedule_params->Tr0_trips = &s->Tr0_trips[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip = &s->Tvm_trips_flip[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip = &s->Tr0_trips_flip[k]; + CalculatePrefetchSchedule_params->Tvm_trips_flip_rounded = &s->Tvm_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->Tr0_trips_flip_rounded = &s->Tr0_trips_flip_rounded[k]; + CalculatePrefetchSchedule_params->VUpdateOffsetPix = &mode_lib->mp.VUpdateOffsetPix[k]; + CalculatePrefetchSchedule_params->VUpdateWidthPix = &mode_lib->mp.VUpdateWidthPix[k]; + CalculatePrefetchSchedule_params->VReadyOffsetPix = &mode_lib->mp.VReadyOffsetPix[k]; + CalculatePrefetchSchedule_params->prefetch_cursor_bw = &mode_lib->mp.prefetch_cursor_bw[k]; + CalculatePrefetchSchedule_params->prefetch_sw_bytes = &s->prefetch_sw_bytes[k]; + CalculatePrefetchSchedule_params->Tpre_rounded = &s->Tpre_rounded[k]; + CalculatePrefetchSchedule_params->Tpre_oto = &s->Tpre_oto[k]; + CalculatePrefetchSchedule_params->prefetch_swath_time_us = &s->dummy_single[0]; + + mode_lib->mp.NoTimeToPrefetch[k] = CalculatePrefetchSchedule(&mode_lib->scratch, CalculatePrefetchSchedule_params); + + if (s->impacted_dst_y_pre[k] > 0) + mode_lib->mp.impacted_prefetch_margin_us[k] = (mode_lib->mp.dst_y_prefetch[k] - s->impacted_dst_y_pre[k]) * s->line_times[k]; + else + mode_lib->mp.impacted_prefetch_margin_us[k] = 0; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%0u NoTimeToPrefetch=%0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); +#endif + mode_lib->mp.VStartupMin[k] = s->MaxVStartupLines[k]; + } // for k + + mode_lib->mp.PrefetchModeSupported = true; + for (k = 0; k < s->num_active_planes; ++k) { + if (mode_lib->mp.NoTimeToPrefetch[k] == true || + mode_lib->mp.NotEnoughTimeForDynamicMetadata[k] || + mode_lib->mp.DSTYAfterScaler[k] > 8) { + DML_LOG_VERBOSE("DML::%s: k=%u, NoTimeToPrefetch = %0d\n", __func__, k, mode_lib->mp.NoTimeToPrefetch[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughTimeForDynamicMetadata=%u\n", __func__, k, mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, DSTYAfterScaler=%u (should be <= 0)\n", __func__, k, mode_lib->mp.DSTYAfterScaler[k]); + mode_lib->mp.PrefetchModeSupported = false; + } + if (mode_lib->mp.dst_y_prefetch[k] < 2) + s->DestinationLineTimesForPrefetchLessThan2 = true; + + if (mode_lib->mp.VRatioPrefetchY[k] > __DML2_CALCS_MAX_VRATIO_PRE__ || + mode_lib->mp.VRatioPrefetchC[k] > __DML2_CALCS_MAX_VRATIO_PRE__) { + s->VRatioPrefetchMoreThanMax = true; + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchY=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: k=%d, VRatioPrefetchC=%f (should not be < %f)\n", __func__, k, mode_lib->mp.VRatioPrefetchC[k], __DML2_CALCS_MAX_VRATIO_PRE__); + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); + } + + if (mode_lib->mp.NotEnoughUrgentLatencyHiding[k]) { + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHiding = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHiding[k]); + mode_lib->mp.PrefetchModeSupported = false; + } + } + + if (s->VRatioPrefetchMoreThanMax == true || s->DestinationLineTimesForPrefetchLessThan2 == true) { + DML_LOG_VERBOSE("DML::%s: VRatioPrefetchMoreThanMax = %u\n", __func__, s->VRatioPrefetchMoreThanMax); + DML_LOG_VERBOSE("DML::%s: DestinationLineTimesForPrefetchLessThan2 = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2); + mode_lib->mp.PrefetchModeSupported = false; + } + + DML_LOG_VERBOSE("DML::%s: Prefetch schedule is %sOK at vstartup = %u\n", __func__, + mode_lib->mp.PrefetchModeSupported ? "" : "NOT ", CalculatePrefetchSchedule_params->VStartup); + + // Prefetch schedule OK, now check prefetch bw + if (mode_lib->mp.PrefetchModeSupported == true) { + for (k = 0; k < s->num_active_planes; ++k) { + double line_time_us = display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + CalculateUrgentBurstFactor( + &display_cfg->plane_descriptors[k], + mode_lib->mp.swath_width_luma_ub[k], + mode_lib->mp.swath_width_chroma_ub[k], + mode_lib->mp.SwathHeightY[k], + mode_lib->mp.SwathHeightC[k], + line_time_us, + mode_lib->mp.UrgentLatency, + mode_lib->mp.VRatioPrefetchY[k], + mode_lib->mp.VRatioPrefetchC[k], + mode_lib->mp.BytePerPixelInDETY[k], + mode_lib->mp.BytePerPixelInDETC[k], + mode_lib->mp.DETBufferSizeY[k], + mode_lib->mp.DETBufferSizeC[k], + /* Output */ + &mode_lib->mp.UrgentBurstFactorLumaPre[k], + &mode_lib->mp.UrgentBurstFactorChromaPre[k], + &mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%0u DPPPerSurface=%u\n", __func__, k, mode_lib->mp.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLuma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChroma=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorLumaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorLumaPre[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u UrgentBurstFactorChromaPre=%f\n", __func__, k, mode_lib->mp.UrgentBurstFactorChromaPre[k]); + + DML_LOG_VERBOSE("DML::%s: k=%0u VRatioPrefetchY=%f\n", __func__, k, mode_lib->mp.VRatioPrefetchY[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u VRatioY=%f\n", __func__, k, display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio); + + DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_vmrow_bw=%f\n", __func__, k, mode_lib->mp.prefetch_vmrow_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_l=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u vactive_sw_bw_c=%f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u cursor_bw=%f\n", __func__, k, mode_lib->mp.cursor_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u dpte_row_bw=%f\n", __func__, k, mode_lib->mp.dpte_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u meta_row_bw=%f\n", __func__, k, mode_lib->mp.meta_row_bw[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWLuma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u RequiredPrefetchPixelDataBWChroma=%f\n", __func__, k, mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]); + DML_LOG_VERBOSE("DML::%s: k=%0u prefetch_cursor_bw=%f\n", __func__, k, mode_lib->mp.prefetch_cursor_bw[k]); +#endif + } + + for (k = 0; k <= s->num_active_planes - 1; k++) + mode_lib->mp.final_flip_bw[k] = 0; + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = mode_lib->mp.urg_vactive_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->mp.urg_bandwidth_required; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = mode_lib->mp.urg_bandwidth_required_qual; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->mp.non_urg_bandwidth_required; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = s->surface_dummy_bw0; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 0; + calculate_peak_bandwidth_params->num_active_planes = s->num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->mp.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->mp.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->mp.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->mp.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->mp.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->mp.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->mp.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + // Check urg peak bandwidth against available urg bw + // check at SDP and DRAM, for all soc states (SVP prefetch an Sys Active) + check_urgent_bandwidth_support( + &mode_lib->mp.FractionOfUrgentBandwidth, // double* frac_urg_bandwidth + &mode_lib->mp.FractionOfUrgentBandwidthMALL, // double* frac_urg_bandwidth_mall + &s->dummy_boolean[1], // vactive bw ok + &mode_lib->mp.PrefetchModeSupported, // prefetch bw ok + + mode_lib->soc.mall_allocated_for_dcn_mbytes, + mode_lib->mp.non_urg_bandwidth_required, + mode_lib->mp.urg_vactive_bandwidth_required, + mode_lib->mp.urg_bandwidth_required, + mode_lib->mp.urg_bandwidth_available); + + if (!mode_lib->mp.PrefetchModeSupported) + DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for prefetch!\n", __func__); + + for (k = 0; k < s->num_active_planes; ++k) { + if (mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]) { + DML_LOG_VERBOSE("DML::%s: k=%u, NotEnoughUrgentLatencyHidingPre = %u\n", __func__, k, mode_lib->mp.NotEnoughUrgentLatencyHidingPre[k]); + mode_lib->mp.PrefetchModeSupported = false; + } + } + } // prefetch schedule ok + + // Prefetch schedule and prefetch bw ok, now check flip bw + if (mode_lib->mp.PrefetchModeSupported == true) { // prefetch schedule and prefetch bw ok, now check flip bw + + mode_lib->mp.BandwidthAvailableForImmediateFlip = + get_bandwidth_available_for_immediate_flip( + dml2_core_internal_soc_state_sys_active, + mode_lib->mp.urg_bandwidth_required_qual, // no flip + mode_lib->mp.urg_bandwidth_available); + mode_lib->mp.TotImmediateFlipBytes = 0; + for (k = 0; k < s->num_active_planes; ++k) { + if (display_cfg->plane_descriptors[k].immediate_flip) { + s->per_pipe_flip_bytes[k] = get_pipe_flip_bytes(s->HostVMInefficiencyFactor, + mode_lib->mp.vm_bytes[k], + mode_lib->mp.PixelPTEBytesPerRow[k], + mode_lib->mp.meta_row_bytes[k]); + } else { + s->per_pipe_flip_bytes[k] = 0; + } + mode_lib->mp.TotImmediateFlipBytes += s->per_pipe_flip_bytes[k] * mode_lib->mp.NoOfDPP[k]; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k = %u\n", __func__, k); + DML_LOG_VERBOSE("DML::%s: DPPPerSurface = %u\n", __func__, mode_lib->mp.NoOfDPP[k]); + DML_LOG_VERBOSE("DML::%s: vm_bytes = %u\n", __func__, mode_lib->mp.vm_bytes[k]); + DML_LOG_VERBOSE("DML::%s: PixelPTEBytesPerRow = %u\n", __func__, mode_lib->mp.PixelPTEBytesPerRow[k]); + DML_LOG_VERBOSE("DML::%s: meta_row_bytes = %u\n", __func__, mode_lib->mp.meta_row_bytes[k]); + DML_LOG_VERBOSE("DML::%s: TotImmediateFlipBytes = %u\n", __func__, mode_lib->mp.TotImmediateFlipBytes); +#endif + } + for (k = 0; k < s->num_active_planes; ++k) { + CalculateFlipSchedule( + &mode_lib->scratch, + display_cfg->plane_descriptors[k].immediate_flip, + 0, // use_lb_flip_bw + s->HostVMInefficiencyFactor, + s->Tvm_trips_flip[k], + s->Tr0_trips_flip[k], + s->Tvm_trips_flip_rounded[k], + s->Tr0_trips_flip_rounded[k], + display_cfg->gpuvm_enable, + mode_lib->mp.vm_bytes[k], + mode_lib->mp.PixelPTEBytesPerRow[k], + mode_lib->mp.BandwidthAvailableForImmediateFlip, + mode_lib->mp.TotImmediateFlipBytes, + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), + display_cfg->plane_descriptors[k].composition.scaler_info.plane0.v_ratio, + display_cfg->plane_descriptors[k].composition.scaler_info.plane1.v_ratio, + mode_lib->mp.Tno_bw[k], + mode_lib->mp.dpte_row_height[k], + mode_lib->mp.dpte_row_height_chroma[k], + mode_lib->mp.use_one_row_for_frame_flip[k], + mode_lib->ip.max_flip_time_us, + mode_lib->ip.max_flip_time_lines, + s->per_pipe_flip_bytes[k], + mode_lib->mp.meta_row_bytes[k], + mode_lib->mp.meta_row_height[k], + mode_lib->mp.meta_row_height_chroma[k], + mode_lib->ip.dcn_mrq_present && display_cfg->plane_descriptors[k].surface.dcc.enable, + + // Output + &mode_lib->mp.dst_y_per_vm_flip[k], + &mode_lib->mp.dst_y_per_row_flip[k], + &mode_lib->mp.final_flip_bw[k], + &mode_lib->mp.ImmediateFlipSupportedForPipe[k]); + } + + calculate_peak_bandwidth_params->urg_vactive_bandwidth_required = s->dummy_bw; + calculate_peak_bandwidth_params->urg_bandwidth_required = mode_lib->mp.urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->urg_bandwidth_required_qual = s->dummy_bw; + calculate_peak_bandwidth_params->non_urg_bandwidth_required = mode_lib->mp.non_urg_bandwidth_required_flip; + calculate_peak_bandwidth_params->surface_avg_vactive_required_bw = s->surface_dummy_bw; + calculate_peak_bandwidth_params->surface_peak_required_bw = s->surface_dummy_bw0; + + calculate_peak_bandwidth_params->display_cfg = display_cfg; + calculate_peak_bandwidth_params->inc_flip_bw = 1; + calculate_peak_bandwidth_params->num_active_planes = s->num_active_planes; + calculate_peak_bandwidth_params->num_of_dpp = mode_lib->mp.NoOfDPP; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_nom_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_nom_overhead_factor_p1; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p0 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p0; + calculate_peak_bandwidth_params->dcc_dram_bw_pref_overhead_factor_p1 = mode_lib->mp.dcc_dram_bw_pref_overhead_factor_p1; + calculate_peak_bandwidth_params->mall_prefetch_sdp_overhead_factor = mode_lib->mp.mall_prefetch_sdp_overhead_factor; + calculate_peak_bandwidth_params->mall_prefetch_dram_overhead_factor = mode_lib->mp.mall_prefetch_dram_overhead_factor; + + calculate_peak_bandwidth_params->surface_read_bandwidth_l = mode_lib->mp.vactive_sw_bw_l; + calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; + calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; + calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; + calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; + calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; + calculate_peak_bandwidth_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; + calculate_peak_bandwidth_params->meta_row_bw = mode_lib->mp.meta_row_bw; + calculate_peak_bandwidth_params->prefetch_cursor_bw = mode_lib->mp.prefetch_cursor_bw; + calculate_peak_bandwidth_params->prefetch_vmrow_bw = mode_lib->mp.prefetch_vmrow_bw; + calculate_peak_bandwidth_params->prefetch_bandwidth_max = s->dummy_single_array[0]; + calculate_peak_bandwidth_params->flip_bw = mode_lib->mp.final_flip_bw; + calculate_peak_bandwidth_params->urgent_burst_factor_l = mode_lib->mp.UrgentBurstFactorLuma; + calculate_peak_bandwidth_params->urgent_burst_factor_c = mode_lib->mp.UrgentBurstFactorChroma; + calculate_peak_bandwidth_params->urgent_burst_factor_cursor = mode_lib->mp.UrgentBurstFactorCursor; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_l = mode_lib->mp.UrgentBurstFactorLumaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_c = mode_lib->mp.UrgentBurstFactorChromaPre; + calculate_peak_bandwidth_params->urgent_burst_factor_prefetch_cursor = mode_lib->mp.UrgentBurstFactorCursorPre; + + calculate_peak_bandwidth_required( + &mode_lib->scratch, + calculate_peak_bandwidth_params); + + calculate_immediate_flip_bandwidth_support( + &mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip, // double* frac_urg_bandwidth_flip + &mode_lib->mp.ImmediateFlipSupported, // bool* flip_bandwidth_support_ok + + dml2_core_internal_soc_state_sys_active, + mode_lib->mp.urg_bandwidth_required_flip, + mode_lib->mp.non_urg_bandwidth_required_flip, + mode_lib->mp.urg_bandwidth_available); + + if (!mode_lib->mp.ImmediateFlipSupported) + DML_LOG_VERBOSE("DML::%s: Bandwidth not sufficient for flip!", __func__); + + for (k = 0; k < s->num_active_planes; ++k) { + if (display_cfg->plane_descriptors[k].immediate_flip && mode_lib->mp.ImmediateFlipSupportedForPipe[k] == false) { + mode_lib->mp.ImmediateFlipSupported = false; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Pipe %0d not supporting iflip!\n", __func__, k); +#endif + } + } + } else { // flip or prefetch not support + mode_lib->mp.ImmediateFlipSupported = false; + } + + // consider flip support is okay if the flip bw is ok or (when user does't require a iflip and there is no host vm) + must_support_iflip = display_cfg->hostvm_enable || s->immediate_flip_required; + mode_lib->mp.PrefetchAndImmediateFlipSupported = (mode_lib->mp.PrefetchModeSupported == true && (!must_support_iflip || mode_lib->mp.ImmediateFlipSupported)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: PrefetchModeSupported = %u\n", __func__, mode_lib->mp.PrefetchModeSupported); + for (k = 0; k < s->num_active_planes; ++k) + DML_LOG_VERBOSE("DML::%s: immediate_flip_required[%u] = %u\n", __func__, k, display_cfg->plane_descriptors[k].immediate_flip); + DML_LOG_VERBOSE("DML::%s: HostVMEnable = %u\n", __func__, display_cfg->hostvm_enable); + DML_LOG_VERBOSE("DML::%s: ImmediateFlipSupported = %u\n", __func__, mode_lib->mp.ImmediateFlipSupported); + DML_LOG_VERBOSE("DML::%s: PrefetchAndImmediateFlipSupported = %u\n", __func__, mode_lib->mp.PrefetchAndImmediateFlipSupported); +#endif + DML_LOG_VERBOSE("DML::%s: Done one iteration: k=%d, MaxVStartupLines=%u\n", __func__, k, s->MaxVStartupLines[k]); + } + + for (k = 0; k < s->num_active_planes; ++k) + DML_LOG_VERBOSE("DML::%s: k=%d MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + + if (!mode_lib->mp.PrefetchAndImmediateFlipSupported) { + DML_LOG_VERBOSE("DML::%s: Bad, Prefetch and flip scheduling solution NOT found!\n", __func__); + } else { + DML_LOG_VERBOSE("DML::%s: Good, Prefetch and flip scheduling solution found\n", __func__); + + // DCC Configuration + for (k = 0; k < s->num_active_planes; ++k) { +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: Calculate DCC configuration for surface k=%u\n", __func__, k); +#endif + CalculateDCCConfiguration( + display_cfg->plane_descriptors[k].surface.dcc.enable, + display_cfg->overrides.dcc_programming_assumes_scan_direction_unknown, + display_cfg->plane_descriptors[k].pixel_format, + display_cfg->plane_descriptors[k].surface.plane0.width, + display_cfg->plane_descriptors[k].surface.plane1.width, + display_cfg->plane_descriptors[k].surface.plane0.height, + display_cfg->plane_descriptors[k].surface.plane1.height, + s->NomDETInKByte, + mode_lib->mp.Read256BlockHeightY[k], + mode_lib->mp.Read256BlockHeightC[k], + display_cfg->plane_descriptors[k].surface.tiling, + mode_lib->mp.BytePerPixelY[k], + mode_lib->mp.BytePerPixelC[k], + mode_lib->mp.BytePerPixelInDETY[k], + mode_lib->mp.BytePerPixelInDETC[k], + display_cfg->plane_descriptors[k].composition.rotation_angle, + + /* Output */ + &mode_lib->mp.RequestLuma[k], + &mode_lib->mp.RequestChroma[k], + &mode_lib->mp.DCCYMaxUncompressedBlock[k], + &mode_lib->mp.DCCCMaxUncompressedBlock[k], + &mode_lib->mp.DCCYMaxCompressedBlock[k], + &mode_lib->mp.DCCCMaxCompressedBlock[k], + &mode_lib->mp.DCCYIndependentBlock[k], + &mode_lib->mp.DCCCIndependentBlock[k]); + } + + //Watermarks and NB P-State/DRAM Clock Change Support + s->mmSOCParameters.UrgentLatency = mode_lib->mp.UrgentLatency; + s->mmSOCParameters.ExtraLatency = mode_lib->mp.ExtraLatency; + s->mmSOCParameters.ExtraLatency_sr = mode_lib->mp.ExtraLatency_sr; + s->mmSOCParameters.WritebackLatency = mode_lib->soc.qos_parameters.writeback.base_latency_us; + s->mmSOCParameters.DRAMClockChangeLatency = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + s->mmSOCParameters.FCLKChangeLatency = mode_lib->soc.power_management_parameters.fclk_change_blackout_us; + s->mmSOCParameters.SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; + s->mmSOCParameters.SREnterPlusExitTime = mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us; + s->mmSOCParameters.SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; + s->mmSOCParameters.SREnterPlusExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_enter_plus_exit_latency_us; + s->mmSOCParameters.USRRetrainingLatency = 0; + s->mmSOCParameters.SMNLatency = 0; + s->mmSOCParameters.g6_temp_read_blackout_us = get_g6_temp_read_blackout_us(&mode_lib->soc, (unsigned int)(mode_lib->mp.uclk_freq_mhz * 1000), in_out_params->min_clk_index); + s->mmSOCParameters.max_urgent_latency_us = get_max_urgent_latency_us(&mode_lib->soc.qos_parameters.qos_params.dcn4x, mode_lib->mp.uclk_freq_mhz, mode_lib->mp.FabricClock, in_out_params->min_clk_index); + s->mmSOCParameters.df_response_time_us = mode_lib->soc.qos_parameters.qos_params.dcn4x.df_qos_response_time_fclk_cycles / mode_lib->mp.FabricClock; + s->mmSOCParameters.qos_type = mode_lib->soc.qos_parameters.qos_type; + + CalculateWatermarks_params->display_cfg = display_cfg; + CalculateWatermarks_params->USRRetrainingRequired = false; + CalculateWatermarks_params->NumberOfActiveSurfaces = s->num_active_planes; + CalculateWatermarks_params->MaxLineBufferLines = mode_lib->ip.max_line_buffer_lines; + CalculateWatermarks_params->LineBufferSize = mode_lib->ip.line_buffer_size_bits; + CalculateWatermarks_params->WritebackInterfaceBufferSize = mode_lib->ip.writeback_interface_buffer_size_kbytes; + CalculateWatermarks_params->DCFCLK = mode_lib->mp.Dcfclk; + CalculateWatermarks_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; + CalculateWatermarks_params->SynchronizeDRRDisplaysForUCLKPStateChange = display_cfg->overrides.synchronize_ddr_displays_for_uclk_pstate_change; + CalculateWatermarks_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; + CalculateWatermarks_params->mmSOCParameters = s->mmSOCParameters; + CalculateWatermarks_params->WritebackChunkSize = mode_lib->ip.writeback_chunk_size_kbytes; + CalculateWatermarks_params->SOCCLK = s->SOCCLK; + CalculateWatermarks_params->DCFClkDeepSleep = mode_lib->mp.DCFCLKDeepSleep; + CalculateWatermarks_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; + CalculateWatermarks_params->DETBufferSizeC = mode_lib->mp.DETBufferSizeC; + CalculateWatermarks_params->SwathHeightY = mode_lib->mp.SwathHeightY; + CalculateWatermarks_params->SwathHeightC = mode_lib->mp.SwathHeightC; + CalculateWatermarks_params->SwathWidthY = mode_lib->mp.SwathWidthY; + CalculateWatermarks_params->SwathWidthC = mode_lib->mp.SwathWidthC; + CalculateWatermarks_params->BytePerPixelDETY = mode_lib->mp.BytePerPixelInDETY; + CalculateWatermarks_params->BytePerPixelDETC = mode_lib->mp.BytePerPixelInDETC; + CalculateWatermarks_params->DSTXAfterScaler = mode_lib->mp.DSTXAfterScaler; + CalculateWatermarks_params->DSTYAfterScaler = mode_lib->mp.DSTYAfterScaler; + CalculateWatermarks_params->UnboundedRequestEnabled = mode_lib->mp.UnboundedRequestEnabled; + CalculateWatermarks_params->CompressedBufferSizeInkByte = mode_lib->mp.CompressedBufferSizeInkByte; + CalculateWatermarks_params->meta_row_height_l = mode_lib->mp.meta_row_height; + CalculateWatermarks_params->meta_row_height_c = mode_lib->mp.meta_row_height_chroma; + CalculateWatermarks_params->DPPPerSurface = mode_lib->mp.NoOfDPP; + + // Output + CalculateWatermarks_params->Watermark = &mode_lib->mp.Watermark; + CalculateWatermarks_params->DRAMClockChangeSupport = mode_lib->mp.DRAMClockChangeSupport; + CalculateWatermarks_params->global_dram_clock_change_supported = &mode_lib->mp.global_dram_clock_change_supported; + CalculateWatermarks_params->MaxActiveDRAMClockChangeLatencySupported = mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported; + CalculateWatermarks_params->SubViewportLinesNeededInMALL = mode_lib->mp.SubViewportLinesNeededInMALL; + CalculateWatermarks_params->FCLKChangeSupport = mode_lib->mp.FCLKChangeSupport; + CalculateWatermarks_params->global_fclk_change_supported = &mode_lib->mp.global_fclk_change_supported; + CalculateWatermarks_params->MaxActiveFCLKChangeLatencySupported = &mode_lib->mp.MaxActiveFCLKChangeLatencySupported; + CalculateWatermarks_params->USRRetrainingSupport = &mode_lib->mp.USRRetrainingSupport; + CalculateWatermarks_params->g6_temp_read_support = &mode_lib->mp.g6_temp_read_support; + CalculateWatermarks_params->VActiveLatencyHidingMargin = 0; + CalculateWatermarks_params->VActiveLatencyHidingUs = 0; + + CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport(&mode_lib->scratch, CalculateWatermarks_params); + + for (k = 0; k < s->num_active_planes; ++k) { + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].writeback.active_writebacks_per_stream > 0) { + mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k] = math_max2(0, mode_lib->mp.VStartupMin[k] * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) - mode_lib->mp.Watermark.WritebackDRAMClockChangeWatermark); + mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k] = math_max2(0, mode_lib->mp.VStartupMin[k] * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000) - mode_lib->mp.Watermark.WritebackFCLKChangeWatermark); + } else { + mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k] = 0; + mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k] = 0; + } + } + + calculate_pstate_keepout_dst_lines(display_cfg, &mode_lib->mp.Watermark, mode_lib->mp.pstate_keepout_dst_lines); + + DML_LOG_VERBOSE("DML::%s: DEBUG stream_index = %0d\n", __func__, display_cfg->plane_descriptors[0].stream_index); + DML_LOG_VERBOSE("DML::%s: DEBUG PixelClock = %ld kHz\n", __func__, (display_cfg->stream_descriptors[display_cfg->plane_descriptors[0].stream_index].timing.pixel_clock_khz)); + + //Display Pipeline Delivery Time in Prefetch, Groups + CalculatePixelDeliveryTimes( + display_cfg, + cfg_support_info, + s->num_active_planes, + mode_lib->mp.VRatioPrefetchY, + mode_lib->mp.VRatioPrefetchC, + mode_lib->mp.swath_width_luma_ub, + mode_lib->mp.swath_width_chroma_ub, + mode_lib->mp.PSCL_THROUGHPUT, + mode_lib->mp.PSCL_THROUGHPUT_CHROMA, + mode_lib->mp.Dppclk, + mode_lib->mp.BytePerPixelC, + mode_lib->mp.req_per_swath_ub_l, + mode_lib->mp.req_per_swath_ub_c, + + /* Output */ + mode_lib->mp.DisplayPipeLineDeliveryTimeLuma, + mode_lib->mp.DisplayPipeLineDeliveryTimeChroma, + mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch, + mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch, + mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma, + mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma, + mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch, + mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch); + + CalculateMetaAndPTETimes_params->scratch = &mode_lib->scratch; + CalculateMetaAndPTETimes_params->display_cfg = display_cfg; + CalculateMetaAndPTETimes_params->NumberOfActiveSurfaces = s->num_active_planes; + CalculateMetaAndPTETimes_params->use_one_row_for_frame = mode_lib->mp.use_one_row_for_frame; + CalculateMetaAndPTETimes_params->dst_y_per_row_vblank = mode_lib->mp.dst_y_per_row_vblank; + CalculateMetaAndPTETimes_params->dst_y_per_row_flip = mode_lib->mp.dst_y_per_row_flip; + CalculateMetaAndPTETimes_params->BytePerPixelY = mode_lib->mp.BytePerPixelY; + CalculateMetaAndPTETimes_params->BytePerPixelC = mode_lib->mp.BytePerPixelC; + CalculateMetaAndPTETimes_params->dpte_row_height = mode_lib->mp.dpte_row_height; + CalculateMetaAndPTETimes_params->dpte_row_height_chroma = mode_lib->mp.dpte_row_height_chroma; + CalculateMetaAndPTETimes_params->dpte_group_bytes = mode_lib->mp.dpte_group_bytes; + CalculateMetaAndPTETimes_params->PTERequestSizeY = mode_lib->mp.PTERequestSizeY; + CalculateMetaAndPTETimes_params->PTERequestSizeC = mode_lib->mp.PTERequestSizeC; + CalculateMetaAndPTETimes_params->PixelPTEReqWidthY = mode_lib->mp.PixelPTEReqWidthY; + CalculateMetaAndPTETimes_params->PixelPTEReqHeightY = mode_lib->mp.PixelPTEReqHeightY; + CalculateMetaAndPTETimes_params->PixelPTEReqWidthC = mode_lib->mp.PixelPTEReqWidthC; + CalculateMetaAndPTETimes_params->PixelPTEReqHeightC = mode_lib->mp.PixelPTEReqHeightC; + CalculateMetaAndPTETimes_params->dpte_row_width_luma_ub = mode_lib->mp.dpte_row_width_luma_ub; + CalculateMetaAndPTETimes_params->dpte_row_width_chroma_ub = mode_lib->mp.dpte_row_width_chroma_ub; + CalculateMetaAndPTETimes_params->tdlut_groups_per_2row_ub = s->tdlut_groups_per_2row_ub; + CalculateMetaAndPTETimes_params->mrq_present = mode_lib->ip.dcn_mrq_present; + + CalculateMetaAndPTETimes_params->MetaChunkSize = mode_lib->ip.meta_chunk_size_kbytes; + CalculateMetaAndPTETimes_params->MinMetaChunkSizeBytes = mode_lib->ip.min_meta_chunk_size_bytes; + CalculateMetaAndPTETimes_params->meta_row_width = mode_lib->mp.meta_row_width; + CalculateMetaAndPTETimes_params->meta_row_width_chroma = mode_lib->mp.meta_row_width_chroma; + CalculateMetaAndPTETimes_params->meta_row_height = mode_lib->mp.meta_row_height; + CalculateMetaAndPTETimes_params->meta_row_height_chroma = mode_lib->mp.meta_row_height_chroma; + CalculateMetaAndPTETimes_params->meta_req_width = mode_lib->mp.meta_req_width; + CalculateMetaAndPTETimes_params->meta_req_width_chroma = mode_lib->mp.meta_req_width_chroma; + CalculateMetaAndPTETimes_params->meta_req_height = mode_lib->mp.meta_req_height; + CalculateMetaAndPTETimes_params->meta_req_height_chroma = mode_lib->mp.meta_req_height_chroma; + + CalculateMetaAndPTETimes_params->time_per_tdlut_group = mode_lib->mp.time_per_tdlut_group; + CalculateMetaAndPTETimes_params->DST_Y_PER_PTE_ROW_NOM_L = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L; + CalculateMetaAndPTETimes_params->DST_Y_PER_PTE_ROW_NOM_C = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C; + CalculateMetaAndPTETimes_params->time_per_pte_group_nom_luma = mode_lib->mp.time_per_pte_group_nom_luma; + CalculateMetaAndPTETimes_params->time_per_pte_group_vblank_luma = mode_lib->mp.time_per_pte_group_vblank_luma; + CalculateMetaAndPTETimes_params->time_per_pte_group_flip_luma = mode_lib->mp.time_per_pte_group_flip_luma; + CalculateMetaAndPTETimes_params->time_per_pte_group_nom_chroma = mode_lib->mp.time_per_pte_group_nom_chroma; + CalculateMetaAndPTETimes_params->time_per_pte_group_vblank_chroma = mode_lib->mp.time_per_pte_group_vblank_chroma; + CalculateMetaAndPTETimes_params->time_per_pte_group_flip_chroma = mode_lib->mp.time_per_pte_group_flip_chroma; + CalculateMetaAndPTETimes_params->DST_Y_PER_META_ROW_NOM_L = mode_lib->mp.DST_Y_PER_META_ROW_NOM_L; + CalculateMetaAndPTETimes_params->DST_Y_PER_META_ROW_NOM_C = mode_lib->mp.DST_Y_PER_META_ROW_NOM_C; + CalculateMetaAndPTETimes_params->TimePerMetaChunkNominal = mode_lib->mp.TimePerMetaChunkNominal; + CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkNominal = mode_lib->mp.TimePerChromaMetaChunkNominal; + CalculateMetaAndPTETimes_params->TimePerMetaChunkVBlank = mode_lib->mp.TimePerMetaChunkVBlank; + CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkVBlank = mode_lib->mp.TimePerChromaMetaChunkVBlank; + CalculateMetaAndPTETimes_params->TimePerMetaChunkFlip = mode_lib->mp.TimePerMetaChunkFlip; + CalculateMetaAndPTETimes_params->TimePerChromaMetaChunkFlip = mode_lib->mp.TimePerChromaMetaChunkFlip; + + CalculateMetaAndPTETimes(CalculateMetaAndPTETimes_params); + + CalculateVMGroupAndRequestTimes( + display_cfg, + s->num_active_planes, + mode_lib->mp.BytePerPixelC, + mode_lib->mp.dst_y_per_vm_vblank, + mode_lib->mp.dst_y_per_vm_flip, + mode_lib->mp.dpte_row_width_luma_ub, + mode_lib->mp.dpte_row_width_chroma_ub, + mode_lib->mp.vm_group_bytes, + mode_lib->mp.dpde0_bytes_per_frame_ub_l, + mode_lib->mp.dpde0_bytes_per_frame_ub_c, + s->tdlut_pte_bytes_per_frame, + mode_lib->mp.meta_pte_bytes_per_frame_ub_l, + mode_lib->mp.meta_pte_bytes_per_frame_ub_c, + mode_lib->ip.dcn_mrq_present, + + /* Output */ + mode_lib->mp.TimePerVMGroupVBlank, + mode_lib->mp.TimePerVMGroupFlip, + mode_lib->mp.TimePerVMRequestVBlank, + mode_lib->mp.TimePerVMRequestFlip); + + // VStartup Adjustment + for (k = 0; k < s->num_active_planes; ++k) { + bool isInterlaceTiming; + + mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.TWait[k] + mode_lib->mp.ExtraLatency; + if (!display_cfg->plane_descriptors[k].dynamic_meta_data.enable) + mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.TCalc + mode_lib->mp.MinTTUVBlank[k]; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f (before vstartup margin)\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); +#endif + s->Tvstartup_margin = (s->MaxVStartupLines[k] - mode_lib->mp.VStartupMin[k]) * display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000); + mode_lib->mp.MinTTUVBlank[k] = mode_lib->mp.MinTTUVBlank[k] + s->Tvstartup_margin; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, Tvstartup_margin = %f\n", __func__, k, s->Tvstartup_margin); + DML_LOG_VERBOSE("DML::%s: k=%u, MaxVStartupLines = %u\n", __func__, k, s->MaxVStartupLines[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MinTTUVBlank = %f\n", __func__, k, mode_lib->mp.MinTTUVBlank[k]); +#endif + + mode_lib->mp.Tdmdl[k] = mode_lib->mp.Tdmdl[k] + s->Tvstartup_margin; + if (display_cfg->plane_descriptors[k].dynamic_meta_data.enable && mode_lib->ip.dynamic_metadata_vm_enabled) { + mode_lib->mp.Tdmdl_vm[k] = mode_lib->mp.Tdmdl_vm[k] + s->Tvstartup_margin; + } + + isInterlaceTiming = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.interlaced && !mode_lib->ip.ptoi_supported); + + // The actual positioning of the vstartup + mode_lib->mp.VStartup[k] = (isInterlaceTiming ? (2 * s->MaxVStartupLines[k]) : s->MaxVStartupLines[k]); + + s->dlg_vblank_start = ((isInterlaceTiming ? math_floor2((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch) / 2.0, 1.0) : + display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total) - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); + s->LSetup = math_floor2(4.0 * mode_lib->mp.TSetup[k] / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total / ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000)), 1.0) / 4.0; + s->blank_lines_remaining = (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active) - mode_lib->mp.VStartup[k]; + + if (s->blank_lines_remaining < 0) { + DML_LOG_VERBOSE("ERROR: Vstartup is larger than vblank!?\n"); + s->blank_lines_remaining = 0; + DML_ASSERT(0); + } + mode_lib->mp.MIN_DST_Y_NEXT_START[k] = s->dlg_vblank_start + s->blank_lines_remaining + s->LSetup; + + // debug only + if (((mode_lib->mp.VUpdateOffsetPix[k] + mode_lib->mp.VUpdateWidthPix[k] + mode_lib->mp.VReadyOffsetPix[k]) / (double) display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total) <= + (isInterlaceTiming ? + math_floor2((display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch - mode_lib->mp.VStartup[k]) / 2.0, 1.0) : + (int)(display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active - display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch - mode_lib->mp.VStartup[k]))) { + mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k] = true; + } else { + mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k] = false; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, VStartup = %u (max)\n", __func__, k, mode_lib->mp.VStartup[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VStartupMin = %u (max)\n", __func__, k, mode_lib->mp.VStartupMin[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateOffsetPix = %u\n", __func__, k, mode_lib->mp.VUpdateOffsetPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VUpdateWidthPix = %u\n", __func__, k, mode_lib->mp.VUpdateWidthPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VReadyOffsetPix = %u\n", __func__, k, mode_lib->mp.VReadyOffsetPix[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, HTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.h_total); + DML_LOG_VERBOSE("DML::%s: k=%u, VTotal = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_total); + DML_LOG_VERBOSE("DML::%s: k=%u, VActive = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_active); + DML_LOG_VERBOSE("DML::%s: k=%u, VFrontPorch = %lu\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.v_front_porch); + DML_LOG_VERBOSE("DML::%s: k=%u, TSetup = %f\n", __func__, k, mode_lib->mp.TSetup[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, MIN_DST_Y_NEXT_START = %f\n", __func__, k, mode_lib->mp.MIN_DST_Y_NEXT_START[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, VREADY_AT_OR_AFTER_VSYNC = %u\n", __func__, k, mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]); +#endif + } + + //Maximum Bandwidth Used + mode_lib->mp.TotalWRBandwidth = 0; + for (k = 0; k < display_cfg->num_streams; ++k) { + s->WRBandwidth = 0; + if (display_cfg->stream_descriptors[k].writeback.active_writebacks_per_stream > 0) { + s->WRBandwidth = display_cfg->stream_descriptors[k].writeback.writeback_stream[0].output_height + * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].output_width / + (display_cfg->stream_descriptors[k].timing.h_total * display_cfg->stream_descriptors[k].writeback.writeback_stream[0].input_height + / ((double)display_cfg->stream_descriptors[k].timing.pixel_clock_khz / 1000)) + * (display_cfg->stream_descriptors[k].writeback.writeback_stream[0].pixel_format == dml2_444_32 ? 4.0 : 8.0); + mode_lib->mp.TotalWRBandwidth = mode_lib->mp.TotalWRBandwidth + s->WRBandwidth; + } + } + + mode_lib->mp.TotalDataReadBandwidth = 0; + for (k = 0; k < s->num_active_planes; ++k) { + mode_lib->mp.TotalDataReadBandwidth = mode_lib->mp.TotalDataReadBandwidth + mode_lib->mp.vactive_sw_bw_l[k] + mode_lib->mp.vactive_sw_bw_c[k]; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%u, TotalDataReadBandwidth = %f\n", __func__, k, mode_lib->mp.TotalDataReadBandwidth); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_l = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_l[k]); + DML_LOG_VERBOSE("DML::%s: k=%u, vactive_sw_bw_c = %f\n", __func__, k, mode_lib->mp.vactive_sw_bw_c[k]); +#endif + } + + CalculateStutterEfficiency_params->display_cfg = display_cfg; + CalculateStutterEfficiency_params->CompressedBufferSizeInkByte = mode_lib->mp.CompressedBufferSizeInkByte; + CalculateStutterEfficiency_params->UnboundedRequestEnabled = mode_lib->mp.UnboundedRequestEnabled; + CalculateStutterEfficiency_params->MetaFIFOSizeInKEntries = mode_lib->ip.meta_fifo_size_in_kentries; + CalculateStutterEfficiency_params->ZeroSizeBufferEntries = mode_lib->ip.zero_size_buffer_entries; + CalculateStutterEfficiency_params->PixelChunkSizeInKByte = mode_lib->ip.pixel_chunk_size_kbytes; + CalculateStutterEfficiency_params->NumberOfActiveSurfaces = s->num_active_planes; + CalculateStutterEfficiency_params->ROBBufferSizeInKByte = mode_lib->ip.rob_buffer_size_kbytes; + CalculateStutterEfficiency_params->TotalDataReadBandwidth = mode_lib->mp.TotalDataReadBandwidth; + CalculateStutterEfficiency_params->DCFCLK = mode_lib->mp.Dcfclk; + CalculateStutterEfficiency_params->ReturnBW = mode_lib->mp.urg_bandwidth_available_min[dml2_core_internal_soc_state_sys_active]; + CalculateStutterEfficiency_params->CompbufReservedSpace64B = mode_lib->mp.compbuf_reserved_space_64b; + CalculateStutterEfficiency_params->CompbufReservedSpaceZs = mode_lib->ip.compbuf_reserved_space_zs; + CalculateStutterEfficiency_params->SRExitTime = mode_lib->soc.power_management_parameters.stutter_exit_latency_us; + CalculateStutterEfficiency_params->SRExitZ8Time = mode_lib->soc.power_management_parameters.z8_stutter_exit_latency_us; + CalculateStutterEfficiency_params->SynchronizeTimings = display_cfg->overrides.synchronize_timings; + CalculateStutterEfficiency_params->StutterEnterPlusExitWatermark = mode_lib->mp.Watermark.StutterEnterPlusExitWatermark; + CalculateStutterEfficiency_params->Z8StutterEnterPlusExitWatermark = mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark; + CalculateStutterEfficiency_params->ProgressiveToInterlaceUnitInOPP = mode_lib->ip.ptoi_supported; + CalculateStutterEfficiency_params->MinTTUVBlank = mode_lib->mp.MinTTUVBlank; + CalculateStutterEfficiency_params->DPPPerSurface = mode_lib->mp.NoOfDPP; + CalculateStutterEfficiency_params->DETBufferSizeY = mode_lib->mp.DETBufferSizeY; + CalculateStutterEfficiency_params->BytePerPixelY = mode_lib->mp.BytePerPixelY; + CalculateStutterEfficiency_params->BytePerPixelDETY = mode_lib->mp.BytePerPixelInDETY; + CalculateStutterEfficiency_params->SwathWidthY = mode_lib->mp.SwathWidthY; + CalculateStutterEfficiency_params->SwathHeightY = mode_lib->mp.SwathHeightY; + CalculateStutterEfficiency_params->SwathHeightC = mode_lib->mp.SwathHeightC; + CalculateStutterEfficiency_params->BlockHeight256BytesY = mode_lib->mp.Read256BlockHeightY; + CalculateStutterEfficiency_params->BlockWidth256BytesY = mode_lib->mp.Read256BlockWidthY; + CalculateStutterEfficiency_params->BlockHeight256BytesC = mode_lib->mp.Read256BlockHeightC; + CalculateStutterEfficiency_params->BlockWidth256BytesC = mode_lib->mp.Read256BlockWidthC; + CalculateStutterEfficiency_params->DCCYMaxUncompressedBlock = mode_lib->mp.DCCYMaxUncompressedBlock; + CalculateStutterEfficiency_params->DCCCMaxUncompressedBlock = mode_lib->mp.DCCCMaxUncompressedBlock; + CalculateStutterEfficiency_params->ReadBandwidthSurfaceLuma = mode_lib->mp.vactive_sw_bw_l; + CalculateStutterEfficiency_params->ReadBandwidthSurfaceChroma = mode_lib->mp.vactive_sw_bw_c; + CalculateStutterEfficiency_params->dpte_row_bw = mode_lib->mp.dpte_row_bw; + CalculateStutterEfficiency_params->meta_row_bw = mode_lib->mp.meta_row_bw; + CalculateStutterEfficiency_params->rob_alloc_compressed = mode_lib->ip.dcn_mrq_present; + + // output + CalculateStutterEfficiency_params->StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.StutterEfficiencyNotIncludingVBlank; + CalculateStutterEfficiency_params->StutterEfficiency = &mode_lib->mp.StutterEfficiency; + CalculateStutterEfficiency_params->NumberOfStutterBurstsPerFrame = &mode_lib->mp.NumberOfStutterBurstsPerFrame; + CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlank; + CalculateStutterEfficiency_params->Z8StutterEfficiency = &mode_lib->mp.Z8StutterEfficiency; + CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &mode_lib->mp.Z8NumberOfStutterBurstsPerFrame; + CalculateStutterEfficiency_params->StutterPeriod = &mode_lib->mp.StutterPeriod; + CalculateStutterEfficiency_params->DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE = &mode_lib->mp.DCHUBBUB_ARB_CSTATE_MAX_CAP_MODE; + + // Stutter Efficiency + CalculateStutterEfficiency(&mode_lib->scratch, CalculateStutterEfficiency_params); + +#ifdef __DML_VBA_ALLOW_DELTA__ + // Calculate z8 stutter eff assuming 0 reserved space + CalculateStutterEfficiency_params->CompbufReservedSpace64B = 0; + CalculateStutterEfficiency_params->CompbufReservedSpaceZs = 0; + + CalculateStutterEfficiency_params->Z8StutterEfficiencyNotIncludingVBlank = &mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlankBestCase; + CalculateStutterEfficiency_params->Z8StutterEfficiency = &mode_lib->mp.Z8StutterEfficiencyBestCase; + CalculateStutterEfficiency_params->Z8NumberOfStutterBurstsPerFrame = &mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase; + CalculateStutterEfficiency_params->StutterPeriod = &mode_lib->mp.StutterPeriodBestCase; + + // Stutter Efficiency + CalculateStutterEfficiency(&mode_lib->scratch, CalculateStutterEfficiency_params); +#else + mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlankBestCase = mode_lib->mp.Z8StutterEfficiencyNotIncludingVBlank; + mode_lib->mp.Z8StutterEfficiencyBestCase = mode_lib->mp.Z8StutterEfficiency; + mode_lib->mp.Z8NumberOfStutterBurstsPerFrameBestCase = mode_lib->mp.Z8NumberOfStutterBurstsPerFrame; + mode_lib->mp.StutterPeriodBestCase = mode_lib->mp.StutterPeriod; +#endif + } // PrefetchAndImmediateFlipSupported + + max_uclk_mhz = mode_lib->soc.clk_table.uclk.clk_values_khz[mode_lib->soc.clk_table.uclk.num_clk_values - 1] / 1000.0; + min_return_latency_in_DCFCLK_cycles = (min_return_uclk_cycles / max_uclk_mhz + min_return_fclk_cycles / max_fclk_mhz) * hard_minimum_dcfclk_mhz; + mode_lib->mp.min_return_latency_in_dcfclk = (unsigned int)min_return_latency_in_DCFCLK_cycles; + mode_lib->mp.dcfclk_deep_sleep_hysteresis = (unsigned int)math_max2(32, (double)mode_lib->ip.pixel_chunk_size_kbytes * 1024 * 3 / 4 / 64 - min_return_latency_in_DCFCLK_cycles); + DML_ASSERT(mode_lib->mp.dcfclk_deep_sleep_hysteresis < 256); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: max_fclk_mhz = %f\n", __func__, max_fclk_mhz); + DML_LOG_VERBOSE("DML::%s: max_uclk_mhz = %f\n", __func__, max_uclk_mhz); + DML_LOG_VERBOSE("DML::%s: hard_minimum_dcfclk_mhz = %f\n", __func__, hard_minimum_dcfclk_mhz); + DML_LOG_VERBOSE("DML::%s: min_return_uclk_cycles = %ld\n", __func__, min_return_uclk_cycles); + DML_LOG_VERBOSE("DML::%s: min_return_fclk_cycles = %ld\n", __func__, min_return_fclk_cycles); + DML_LOG_VERBOSE("DML::%s: min_return_latency_in_DCFCLK_cycles = %f\n", __func__, min_return_latency_in_DCFCLK_cycles); + DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d \n", __func__, mode_lib->mp.dcfclk_deep_sleep_hysteresis); + DML_LOG_VERBOSE("DML::%s: --- END --- \n", __func__); +#endif + return (in_out_params->mode_lib->mp.PrefetchAndImmediateFlipSupported); +} + +bool dml2_core_calcs_mode_programming_ex(struct dml2_core_calcs_mode_programming_ex *in_out_params) +{ + DML_LOG_VERBOSE("DML::%s: ------------- START ----------\n", __func__); + bool result = dml_core_mode_programming(in_out_params); + + DML_LOG_VERBOSE("DML::%s: result = %0d\n", __func__, result); + DML_LOG_VERBOSE("DML::%s: ------------- DONE ----------\n", __func__); + return result; +} + +void dml2_core_calcs_get_dpte_row_height( + unsigned int *dpte_row_height, + struct dml2_core_internal_display_mode_lib *mode_lib, + bool is_plane1, + enum dml2_source_format_class SourcePixelFormat, + enum dml2_swizzle_mode SurfaceTiling, + enum dml2_rotation_angle ScanDirection, + unsigned int pitch, + unsigned int GPUVMMinPageSizeKBytes) +{ + unsigned int BytePerPixelY; + unsigned int BytePerPixelC; + double BytePerPixelInDETY; + double BytePerPixelInDETC; + unsigned int BlockHeight256BytesY; + unsigned int BlockHeight256BytesC; + unsigned int BlockWidth256BytesY; + unsigned int BlockWidth256BytesC; + unsigned int MacroTileWidthY; + unsigned int MacroTileWidthC; + unsigned int MacroTileHeightY; + unsigned int MacroTileHeightC; + bool surf_linear_128_l = false; + bool surf_linear_128_c = false; + + CalculateBytePerPixelAndBlockSizes( + SourcePixelFormat, + SurfaceTiling, + pitch, + pitch, + + /* Output */ + &BytePerPixelY, + &BytePerPixelC, + &BytePerPixelInDETY, + &BytePerPixelInDETC, + &BlockHeight256BytesY, + &BlockHeight256BytesC, + &BlockWidth256BytesY, + &BlockWidth256BytesC, + &MacroTileHeightY, + &MacroTileHeightC, + &MacroTileWidthY, + &MacroTileWidthC, + &surf_linear_128_l, + &surf_linear_128_c); + + unsigned int BytePerPixel = is_plane1 ? BytePerPixelC : BytePerPixelY; + unsigned int BlockHeight256Bytes = is_plane1 ? BlockHeight256BytesC : BlockHeight256BytesY; + unsigned int BlockWidth256Bytes = is_plane1 ? BlockWidth256BytesC : BlockWidth256BytesY; + unsigned int MacroTileWidth = is_plane1 ? MacroTileWidthC : MacroTileWidthY; + unsigned int MacroTileHeight = is_plane1 ? MacroTileHeightC : MacroTileHeightY; + unsigned int PTEBufferSizeInRequests = is_plane1 ? mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma : mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma; +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML: %s: is_plane1 = %u\n", __func__, is_plane1); + DML_LOG_VERBOSE("DML: %s: BytePerPixel = %u\n", __func__, BytePerPixel); + DML_LOG_VERBOSE("DML: %s: BlockHeight256Bytes = %u\n", __func__, BlockHeight256Bytes); + DML_LOG_VERBOSE("DML: %s: BlockWidth256Bytes = %u\n", __func__, BlockWidth256Bytes); + DML_LOG_VERBOSE("DML: %s: MacroTileWidth = %u\n", __func__, MacroTileWidth); + DML_LOG_VERBOSE("DML: %s: MacroTileHeight = %u\n", __func__, MacroTileHeight); + DML_LOG_VERBOSE("DML: %s: PTEBufferSizeInRequests = %u\n", __func__, PTEBufferSizeInRequests); + DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_luma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_luma); + DML_LOG_VERBOSE("DML: %s: dpte_buffer_size_in_pte_reqs_chroma = %u\n", __func__, mode_lib->ip.dpte_buffer_size_in_pte_reqs_chroma); + DML_LOG_VERBOSE("DML: %s: GPUVMMinPageSizeKBytes = %u\n", __func__, GPUVMMinPageSizeKBytes); +#endif + unsigned int dummy_integer[21]; + + mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportStationary = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.DCCEnable = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.NumberOfDPPs = 1; + mode_lib->scratch.calculate_vm_and_row_bytes_params.BlockHeight256Bytes = BlockHeight256Bytes; + mode_lib->scratch.calculate_vm_and_row_bytes_params.BlockWidth256Bytes = BlockWidth256Bytes; + mode_lib->scratch.calculate_vm_and_row_bytes_params.SourcePixelFormat = SourcePixelFormat; + mode_lib->scratch.calculate_vm_and_row_bytes_params.SurfaceTiling = SurfaceTiling; + mode_lib->scratch.calculate_vm_and_row_bytes_params.BytePerPixel = BytePerPixel; + mode_lib->scratch.calculate_vm_and_row_bytes_params.RotationAngle = ScanDirection; + mode_lib->scratch.calculate_vm_and_row_bytes_params.SwathWidth = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportHeight = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportXStart = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.ViewportYStart = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMEnable = 1; + mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMMaxPageTableLevels = 4; + mode_lib->scratch.calculate_vm_and_row_bytes_params.GPUVMMinPageSizeKBytes = GPUVMMinPageSizeKBytes; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PTEBufferSizeInRequests = PTEBufferSizeInRequests; + mode_lib->scratch.calculate_vm_and_row_bytes_params.Pitch = pitch; + mode_lib->scratch.calculate_vm_and_row_bytes_params.MacroTileWidth = MacroTileWidth; + mode_lib->scratch.calculate_vm_and_row_bytes_params.MacroTileHeight = MacroTileHeight; + mode_lib->scratch.calculate_vm_and_row_bytes_params.is_phantom = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.DCCMetaPitch = 0; + mode_lib->scratch.calculate_vm_and_row_bytes_params.mrq_present = 0; + + mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow = &dummy_integer[1]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRowStorage = &dummy_integer[2]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_width_ub = &dummy_integer[3]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height = dpte_row_height; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height_linear = &dummy_integer[4]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEBytesPerRow_one_row_per_frame = &dummy_integer[5]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_width_ub_one_row_per_frame = &dummy_integer[6]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpte_row_height_one_row_per_frame = &dummy_integer[7]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.vmpg_width = &dummy_integer[8]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.vmpg_height = &dummy_integer[9]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEReqWidth = &dummy_integer[11]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PixelPTEReqHeight = &dummy_integer[12]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.PTERequestSize = &dummy_integer[13]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.dpde0_bytes_per_frame_ub = &dummy_integer[14]; + + mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_bytes = &dummy_integer[15]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.MetaRequestWidth = &dummy_integer[16]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.MetaRequestHeight = &dummy_integer[17]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_width = &dummy_integer[18]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_row_height = &dummy_integer[19]; + mode_lib->scratch.calculate_vm_and_row_bytes_params.meta_pte_bytes_per_frame_ub = &dummy_integer[20]; + + // just supply with enough parameters to calculate dpte + CalculateVMAndRowBytes(&mode_lib->scratch.calculate_vm_and_row_bytes_params); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML: %s: dpte_row_height = %u\n", __func__, *dpte_row_height); +#endif +} + +static bool is_dual_plane(enum dml2_source_format_class source_format) +{ + bool ret_val = false; + + if ((source_format == dml2_420_12) || (source_format == dml2_420_8) || (source_format == dml2_420_10) || (source_format == dml2_rgbe_alpha)) + ret_val = true; + + return ret_val; +} + +static unsigned int dml_get_plane_idx(const struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int pipe_idx) +{ + unsigned int plane_idx = mode_lib->mp.pipe_plane[pipe_idx]; + return plane_idx; +} + +static void rq_dlg_get_wm_regs(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *wm_regs) +{ + double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; + + wm_regs->fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); + wm_regs->sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + wm_regs->sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + wm_regs->sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + wm_regs->sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); + wm_regs->temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); + wm_regs->uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); + wm_regs->urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + wm_regs->usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); + wm_regs->refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.UrgentLatency * refclk_freq_in_mhz); + wm_regs->refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.MetaTripToMemory * refclk_freq_in_mhz); + wm_regs->frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); + wm_regs->frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); + wm_regs->frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); +} + +static unsigned int log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend) +{ + if (a == 0) + return 0; + + return (math_log2_approx(a) - subtrahend); +} + +void dml2_core_calcs_cursor_dlg_reg(struct dml2_cursor_dlg_regs *cursor_dlg_regs, const struct dml2_get_cursor_dlg_reg *p) +{ + int dst_x_offset = (int) ((p->cursor_x_position + (p->cursor_stereo_en == 0 ? 0 : math_max2(p->cursor_primary_offset, p->cursor_secondary_offset)) - + (p->cursor_hotspot_x * (p->cursor_2x_magnify == 0 ? 1 : 2))) * p->dlg_refclk_mhz / p->pixel_rate_mhz / p->hratio); + cursor_dlg_regs->dst_x_offset = (unsigned int) ((dst_x_offset > 0) ? dst_x_offset : 0); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG::%s: cursor_x_position=%d\n", __func__, p->cursor_x_position); + DML_LOG_VERBOSE("DML_DLG::%s: dlg_refclk_mhz=%f\n", __func__, p->dlg_refclk_mhz); + DML_LOG_VERBOSE("DML_DLG::%s: pixel_rate_mhz=%f\n", __func__, p->pixel_rate_mhz); + DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d\n", __func__, dst_x_offset); + DML_LOG_VERBOSE("DML_DLG::%s: dst_x_offset=%d (reg)\n", __func__, cursor_dlg_regs->dst_x_offset); +#endif + + cursor_dlg_regs->chunk_hdl_adjust = 3; + cursor_dlg_regs->dst_y_offset = 0; + + cursor_dlg_regs->qos_level_fixed = 8; + cursor_dlg_regs->qos_ramp_disable = 0; +} + +static void rq_dlg_get_rq_reg(struct dml2_display_rq_regs *rq_regs, + const struct dml2_display_cfg *display_cfg, + const struct dml2_core_internal_display_mode_lib *mode_lib, + unsigned int pipe_idx) +{ + unsigned int plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); + enum dml2_source_format_class source_format = display_cfg->plane_descriptors[plane_idx].pixel_format; + enum dml2_swizzle_mode sw_mode = display_cfg->plane_descriptors[plane_idx].surface.tiling; + bool dual_plane = is_dual_plane((enum dml2_source_format_class)(source_format)); + + unsigned int pixel_chunk_bytes = 0; + unsigned int min_pixel_chunk_bytes = 0; + unsigned int dpte_group_bytes = 0; + unsigned int mpte_group_bytes = 0; + + unsigned int p1_pixel_chunk_bytes = 0; + unsigned int p1_min_pixel_chunk_bytes = 0; + unsigned int p1_dpte_group_bytes = 0; + unsigned int p1_mpte_group_bytes = 0; + + unsigned int detile_buf_plane1_addr = 0; + unsigned int detile_buf_size_in_bytes; + double stored_swath_l_bytes; + double stored_swath_c_bytes; + bool is_phantom_pipe; + + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); + + pixel_chunk_bytes = (unsigned int)(mode_lib->ip.pixel_chunk_size_kbytes * 1024); + min_pixel_chunk_bytes = (unsigned int)(mode_lib->ip.min_pixel_chunk_size_bytes); + + if (pixel_chunk_bytes == 64 * 1024) + min_pixel_chunk_bytes = 0; + + dpte_group_bytes = (unsigned int)(dml_get_dpte_group_size_in_bytes(mode_lib, pipe_idx)); + mpte_group_bytes = (unsigned int)(dml_get_vm_group_size_in_bytes(mode_lib, pipe_idx)); + + p1_pixel_chunk_bytes = pixel_chunk_bytes; + p1_min_pixel_chunk_bytes = min_pixel_chunk_bytes; + p1_dpte_group_bytes = dpte_group_bytes; + p1_mpte_group_bytes = mpte_group_bytes; + + if (source_format == dml2_rgbe_alpha) + p1_pixel_chunk_bytes = (unsigned int)(mode_lib->ip.alpha_pixel_chunk_size_kbytes * 1024); + + rq_regs->unbounded_request_enabled = dml_get_unbounded_request_enabled(mode_lib); + rq_regs->rq_regs_l.chunk_size = log_and_substract_if_non_zero(pixel_chunk_bytes, 10); + rq_regs->rq_regs_c.chunk_size = log_and_substract_if_non_zero(p1_pixel_chunk_bytes, 10); + + if (min_pixel_chunk_bytes == 0) + rq_regs->rq_regs_l.min_chunk_size = 0; + else + rq_regs->rq_regs_l.min_chunk_size = log_and_substract_if_non_zero(min_pixel_chunk_bytes, 8 - 1); + + if (p1_min_pixel_chunk_bytes == 0) + rq_regs->rq_regs_c.min_chunk_size = 0; + else + rq_regs->rq_regs_c.min_chunk_size = log_and_substract_if_non_zero(p1_min_pixel_chunk_bytes, 8 - 1); + + rq_regs->rq_regs_l.dpte_group_size = log_and_substract_if_non_zero(dpte_group_bytes, 6); + rq_regs->rq_regs_l.mpte_group_size = log_and_substract_if_non_zero(mpte_group_bytes, 6); + rq_regs->rq_regs_c.dpte_group_size = log_and_substract_if_non_zero(p1_dpte_group_bytes, 6); + rq_regs->rq_regs_c.mpte_group_size = log_and_substract_if_non_zero(p1_mpte_group_bytes, 6); + + detile_buf_size_in_bytes = (unsigned int)(dml_get_det_buffer_size_kbytes(mode_lib, pipe_idx) * 1024); + + if (sw_mode == dml2_sw_linear && display_cfg->gpuvm_enable) { + unsigned int p0_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_l(mode_lib, pipe_idx)); +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG: %s: p0_pte_row_height_linear = %u\n", __func__, p0_pte_row_height_linear); +#endif + DML_ASSERT(p0_pte_row_height_linear >= 8); + + rq_regs->rq_regs_l.pte_row_height_linear = math_log2_approx(p0_pte_row_height_linear) - 3; + if (dual_plane) { + unsigned int p1_pte_row_height_linear = (unsigned int)(dml_get_dpte_row_height_linear_c(mode_lib, pipe_idx)); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG: %s: p1_pte_row_height_linear = %u\n", __func__, p1_pte_row_height_linear); +#endif + if (sw_mode == dml2_sw_linear) { + DML_ASSERT(p1_pte_row_height_linear >= 8); + } + rq_regs->rq_regs_c.pte_row_height_linear = math_log2_approx(p1_pte_row_height_linear) - 3; + } + } else { + rq_regs->rq_regs_l.pte_row_height_linear = 0; + rq_regs->rq_regs_c.pte_row_height_linear = 0; + } + + rq_regs->rq_regs_l.swath_height = log_and_substract_if_non_zero(dml_get_swath_height_l(mode_lib, pipe_idx), 0); + rq_regs->rq_regs_c.swath_height = log_and_substract_if_non_zero(dml_get_swath_height_c(mode_lib, pipe_idx), 0); + + // FIXME_DCN4, programming guide has dGPU condition + if (pixel_chunk_bytes >= 32 * 1024 || (dual_plane && p1_pixel_chunk_bytes >= 32 * 1024)) { //32kb + rq_regs->drq_expansion_mode = 0; + } else { + rq_regs->drq_expansion_mode = 2; + } + rq_regs->prq_expansion_mode = 1; + rq_regs->crq_expansion_mode = 1; + rq_regs->mrq_expansion_mode = 1; + + stored_swath_l_bytes = dml_get_det_stored_buffer_size_l_bytes(mode_lib, pipe_idx); + stored_swath_c_bytes = dml_get_det_stored_buffer_size_c_bytes(mode_lib, pipe_idx); + is_phantom_pipe = dml_get_is_phantom_pipe(display_cfg, mode_lib, pipe_idx); + + // Note: detile_buf_plane1_addr is in unit of 1KB + if (dual_plane) { + if (is_phantom_pipe) { + detile_buf_plane1_addr = (unsigned int)((1024.0 * 1024.0) / 2.0 / 1024.0); // half to chroma + } else { + if (stored_swath_l_bytes / stored_swath_c_bytes <= 1.5) { + detile_buf_plane1_addr = (unsigned int)(detile_buf_size_in_bytes / 2.0 / 1024.0); // half to chroma +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); +#endif + } else { + detile_buf_plane1_addr = (unsigned int)(dml_round_to_multiple((unsigned int)((2.0 * detile_buf_size_in_bytes) / 3.0), 1024, 0) / 1024.0); // 2/3 to luma +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); +#endif + } + } + } + rq_regs->plane1_base_address = detile_buf_plane1_addr; + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); + DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); + DML_LOG_VERBOSE("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); + DML_LOG_VERBOSE("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); +#endif + //DML_LOG_VERBOSE_rq_regs_st(rq_regs); + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); +} + +static void rq_dlg_get_dlg_reg( + struct dml2_core_internal_scratch *s, + struct dml2_display_dlg_regs *disp_dlg_regs, + struct dml2_display_ttu_regs *disp_ttu_regs, + const struct dml2_display_cfg *display_cfg, + const struct dml2_core_internal_display_mode_lib *mode_lib, + const unsigned int pipe_idx) +{ + struct dml2_core_shared_rq_dlg_get_dlg_reg_locals *l = &s->rq_dlg_get_dlg_reg_locals; + + memset(l, 0, sizeof(struct dml2_core_shared_rq_dlg_get_dlg_reg_locals)); + + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); + + l->plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); + DML_ASSERT(l->plane_idx < DML2_MAX_PLANES); + + l->source_format = dml2_444_8; + l->odm_mode = dml2_odm_mode_bypass; + l->dual_plane = false; + l->htotal = 0; + l->hactive = 0; + l->hblank_end = 0; + l->vblank_end = 0; + l->interlaced = false; + l->pclk_freq_in_mhz = 0.0; + l->refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; + l->ref_freq_to_pix_freq = 0.0; + + if (l->plane_idx < DML2_MAX_PLANES) { + + l->timing = &display_cfg->stream_descriptors[display_cfg->plane_descriptors[l->plane_idx].stream_index].timing; + l->source_format = display_cfg->plane_descriptors[l->plane_idx].pixel_format; + l->odm_mode = mode_lib->mp.ODMMode[l->plane_idx]; + + l->dual_plane = is_dual_plane(l->source_format); + + l->htotal = l->timing->h_total; + l->hactive = l->timing->h_active; + l->hblank_end = l->timing->h_blank_end; + l->vblank_end = l->timing->v_blank_end; + l->interlaced = l->timing->interlaced; + l->pclk_freq_in_mhz = (double)l->timing->pixel_clock_khz / 1000; + l->ref_freq_to_pix_freq = l->refclk_freq_in_mhz / l->pclk_freq_in_mhz; + + DML_LOG_VERBOSE("DML_DLG::%s: plane_idx = %d\n", __func__, l->plane_idx); + DML_LOG_VERBOSE("DML_DLG: %s: htotal = %d\n", __func__, l->htotal); + DML_LOG_VERBOSE("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, l->refclk_freq_in_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: dlg_ref_clk_mhz = %3.2f\n", __func__, display_cfg->overrides.hw.dlg_ref_clk_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: soc.refclk_mhz = %d\n", __func__, mode_lib->soc.dchub_refclk_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, l->pclk_freq_in_mhz); + DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); + DML_LOG_VERBOSE("DML_DLG: %s: interlaced = %d\n", __func__, l->interlaced); + + DML_ASSERT(l->refclk_freq_in_mhz != 0); + DML_ASSERT(l->pclk_freq_in_mhz != 0); + DML_ASSERT(l->ref_freq_to_pix_freq < 4.0); + + // Need to figure out which side of odm combine we're in + // Assume the pipe instance under the same plane is in order + + if (l->odm_mode == dml2_odm_mode_bypass) { + disp_dlg_regs->refcyc_h_blank_end = (unsigned int)((double)l->hblank_end * l->ref_freq_to_pix_freq); + } else if (l->odm_mode == dml2_odm_mode_combine_2to1 || l->odm_mode == dml2_odm_mode_combine_3to1 || l->odm_mode == dml2_odm_mode_combine_4to1) { + // find out how many pipe are in this plane + l->num_active_pipes = mode_lib->mp.num_active_pipes; + l->first_pipe_idx_in_plane = DML2_MAX_PLANES; + l->pipe_idx_in_combine = 0; // pipe index within the plane + l->odm_combine_factor = 2; + + if (l->odm_mode == dml2_odm_mode_combine_3to1) + l->odm_combine_factor = 3; + else if (l->odm_mode == dml2_odm_mode_combine_4to1) + l->odm_combine_factor = 4; + + for (unsigned int i = 0; i < l->num_active_pipes; i++) { + if (dml_get_plane_idx(mode_lib, i) == l->plane_idx) { + if (i < l->first_pipe_idx_in_plane) { + l->first_pipe_idx_in_plane = i; + } + } + } + l->pipe_idx_in_combine = pipe_idx - l->first_pipe_idx_in_plane; // DML assumes the pipes in the same plane will have continuous indexing (i.e. plane 0 use pipe 0, 1, and plane 1 uses pipe 2, 3, etc.) + + disp_dlg_regs->refcyc_h_blank_end = (unsigned int)(((double)l->hblank_end + (double)l->pipe_idx_in_combine * (double)l->hactive / (double)l->odm_combine_factor) * l->ref_freq_to_pix_freq); + DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); + DML_LOG_VERBOSE("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, l->first_pipe_idx_in_plane); + DML_LOG_VERBOSE("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, l->pipe_idx_in_combine); + DML_LOG_VERBOSE("DML_DLG: %s: odm_combine_factor = %d\n", __func__, l->odm_combine_factor); + } + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); + + DML_ASSERT(disp_dlg_regs->refcyc_h_blank_end < (unsigned int)math_pow(2, 13)); + + disp_dlg_regs->ref_freq_to_pix_freq = (unsigned int)(l->ref_freq_to_pix_freq * math_pow(2, 19)); + disp_dlg_regs->refcyc_per_htotal = (unsigned int)(l->ref_freq_to_pix_freq * (double)l->htotal * math_pow(2, 8)); + disp_dlg_regs->dlg_vblank_end = l->interlaced ? (l->vblank_end / 2) : l->vblank_end; // 15 bits + + l->min_ttu_vblank = mode_lib->mp.MinTTUVBlank[mode_lib->mp.pipe_plane[pipe_idx]]; + l->min_dst_y_next_start = (unsigned int)(mode_lib->mp.MIN_DST_Y_NEXT_START[mode_lib->mp.pipe_plane[pipe_idx]]); + + DML_LOG_VERBOSE("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, l->min_ttu_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, l->min_dst_y_next_start); + DML_LOG_VERBOSE("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, l->ref_freq_to_pix_freq); + + l->vready_after_vcount0 = (unsigned int)(mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[mode_lib->mp.pipe_plane[pipe_idx]]); + disp_dlg_regs->vready_after_vcount0 = l->vready_after_vcount0; + + DML_LOG_VERBOSE("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); + + l->dst_x_after_scaler = (unsigned int)(mode_lib->mp.DSTXAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); + l->dst_y_after_scaler = (unsigned int)(mode_lib->mp.DSTYAfterScaler[mode_lib->mp.pipe_plane[pipe_idx]]); + + DML_LOG_VERBOSE("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, l->dst_x_after_scaler); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, l->dst_y_after_scaler); + + l->dst_y_prefetch = mode_lib->mp.dst_y_prefetch[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_vm_vblank = mode_lib->mp.dst_y_per_vm_vblank[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_row_vblank = mode_lib->mp.dst_y_per_row_vblank[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_vm_flip = mode_lib->mp.dst_y_per_vm_flip[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_row_flip = mode_lib->mp.dst_y_per_row_flip[mode_lib->mp.pipe_plane[pipe_idx]]; + + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, l->dst_y_prefetch); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, l->dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, l->dst_y_per_row_flip); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, l->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, l->dst_y_per_row_vblank); + + if (l->dst_y_prefetch > 0 && l->dst_y_per_vm_vblank > 0 && l->dst_y_per_row_vblank > 0) { + DML_ASSERT(l->dst_y_prefetch > (l->dst_y_per_vm_vblank + l->dst_y_per_row_vblank)); + } + + l->vratio_pre_l = mode_lib->mp.VRatioPrefetchY[mode_lib->mp.pipe_plane[pipe_idx]]; + l->vratio_pre_c = mode_lib->mp.VRatioPrefetchC[mode_lib->mp.pipe_plane[pipe_idx]]; + + DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, l->vratio_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, l->vratio_pre_c); + + // Active + l->refcyc_per_line_delivery_pre_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_line_delivery_l = mode_lib->mp.DisplayPipeLineDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, l->refcyc_per_line_delivery_l); + + l->refcyc_per_line_delivery_pre_c = 0.0; + l->refcyc_per_line_delivery_c = 0.0; + + if (l->dual_plane) { + l->refcyc_per_line_delivery_pre_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_line_delivery_c = mode_lib->mp.DisplayPipeLineDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_pre_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, l->refcyc_per_line_delivery_c); + } + + disp_dlg_regs->refcyc_per_vm_dmdata = (unsigned int)(mode_lib->mp.Tdmdl_vm[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); + disp_dlg_regs->dmdata_dl_delta = (unsigned int)(mode_lib->mp.Tdmdl[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); + + l->refcyc_per_req_delivery_pre_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_req_delivery_l = mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_l); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, l->refcyc_per_req_delivery_l); + + l->refcyc_per_req_delivery_pre_c = 0.0; + l->refcyc_per_req_delivery_c = 0.0; + if (l->dual_plane) { + l->refcyc_per_req_delivery_pre_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_req_delivery_c = mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_pre_c); + DML_LOG_VERBOSE("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, l->refcyc_per_req_delivery_c); + } + + // TTU - Cursor + DML_ASSERT(display_cfg->plane_descriptors[l->plane_idx].cursor.num_cursors <= 1); + + // Assign to register structures + disp_dlg_regs->min_dst_y_next_start = (unsigned int)((double)l->min_dst_y_next_start * math_pow(2, 2)); + DML_ASSERT(disp_dlg_regs->min_dst_y_next_start < (unsigned int)math_pow(2, 18)); + + disp_dlg_regs->dst_y_after_scaler = l->dst_y_after_scaler; // in terms of line + disp_dlg_regs->refcyc_x_after_scaler = (unsigned int)((double)l->dst_x_after_scaler * l->ref_freq_to_pix_freq); // in terms of refclk + disp_dlg_regs->dst_y_prefetch = (unsigned int)(l->dst_y_prefetch * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_vm_vblank = (unsigned int)(l->dst_y_per_vm_vblank * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_row_vblank = (unsigned int)(l->dst_y_per_row_vblank * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_vm_flip = (unsigned int)(l->dst_y_per_vm_flip * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_row_flip = (unsigned int)(l->dst_y_per_row_flip * math_pow(2, 2)); + + disp_dlg_regs->vratio_prefetch = (unsigned int)(l->vratio_pre_l * math_pow(2, 19)); + disp_dlg_regs->vratio_prefetch_c = (unsigned int)(l->vratio_pre_c * math_pow(2, 19)); + + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); + DML_LOG_VERBOSE("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); + + disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(mode_lib->mp.TimePerVMGroupVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); + disp_dlg_regs->refcyc_per_vm_group_flip = (unsigned int)(mode_lib->mp.TimePerVMGroupFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz); + disp_dlg_regs->refcyc_per_vm_req_vblank = (unsigned int)(mode_lib->mp.TimePerVMRequestVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz * math_pow(2, 10)); + disp_dlg_regs->refcyc_per_vm_req_flip = (unsigned int)(mode_lib->mp.TimePerVMRequestFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz * math_pow(2, 10)); + + l->dst_y_per_pte_row_nom_l = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_pte_row_nom_c = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C[mode_lib->mp.pipe_plane[pipe_idx]]; + l->refcyc_per_pte_group_nom_l = mode_lib->mp.time_per_pte_group_nom_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_pte_group_nom_c = mode_lib->mp.time_per_pte_group_nom_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_pte_group_vblank_l = mode_lib->mp.time_per_pte_group_vblank_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_pte_group_vblank_c = mode_lib->mp.time_per_pte_group_vblank_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_pte_group_flip_l = mode_lib->mp.time_per_pte_group_flip_luma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_pte_group_flip_c = mode_lib->mp.time_per_pte_group_flip_chroma[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_tdlut_group = mode_lib->mp.time_per_tdlut_group[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + disp_dlg_regs->dst_y_per_pte_row_nom_l = (unsigned int)(l->dst_y_per_pte_row_nom_l * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_pte_row_nom_c = (unsigned int)(l->dst_y_per_pte_row_nom_c * math_pow(2, 2)); + + disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int)(l->refcyc_per_pte_group_nom_l); + disp_dlg_regs->refcyc_per_pte_group_nom_c = (unsigned int)(l->refcyc_per_pte_group_nom_c); + disp_dlg_regs->refcyc_per_pte_group_vblank_l = (unsigned int)(l->refcyc_per_pte_group_vblank_l); + disp_dlg_regs->refcyc_per_pte_group_vblank_c = (unsigned int)(l->refcyc_per_pte_group_vblank_c); + disp_dlg_regs->refcyc_per_pte_group_flip_l = (unsigned int)(l->refcyc_per_pte_group_flip_l); + disp_dlg_regs->refcyc_per_pte_group_flip_c = (unsigned int)(l->refcyc_per_pte_group_flip_c); + disp_dlg_regs->refcyc_per_line_delivery_pre_l = (unsigned int)math_floor2(l->refcyc_per_line_delivery_pre_l, 1); + disp_dlg_regs->refcyc_per_line_delivery_l = (unsigned int)math_floor2(l->refcyc_per_line_delivery_l, 1); + disp_dlg_regs->refcyc_per_line_delivery_pre_c = (unsigned int)math_floor2(l->refcyc_per_line_delivery_pre_c, 1); + disp_dlg_regs->refcyc_per_line_delivery_c = (unsigned int)math_floor2(l->refcyc_per_line_delivery_c, 1); + + l->dst_y_per_meta_row_nom_l = mode_lib->mp.DST_Y_PER_META_ROW_NOM_L[mode_lib->mp.pipe_plane[pipe_idx]]; + l->dst_y_per_meta_row_nom_c = mode_lib->mp.DST_Y_PER_META_ROW_NOM_C[mode_lib->mp.pipe_plane[pipe_idx]]; + l->refcyc_per_meta_chunk_nom_l = mode_lib->mp.TimePerMetaChunkNominal[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_meta_chunk_nom_c = mode_lib->mp.TimePerChromaMetaChunkNominal[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_meta_chunk_vblank_l = mode_lib->mp.TimePerMetaChunkVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_meta_chunk_vblank_c = mode_lib->mp.TimePerChromaMetaChunkVBlank[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_meta_chunk_flip_l = mode_lib->mp.TimePerMetaChunkFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + l->refcyc_per_meta_chunk_flip_c = mode_lib->mp.TimePerChromaMetaChunkFlip[mode_lib->mp.pipe_plane[pipe_idx]] * l->refclk_freq_in_mhz; + + disp_dlg_regs->dst_y_per_meta_row_nom_l = (unsigned int)(l->dst_y_per_meta_row_nom_l * math_pow(2, 2)); + disp_dlg_regs->dst_y_per_meta_row_nom_c = (unsigned int)(l->dst_y_per_meta_row_nom_c * math_pow(2, 2)); + disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (unsigned int)(l->refcyc_per_meta_chunk_nom_l); + disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (unsigned int)(l->refcyc_per_meta_chunk_nom_c); + disp_dlg_regs->refcyc_per_meta_chunk_vblank_l = (unsigned int)(l->refcyc_per_meta_chunk_vblank_l); + disp_dlg_regs->refcyc_per_meta_chunk_vblank_c = (unsigned int)(l->refcyc_per_meta_chunk_vblank_c); + disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (unsigned int)(l->refcyc_per_meta_chunk_flip_l); + disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (unsigned int)(l->refcyc_per_meta_chunk_flip_c); + + disp_dlg_regs->refcyc_per_tdlut_group = (unsigned int)(l->refcyc_per_tdlut_group); + disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off + + disp_ttu_regs->refcyc_per_req_delivery_pre_l = (unsigned int)(l->refcyc_per_req_delivery_pre_l * math_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_l = (unsigned int)(l->refcyc_per_req_delivery_l * math_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_pre_c = (unsigned int)(l->refcyc_per_req_delivery_pre_c * math_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_c = (unsigned int)(l->refcyc_per_req_delivery_c * math_pow(2, 10)); + disp_ttu_regs->qos_level_low_wm = 0; + + disp_ttu_regs->qos_level_high_wm = (unsigned int)(4.0 * (double)l->htotal * l->ref_freq_to_pix_freq); + + disp_ttu_regs->qos_level_flip = 14; + disp_ttu_regs->qos_level_fixed_l = 8; + disp_ttu_regs->qos_level_fixed_c = 8; + disp_ttu_regs->qos_ramp_disable_l = 0; + disp_ttu_regs->qos_ramp_disable_c = 0; + disp_ttu_regs->min_ttu_vblank = (unsigned int)(l->min_ttu_vblank * l->refclk_freq_in_mhz); + + // CHECK for HW registers' range, DML_ASSERT or clamp + DML_ASSERT(l->refcyc_per_req_delivery_pre_l < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_l < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_pre_c < math_pow(2, 13)); + DML_ASSERT(l->refcyc_per_req_delivery_c < math_pow(2, 13)); + if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_group_vblank = (unsigned int)(math_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_group_flip >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_group_flip = (unsigned int)(math_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_req_vblank >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_req_vblank = (unsigned int)(math_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_req_flip >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_req_flip = (unsigned int)(math_pow(2, 23) - 1); + + + DML_ASSERT(disp_dlg_regs->dst_y_after_scaler < (unsigned int)8); + DML_ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (unsigned int)math_pow(2, 13)); + + if (disp_dlg_regs->dst_y_per_pte_row_nom_l >= (unsigned int)math_pow(2, 17)) { + DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_L %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_l, (unsigned int)math_pow(2, 17) - 1); + l->dst_y_per_pte_row_nom_l = (unsigned int)math_pow(2, 17) - 1; + } + if (l->dual_plane) { + if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (unsigned int)math_pow(2, 17)) { + DML_LOG_VERBOSE("DML_DLG: %s: Warning DST_Y_PER_PTE_ROW_NOM_C %u > register max U15.2 %u, clamp to max\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (unsigned int)math_pow(2, 17) - 1); + l->dst_y_per_pte_row_nom_c = (unsigned int)math_pow(2, 17) - 1; + } + } + + if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_pte_group_nom_l = (unsigned int)(math_pow(2, 23) - 1); + if (l->dual_plane) { + if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (unsigned int)math_pow(2, 23)) + disp_dlg_regs->refcyc_per_pte_group_nom_c = (unsigned int)(math_pow(2, 23) - 1); + } + DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (unsigned int)math_pow(2, 13)); + if (l->dual_plane) { + DML_ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (unsigned int)math_pow(2, 13)); + } + + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (unsigned int)math_pow(2, 13)); + DML_ASSERT(disp_ttu_regs->qos_level_low_wm < (unsigned int)math_pow(2, 14)); + DML_ASSERT(disp_ttu_regs->qos_level_high_wm < (unsigned int)math_pow(2, 14)); + DML_ASSERT(disp_ttu_regs->min_ttu_vblank < (unsigned int)math_pow(2, 24)); + + DML_LOG_VERBOSE("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); + + } +} + +static void rq_dlg_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *arb_param) +{ + double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; + + arb_param->max_req_outstanding = mode_lib->soc.max_outstanding_reqs; + arb_param->min_req_outstanding = mode_lib->soc.max_outstanding_reqs; // turn off the sat level feature if this set to max + arb_param->sdpif_request_rate_limit = (3 * mode_lib->ip.words_per_channel * mode_lib->soc.clk_table.dram_config.channel_count) / 4; + arb_param->sdpif_request_rate_limit = arb_param->sdpif_request_rate_limit < 96 ? 96 : arb_param->sdpif_request_rate_limit; + arb_param->sat_level_us = 60; + arb_param->hvm_max_qos_commit_threshold = 0xf; + arb_param->hvm_min_req_outstand_commit_threshold = 0xa; + arb_param->compbuf_reserved_space_kbytes = dml_get_compbuf_reserved_space_64b(mode_lib) * 64 / 1024; + arb_param->compbuf_size = mode_lib->mp.CompressedBufferSizeInkByte / mode_lib->ip.compressed_buffer_segment_size_in_kbytes; + arb_param->allow_sdpif_rate_limit_when_cstate_req = dml_get_hw_debug5(mode_lib); + arb_param->dcfclk_deep_sleep_hysteresis = dml_get_dcfclk_deep_sleep_hysteresis(mode_lib); + arb_param->pstate_stall_threshold = (unsigned int)(mode_lib->ip_caps.fams2.max_allow_delay_us * refclk_freq_in_mhz); + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: max_req_outstanding = %d\n", __func__, arb_param->max_req_outstanding); + DML_LOG_VERBOSE("DML::%s: sdpif_request_rate_limit = %d\n", __func__, arb_param->sdpif_request_rate_limit); + DML_LOG_VERBOSE("DML::%s: compbuf_reserved_space_kbytes = %d\n", __func__, arb_param->compbuf_reserved_space_kbytes); + DML_LOG_VERBOSE("DML::%s: allow_sdpif_rate_limit_when_cstate_req = %d\n", __func__, arb_param->allow_sdpif_rate_limit_when_cstate_req); + DML_LOG_VERBOSE("DML::%s: dcfclk_deep_sleep_hysteresis = %d\n", __func__, arb_param->dcfclk_deep_sleep_hysteresis); +#endif + +} + +void dml2_core_calcs_get_watermarks(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *out) +{ + rq_dlg_get_wm_regs(display_cfg, mode_lib, out); +} + +void dml2_core_calcs_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *out) +{ + rq_dlg_get_arb_params(display_cfg, mode_lib, out); +} + +void dml2_core_calcs_get_pipe_regs(const struct dml2_display_cfg *display_cfg, + struct dml2_core_internal_display_mode_lib *mode_lib, + struct dml2_dchub_per_pipe_register_set *out, int pipe_index) +{ + rq_dlg_get_rq_reg(&out->rq_regs, display_cfg, mode_lib, pipe_index); + rq_dlg_get_dlg_reg(&mode_lib->scratch, &out->dlg_regs, &out->ttu_regs, display_cfg, mode_lib, pipe_index); + out->det_size = dml_get_det_buffer_size_kbytes(mode_lib, pipe_index) / mode_lib->ip.config_return_buffer_segment_size_in_kbytes; +} + +void dml2_core_calcs_get_global_sync_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, union dml2_global_sync_programming *out, int pipe_index) +{ + out->dcn4x.vready_offset_pixels = dml_get_vready_offset(mode_lib, pipe_index); + out->dcn4x.vstartup_lines = dml_get_vstartup_calculated(mode_lib, pipe_index); + out->dcn4x.vupdate_offset_pixels = dml_get_vupdate_offset(mode_lib, pipe_index); + out->dcn4x.vupdate_vupdate_width_pixels = dml_get_vupdate_width(mode_lib, pipe_index); + out->dcn4x.pstate_keepout_start_lines = dml_get_pstate_keepout_dst_lines(mode_lib, pipe_index); +} + +void dml2_core_calcs_get_stream_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_per_stream_programming *out, int pipe_index) +{ + dml2_core_calcs_get_global_sync_programming(mode_lib, &out->global_sync, pipe_index); +} + +void dml2_core_calcs_get_global_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, + const struct display_configuation_with_meta *display_cfg, + struct dmub_cmd_fams2_global_config *fams2_global_config) +{ + fams2_global_config->features.bits.enable = display_cfg->stage3.fams2_required; + + if (fams2_global_config->features.bits.enable) { + fams2_global_config->features.bits.enable_stall_recovery = true; + fams2_global_config->features.bits.allow_delay_check_mode = FAMS2_ALLOW_DELAY_CHECK_FROM_START; + + fams2_global_config->max_allow_delay_us = mode_lib->ip_caps.fams2.max_allow_delay_us; + fams2_global_config->lock_wait_time_us = mode_lib->ip_caps.fams2.lock_timeout_us; + fams2_global_config->recovery_timeout_us = mode_lib->ip_caps.fams2.recovery_timeout_us; + fams2_global_config->hwfq_flip_programming_delay_us = mode_lib->ip_caps.fams2.flip_programming_delay_us; + + fams2_global_config->num_streams = display_cfg->display_config.num_streams; + } +} + +void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, + const struct display_configuation_with_meta *display_cfg, + union dmub_cmd_fams2_config *fams2_base_programming, + union dmub_cmd_fams2_config *fams2_sub_programming, + enum dml2_pstate_method pstate_method, + int plane_index) +{ + const struct dml2_plane_parameters *plane_descriptor = &display_cfg->display_config.plane_descriptors[plane_index]; + const struct dml2_stream_parameters *stream_descriptor = &display_cfg->display_config.stream_descriptors[plane_descriptor->stream_index]; + const struct dml2_pstate_meta *stream_pstate_meta = &display_cfg->stage3.stream_pstate_meta[plane_descriptor->stream_index]; + + struct dmub_fams2_cmd_stream_static_base_state *base_programming = &fams2_base_programming->stream_v1.base; + union dmub_fams2_cmd_stream_static_sub_state *sub_programming = &fams2_sub_programming->stream_v1.sub_state; + + unsigned int i; + + if (display_cfg->display_config.overrides.all_streams_blanked) { + /* stream is blanked, so do nothing */ + return; + } + + /* from display configuration */ + base_programming->htotal = (uint16_t)stream_descriptor->timing.h_total; + base_programming->vtotal = (uint16_t)stream_descriptor->timing.v_total; + base_programming->vblank_start = (uint16_t)(stream_pstate_meta->nom_vtotal - + stream_descriptor->timing.v_front_porch); + base_programming->vblank_end = (uint16_t)(stream_pstate_meta->nom_vtotal - + stream_descriptor->timing.v_front_porch - + stream_descriptor->timing.v_active); + base_programming->config.bits.is_drr = stream_descriptor->timing.drr_config.enabled; + + /* from meta */ + base_programming->otg_vline_time_ns = + (unsigned int)(stream_pstate_meta->otg_vline_time_us * 1000.0); + base_programming->scheduling_delay_otg_vlines = (uint8_t)stream_pstate_meta->scheduling_delay_otg_vlines; + base_programming->contention_delay_otg_vlines = (uint8_t)stream_pstate_meta->contention_delay_otg_vlines; + base_programming->vline_int_ack_delay_otg_vlines = (uint8_t)stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines; + base_programming->drr_keepout_otg_vline = (uint16_t)(stream_pstate_meta->nom_vtotal - + stream_descriptor->timing.v_front_porch - + stream_pstate_meta->method_drr.programming_delay_otg_vlines); + base_programming->allow_to_target_delay_otg_vlines = (uint8_t)stream_pstate_meta->allow_to_target_delay_otg_vlines; + base_programming->max_vtotal = (uint16_t)stream_pstate_meta->max_vtotal; + + /* from core */ + base_programming->config.bits.min_ttu_vblank_usable = true; + for (i = 0; i < display_cfg->display_config.num_planes; i++) { + /* check if all planes support p-state in blank */ + if (display_cfg->display_config.plane_descriptors[i].stream_index == plane_descriptor->stream_index && + mode_lib->mp.MinTTUVBlank[i] <= mode_lib->mp.Watermark.DRAMClockChangeWatermark) { + base_programming->config.bits.min_ttu_vblank_usable = false; + break; + } + } + + switch (pstate_method) { + case dml2_pstate_method_vactive: + case dml2_pstate_method_fw_vactive_drr: + /* legacy vactive */ + base_programming->type = FAMS2_STREAM_TYPE_VACTIVE; + sub_programming->legacy.vactive_det_fill_delay_otg_vlines = + (uint8_t)stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; + base_programming->allow_start_otg_vline = + (uint16_t)stream_pstate_meta->method_vactive.common.allow_start_otg_vline; + base_programming->allow_end_otg_vline = + (uint16_t)stream_pstate_meta->method_vactive.common.allow_end_otg_vline; + base_programming->config.bits.clamp_vtotal_min = true; + break; + case dml2_pstate_method_vblank: + case dml2_pstate_method_fw_vblank_drr: + /* legacy vblank */ + base_programming->type = FAMS2_STREAM_TYPE_VBLANK; + base_programming->allow_start_otg_vline = + (uint16_t)stream_pstate_meta->method_vblank.common.allow_start_otg_vline; + base_programming->allow_end_otg_vline = + (uint16_t)stream_pstate_meta->method_vblank.common.allow_end_otg_vline; + base_programming->config.bits.clamp_vtotal_min = true; + break; + case dml2_pstate_method_fw_drr: + /* drr */ + base_programming->type = FAMS2_STREAM_TYPE_DRR; + sub_programming->drr.programming_delay_otg_vlines = + (uint8_t)stream_pstate_meta->method_drr.programming_delay_otg_vlines; + sub_programming->drr.nom_stretched_vtotal = + (uint16_t)stream_pstate_meta->method_drr.stretched_vtotal; + base_programming->allow_start_otg_vline = + (uint16_t)stream_pstate_meta->method_drr.common.allow_start_otg_vline; + base_programming->allow_end_otg_vline = + (uint16_t)stream_pstate_meta->method_drr.common.allow_end_otg_vline; + /* drr only clamps to vtotal min for single display */ + base_programming->config.bits.clamp_vtotal_min = display_cfg->display_config.num_streams == 1; + sub_programming->drr.only_stretch_if_required = true; + break; + case dml2_pstate_method_fw_svp: + case dml2_pstate_method_fw_svp_drr: + /* subvp */ + base_programming->type = FAMS2_STREAM_TYPE_SUBVP; + sub_programming->subvp.vratio_numerator = + (uint16_t)(plane_descriptor->composition.scaler_info.plane0.v_ratio * 1000.0); + sub_programming->subvp.vratio_denominator = 1000; + sub_programming->subvp.programming_delay_otg_vlines = + (uint8_t)stream_pstate_meta->method_subvp.programming_delay_otg_vlines; + sub_programming->subvp.prefetch_to_mall_otg_vlines = + (uint8_t)stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines; + sub_programming->subvp.phantom_vtotal = + (uint16_t)stream_pstate_meta->method_subvp.phantom_vtotal; + sub_programming->subvp.phantom_vactive = + (uint16_t)stream_pstate_meta->method_subvp.phantom_vactive; + sub_programming->subvp.config.bits.is_multi_planar = + plane_descriptor->surface.plane1.height > 0; + sub_programming->subvp.config.bits.is_yuv420 = + plane_descriptor->pixel_format == dml2_420_8 || + plane_descriptor->pixel_format == dml2_420_10 || + plane_descriptor->pixel_format == dml2_420_12; + + base_programming->allow_start_otg_vline = + (uint16_t)stream_pstate_meta->method_subvp.common.allow_start_otg_vline; + base_programming->allow_end_otg_vline = + (uint16_t)stream_pstate_meta->method_subvp.common.allow_end_otg_vline; + base_programming->config.bits.clamp_vtotal_min = true; + break; + case dml2_pstate_method_reserved_hw: + case dml2_pstate_method_reserved_fw: + case dml2_pstate_method_reserved_fw_drr_clamped: + case dml2_pstate_method_reserved_fw_drr_var: + case dml2_pstate_method_na: + case dml2_pstate_method_count: + default: + /* this should never happen */ + break; + } +} + +void dml2_core_calcs_get_mcache_allocation(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_mcache_surface_allocation *out, int plane_idx) +{ + unsigned int n; + + out->num_mcaches_plane0 = dml_get_plane_num_mcaches_plane0(mode_lib, plane_idx); + out->num_mcaches_plane1 = dml_get_plane_num_mcaches_plane1(mode_lib, plane_idx); + out->shift_granularity.p0 = dml_get_plane_mcache_shift_granularity_plane0(mode_lib, plane_idx); + out->shift_granularity.p1 = dml_get_plane_mcache_shift_granularity_plane1(mode_lib, plane_idx); + + for (n = 0; n < out->num_mcaches_plane0; n++) + out->mcache_x_offsets_plane0[n] = dml_get_plane_array_mcache_offsets_plane0(mode_lib, plane_idx, n); + + for (n = 0; n < out->num_mcaches_plane1; n++) + out->mcache_x_offsets_plane1[n] = dml_get_plane_array_mcache_offsets_plane1(mode_lib, plane_idx, n); + + out->last_slice_sharing.mall_comb_mcache_p0 = dml_get_plane_mall_comb_mcache_l(mode_lib, plane_idx); + out->last_slice_sharing.mall_comb_mcache_p1 = dml_get_plane_mall_comb_mcache_c(mode_lib, plane_idx); + out->last_slice_sharing.plane0_plane1 = dml_get_plane_lc_comb_mcache(mode_lib, plane_idx); + out->informative.meta_row_bytes_plane0 = dml_get_plane_mcache_row_bytes_plane0(mode_lib, plane_idx); + out->informative.meta_row_bytes_plane1 = dml_get_plane_mcache_row_bytes_plane1(mode_lib, plane_idx); + + out->valid = true; +} + +void dml2_core_calcs_get_mall_allocation(struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int *out, int pipe_index) +{ + *out = dml_get_surface_size_in_mall_bytes(mode_lib, pipe_index); +} + +void dml2_core_calcs_get_plane_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_plane_support_info *out, int plane_idx) +{ + out->mall_svp_size_requirement_ways = 0; + + out->nominal_vblank_pstate_latency_hiding_us = + (int)(display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_idx].stream_index].timing.h_total / + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_idx].stream_index].timing.pixel_clock_khz / 1000) * mode_lib->ms.TWait[plane_idx]); + + out->dram_change_latency_hiding_margin_in_active = (int)mode_lib->ms.VActiveLatencyHidingMargin[plane_idx]; + + out->active_latency_hiding_us = (int)mode_lib->ms.VActiveLatencyHidingUs[plane_idx]; + + out->dram_change_vactive_det_fill_delay_us = (unsigned int)math_ceil(mode_lib->ms.dram_change_vactive_det_fill_delay_us[plane_idx]); +} + +void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_stream_support_info *out, int plane_index) +{ + double phantom_processing_delay_pix; + unsigned int phantom_processing_delay_lines; + unsigned int phantom_min_v_active_lines; + unsigned int phantom_v_active_lines; + unsigned int phantom_v_startup_lines; + unsigned int phantom_v_blank_lines; + unsigned int main_v_blank_lines; + unsigned int rem; + + phantom_processing_delay_pix = (double)((mode_lib->ip.subvp_fw_processing_delay_us + mode_lib->ip.subvp_pstate_allow_width_us) * + ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.pixel_clock_khz / 1000)); + phantom_processing_delay_lines = (unsigned int)(phantom_processing_delay_pix / (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.h_total); + dml2_core_div_rem(phantom_processing_delay_pix, + display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.h_total, + &rem); + if (rem) + phantom_processing_delay_lines++; + + phantom_v_startup_lines = dml_get_plane_max_vstartup_lines(mode_lib, plane_index); + phantom_min_v_active_lines = (unsigned int)math_ceil((double)dml_get_plane_subviewport_lines_needed_in_mall(mode_lib, plane_index) / + display_cfg->plane_descriptors[plane_index].composition.scaler_info.plane0.v_ratio); + phantom_v_active_lines = phantom_processing_delay_lines + phantom_min_v_active_lines + mode_lib->ip.subvp_swath_height_margin_lines; + + // phantom_vblank = max(vbp(vstartup) + vactive + vfp(always 1) + vsync(can be 1), main_vblank) + phantom_v_blank_lines = phantom_v_startup_lines + 1 + 1; + main_v_blank_lines = display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.v_total - display_cfg->stream_descriptors[display_cfg->plane_descriptors[plane_index].stream_index].timing.v_active; + if (phantom_v_blank_lines > main_v_blank_lines) + phantom_v_blank_lines = main_v_blank_lines; + + out->phantom_v_active = phantom_v_active_lines; + // phantom_vtotal = vactive + vblank + out->phantom_v_total = phantom_v_active_lines + phantom_v_blank_lines; + + out->phantom_min_v_active = phantom_min_v_active_lines; + out->phantom_v_startup = phantom_v_startup_lines; + + out->vblank_reserved_time_us = display_cfg->plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000; +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: subvp_fw_processing_delay_us = %d\n", __func__, mode_lib->ip.subvp_fw_processing_delay_us); + DML_LOG_VERBOSE("DML::%s: subvp_pstate_allow_width_us = %d\n", __func__, mode_lib->ip.subvp_pstate_allow_width_us); + DML_LOG_VERBOSE("DML::%s: subvp_swath_height_margin_lines = %d\n", __func__, mode_lib->ip.subvp_swath_height_margin_lines); + DML_LOG_VERBOSE("DML::%s: vblank_reserved_time_us = %u\n", __func__, out->vblank_reserved_time_us); +#endif +} + +void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_cfg_programming *out) +{ + unsigned int k, n; + + out->informative.mode_support_info.ModeIsSupported = mode_lib->ms.support.ModeSupport; + out->informative.mode_support_info.ImmediateFlipSupport = mode_lib->ms.support.ImmediateFlipSupport; + out->informative.mode_support_info.WritebackLatencySupport = mode_lib->ms.support.WritebackLatencySupport; + out->informative.mode_support_info.ScaleRatioAndTapsSupport = mode_lib->ms.support.ScaleRatioAndTapsSupport; + out->informative.mode_support_info.SourceFormatPixelAndScanSupport = mode_lib->ms.support.SourceFormatPixelAndScanSupport; + out->informative.mode_support_info.P2IWith420 = mode_lib->ms.support.P2IWith420; + out->informative.mode_support_info.DSCOnlyIfNecessaryWithBPP = false; + out->informative.mode_support_info.DSC422NativeNotSupported = mode_lib->ms.support.DSC422NativeNotSupported; + out->informative.mode_support_info.LinkRateDoesNotMatchDPVersion = mode_lib->ms.support.LinkRateDoesNotMatchDPVersion; + out->informative.mode_support_info.LinkRateForMultistreamNotIndicated = mode_lib->ms.support.LinkRateForMultistreamNotIndicated; + out->informative.mode_support_info.BPPForMultistreamNotIndicated = mode_lib->ms.support.BPPForMultistreamNotIndicated; + out->informative.mode_support_info.MultistreamWithHDMIOreDP = mode_lib->ms.support.MultistreamWithHDMIOreDP; + out->informative.mode_support_info.MSOOrODMSplitWithNonDPLink = mode_lib->ms.support.MSOOrODMSplitWithNonDPLink; + out->informative.mode_support_info.NotEnoughLanesForMSO = mode_lib->ms.support.NotEnoughLanesForMSO; + out->informative.mode_support_info.NumberOfOTGSupport = mode_lib->ms.support.NumberOfOTGSupport; + out->informative.mode_support_info.NumberOfHDMIFRLSupport = mode_lib->ms.support.NumberOfHDMIFRLSupport; + out->informative.mode_support_info.NumberOfDP2p0Support = mode_lib->ms.support.NumberOfDP2p0Support; + out->informative.mode_support_info.WritebackScaleRatioAndTapsSupport = mode_lib->ms.support.WritebackScaleRatioAndTapsSupport; + out->informative.mode_support_info.CursorSupport = mode_lib->ms.support.CursorSupport; + out->informative.mode_support_info.PitchSupport = mode_lib->ms.support.PitchSupport; + out->informative.mode_support_info.ViewportExceedsSurface = mode_lib->ms.support.ViewportExceedsSurface; + out->informative.mode_support_info.ImmediateFlipRequiredButTheRequirementForEachSurfaceIsNotSpecified = false; + out->informative.mode_support_info.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = mode_lib->ms.support.ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe; + out->informative.mode_support_info.InvalidCombinationOfMALLUseForPStateAndStaticScreen = mode_lib->ms.support.InvalidCombinationOfMALLUseForPStateAndStaticScreen; + out->informative.mode_support_info.InvalidCombinationOfMALLUseForPState = mode_lib->ms.support.InvalidCombinationOfMALLUseForPState; + out->informative.mode_support_info.ExceededMALLSize = mode_lib->ms.support.ExceededMALLSize; + out->informative.mode_support_info.EnoughWritebackUnits = mode_lib->ms.support.EnoughWritebackUnits; + out->informative.mode_support_info.temp_read_or_ppt_support = mode_lib->ms.support.temp_read_or_ppt_support; + out->informative.mode_support_info.g6_temp_read_support = mode_lib->ms.support.g6_temp_read_support; + + out->informative.mode_support_info.ExceededMultistreamSlots = mode_lib->ms.support.ExceededMultistreamSlots; + out->informative.mode_support_info.NotEnoughDSCUnits = mode_lib->ms.support.NotEnoughDSCUnits; + out->informative.mode_support_info.NotEnoughDSCSlices = mode_lib->ms.support.NotEnoughDSCSlices; + out->informative.mode_support_info.PixelsPerLinePerDSCUnitSupport = mode_lib->ms.support.PixelsPerLinePerDSCUnitSupport; + out->informative.mode_support_info.DSCCLKRequiredMoreThanSupported = mode_lib->ms.support.DSCCLKRequiredMoreThanSupported; + out->informative.mode_support_info.DTBCLKRequiredMoreThanSupported = mode_lib->ms.support.DTBCLKRequiredMoreThanSupported; + out->informative.mode_support_info.LinkCapacitySupport = mode_lib->ms.support.LinkCapacitySupport; + + out->informative.mode_support_info.ROBSupport = mode_lib->ms.support.ROBSupport; + out->informative.mode_support_info.OutstandingRequestsSupport = mode_lib->ms.support.OutstandingRequestsSupport; + out->informative.mode_support_info.OutstandingRequestsUrgencyAvoidance = mode_lib->ms.support.OutstandingRequestsUrgencyAvoidance; + out->informative.mode_support_info.PTEBufferSizeNotExceeded = mode_lib->ms.support.PTEBufferSizeNotExceeded; + out->informative.mode_support_info.DCCMetaBufferSizeNotExceeded = mode_lib->ms.support.DCCMetaBufferSizeNotExceeded; + + out->informative.mode_support_info.TotalVerticalActiveBandwidthSupport = mode_lib->ms.support.AvgBandwidthSupport; + out->informative.mode_support_info.VActiveBandwidthSupport = mode_lib->ms.support.UrgVactiveBandwidthSupport; + out->informative.mode_support_info.USRRetrainingSupport = mode_lib->ms.support.USRRetrainingSupport; + + out->informative.mode_support_info.PrefetchSupported = mode_lib->ms.support.PrefetchSupported; + out->informative.mode_support_info.DynamicMetadataSupported = mode_lib->ms.support.DynamicMetadataSupported; + out->informative.mode_support_info.VRatioInPrefetchSupported = mode_lib->ms.support.VRatioInPrefetchSupported; + out->informative.mode_support_info.DISPCLK_DPPCLK_Support = mode_lib->ms.support.DISPCLK_DPPCLK_Support; + out->informative.mode_support_info.TotalAvailablePipesSupport = mode_lib->ms.support.TotalAvailablePipesSupport; + out->informative.mode_support_info.NumberOfTDLUT33cubeSupport = mode_lib->ms.support.NumberOfTDLUT33cubeSupport; + out->informative.mode_support_info.ViewportSizeSupport = mode_lib->ms.support.ViewportSizeSupport; + out->informative.mode_support_info.qos_bandwidth_support = mode_lib->ms.support.qos_bandwidth_support; + out->informative.mode_support_info.dcfclk_support = mode_lib->ms.support.dcfclk_support; + + for (k = 0; k < out->display_config.num_planes; k++) { + + out->informative.mode_support_info.FCLKChangeSupport[k] = mode_lib->ms.support.FCLKChangeSupport[k]; + out->informative.mode_support_info.MPCCombineEnable[k] = mode_lib->ms.support.MPCCombineEnable[k]; + out->informative.mode_support_info.ODMMode[k] = mode_lib->ms.support.ODMMode[k]; + out->informative.mode_support_info.DPPPerSurface[k] = mode_lib->ms.support.DPPPerSurface[k]; + out->informative.mode_support_info.DSCEnabled[k] = mode_lib->ms.support.DSCEnabled[k]; + out->informative.mode_support_info.FECEnabled[k] = mode_lib->ms.support.FECEnabled[k]; + out->informative.mode_support_info.NumberOfDSCSlices[k] = mode_lib->ms.support.NumberOfDSCSlices[k]; + out->informative.mode_support_info.OutputBpp[k] = mode_lib->ms.support.OutputBpp[k]; + + if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_unknown) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_unknown; + else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_dp) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_dp; + else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_edp) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_edp; + else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_dp2p0) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_dp2p0; + else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_hdmi) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_hdmi; + else if (mode_lib->ms.support.OutputType[k] == dml2_core_internal_output_type_hdmifrl) + out->informative.mode_support_info.OutputType[k] = dml2_output_type_hdmifrl; + + if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_unknown) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_unknown; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr2) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr2; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_hbr3) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_hbr3; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr10) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr10; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr13p5) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr13p5; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_dp_rate_uhbr20) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_dp_rate_uhbr20; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_3x3) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_3x3; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_6x3) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_6x3; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_6x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_6x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_8x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_8x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_10x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_10x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_12x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_12x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_16x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_16x4; + else if (mode_lib->ms.support.OutputRate[k] == dml2_core_internal_output_rate_hdmi_rate_20x4) + out->informative.mode_support_info.OutputRate[k] = dml2_output_rate_hdmi_rate_20x4; + + out->informative.mode_support_info.AlignedYPitch[k] = mode_lib->ms.support.AlignedYPitch[k]; + out->informative.mode_support_info.AlignedCPitch[k] = mode_lib->ms.support.AlignedCPitch[k]; + } + + out->informative.watermarks.urgent_us = dml_get_wm_urgent(mode_lib); + out->informative.watermarks.writeback_urgent_us = dml_get_wm_writeback_urgent(mode_lib); + out->informative.watermarks.writeback_pstate_us = dml_get_wm_writeback_dram_clock_change(mode_lib); + out->informative.watermarks.writeback_fclk_pstate_us = dml_get_wm_writeback_fclk_change(mode_lib); + + out->informative.watermarks.cstate_exit_us = dml_get_wm_stutter_exit(mode_lib); + out->informative.watermarks.cstate_enter_plus_exit_us = dml_get_wm_stutter_enter_exit(mode_lib); + out->informative.watermarks.z8_cstate_exit_us = dml_get_wm_z8_stutter_exit(mode_lib); + out->informative.watermarks.z8_cstate_enter_plus_exit_us = dml_get_wm_z8_stutter_enter_exit(mode_lib); + out->informative.watermarks.pstate_change_us = dml_get_wm_dram_clock_change(mode_lib); + out->informative.watermarks.fclk_pstate_change_us = dml_get_wm_fclk_change(mode_lib); + out->informative.watermarks.usr_retraining_us = dml_get_wm_usr_retraining(mode_lib); + out->informative.watermarks.temp_read_or_ppt_watermark_us = dml_get_wm_temp_read_or_ppt(mode_lib); + + out->informative.mall.total_surface_size_in_mall_bytes = 0; + out->informative.dpp.total_num_dpps_required = 0; + for (k = 0; k < out->display_config.num_planes; ++k) { + out->informative.mall.total_surface_size_in_mall_bytes += mode_lib->mp.SurfaceSizeInTheMALL[k]; + out->informative.dpp.total_num_dpps_required += mode_lib->mp.NoOfDPP[k]; + } + + out->informative.qos.min_return_latency_in_dcfclk = mode_lib->mp.min_return_latency_in_dcfclk; + out->informative.qos.urgent_latency_us = dml_get_urgent_latency(mode_lib); + + out->informative.qos.max_urgent_latency_us = dml_get_max_urgent_latency_us(mode_lib); + out->informative.qos.avg_non_urgent_latency_us = dml_get_avg_non_urgent_latency_us(mode_lib); + out->informative.qos.avg_urgent_latency_us = dml_get_avg_urgent_latency_us(mode_lib); + + out->informative.qos.wm_memory_trip_us = dml_get_wm_memory_trip(mode_lib); + out->informative.qos.meta_trip_memory_us = dml_get_meta_trip_memory_us(mode_lib); + out->informative.qos.fraction_of_urgent_bandwidth = dml_get_fraction_of_urgent_bandwidth(mode_lib); + out->informative.qos.fraction_of_urgent_bandwidth_immediate_flip = dml_get_fraction_of_urgent_bandwidth_imm_flip(mode_lib); + out->informative.qos.fraction_of_urgent_bandwidth_mall = dml_get_fraction_of_urgent_bandwidth_mall(mode_lib); + + out->informative.qos.avg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_avg_bw_required_sdp(mode_lib); + out->informative.qos.avg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_avg_bw_required_dram(mode_lib); + out->informative.qos.avg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_avg_bw_required_sdp(mode_lib); + out->informative.qos.avg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_avg_bw_required_dram(mode_lib); + + out->informative.qos.avg_bw_available.sys_active.sdp_bw_mbps = dml_get_sys_active_avg_bw_available_sdp(mode_lib); + out->informative.qos.avg_bw_available.sys_active.dram_bw_mbps = dml_get_sys_active_avg_bw_available_dram(mode_lib); + out->informative.qos.avg_bw_available.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_avg_bw_available_sdp(mode_lib); + out->informative.qos.avg_bw_available.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_avg_bw_available_dram(mode_lib); + + out->informative.qos.urg_bw_available.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_available_sdp(mode_lib); + out->informative.qos.urg_bw_available.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_available_dram(mode_lib); + out->informative.qos.urg_bw_available.sys_active.dram_vm_only_bw_mbps = dml_get_sys_active_urg_bw_available_dram_vm_only(mode_lib); + + out->informative.qos.urg_bw_available.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_available_sdp(mode_lib); + out->informative.qos.urg_bw_available.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_available_dram(mode_lib); + out->informative.qos.urg_bw_available.svp_prefetch.dram_vm_only_bw_mbps = dml_get_svp_prefetch_urg_bw_available_dram_vm_only(mode_lib); + + out->informative.qos.urg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_required_sdp(mode_lib); + out->informative.qos.urg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_required_dram(mode_lib); + out->informative.qos.urg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_required_sdp(mode_lib); + out->informative.qos.urg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_required_dram(mode_lib); + + out->informative.qos.non_urg_bw_required.sys_active.sdp_bw_mbps = dml_get_sys_active_non_urg_required_sdp(mode_lib); + out->informative.qos.non_urg_bw_required.sys_active.dram_bw_mbps = dml_get_sys_active_non_urg_required_dram(mode_lib); + out->informative.qos.non_urg_bw_required.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_sdp(mode_lib); + out->informative.qos.non_urg_bw_required.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_dram(mode_lib); + + out->informative.qos.urg_bw_required_with_flip.sys_active.sdp_bw_mbps = dml_get_sys_active_urg_bw_required_sdp_flip(mode_lib); + out->informative.qos.urg_bw_required_with_flip.sys_active.dram_bw_mbps = dml_get_sys_active_urg_bw_required_dram_flip(mode_lib); + out->informative.qos.urg_bw_required_with_flip.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_urg_bw_required_sdp_flip(mode_lib); + out->informative.qos.urg_bw_required_with_flip.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_urg_bw_required_dram_flip(mode_lib); + + out->informative.qos.non_urg_bw_required_with_flip.sys_active.sdp_bw_mbps = dml_get_sys_active_non_urg_required_sdp_flip(mode_lib); + out->informative.qos.non_urg_bw_required_with_flip.sys_active.dram_bw_mbps = dml_get_sys_active_non_urg_required_dram_flip(mode_lib); + out->informative.qos.non_urg_bw_required_with_flip.svp_prefetch.sdp_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_sdp_flip(mode_lib); + out->informative.qos.non_urg_bw_required_with_flip.svp_prefetch.dram_bw_mbps = dml_get_svp_prefetch_non_urg_bw_required_dram_flip(mode_lib); + + out->informative.crb.comp_buffer_size_kbytes = dml_get_comp_buffer_size_kbytes(mode_lib); + out->informative.crb.UnboundedRequestEnabled = dml_get_unbounded_request_enabled(mode_lib); + + out->informative.crb.compbuf_reserved_space_64b = dml_get_compbuf_reserved_space_64b(mode_lib); + out->informative.misc.hw_debug5 = dml_get_hw_debug5(mode_lib); + out->informative.misc.dcfclk_deep_sleep_hysteresis = dml_get_dcfclk_deep_sleep_hysteresis(mode_lib); + + out->informative.power_management.stutter_efficiency = dml_get_stutter_efficiency_no_vblank(mode_lib); + out->informative.power_management.stutter_efficiency_with_vblank = dml_get_stutter_efficiency(mode_lib); + out->informative.power_management.stutter_num_bursts = dml_get_stutter_num_bursts(mode_lib); + + out->informative.power_management.z8.stutter_efficiency = dml_get_stutter_efficiency_z8(mode_lib); + out->informative.power_management.z8.stutter_efficiency_with_vblank = dml_get_stutter_efficiency(mode_lib); + out->informative.power_management.z8.stutter_num_bursts = dml_get_stutter_num_bursts_z8(mode_lib); + out->informative.power_management.z8.stutter_period = dml_get_stutter_period(mode_lib); + + out->informative.power_management.z8.bestcase.stutter_efficiency = dml_get_stutter_efficiency_z8_bestcase(mode_lib); + out->informative.power_management.z8.bestcase.stutter_num_bursts = dml_get_stutter_num_bursts_z8_bestcase(mode_lib); + out->informative.power_management.z8.bestcase.stutter_period = dml_get_stutter_period_bestcase(mode_lib); + + out->informative.misc.cstate_max_cap_mode = dml_get_cstate_max_cap_mode(mode_lib); + + out->min_clocks.dcn4x.dpprefclk_khz = (int unsigned)dml_get_global_dppclk_khz(mode_lib); + + out->informative.qos.max_active_fclk_change_latency_supported = dml_get_fclk_change_latency(mode_lib); + + out->informative.misc.LowestPrefetchMargin = 10 * 1000 * 1000; + + for (k = 0; k < out->display_config.num_planes; k++) { + + if ((out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us) + && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.fclk_change_blackout_us) + && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us)) + out->informative.misc.PrefetchMode[k] = 0; + else if ((out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.fclk_change_blackout_us) + && (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us)) + out->informative.misc.PrefetchMode[k] = 1; + else if (out->display_config.plane_descriptors->overrides.reserved_vblank_time_ns >= 1000.0 * mode_lib->soc.power_management_parameters.stutter_enter_plus_exit_latency_us) + out->informative.misc.PrefetchMode[k] = 2; + else + out->informative.misc.PrefetchMode[k] = 3; + + out->informative.misc.min_ttu_vblank_us[k] = mode_lib->mp.MinTTUVBlank[k]; + out->informative.mall.subviewport_lines_needed_in_mall[k] = mode_lib->mp.SubViewportLinesNeededInMALL[k]; + out->informative.crb.det_size_in_kbytes[k] = mode_lib->mp.DETBufferSizeInKByte[k]; + out->informative.crb.DETBufferSizeY[k] = mode_lib->mp.DETBufferSizeY[k]; + out->informative.misc.ImmediateFlipSupportedForPipe[k] = mode_lib->mp.ImmediateFlipSupportedForPipe[k]; + out->informative.misc.UsesMALLForStaticScreen[k] = mode_lib->mp.is_using_mall_for_ss[k]; + out->informative.plane_info[k].dpte_row_height_plane0 = mode_lib->mp.dpte_row_height[k]; + out->informative.plane_info[k].dpte_row_height_plane1 = mode_lib->mp.dpte_row_height_chroma[k]; + out->informative.plane_info[k].meta_row_height_plane0 = mode_lib->mp.meta_row_height[k]; + out->informative.plane_info[k].meta_row_height_plane1 = mode_lib->mp.meta_row_height_chroma[k]; + out->informative.dcc_control[k].max_uncompressed_block_plane0 = mode_lib->mp.DCCYMaxUncompressedBlock[k]; + out->informative.dcc_control[k].max_compressed_block_plane0 = mode_lib->mp.DCCYMaxCompressedBlock[k]; + out->informative.dcc_control[k].independent_block_plane0 = mode_lib->mp.DCCYIndependentBlock[k]; + out->informative.dcc_control[k].max_uncompressed_block_plane1 = mode_lib->mp.DCCCMaxUncompressedBlock[k]; + out->informative.dcc_control[k].max_compressed_block_plane1 = mode_lib->mp.DCCCMaxCompressedBlock[k]; + out->informative.dcc_control[k].independent_block_plane1 = mode_lib->mp.DCCCIndependentBlock[k]; + out->informative.misc.dst_x_after_scaler[k] = mode_lib->mp.DSTXAfterScaler[k]; + out->informative.misc.dst_y_after_scaler[k] = mode_lib->mp.DSTYAfterScaler[k]; + out->informative.misc.prefetch_source_lines_plane0[k] = mode_lib->mp.PrefetchSourceLinesY[k]; + out->informative.misc.prefetch_source_lines_plane1[k] = mode_lib->mp.PrefetchSourceLinesC[k]; + out->informative.misc.vready_at_or_after_vsync[k] = mode_lib->mp.VREADY_AT_OR_AFTER_VSYNC[k]; + out->informative.misc.min_dst_y_next_start[k] = mode_lib->mp.MIN_DST_Y_NEXT_START[k]; + out->informative.plane_info[k].swath_width_plane0 = mode_lib->mp.SwathWidthY[k]; + out->informative.plane_info[k].swath_height_plane0 = mode_lib->mp.SwathHeightY[k]; + out->informative.plane_info[k].swath_height_plane1 = mode_lib->mp.SwathHeightC[k]; + out->informative.misc.CursorDstXOffset[k] = mode_lib->mp.CursorDstXOffset[k]; + out->informative.misc.CursorDstYOffset[k] = mode_lib->mp.CursorDstYOffset[k]; + out->informative.misc.CursorChunkHDLAdjust[k] = mode_lib->mp.CursorChunkHDLAdjust[k]; + out->informative.misc.dpte_group_bytes[k] = mode_lib->mp.dpte_group_bytes[k]; + out->informative.misc.vm_group_bytes[k] = mode_lib->mp.vm_group_bytes[k]; + out->informative.misc.DisplayPipeRequestDeliveryTimeLuma[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeLuma[k]; + out->informative.misc.DisplayPipeRequestDeliveryTimeChroma[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeChroma[k]; + out->informative.misc.DisplayPipeRequestDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeLumaPrefetch[k]; + out->informative.misc.DisplayPipeRequestDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeRequestDeliveryTimeChromaPrefetch[k]; + out->informative.misc.TimePerVMGroupVBlank[k] = mode_lib->mp.TimePerVMGroupVBlank[k]; + out->informative.misc.TimePerVMGroupFlip[k] = mode_lib->mp.TimePerVMGroupFlip[k]; + out->informative.misc.TimePerVMRequestVBlank[k] = mode_lib->mp.TimePerVMRequestVBlank[k]; + out->informative.misc.TimePerVMRequestFlip[k] = mode_lib->mp.TimePerVMRequestFlip[k]; + out->informative.misc.Tdmdl_vm[k] = mode_lib->mp.Tdmdl_vm[k]; + out->informative.misc.Tdmdl[k] = mode_lib->mp.Tdmdl[k]; + out->informative.misc.VStartup[k] = mode_lib->mp.VStartup[k]; + out->informative.misc.VUpdateOffsetPix[k] = mode_lib->mp.VUpdateOffsetPix[k]; + out->informative.misc.VUpdateWidthPix[k] = mode_lib->mp.VUpdateWidthPix[k]; + out->informative.misc.VReadyOffsetPix[k] = mode_lib->mp.VReadyOffsetPix[k]; + + out->informative.misc.DST_Y_PER_PTE_ROW_NOM_L[k] = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_L[k]; + out->informative.misc.DST_Y_PER_PTE_ROW_NOM_C[k] = mode_lib->mp.DST_Y_PER_PTE_ROW_NOM_C[k]; + out->informative.misc.time_per_pte_group_nom_luma[k] = mode_lib->mp.time_per_pte_group_nom_luma[k]; + out->informative.misc.time_per_pte_group_nom_chroma[k] = mode_lib->mp.time_per_pte_group_nom_chroma[k]; + out->informative.misc.time_per_pte_group_vblank_luma[k] = mode_lib->mp.time_per_pte_group_vblank_luma[k]; + out->informative.misc.time_per_pte_group_vblank_chroma[k] = mode_lib->mp.time_per_pte_group_vblank_chroma[k]; + out->informative.misc.time_per_pte_group_flip_luma[k] = mode_lib->mp.time_per_pte_group_flip_luma[k]; + out->informative.misc.time_per_pte_group_flip_chroma[k] = mode_lib->mp.time_per_pte_group_flip_chroma[k]; + out->informative.misc.VRatioPrefetchY[k] = mode_lib->mp.VRatioPrefetchY[k]; + out->informative.misc.VRatioPrefetchC[k] = mode_lib->mp.VRatioPrefetchC[k]; + out->informative.misc.DestinationLinesForPrefetch[k] = mode_lib->mp.dst_y_prefetch[k]; + out->informative.misc.DestinationLinesToRequestVMInVBlank[k] = mode_lib->mp.dst_y_per_vm_vblank[k]; + out->informative.misc.DestinationLinesToRequestRowInVBlank[k] = mode_lib->mp.dst_y_per_row_vblank[k]; + out->informative.misc.DestinationLinesToRequestVMInImmediateFlip[k] = mode_lib->mp.dst_y_per_vm_flip[k]; + out->informative.misc.DestinationLinesToRequestRowInImmediateFlip[k] = mode_lib->mp.dst_y_per_row_flip[k]; + out->informative.misc.DisplayPipeLineDeliveryTimeLuma[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLuma[k]; + out->informative.misc.DisplayPipeLineDeliveryTimeChroma[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChroma[k]; + out->informative.misc.DisplayPipeLineDeliveryTimeLumaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeLumaPrefetch[k]; + out->informative.misc.DisplayPipeLineDeliveryTimeChromaPrefetch[k] = mode_lib->mp.DisplayPipeLineDeliveryTimeChromaPrefetch[k]; + + out->informative.misc.WritebackRequiredBandwidth = mode_lib->mp.TotalWRBandwidth / 1000.0; + out->informative.misc.WritebackAllowDRAMClockChangeEndPosition[k] = mode_lib->mp.WritebackAllowDRAMClockChangeEndPosition[k]; + out->informative.misc.WritebackAllowFCLKChangeEndPosition[k] = mode_lib->mp.WritebackAllowFCLKChangeEndPosition[k]; + out->informative.misc.DSCCLK_calculated[k] = mode_lib->mp.DSCCLK[k]; + out->informative.misc.BIGK_FRAGMENT_SIZE[k] = mode_lib->mp.BIGK_FRAGMENT_SIZE[k]; + out->informative.misc.PTE_BUFFER_MODE[k] = mode_lib->mp.PTE_BUFFER_MODE[k]; + out->informative.misc.DSCDelay[k] = mode_lib->mp.DSCDelay[k]; + out->informative.misc.MaxActiveDRAMClockChangeLatencySupported[k] = mode_lib->mp.MaxActiveDRAMClockChangeLatencySupported[k]; + + if (mode_lib->mp.impacted_prefetch_margin_us[k] < out->informative.misc.LowestPrefetchMargin) + out->informative.misc.LowestPrefetchMargin = mode_lib->mp.impacted_prefetch_margin_us[k]; + } + + // For this DV informative layer, all pipes in the same planes will just use the same id + // will have the optimization and helper layer later on + // only work when we can have high "mcache" that fit everything without thrashing the cache + for (k = 0; k < out->display_config.num_planes; k++) { + out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane0 = dml_get_plane_num_mcaches_plane0(mode_lib, k); + out->informative.non_optimized_mcache_allocation[k].informative.meta_row_bytes_plane0 = dml_get_plane_mcache_row_bytes_plane0(mode_lib, k); + + for (n = 0; n < out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane0; n++) { + out->informative.non_optimized_mcache_allocation[k].mcache_x_offsets_plane0[n] = dml_get_plane_array_mcache_offsets_plane0(mode_lib, k, n); + out->informative.non_optimized_mcache_allocation[k].global_mcache_ids_plane0[n] = k; + } + + out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane1 = dml_get_plane_num_mcaches_plane1(mode_lib, k); + out->informative.non_optimized_mcache_allocation[k].informative.meta_row_bytes_plane1 = dml_get_plane_mcache_row_bytes_plane1(mode_lib, k); + + for (n = 0; n < out->informative.non_optimized_mcache_allocation[k].num_mcaches_plane1; n++) { + out->informative.non_optimized_mcache_allocation[k].mcache_x_offsets_plane1[n] = dml_get_plane_array_mcache_offsets_plane1(mode_lib, k, n); + out->informative.non_optimized_mcache_allocation[k].global_mcache_ids_plane1[n] = k; + } + } + out->informative.qos.max_non_urgent_latency_us = dml_get_max_non_urgent_latency_us(mode_lib); + + if (mode_lib->soc.qos_parameters.qos_type == dml2_qos_param_type_dcn4x) { + if (((mode_lib->ip.rob_buffer_size_kbytes - mode_lib->ip.pixel_chunk_size_kbytes) * 1024 + / mode_lib->ms.support.non_urg_bandwidth_required[dml2_core_internal_soc_state_sys_active][dml2_core_internal_bw_sdp]) >= out->informative.qos.max_non_urgent_latency_us) { + out->informative.misc.ROBUrgencyAvoidance = true; + } else { + out->informative.misc.ROBUrgencyAvoidance = false; + } + } else { + out->informative.misc.ROBUrgencyAvoidance = true; + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h new file mode 100644 index 000000000000..27ef0e096b25 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h @@ -0,0 +1,39 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_CORE_DCN4_CALCS_H__ +#define __DML2_CORE_DCN4_CALCS_H__ + +#include "dml2_core_shared_types.h" + +struct dml2_dchub_watermark_regs; +struct dml2_display_arb_regs; +struct dml2_per_stream_programming; +struct dml2_dchub_per_pipe_register_set; +struct core_plane_support_info; +struct core_stream_support_info; +struct dml2_cursor_dlg_regs; +struct display_configuation_with_meta; + +unsigned int dml2_core_calcs_mode_support_ex(struct dml2_core_calcs_mode_support_ex *in_out_params); +bool dml2_core_calcs_mode_programming_ex(struct dml2_core_calcs_mode_programming_ex *in_out_params); +void dml2_core_calcs_get_watermarks(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_watermark_regs *out); +void dml2_core_calcs_get_arb_params(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_arb_regs *out); +void dml2_core_calcs_get_pipe_regs(const struct dml2_display_cfg *dml2_display_cfg, struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_dchub_per_pipe_register_set *out, int pipe_index); +void dml2_core_calcs_get_stream_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_per_stream_programming *out, int pipe_index); +void dml2_core_calcs_get_global_sync_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, union dml2_global_sync_programming *out, int pipe_index); +void dml2_core_calcs_get_mcache_allocation(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_mcache_surface_allocation *out, int plane_index); +void dml2_core_calcs_get_plane_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_plane_support_info *out, int plane_index); +void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mode_lib *mode_lib, struct dml2_display_cfg_programming *out); +void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_stream_support_info *out, int plane_index); +void dml2_core_calcs_get_mall_allocation(struct dml2_core_internal_display_mode_lib *mode_lib, unsigned int *out, int pipe_index); +void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, const struct display_configuation_with_meta *display_cfg, union dmub_cmd_fams2_config *fams2_base_programming, union dmub_cmd_fams2_config *fams2_sub_programming, enum dml2_pstate_method pstate_method, int plane_index); +void dml2_core_calcs_get_global_fams2_programming(const struct dml2_core_internal_display_mode_lib *mode_lib, const struct display_configuation_with_meta *display_cfg, struct dmub_cmd_fams2_global_config *fams2_global_config); + +void dml2_core_calcs_get_dpte_row_height(unsigned int *dpte_row_height, struct dml2_core_internal_display_mode_lib *mode_lib, bool is_plane1, enum dml2_source_format_class SourcePixelFormat, enum dml2_swizzle_mode SurfaceTiling, enum dml2_rotation_angle ScanDirection, unsigned int pitch, unsigned int GPUVMMinPageSizeKBytes); +void dml2_core_calcs_cursor_dlg_reg(struct dml2_cursor_dlg_regs *cursor_dlg_regs, const struct dml2_get_cursor_dlg_reg *p); +const char *dml2_core_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type); +const char *dml2_core_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c new file mode 100644 index 000000000000..cc4f0663c6d6 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c @@ -0,0 +1,39 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_core_factory.h" +#include "dml2_core_dcn4.h" +#include "dml2_external_lib_deps.h" + +bool dml2_core_create(enum dml2_project_id project_id, struct dml2_core_instance *out) +{ + bool result = false; + + if (!out) + return false; + + memset(out, 0, sizeof(struct dml2_core_instance)); + + out->project_id = project_id; + + switch (project_id) { + case dml2_project_dcn4x_stage1: + result = false; + break; + case dml2_project_dcn4x_stage2: + case dml2_project_dcn4x_stage2_auto_drr_svp: + out->initialize = &core_dcn4_initialize; + out->mode_support = &core_dcn4_mode_support; + out->mode_programming = &core_dcn4_mode_programming; + out->populate_informative = &core_dcn4_populate_informative; + out->calculate_mcache_allocation = &core_dcn4_calculate_mcache_allocation; + result = true; + break; + case dml2_project_invalid: + default: + break; + } + + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h new file mode 100644 index 000000000000..411c514fe65c --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_CORE_FACTORY_H__ +#define __DML2_CORE_FACTORY_H__ + +#include "dml2_internal_shared_types.h" +#include "dml_top_types.h" + +bool dml2_core_create(enum dml2_project_id project_id, struct dml2_core_instance *out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h new file mode 100644 index 000000000000..051c31ec2f0e --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h @@ -0,0 +1,2335 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_CORE_SHARED_TYPES_H__ +#define __DML2_CORE_SHARED_TYPES_H__ + +#include "dml2_external_lib_deps.h" +#include "dml_top_display_cfg_types.h" +#include "dml_top_types.h" + +#define __DML_VBA_DEBUG__ +#define __DML2_CALCS_MAX_VRATIO_PRE_OTO__ 4.0 // 0); + return dividend / divisor; + +} + +const char *dml2_core_utils_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type) +{ + switch (bw_type) { + case (dml2_core_internal_bw_sdp): + return("dml2_core_internal_bw_sdp"); + case (dml2_core_internal_bw_dram): + return("dml2_core_internal_bw_dram"); + case (dml2_core_internal_bw_max): + return("dml2_core_internal_bw_max"); + default: + return("dml2_core_internal_bw_unknown"); + } +} + +bool dml2_core_utils_is_420(enum dml2_source_format_class source_format) +{ + bool val = false; + + switch (source_format) { + case dml2_444_8: + val = 0; + break; + case dml2_444_16: + val = 0; + break; + case dml2_444_32: + val = 0; + break; + case dml2_444_64: + val = 0; + break; + case dml2_420_8: + val = 1; + break; + case dml2_420_10: + val = 1; + break; + case dml2_420_12: + val = 1; + break; + case dml2_rgbe_alpha: + val = 0; + break; + case dml2_rgbe: + val = 0; + break; + case dml2_mono_8: + val = 0; + break; + case dml2_mono_16: + val = 0; + break; + case dml2_422_planar_8: + val = 0; + break; + case dml2_422_planar_10: + val = 0; + break; + case dml2_422_planar_12: + val = 0; + break; + case dml2_422_packed_8: + val = 0; + break; + case dml2_422_packed_10: + val = 0; + break; + case dml2_422_packed_12: + val = 0; + break; + default: + DML_ASSERT(0); + break; + } + return val; +} + +bool dml2_core_utils_is_422_planar(enum dml2_source_format_class source_format) +{ + bool val = false; + + switch (source_format) { + case dml2_444_8: + val = 0; + break; + case dml2_444_16: + val = 0; + break; + case dml2_444_32: + val = 0; + break; + case dml2_444_64: + val = 0; + break; + case dml2_420_8: + val = 0; + break; + case dml2_420_10: + val = 0; + break; + case dml2_420_12: + val = 0; + break; + case dml2_rgbe_alpha: + val = 0; + break; + case dml2_rgbe: + val = 0; + break; + case dml2_mono_8: + val = 0; + break; + case dml2_mono_16: + val = 0; + break; + case dml2_422_planar_8: + val = 1; + break; + case dml2_422_planar_10: + val = 1; + break; + case dml2_422_planar_12: + val = 1; + break; + case dml2_422_packed_8: + val = 0; + break; + case dml2_422_packed_10: + val = 0; + break; + case dml2_422_packed_12: + val = 0; + break; + default: + DML_ASSERT(0); + break; + } + return val; +} + +bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format) +{ + bool val = false; + + switch (source_format) { + case dml2_444_8: + val = 0; + break; + case dml2_444_16: + val = 0; + break; + case dml2_444_32: + val = 0; + break; + case dml2_444_64: + val = 0; + break; + case dml2_420_8: + val = 0; + break; + case dml2_420_10: + val = 0; + break; + case dml2_420_12: + val = 0; + break; + case dml2_rgbe_alpha: + val = 0; + break; + case dml2_rgbe: + val = 0; + break; + case dml2_mono_8: + val = 0; + break; + case dml2_mono_16: + val = 0; + break; + case dml2_422_planar_8: + val = 0; + break; + case dml2_422_planar_10: + val = 0; + break; + case dml2_422_planar_12: + val = 0; + break; + case dml2_422_packed_8: + val = 1; + break; + case dml2_422_packed_10: + val = 1; + break; + case dml2_422_packed_12: + val = 1; + break; + default: + DML_ASSERT(0); + break; + } + return val; +} + +void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only) +{ + DML_LOG_VERBOSE("DML: ===================================== \n"); + DML_LOG_VERBOSE("DML: DML_MODE_SUPPORT_INFO_ST\n"); + if (!fail_only || support->ScaleRatioAndTapsSupport == 0) + DML_LOG_VERBOSE("DML: support: ScaleRatioAndTapsSupport = %d\n", support->ScaleRatioAndTapsSupport); + if (!fail_only || support->SourceFormatPixelAndScanSupport == 0) + DML_LOG_VERBOSE("DML: support: SourceFormatPixelAndScanSupport = %d\n", support->SourceFormatPixelAndScanSupport); + if (!fail_only || support->ViewportSizeSupport == 0) + DML_LOG_VERBOSE("DML: support: ViewportSizeSupport = %d\n", support->ViewportSizeSupport); + if (!fail_only || support->LinkRateDoesNotMatchDPVersion == 1) + DML_LOG_VERBOSE("DML: support: LinkRateDoesNotMatchDPVersion = %d\n", support->LinkRateDoesNotMatchDPVersion); + if (!fail_only || support->LinkRateForMultistreamNotIndicated == 1) + DML_LOG_VERBOSE("DML: support: LinkRateForMultistreamNotIndicated = %d\n", support->LinkRateForMultistreamNotIndicated); + if (!fail_only || support->BPPForMultistreamNotIndicated == 1) + DML_LOG_VERBOSE("DML: support: BPPForMultistreamNotIndicated = %d\n", support->BPPForMultistreamNotIndicated); + if (!fail_only || support->MultistreamWithHDMIOreDP == 1) + DML_LOG_VERBOSE("DML: support: MultistreamWithHDMIOreDP = %d\n", support->MultistreamWithHDMIOreDP); + if (!fail_only || support->ExceededMultistreamSlots == 1) + DML_LOG_VERBOSE("DML: support: ExceededMultistreamSlots = %d\n", support->ExceededMultistreamSlots); + if (!fail_only || support->MSOOrODMSplitWithNonDPLink == 1) + DML_LOG_VERBOSE("DML: support: MSOOrODMSplitWithNonDPLink = %d\n", support->MSOOrODMSplitWithNonDPLink); + if (!fail_only || support->NotEnoughLanesForMSO == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughLanesForMSO = %d\n", support->NotEnoughLanesForMSO); + if (!fail_only || support->P2IWith420 == 1) + DML_LOG_VERBOSE("DML: support: P2IWith420 = %d\n", support->P2IWith420); + if (!fail_only || support->DSC422NativeNotSupported == 1) + DML_LOG_VERBOSE("DML: support: DSC422NativeNotSupported = %d\n", support->DSC422NativeNotSupported); + if (!fail_only || support->DSCSlicesODMModeSupported == 0) + DML_LOG_VERBOSE("DML: support: DSCSlicesODMModeSupported = %d\n", support->DSCSlicesODMModeSupported); + if (!fail_only || support->NotEnoughDSCUnits == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughDSCUnits = %d\n", support->NotEnoughDSCUnits); + if (!fail_only || support->NotEnoughDSCSlices == 1) + DML_LOG_VERBOSE("DML: support: NotEnoughDSCSlices = %d\n", support->NotEnoughDSCSlices); + if (!fail_only || support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe == 1) + DML_LOG_VERBOSE("DML: support: ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe = %d\n", support->ImmediateFlipOrHostVMAndPStateWithMALLFullFrameOrPhantomPipe); + if (!fail_only || support->InvalidCombinationOfMALLUseForPStateAndStaticScreen == 1) + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPStateAndStaticScreen = %d\n", support->InvalidCombinationOfMALLUseForPStateAndStaticScreen); + if (!fail_only || support->DSCCLKRequiredMoreThanSupported == 1) + DML_LOG_VERBOSE("DML: support: DSCCLKRequiredMoreThanSupported = %d\n", support->DSCCLKRequiredMoreThanSupported); + if (!fail_only || support->PixelsPerLinePerDSCUnitSupport == 0) + DML_LOG_VERBOSE("DML: support: PixelsPerLinePerDSCUnitSupport = %d\n", support->PixelsPerLinePerDSCUnitSupport); + if (!fail_only || support->DTBCLKRequiredMoreThanSupported == 1) + DML_LOG_VERBOSE("DML: support: DTBCLKRequiredMoreThanSupported = %d\n", support->DTBCLKRequiredMoreThanSupported); + if (!fail_only || support->InvalidCombinationOfMALLUseForPState == 1) + DML_LOG_VERBOSE("DML: support: InvalidCombinationOfMALLUseForPState = %d\n", support->InvalidCombinationOfMALLUseForPState); + if (!fail_only || support->ROBSupport == 0) + DML_LOG_VERBOSE("DML: support: ROBSupport = %d\n", support->ROBSupport); + if (!fail_only || support->OutstandingRequestsSupport == 0) + DML_LOG_VERBOSE("DML: support: OutstandingRequestsSupport = %d\n", support->OutstandingRequestsSupport); + if (!fail_only || support->OutstandingRequestsUrgencyAvoidance == 0) + DML_LOG_VERBOSE("DML: support: OutstandingRequestsUrgencyAvoidance = %d\n", support->OutstandingRequestsUrgencyAvoidance); + if (!fail_only || support->DISPCLK_DPPCLK_Support == 0) + DML_LOG_VERBOSE("DML: support: DISPCLK_DPPCLK_Support = %d\n", support->DISPCLK_DPPCLK_Support); + if (!fail_only || support->TotalAvailablePipesSupport == 0) + DML_LOG_VERBOSE("DML: support: TotalAvailablePipesSupport = %d\n", support->TotalAvailablePipesSupport); + if (!fail_only || support->NumberOfOTGSupport == 0) + DML_LOG_VERBOSE("DML: support: NumberOfOTGSupport = %d\n", support->NumberOfOTGSupport); + if (!fail_only || support->NumberOfHDMIFRLSupport == 0) + DML_LOG_VERBOSE("DML: support: NumberOfHDMIFRLSupport = %d\n", support->NumberOfHDMIFRLSupport); + if (!fail_only || support->NumberOfDP2p0Support == 0) + DML_LOG_VERBOSE("DML: support: NumberOfDP2p0Support = %d\n", support->NumberOfDP2p0Support); + if (!fail_only || support->EnoughWritebackUnits == 0) + DML_LOG_VERBOSE("DML: support: EnoughWritebackUnits = %d\n", support->EnoughWritebackUnits); + if (!fail_only || support->WritebackScaleRatioAndTapsSupport == 0) + DML_LOG_VERBOSE("DML: support: WritebackScaleRatioAndTapsSupport = %d\n", support->WritebackScaleRatioAndTapsSupport); + if (!fail_only || support->WritebackLatencySupport == 0) + DML_LOG_VERBOSE("DML: support: WritebackLatencySupport = %d\n", support->WritebackLatencySupport); + if (!fail_only || support->CursorSupport == 0) + DML_LOG_VERBOSE("DML: support: CursorSupport = %d\n", support->CursorSupport); + if (!fail_only || support->PitchSupport == 0) + DML_LOG_VERBOSE("DML: support: PitchSupport = %d\n", support->PitchSupport); + if (!fail_only || support->ViewportExceedsSurface == 1) + DML_LOG_VERBOSE("DML: support: ViewportExceedsSurface = %d\n", support->ViewportExceedsSurface); + if (!fail_only || support->PrefetchSupported == 0) + DML_LOG_VERBOSE("DML: support: PrefetchSupported = %d\n", support->PrefetchSupported); + if (!fail_only || support->EnoughUrgentLatencyHidingSupport == 0) + DML_LOG_VERBOSE("DML: support: EnoughUrgentLatencyHidingSupport = %d\n", support->EnoughUrgentLatencyHidingSupport); + if (!fail_only || support->AvgBandwidthSupport == 0) + DML_LOG_VERBOSE("DML: support: AvgBandwidthSupport = %d\n", support->AvgBandwidthSupport); + if (!fail_only || support->DynamicMetadataSupported == 0) + DML_LOG_VERBOSE("DML: support: DynamicMetadataSupported = %d\n", support->DynamicMetadataSupported); + if (!fail_only || support->VRatioInPrefetchSupported == 0) + DML_LOG_VERBOSE("DML: support: VRatioInPrefetchSupported = %d\n", support->VRatioInPrefetchSupported); + if (!fail_only || support->PTEBufferSizeNotExceeded == 0) + DML_LOG_VERBOSE("DML: support: PTEBufferSizeNotExceeded = %d\n", support->PTEBufferSizeNotExceeded); + if (!fail_only || support->DCCMetaBufferSizeNotExceeded == 0) + DML_LOG_VERBOSE("DML: support: DCCMetaBufferSizeNotExceeded = %d\n", support->DCCMetaBufferSizeNotExceeded); + if (!fail_only || support->ExceededMALLSize == 1) + DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); + if (!fail_only || support->g6_temp_read_support == 0) + DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); + if (!fail_only || support->ImmediateFlipSupport == 0) + DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); + if (!fail_only || support->LinkCapacitySupport == 0) + DML_LOG_VERBOSE("DML: support: LinkCapacitySupport = %d\n", support->LinkCapacitySupport); + + if (!fail_only || support->ModeSupport == 0) + DML_LOG_VERBOSE("DML: support: ModeSupport = %d\n", support->ModeSupport); + DML_LOG_VERBOSE("DML: ===================================== \n"); +} + +const char *dml2_core_utils_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type) +{ + switch (dml2_core_internal_soc_state_type) { + case (dml2_core_internal_soc_state_sys_idle): + return("dml2_core_internal_soc_state_sys_idle"); + case (dml2_core_internal_soc_state_sys_active): + return("dml2_core_internal_soc_state_sys_active"); + case (dml2_core_internal_soc_state_svp_prefetch): + return("dml2_core_internal_soc_state_svp_prefetch"); + case dml2_core_internal_soc_state_max: + default: + return("dml2_core_internal_soc_state_unknown"); + } +} + + +void dml2_core_utils_get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg) +{ + for (unsigned int k = 0; k < display_cfg->num_planes; k++) { + double bpc = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.bpc; + if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_disable) { + switch (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].output.output_format) { + case dml2_444: + out_bpp[k] = bpc * 3; + break; + case dml2_s422: + out_bpp[k] = bpc * 2; + break; + case dml2_n422: + out_bpp[k] = bpc * 2; + break; + case dml2_420: + default: + out_bpp[k] = bpc * 1.5; + break; + } + } else if (display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable == dml2_dsc_enable) { + out_bpp[k] = (double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.dsc_compressed_bpp_x16 / 16; + } else { + out_bpp[k] = 0; + } +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: k=%d bpc=%f\n", __func__, k, bpc); + DML_LOG_VERBOSE("DML::%s: k=%d dsc.enable=%d\n", __func__, k, display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.dsc.enable); + DML_LOG_VERBOSE("DML::%s: k=%d out_bpp=%f\n", __func__, k, out_bpp[k]); +#endif + } +} + +unsigned int dml2_core_utils_round_to_multiple(unsigned int num, unsigned int multiple, bool up) +{ + unsigned int remainder; + + if (multiple == 0) + return num; + + remainder = num % multiple; + if (remainder == 0) + return num; + + if (up) + return (num + multiple - remainder); + else + return (num - remainder); +} + +unsigned int dml2_core_util_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info) +{ + unsigned int num_active_pipes = 0; + + for (unsigned int k = 0; k < num_planes; k++) { + num_active_pipes = num_active_pipes + (unsigned int)cfg_support_info->plane_support_info[k].dpps_used; + } + +#ifdef __DML_VBA_DEBUG__ + DML_LOG_VERBOSE("DML::%s: num_active_pipes = %d\n", __func__, num_active_pipes); +#endif + return num_active_pipes; +} + +void dml2_core_utils_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane) +{ + unsigned int pipe_idx = 0; + + for (unsigned int k = 0; k < DML2_MAX_PLANES; ++k) { + pipe_plane[k] = __DML2_CALCS_PIPE_NO_PLANE__; + } + + for (unsigned int plane_idx = 0; plane_idx < DML2_MAX_PLANES; plane_idx++) { + for (int i = 0; i < cfg_support_info->plane_support_info[plane_idx].dpps_used; i++) { + pipe_plane[pipe_idx] = plane_idx; + pipe_idx++; + } + } +} + +bool dml2_core_utils_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg) +{ + bool is_phantom = false; + + if (plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe || + plane_cfg->overrides.legacy_svp_config == dml2_svp_mode_override_phantom_pipe_no_data_return) { + is_phantom = true; + } + + return is_phantom; +} + +unsigned int dml2_core_utils_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel) +{ + + if (sw_mode == dml2_sw_linear) + return 256; + else if (sw_mode == dml2_sw_256b_2d) + return 256; + else if (sw_mode == dml2_sw_4kb_2d) + return 4096; + else if (sw_mode == dml2_sw_64kb_2d) + return 65536; + else if (sw_mode == dml2_sw_256kb_2d) + return 262144; + else if (sw_mode == dml2_gfx11_sw_linear) + return 256; + else if (sw_mode == dml2_gfx11_sw_64kb_d) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_d_t) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_d_x) + return 65536; + else if (sw_mode == dml2_gfx11_sw_64kb_r_x) + return 65536; + else if (sw_mode == dml2_gfx11_sw_256kb_d_x) + return 262144; + else if (sw_mode == dml2_gfx11_sw_256kb_r_x) + return 262144; + else { + DML_ASSERT(0); + return 256; + }; +} + +bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel) +{ + return (byte_per_pixel != 2); +} + +bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode) +{ + return sw_mode == dml2_sw_linear; +}; + + +bool dml2_core_utils_is_vertical_rotation(enum dml2_rotation_angle Scan) +{ + bool is_vert = false; + if (Scan == dml2_rotation_90 || Scan == dml2_rotation_270) { + is_vert = true; + } else { + is_vert = false; + } + return is_vert; +} + +int unsigned dml2_core_utils_get_gfx_version(enum dml2_swizzle_mode sw_mode) +{ + int unsigned version = 0; + + if (sw_mode == dml2_sw_linear || + sw_mode == dml2_sw_256b_2d || + sw_mode == dml2_sw_4kb_2d || + sw_mode == dml2_sw_64kb_2d || + sw_mode == dml2_sw_256kb_2d) + version = 12; + else if (sw_mode == dml2_gfx11_sw_linear || + sw_mode == dml2_gfx11_sw_64kb_d || + sw_mode == dml2_gfx11_sw_64kb_d_t || + sw_mode == dml2_gfx11_sw_64kb_d_x || + sw_mode == dml2_gfx11_sw_64kb_r_x || + sw_mode == dml2_gfx11_sw_256kb_d_x || + sw_mode == dml2_gfx11_sw_256kb_r_x) + version = 11; + else { + DML_LOG_VERBOSE("ERROR: Invalid sw_mode setting! val=%u\n", sw_mode); + DML_ASSERT(0); + } + + return version; +} + +unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params) +{ + unsigned int i; + unsigned int index = 0; + + for (i = 0; i < DML_MAX_CLK_TABLE_SIZE; i++) { + DML_LOG_VERBOSE("DML::%s: per_uclk_dpm_params[%d].minimum_uclk_khz = %ld\n", __func__, i, per_uclk_dpm_params[i].minimum_uclk_khz); + + if (i == 0) + index = 0; + else + index = i - 1; + + if (uclk_freq_khz < per_uclk_dpm_params[i].minimum_uclk_khz || + per_uclk_dpm_params[i].minimum_uclk_khz == 0) { + break; + } + } +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, index); +#endif + return index; +} + +unsigned int dml2_core_utils_get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table) +{ + unsigned int i; + bool clk_entry_found = false; + + for (i = 0; i < clk_table->uclk.num_clk_values; i++) { + DML_LOG_VERBOSE("DML::%s: clk_table.uclk.clk_values_khz[%d] = %ld\n", __func__, i, clk_table->uclk.clk_values_khz[i]); + + if (uclk_freq_khz == clk_table->uclk.clk_values_khz[i]) { + clk_entry_found = true; + break; + } + } + + if (!clk_entry_found) + DML_ASSERT(clk_entry_found); +#if defined(__DML_VBA_DEBUG__) + DML_LOG_VERBOSE("DML::%s: uclk_freq_khz = %ld\n", __func__, uclk_freq_khz); + DML_LOG_VERBOSE("DML::%s: index = %d\n", __func__, i); +#endif + return i; +} + +bool dml2_core_utils_is_dual_plane(enum dml2_source_format_class source_format) +{ + bool ret_val = false; + + if (dml2_core_utils_is_420(source_format) || dml2_core_utils_is_422_planar(source_format) || (source_format == dml2_rgbe_alpha)) + ret_val = true; + + return ret_val; +} + +unsigned int dml2_core_utils_log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend) +{ + if (a == 0) + return 0; + + return (math_log2_approx(a) - subtrahend); +} + +static void create_phantom_stream_from_main_stream(struct dml2_stream_parameters *phantom, const struct dml2_stream_parameters *main, + const struct dml2_implicit_svp_meta *meta) +{ + memcpy(phantom, main, sizeof(struct dml2_stream_parameters)); + + phantom->timing.v_total = meta->v_total; + phantom->timing.v_active = meta->v_active; + phantom->timing.v_front_porch = meta->v_front_porch; + phantom->timing.v_blank_end = phantom->timing.v_total - phantom->timing.v_front_porch - phantom->timing.v_active; + phantom->timing.vblank_nom = phantom->timing.v_total - phantom->timing.v_active; + phantom->timing.drr_config.enabled = false; +} + +static void create_phantom_plane_from_main_plane(struct dml2_plane_parameters *phantom, const struct dml2_plane_parameters *main, + const struct dml2_stream_parameters *phantom_stream, int phantom_stream_index, const struct dml2_stream_parameters *main_stream) +{ + memcpy(phantom, main, sizeof(struct dml2_plane_parameters)); + + phantom->stream_index = phantom_stream_index; + phantom->overrides.refresh_from_mall = dml2_refresh_from_mall_mode_override_force_disable; + phantom->overrides.legacy_svp_config = dml2_svp_mode_override_phantom_pipe_no_data_return; + phantom->composition.viewport.plane0.height = (long int unsigned) math_min2(math_ceil2( + (double)main->composition.scaler_info.plane0.v_ratio * (double)phantom_stream->timing.v_active, 16.0), + (double)main->composition.viewport.plane0.height); + phantom->composition.viewport.plane1.height = (long int unsigned) math_min2(math_ceil2( + (double)main->composition.scaler_info.plane1.v_ratio * (double)phantom_stream->timing.v_active, 16.0), + (double)main->composition.viewport.plane1.height); + phantom->immediate_flip = false; + phantom->dynamic_meta_data.enable = false; + phantom->cursor.num_cursors = 0; + phantom->cursor.cursor_width = 0; + phantom->tdlut.setup_for_tdlut = false; +} + +void dml2_core_utils_expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, + struct dml2_core_scratch *scratch) +{ + unsigned int stream_index, plane_index; + const struct dml2_plane_parameters *main_plane; + const struct dml2_stream_parameters *main_stream; + const struct dml2_stream_parameters *phantom_stream; + + memcpy(svp_expanded_display_cfg, &display_cfg->display_config, sizeof(struct dml2_display_cfg)); + memset(scratch->main_stream_index_from_svp_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); + memset(scratch->svp_stream_index_from_main_stream_index, 0, sizeof(int) * DML2_MAX_PLANES); + memset(scratch->main_plane_index_to_phantom_plane_index, 0, sizeof(int) * DML2_MAX_PLANES); + + if (!display_cfg->display_config.overrides.enable_subvp_implicit_pmo) + return; + + /* disable unbounded requesting for all planes until stage 3 has been performed */ + if (!display_cfg->stage3.performed) { + svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.enable = true; + svp_expanded_display_cfg->overrides.hw.force_unbounded_requesting.value = false; + } + // Create the phantom streams + for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { + main_stream = &display_cfg->display_config.stream_descriptors[stream_index]; + scratch->main_stream_index_from_svp_stream_index[stream_index] = stream_index; + scratch->svp_stream_index_from_main_stream_index[stream_index] = stream_index; + + if (display_cfg->stage3.stream_svp_meta[stream_index].valid) { + // Create the phantom stream + create_phantom_stream_from_main_stream(&svp_expanded_display_cfg->stream_descriptors[svp_expanded_display_cfg->num_streams], + main_stream, &display_cfg->stage3.stream_svp_meta[stream_index]); + + // Associate this phantom stream to the main stream + scratch->main_stream_index_from_svp_stream_index[svp_expanded_display_cfg->num_streams] = stream_index; + scratch->svp_stream_index_from_main_stream_index[stream_index] = svp_expanded_display_cfg->num_streams; + + // Increment num streams + svp_expanded_display_cfg->num_streams++; + } + } + + // Create the phantom planes + for (plane_index = 0; plane_index < display_cfg->display_config.num_planes; plane_index++) { + main_plane = &display_cfg->display_config.plane_descriptors[plane_index]; + + if (display_cfg->stage3.stream_svp_meta[main_plane->stream_index].valid) { + main_stream = &display_cfg->display_config.stream_descriptors[main_plane->stream_index]; + phantom_stream = &svp_expanded_display_cfg->stream_descriptors[scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index]]; + create_phantom_plane_from_main_plane(&svp_expanded_display_cfg->plane_descriptors[svp_expanded_display_cfg->num_planes], + main_plane, phantom_stream, scratch->svp_stream_index_from_main_stream_index[main_plane->stream_index], main_stream); + + // Associate this phantom plane to the main plane + scratch->phantom_plane_index_to_main_plane_index[svp_expanded_display_cfg->num_planes] = plane_index; + scratch->main_plane_index_to_phantom_plane_index[plane_index] = svp_expanded_display_cfg->num_planes; + + // Increment num planes + svp_expanded_display_cfg->num_planes++; + + // Adjust the main plane settings + svp_expanded_display_cfg->plane_descriptors[plane_index].overrides.legacy_svp_config = dml2_svp_mode_override_main_pipe; + } + } +} + +bool dml2_core_utils_is_stream_encoder_required(const struct dml2_stream_parameters *stream_descriptor) +{ + switch (stream_descriptor->output.output_encoder) { + case dml2_dp: + case dml2_dp2p0: + case dml2_edp: + case dml2_hdmi: + case dml2_hdmifrl: + return true; + case dml2_none: + default: + return false; + } +} +bool dml2_core_utils_is_encoder_dsc_capable(const struct dml2_stream_parameters *stream_descriptor) +{ + switch (stream_descriptor->output.output_encoder) { + case dml2_dp: + case dml2_dp2p0: + case dml2_edp: + case dml2_hdmifrl: + return true; + case dml2_hdmi: + case dml2_none: + default: + return false; + } +} + + +bool dml2_core_utils_is_dio_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) +{ + switch (stream_descriptor->output.output_encoder) { + case dml2_dp: + case dml2_edp: + return true; + case dml2_dp2p0: + case dml2_hdmi: + case dml2_hdmifrl: + case dml2_none: + default: + return false; + } +} + +bool dml2_core_utils_is_hpo_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) +{ + switch (stream_descriptor->output.output_encoder) { + case dml2_dp2p0: + return true; + case dml2_dp: + case dml2_edp: + case dml2_hdmi: + case dml2_hdmifrl: + case dml2_none: + default: + return false; + } +} + +bool dml2_core_utils_is_dp_encoder(const struct dml2_stream_parameters *stream_descriptor) +{ + return dml2_core_utils_is_dio_dp_encoder(stream_descriptor) + || dml2_core_utils_is_hpo_dp_encoder(stream_descriptor); +} + + +bool dml2_core_utils_is_dp_8b_10b_link_rate(enum dml2_output_link_dp_rate rate) +{ + switch (rate) { + case dml2_dp_rate_hbr: + case dml2_dp_rate_hbr2: + case dml2_dp_rate_hbr3: + return true; + case dml2_dp_rate_na: + case dml2_dp_rate_uhbr10: + case dml2_dp_rate_uhbr13p5: + case dml2_dp_rate_uhbr20: + default: + return false; + } +} + +bool dml2_core_utils_is_dp_128b_132b_link_rate(enum dml2_output_link_dp_rate rate) +{ + switch (rate) { + case dml2_dp_rate_uhbr10: + case dml2_dp_rate_uhbr13p5: + case dml2_dp_rate_uhbr20: + return true; + case dml2_dp_rate_hbr: + case dml2_dp_rate_hbr2: + case dml2_dp_rate_hbr3: + case dml2_dp_rate_na: + default: + return false; + } +} + +bool dml2_core_utils_is_odm_split(enum dml2_odm_mode odm_mode) +{ + switch (odm_mode) { + case dml2_odm_mode_split_1to2: + case dml2_odm_mode_mso_1to2: + case dml2_odm_mode_mso_1to4: + return true; + case dml2_odm_mode_auto: + case dml2_odm_mode_bypass: + case dml2_odm_mode_combine_2to1: + case dml2_odm_mode_combine_3to1: + case dml2_odm_mode_combine_4to1: + default: + return false; + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h new file mode 100644 index 000000000000..95f0d017add4 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h @@ -0,0 +1,43 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_CORE_UTILS_H__ +#define __DML2_CORE_UTILS_H__ +#include "dml2_internal_shared_types.h" +#include "dml2_debug.h" +#include "lib_float_math.h" + +double dml2_core_utils_div_rem(double dividend, unsigned int divisor, unsigned int *remainder); +const char *dml2_core_utils_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type); +bool dml2_core_utils_is_420(enum dml2_source_format_class source_format); +bool dml2_core_utils_is_422_planar(enum dml2_source_format_class source_format); +bool dml2_core_utils_is_422_packed(enum dml2_source_format_class source_format); +void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mode_support_info *support, bool fail_only); +const char *dml2_core_utils_internal_soc_state_type_str(enum dml2_core_internal_soc_state_type dml2_core_internal_soc_state_type); +void dml2_core_utils_get_stream_output_bpp(double *out_bpp, const struct dml2_display_cfg *display_cfg); +unsigned int dml2_core_utils_round_to_multiple(unsigned int num, unsigned int multiple, bool up); +unsigned int dml2_core_util_get_num_active_pipes(int unsigned num_planes, const struct core_display_cfg_support_info *cfg_support_info); +void dml2_core_utils_pipe_plane_mapping(const struct core_display_cfg_support_info *cfg_support_info, unsigned int *pipe_plane); +bool dml2_core_utils_is_phantom_pipe(const struct dml2_plane_parameters *plane_cfg); +unsigned int dml2_core_utils_get_tile_block_size_bytes(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel); +bool dml2_core_utils_get_segment_horizontal_contiguous(enum dml2_swizzle_mode sw_mode, unsigned int byte_per_pixel); +bool dml2_core_utils_is_vertical_rotation(enum dml2_rotation_angle Scan); +bool dml2_core_utils_is_linear(enum dml2_swizzle_mode sw_mode); +int unsigned dml2_core_utils_get_gfx_version(enum dml2_swizzle_mode sw_mode); +unsigned int dml2_core_utils_get_qos_param_index(unsigned long uclk_freq_khz, const struct dml2_dcn4_uclk_dpm_dependent_qos_params *per_uclk_dpm_params); +unsigned int dml2_core_utils_get_active_min_uclk_dpm_index(unsigned long uclk_freq_khz, const struct dml2_soc_state_table *clk_table); +bool dml2_core_utils_is_dual_plane(enum dml2_source_format_class source_format); +unsigned int dml2_core_utils_log_and_substract_if_non_zero(unsigned int a, unsigned int subtrahend); +void dml2_core_utils_expand_implict_subvp(const struct display_configuation_with_meta *display_cfg, struct dml2_display_cfg *svp_expanded_display_cfg, + struct dml2_core_scratch *scratch); +bool dml2_core_utils_is_stream_encoder_required(const struct dml2_stream_parameters *stream_descriptor); +bool dml2_core_utils_is_encoder_dsc_capable(const struct dml2_stream_parameters *stream_descriptor); +bool dml2_core_utils_is_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); +bool dml2_core_utils_is_dio_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); +bool dml2_core_utils_is_hpo_dp_encoder(const struct dml2_stream_parameters *stream_descriptor); +bool dml2_core_utils_is_dp_8b_10b_link_rate(enum dml2_output_link_dp_rate rate); +bool dml2_core_utils_is_dp_128b_132b_link_rate(enum dml2_output_link_dp_rate rate); +bool dml2_core_utils_is_odm_split(enum dml2_odm_mode odm_mode); + +#endif /* __DML2_CORE_UTILS_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c new file mode 100644 index 000000000000..22969a533a7b --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c @@ -0,0 +1,785 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_dpmm_dcn4.h" +#include "dml2_internal_shared_types.h" +#include "dml_top_types.h" +#include "lib_float_math.h" + +static double dram_bw_kbps_to_uclk_khz(unsigned long long bandwidth_kbps, const struct dml2_dram_params *dram_config) +{ + double uclk_khz = 0; + unsigned long uclk_mbytes_per_tick = 0; + + uclk_mbytes_per_tick = dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock; + + uclk_khz = (double)bandwidth_kbps / uclk_mbytes_per_tick; + + return uclk_khz; +} + +static void get_minimum_clocks_for_latency(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, + double *uclk, + double *fclk, + double *dcfclk) +{ + int min_clock_index_for_latency; + + if (in_out->display_cfg->stage3.success) + min_clock_index_for_latency = in_out->display_cfg->stage3.min_clk_index_for_latency; + else + min_clock_index_for_latency = in_out->display_cfg->stage1.min_clk_index_for_latency; + + *dcfclk = in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].min_dcfclk_khz; + *fclk = in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].min_fclk_khz; + *uclk = dram_bw_kbps_to_uclk_khz(in_out->min_clk_table->dram_bw_table.entries[min_clock_index_for_latency].pre_derate_dram_bw_kbps, + &in_out->soc_bb->clk_table.dram_config); +} + +static unsigned long dml_round_up(double a) +{ + if (a - (unsigned long)a > 0) { + return ((unsigned long)a) + 1; + } + return (unsigned long)a; +} + +static void calculate_system_active_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + double min_uclk_avg, min_uclk_urgent, min_uclk_bw; + double min_fclk_avg, min_fclk_urgent, min_fclk_bw; + double min_dcfclk_avg, min_dcfclk_urgent, min_dcfclk_bw; + double min_uclk_latency, min_fclk_latency, min_dcfclk_latency; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100); + + min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + if (in_out->display_cfg->display_config.hostvm_enable) + min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel_and_vm / 100); + else + min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100); + + min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; + + min_fclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100); + + min_fclk_urgent = (double)mode_support_result->global.active.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100); + + min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; + + min_dcfclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100); + + min_dcfclk_urgent = (double)mode_support_result->global.active.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100); + + min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; + + get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); + + in_out->programming->min_clocks.dcn4x.active.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); + in_out->programming->min_clocks.dcn4x.active.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); + in_out->programming->min_clocks.dcn4x.active.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); +} + +static void calculate_svp_prefetch_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + double min_uclk_avg, min_uclk_urgent, min_uclk_bw; + double min_fclk_avg, min_fclk_urgent, min_fclk_bw; + double min_dcfclk_avg, min_dcfclk_urgent, min_dcfclk_bw; + double min_fclk_latency, min_dcfclk_latency; + double min_uclk_latency; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + /* assumes DF throttling is enabled */ + min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.dram_derate_percent_pixel / 100); + + min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dram_derate_percent_pixel / 100); + + min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; + + min_fclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.fclk_derate_percent / 100); + + min_fclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.fclk_derate_percent / 100); + + min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; + + min_dcfclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_average.dcfclk_derate_percent / 100); + + min_dcfclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.dcn_mall_prefetch_urgent.dcfclk_derate_percent / 100); + + min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; + + get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); + + in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); + in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); + in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); + + /* assumes DF throttling is disabled */ + min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dram_derate_percent_pixel / 100); + + min_uclk_urgent = dram_bw_kbps_to_uclk_khz(mode_support_result->global.svp_prefetch.urgent_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_urgent = (double)min_uclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100); + + min_uclk_bw = min_uclk_urgent > min_uclk_avg ? min_uclk_urgent : min_uclk_avg; + + min_fclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.fclk_derate_percent / 100); + + min_fclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_urgent = (double)min_fclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100); + + min_fclk_bw = min_fclk_urgent > min_fclk_avg ? min_fclk_urgent : min_fclk_avg; + + min_dcfclk_avg = (double)mode_support_result->global.svp_prefetch.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_average.dcfclk_derate_percent / 100); + + min_dcfclk_urgent = (double)mode_support_result->global.svp_prefetch.urgent_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_urgent = (double)min_dcfclk_urgent / ((double)in_out->soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent / 100); + + min_dcfclk_bw = min_dcfclk_urgent > min_dcfclk_avg ? min_dcfclk_urgent : min_dcfclk_avg; + + get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); + + in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz = dml_round_up(min_uclk_bw > min_uclk_latency ? min_uclk_bw : min_uclk_latency); + in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz = dml_round_up(min_fclk_bw > min_fclk_latency ? min_fclk_bw : min_fclk_latency); + in_out->programming->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz = dml_round_up(min_dcfclk_bw > min_dcfclk_latency ? min_dcfclk_bw : min_dcfclk_latency); +} + +static void calculate_idle_minimums(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + double min_uclk_avg; + double min_fclk_avg; + double min_dcfclk_avg; + double min_uclk_latency, min_fclk_latency, min_dcfclk_latency; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + min_uclk_avg = dram_bw_kbps_to_uclk_khz(mode_support_result->global.active.average_bw_dram_kbps, &in_out->soc_bb->clk_table.dram_config); + min_uclk_avg = (double)min_uclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.dram_derate_percent_pixel / 100); + + min_fclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->fabric_datapath_to_dcn_data_return_bytes; + min_fclk_avg = (double)min_fclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.fclk_derate_percent / 100); + + min_dcfclk_avg = (double)mode_support_result->global.active.average_bw_sdp_kbps / in_out->soc_bb->return_bus_width_bytes; + min_dcfclk_avg = (double)min_dcfclk_avg / ((double)in_out->soc_bb->qos_parameters.derate_table.system_idle_average.dcfclk_derate_percent / 100); + + get_minimum_clocks_for_latency(in_out, &min_uclk_latency, &min_fclk_latency, &min_dcfclk_latency); + + in_out->programming->min_clocks.dcn4x.idle.uclk_khz = dml_round_up(min_uclk_avg > min_uclk_latency ? min_uclk_avg : min_uclk_latency); + in_out->programming->min_clocks.dcn4x.idle.fclk_khz = dml_round_up(min_fclk_avg > min_fclk_latency ? min_fclk_avg : min_fclk_latency); + in_out->programming->min_clocks.dcn4x.idle.dcfclk_khz = dml_round_up(min_dcfclk_avg > min_dcfclk_latency ? min_dcfclk_avg : min_dcfclk_latency); +} + +static bool add_margin_and_round_to_dfs_grainularity(double clock_khz, double margin, unsigned long vco_freq_khz, unsigned long *rounded_khz, uint32_t *divider_id) +{ + enum dentist_divider_range { + DFS_DIVIDER_RANGE_1_START = 8, /* 2.00 */ + DFS_DIVIDER_RANGE_1_STEP = 1, /* 0.25 */ + DFS_DIVIDER_RANGE_2_START = 64, /* 16.00 */ + DFS_DIVIDER_RANGE_2_STEP = 2, /* 0.50 */ + DFS_DIVIDER_RANGE_3_START = 128, /* 32.00 */ + DFS_DIVIDER_RANGE_3_STEP = 4, /* 1.00 */ + DFS_DIVIDER_RANGE_4_START = 248, /* 62.00 */ + DFS_DIVIDER_RANGE_4_STEP = 264, /* 66.00 */ + DFS_DIVIDER_RANGE_SCALE_FACTOR = 4 + }; + + enum DFS_base_divider_id { + DFS_BASE_DID_1 = 0x08, + DFS_BASE_DID_2 = 0x40, + DFS_BASE_DID_3 = 0x60, + DFS_BASE_DID_4 = 0x7e, + DFS_MAX_DID = 0x7f + }; + + unsigned int divider; + + if (clock_khz < 1 || vco_freq_khz < 1 || clock_khz > vco_freq_khz) + return false; + + clock_khz *= 1.0 + margin; + + divider = (unsigned int)((int)DFS_DIVIDER_RANGE_SCALE_FACTOR * (vco_freq_khz / clock_khz)); + + /* we want to floor here to get higher clock than required rather than lower */ + if (divider < DFS_DIVIDER_RANGE_2_START) { + if (divider < DFS_DIVIDER_RANGE_1_START) + *divider_id = DFS_BASE_DID_1; + else + *divider_id = DFS_BASE_DID_1 + ((divider - DFS_DIVIDER_RANGE_1_START) / DFS_DIVIDER_RANGE_1_STEP); + } else if (divider < DFS_DIVIDER_RANGE_3_START) { + *divider_id = DFS_BASE_DID_2 + ((divider - DFS_DIVIDER_RANGE_2_START) / DFS_DIVIDER_RANGE_2_STEP); + } else if (divider < DFS_DIVIDER_RANGE_4_START) { + *divider_id = DFS_BASE_DID_3 + ((divider - DFS_DIVIDER_RANGE_3_START) / DFS_DIVIDER_RANGE_3_STEP); + } else { + *divider_id = DFS_BASE_DID_4 + ((divider - DFS_DIVIDER_RANGE_4_START) / DFS_DIVIDER_RANGE_4_STEP); + if (*divider_id > DFS_MAX_DID) + *divider_id = DFS_MAX_DID; + } + + *rounded_khz = vco_freq_khz * DFS_DIVIDER_RANGE_SCALE_FACTOR / divider; + + return true; +} + +static bool round_to_non_dfs_granularity(unsigned long dispclk_khz, unsigned long dpprefclk_khz, unsigned long dtbrefclk_khz, + unsigned long *rounded_dispclk_khz, unsigned long *rounded_dpprefclk_khz, unsigned long *rounded_dtbrefclk_khz) +{ + unsigned long pll_frequency_khz; + + pll_frequency_khz = (unsigned long) math_max2(600000, math_ceil2(math_max3(dispclk_khz, dpprefclk_khz, dtbrefclk_khz), 1000)); + + *rounded_dispclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dispclk_khz, 32); + + *rounded_dpprefclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dpprefclk_khz, 32); + + if (dtbrefclk_khz > 0) { + *rounded_dtbrefclk_khz = pll_frequency_khz / (unsigned long) math_min2(pll_frequency_khz / dtbrefclk_khz, 32); + } else { + *rounded_dtbrefclk_khz = 0; + } + + return true; +} + +static bool round_up_and_copy_to_next_dpm(unsigned long min_value, unsigned long *rounded_value, const struct dml2_clk_table *clock_table) +{ + bool result = false; + int index = 0; + + if (clock_table->num_clk_values > 2) { + while (index < clock_table->num_clk_values && clock_table->clk_values_khz[index] < min_value) + index++; + + if (index < clock_table->num_clk_values) { + *rounded_value = clock_table->clk_values_khz[index]; + result = true; + } + } else if (clock_table->clk_values_khz[clock_table->num_clk_values - 1] >= min_value) { + *rounded_value = min_value; + result = true; + } + return result; +} + +static bool round_up_to_next_dpm(unsigned long *clock_value, const struct dml2_clk_table *clock_table) +{ + return round_up_and_copy_to_next_dpm(*clock_value, clock_value, clock_table); +} + +static bool map_soc_min_clocks_to_dpm_fine_grained(struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) +{ + bool result; + + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.dcfclk_khz, &state_table->dcfclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.fclk_khz, &state_table->fclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.active.uclk_khz, &state_table->uclk); + + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.dcfclk_khz, &state_table->dcfclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.fclk_khz, &state_table->fclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch.uclk_khz, &state_table->uclk); + + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.dcfclk_khz, &state_table->dcfclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.fclk_khz, &state_table->fclk); + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.idle.uclk_khz, &state_table->uclk); + + /* these clocks are optional, so they can fail to map, in which case map all to 0 */ + if (result) { + if (!round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz, &state_table->dcfclk) || + !round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz, &state_table->fclk) || + !round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz, &state_table->uclk)) { + display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.dcfclk_khz = 0; + display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.fclk_khz = 0; + display_cfg->min_clocks.dcn4x.svp_prefetch_no_throttle.uclk_khz = 0; + } + } + + return result; +} + +static bool map_soc_min_clocks_to_dpm_coarse_grained(struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) +{ + bool result; + int index; + + result = false; + for (index = 0; index < state_table->uclk.num_clk_values; index++) { + if (display_cfg->min_clocks.dcn4x.active.dcfclk_khz <= state_table->dcfclk.clk_values_khz[index] && + display_cfg->min_clocks.dcn4x.active.fclk_khz <= state_table->fclk.clk_values_khz[index] && + display_cfg->min_clocks.dcn4x.active.uclk_khz <= state_table->uclk.clk_values_khz[index]) { + display_cfg->min_clocks.dcn4x.active.dcfclk_khz = state_table->dcfclk.clk_values_khz[index]; + display_cfg->min_clocks.dcn4x.active.fclk_khz = state_table->fclk.clk_values_khz[index]; + display_cfg->min_clocks.dcn4x.active.uclk_khz = state_table->uclk.clk_values_khz[index]; + result = true; + break; + } + } + + if (result) { + result = false; + for (index = 0; index < state_table->uclk.num_clk_values; index++) { + if (display_cfg->min_clocks.dcn4x.idle.dcfclk_khz <= state_table->dcfclk.clk_values_khz[index] && + display_cfg->min_clocks.dcn4x.idle.fclk_khz <= state_table->fclk.clk_values_khz[index] && + display_cfg->min_clocks.dcn4x.idle.uclk_khz <= state_table->uclk.clk_values_khz[index]) { + display_cfg->min_clocks.dcn4x.idle.dcfclk_khz = state_table->dcfclk.clk_values_khz[index]; + display_cfg->min_clocks.dcn4x.idle.fclk_khz = state_table->fclk.clk_values_khz[index]; + display_cfg->min_clocks.dcn4x.idle.uclk_khz = state_table->uclk.clk_values_khz[index]; + result = true; + break; + } + } + } + + // SVP is not supported on any coarse grained SoCs + display_cfg->min_clocks.dcn4x.svp_prefetch.dcfclk_khz = 0; + display_cfg->min_clocks.dcn4x.svp_prefetch.fclk_khz = 0; + display_cfg->min_clocks.dcn4x.svp_prefetch.uclk_khz = 0; + + return result; +} + +static bool map_min_clocks_to_dpm(const struct dml2_core_mode_support_result *mode_support_result, struct dml2_display_cfg_programming *display_cfg, const struct dml2_soc_state_table *state_table) +{ + bool result = false; + bool dcfclk_fine_grained = false, fclk_fine_grained = false, clock_state_count_identical = false; + unsigned int i; + + if (!state_table || !display_cfg) + return false; + + if (state_table->dcfclk.num_clk_values == 2) { + dcfclk_fine_grained = true; + } + + if (state_table->fclk.num_clk_values == 2) { + fclk_fine_grained = true; + } + + if (state_table->fclk.num_clk_values == state_table->dcfclk.num_clk_values && + state_table->fclk.num_clk_values == state_table->uclk.num_clk_values) { + clock_state_count_identical = true; + } + + if (dcfclk_fine_grained || fclk_fine_grained || !clock_state_count_identical) + result = map_soc_min_clocks_to_dpm_fine_grained(display_cfg, state_table); + else + result = map_soc_min_clocks_to_dpm_coarse_grained(display_cfg, state_table); + + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dispclk_khz, &state_table->dispclk); + + for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { + if (result) + result = round_up_to_next_dpm(&display_cfg->plane_programming[i].min_clocks.dcn4x.dppclk_khz, &state_table->dppclk); + } + + for (i = 0; i < display_cfg->display_config.num_streams; i++) { + if (result) + result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].dscclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.dscclk_khz, &state_table->dscclk); + if (result) + result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].dtbclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.dtbclk_khz, &state_table->dtbclk); + if (result) + result = round_up_and_copy_to_next_dpm(mode_support_result->per_stream[i].phyclk_khz, &display_cfg->stream_programming[i].min_clocks.dcn4x.phyclk_khz, &state_table->phyclk); + } + + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dpprefclk_khz, &state_table->dppclk); + + if (result) + result = round_up_to_next_dpm(&display_cfg->min_clocks.dcn4x.dtbrefclk_khz, &state_table->dtbclk); + + return result; +} + +static bool are_timings_trivially_synchronizable(struct dml2_display_cfg *display_config, int mask) +{ + unsigned int i; + bool identical = true; + bool contains_drr = false; + unsigned int remap_array[DML2_MAX_PLANES]; + unsigned int remap_array_size = 0; + + // Create a remap array to enable simple iteration through only masked stream indicies + for (i = 0; i < display_config->num_streams; i++) { + if (mask & (0x1 << i)) { + remap_array[remap_array_size++] = i; + } + } + + // 0 or 1 display is always trivially synchronizable + if (remap_array_size <= 1) + return true; + + // Check that all displays timings are the same + for (i = 1; i < remap_array_size; i++) { + if (memcmp(&display_config->stream_descriptors[remap_array[i - 1]].timing, &display_config->stream_descriptors[remap_array[i]].timing, sizeof(struct dml2_timing_cfg))) { + identical = false; + break; + } + } + + // Check if any displays are drr + for (i = 0; i < remap_array_size; i++) { + if (display_config->stream_descriptors[remap_array[i]].timing.drr_config.enabled) { + contains_drr = true; + break; + } + } + + // Trivial sync is possible if all displays are identical and none are DRR + return !contains_drr && identical; +} + +static int find_smallest_idle_time_in_vblank_us(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int mask) +{ + unsigned int i; + int min_idle_us = 0; + unsigned int remap_array[DML2_MAX_PLANES]; + unsigned int remap_array_size = 0; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + // Create a remap array to enable simple iteration through only masked stream indicies + for (i = 0; i < in_out->programming->display_config.num_streams; i++) { + if (mask & (0x1 << i)) { + remap_array[remap_array_size++] = i; + } + } + + if (remap_array_size == 0) + return 0; + + min_idle_us = mode_support_result->cfg_support_info.stream_support_info[remap_array[0]].vblank_reserved_time_us; + + for (i = 1; i < remap_array_size; i++) { + if (min_idle_us > mode_support_result->cfg_support_info.stream_support_info[remap_array[i]].vblank_reserved_time_us) + min_idle_us = mode_support_result->cfg_support_info.stream_support_info[remap_array[i]].vblank_reserved_time_us; + } + + return min_idle_us; +} + +static bool determine_power_management_features_with_vblank_only(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + int min_idle_us; + + if (are_timings_trivially_synchronizable(&in_out->programming->display_config, 0xF)) { + min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, 0xF); + + if (min_idle_us >= in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us) + in_out->programming->uclk_pstate_supported = true; + + if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) + in_out->programming->fclk_pstate_supported = true; + } + + return true; +} + +static int get_displays_without_vactive_margin_mask(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int latency_hiding_requirement_us) +{ + unsigned int i; + int displays_without_vactive_margin_mask = 0x0; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + for (i = 0; i < in_out->programming->display_config.num_planes; i++) { + if (mode_support_result->cfg_support_info.plane_support_info[i].active_latency_hiding_us + < latency_hiding_requirement_us) + displays_without_vactive_margin_mask |= (0x1 << i); + } + + return displays_without_vactive_margin_mask; +} + +static int get_displays_with_fams_mask(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out, int latency_hiding_requirement_us) +{ + unsigned int i; + int displays_with_fams_mask = 0x0; + + for (i = 0; i < in_out->programming->display_config.num_planes; i++) { + if (in_out->programming->display_config.plane_descriptors->overrides.legacy_svp_config != dml2_svp_mode_override_auto) + displays_with_fams_mask |= (0x1 << i); + } + + return displays_with_fams_mask; +} + +static bool determine_power_management_features_with_vactive_and_vblank(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + int displays_without_vactive_margin_mask = 0x0; + int min_idle_us = 0; + + if (in_out->programming->uclk_pstate_supported == false) { + displays_without_vactive_margin_mask = + get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); + + if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { + min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); + + if (min_idle_us >= in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us) + in_out->programming->uclk_pstate_supported = true; + } + } + + if (in_out->programming->fclk_pstate_supported == false) { + displays_without_vactive_margin_mask = + get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.fclk_change_blackout_us)); + + if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { + min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); + + if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) + in_out->programming->fclk_pstate_supported = true; + } + } + + return true; +} + +static bool determine_power_management_features_with_fams(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + int displays_without_vactive_margin_mask = 0x0; + int displays_without_fams_mask = 0x0; + + displays_without_vactive_margin_mask = + get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); + + displays_without_fams_mask = + get_displays_with_fams_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.dram_clk_change_blackout_us)); + + if ((displays_without_vactive_margin_mask & ~displays_without_fams_mask) == 0) + in_out->programming->uclk_pstate_supported = true; + + return true; +} + +static void clamp_uclk_to_max(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + in_out->programming->min_clocks.dcn4x.active.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; + in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; + in_out->programming->min_clocks.dcn4x.idle.uclk_khz = in_out->soc_bb->clk_table.uclk.clk_values_khz[in_out->soc_bb->clk_table.uclk.num_clk_values - 1]; +} + +static void clamp_fclk_to_max(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + in_out->programming->min_clocks.dcn4x.active.fclk_khz = in_out->soc_bb->clk_table.fclk.clk_values_khz[in_out->soc_bb->clk_table.fclk.num_clk_values - 1]; + in_out->programming->min_clocks.dcn4x.idle.fclk_khz = in_out->soc_bb->clk_table.fclk.clk_values_khz[in_out->soc_bb->clk_table.fclk.num_clk_values - 1]; +} + +static bool map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + int i; + bool result; + double dispclk_khz; + const struct dml2_core_mode_support_result *mode_support_result = &in_out->display_cfg->mode_support_result; + + calculate_system_active_minimums(in_out); + calculate_svp_prefetch_minimums(in_out); + calculate_idle_minimums(in_out); + + // In NV4, there's no support for FCLK or DCFCLK DPM change before SVP prefetch starts, therefore + // active minimums must be boosted to prefetch minimums + if (in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz > in_out->programming->min_clocks.dcn4x.active.uclk_khz) + in_out->programming->min_clocks.dcn4x.active.uclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.uclk_khz; + + if (in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz > in_out->programming->min_clocks.dcn4x.active.fclk_khz) + in_out->programming->min_clocks.dcn4x.active.fclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.fclk_khz; + + if (in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz > in_out->programming->min_clocks.dcn4x.active.dcfclk_khz) + in_out->programming->min_clocks.dcn4x.active.dcfclk_khz = in_out->programming->min_clocks.dcn4x.svp_prefetch.dcfclk_khz; + + // need some massaging for the dispclk ramping cases: + dispclk_khz = mode_support_result->global.dispclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0) * (1.0 + in_out->ip->dispclk_ramp_margin_percent / 100.0); + // ramping margin should not make dispclk exceed the maximum dispclk speed: + dispclk_khz = math_min2(dispclk_khz, in_out->min_clk_table->max_clocks_khz.dispclk); + // but still the required dispclk can be more than the maximum dispclk speed: + dispclk_khz = math_max2(dispclk_khz, mode_support_result->global.dispclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); + + // DPP Ref is always set to max of all DPP clocks + for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { + if (in_out->programming->min_clocks.dcn4x.dpprefclk_khz < mode_support_result->per_plane[i].dppclk_khz) + in_out->programming->min_clocks.dcn4x.dpprefclk_khz = mode_support_result->per_plane[i].dppclk_khz; + } + in_out->programming->min_clocks.dcn4x.dpprefclk_khz = (unsigned long) (in_out->programming->min_clocks.dcn4x.dpprefclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); + + // DTB Ref is always set to max of all DTB clocks + for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { + if (in_out->programming->min_clocks.dcn4x.dtbrefclk_khz < mode_support_result->per_stream[i].dtbclk_khz) + in_out->programming->min_clocks.dcn4x.dtbrefclk_khz = mode_support_result->per_stream[i].dtbclk_khz; + } + in_out->programming->min_clocks.dcn4x.dtbrefclk_khz = (unsigned long)(in_out->programming->min_clocks.dcn4x.dtbrefclk_khz * (1 + in_out->soc_bb->dcn_downspread_percent / 100.0)); + + if (in_out->soc_bb->no_dfs) { + round_to_non_dfs_granularity((unsigned long)dispclk_khz, in_out->programming->min_clocks.dcn4x.dpprefclk_khz, in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, + &in_out->programming->min_clocks.dcn4x.dispclk_khz, &in_out->programming->min_clocks.dcn4x.dpprefclk_khz, &in_out->programming->min_clocks.dcn4x.dtbrefclk_khz); + } else { + add_margin_and_round_to_dfs_grainularity(dispclk_khz, 0.0, + (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dispclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dispclk_did); + + add_margin_and_round_to_dfs_grainularity(in_out->programming->min_clocks.dcn4x.dpprefclk_khz, 0.0, + (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dpprefclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dpprefclk_did); + + add_margin_and_round_to_dfs_grainularity(in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, 0.0, + (unsigned long)(in_out->soc_bb->dispclk_dppclk_vco_speed_mhz * 1000), &in_out->programming->min_clocks.dcn4x.dtbrefclk_khz, &in_out->programming->min_clocks.dcn4x.divider_ids.dtbrefclk_did); + } + + + for (i = 0; i < DML2_MAX_DCN_PIPES; i++) { + in_out->programming->plane_programming[i].min_clocks.dcn4x.dppclk_khz = (unsigned long)(in_out->programming->min_clocks.dcn4x.dpprefclk_khz / 255.0 + * math_ceil2(in_out->display_cfg->mode_support_result.per_plane[i].dppclk_khz * (1.0 + in_out->soc_bb->dcn_downspread_percent / 100.0) * 255.0 / in_out->programming->min_clocks.dcn4x.dpprefclk_khz, 1.0)); + } + + in_out->programming->min_clocks.dcn4x.deepsleep_dcfclk_khz = mode_support_result->global.dcfclk_deepsleep_khz; + in_out->programming->min_clocks.dcn4x.socclk_khz = mode_support_result->global.socclk_khz; + + result = map_min_clocks_to_dpm(mode_support_result, in_out->programming, &in_out->soc_bb->clk_table); + + // By default, all power management features are not enabled + in_out->programming->fclk_pstate_supported = false; + in_out->programming->uclk_pstate_supported = false; + + return result; +} + +bool dpmm_dcn3_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + bool result; + + result = map_mode_to_soc_dpm(in_out); + + // Check if any can be enabled by nominal vblank idle time + determine_power_management_features_with_vblank_only(in_out); + + // Check if any can be enabled in vactive/vblank + determine_power_management_features_with_vactive_and_vblank(in_out); + + // Check if any can be enabled via fams + determine_power_management_features_with_fams(in_out); + + if (in_out->programming->uclk_pstate_supported == false) + clamp_uclk_to_max(in_out); + + if (in_out->programming->fclk_pstate_supported == false) + clamp_fclk_to_max(in_out); + + return result; +} + +bool dpmm_dcn4_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + bool result; + int displays_without_vactive_margin_mask = 0x0; + int min_idle_us = 0; + + result = map_mode_to_soc_dpm(in_out); + + if (in_out->display_cfg->stage3.success) + in_out->programming->uclk_pstate_supported = true; + + displays_without_vactive_margin_mask = + get_displays_without_vactive_margin_mask(in_out, (int)(in_out->soc_bb->power_management_parameters.fclk_change_blackout_us)); + + if (displays_without_vactive_margin_mask == 0) { + in_out->programming->fclk_pstate_supported = true; + } else { + if (are_timings_trivially_synchronizable(&in_out->programming->display_config, displays_without_vactive_margin_mask)) { + min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, displays_without_vactive_margin_mask); + + if (min_idle_us >= in_out->soc_bb->power_management_parameters.fclk_change_blackout_us) + in_out->programming->fclk_pstate_supported = true; + } + } + + if (in_out->programming->uclk_pstate_supported == false) + clamp_uclk_to_max(in_out); + + if (in_out->programming->fclk_pstate_supported == false) + clamp_fclk_to_max(in_out); + + min_idle_us = find_smallest_idle_time_in_vblank_us(in_out, 0xFF); + if (in_out->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && + min_idle_us >= in_out->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us) + in_out->programming->stutter.supported_in_blank = true; + else + in_out->programming->stutter.supported_in_blank = false; + + // TODO: Fix me Sam + if (in_out->soc_bb->power_management_parameters.z8_min_idle_time > 0 && + in_out->programming->informative.power_management.z8.stutter_period >= in_out->soc_bb->power_management_parameters.z8_min_idle_time) + in_out->programming->z8_stutter.meets_eco = true; + else + in_out->programming->z8_stutter.meets_eco = false; + + if (in_out->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && + min_idle_us >= in_out->soc_bb->power_management_parameters.z8_stutter_exit_latency_us) + in_out->programming->z8_stutter.supported_in_blank = true; + else + in_out->programming->z8_stutter.supported_in_blank = false; + + return result; +} + +bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out) +{ + const struct dml2_display_cfg *display_cfg = &in_out->display_cfg->display_config; + const struct dml2_core_internal_display_mode_lib *mode_lib = &in_out->core->clean_me_up.mode_lib; + struct dml2_dchub_global_register_set *dchubbub_regs = &in_out->programming->global_regs; + + double refclk_freq_in_mhz = (display_cfg->overrides.hw.dlg_ref_clk_mhz > 0) ? (double)display_cfg->overrides.hw.dlg_ref_clk_mhz : mode_lib->soc.dchub_refclk_mhz; + + /* set A */ + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_A].frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); + + /* set B */ + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].fclk_pstate = (int unsigned)(mode_lib->mp.Watermark.FCLKChangeWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter = (int unsigned)(mode_lib->mp.Watermark.StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit = (int unsigned)(mode_lib->mp.Watermark.StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_enter_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterEnterPlusExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].sr_exit_z8 = (int unsigned)(mode_lib->mp.Watermark.Z8StutterExitWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].temp_read_or_ppt = (int unsigned)(mode_lib->mp.Watermark.temp_read_or_ppt_watermark_us * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].uclk_pstate = (int unsigned)(mode_lib->mp.Watermark.DRAMClockChangeWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].urgent = (int unsigned)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].usr = (int unsigned)(mode_lib->mp.Watermark.USRRetrainingWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].refcyc_per_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].refcyc_per_meta_trip_to_mem = (unsigned int)(mode_lib->mp.Watermark.UrgentWatermark * refclk_freq_in_mhz); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_flip = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthImmediateFlip * 1000); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_nom = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidth * 1000); + dchubbub_regs->wm_regs[DML2_DCHUB_WATERMARK_SET_B].frac_urg_bw_mall = (unsigned int)(mode_lib->mp.FractionOfUrgentBandwidthMALL * 1000); + + dchubbub_regs->num_watermark_sets = 2; + + return true; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h new file mode 100644 index 000000000000..e7b58f2efda4 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h @@ -0,0 +1,14 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_DPMM_DCN4_H__ +#define __DML2_DPMM_DCN4_H__ + +#include "dml2_internal_shared_types.h" + +bool dpmm_dcn3_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); +bool dpmm_dcn4_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); +bool dpmm_dcn4_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c new file mode 100644 index 000000000000..dfd01440737d --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c @@ -0,0 +1,50 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_dpmm_factory.h" +#include "dml2_dpmm_dcn4.h" +#include "dml2_external_lib_deps.h" + +static bool dummy_map_mode_to_soc_dpm(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out) +{ + return true; +} + +static bool dummy_map_watermarks(struct dml2_dpmm_map_watermarks_params_in_out *in_out) +{ + return true; +} + +bool dml2_dpmm_create(enum dml2_project_id project_id, struct dml2_dpmm_instance *out) +{ + bool result = false; + + if (!out) + return false; + + memset(out, 0, sizeof(struct dml2_dpmm_instance)); + + switch (project_id) { + case dml2_project_dcn4x_stage1: + out->map_mode_to_soc_dpm = &dummy_map_mode_to_soc_dpm; + out->map_watermarks = &dummy_map_watermarks; + result = true; + break; + case dml2_project_dcn4x_stage2: + out->map_mode_to_soc_dpm = &dpmm_dcn3_map_mode_to_soc_dpm; + out->map_watermarks = &dummy_map_watermarks; + result = true; + break; + case dml2_project_dcn4x_stage2_auto_drr_svp: + out->map_mode_to_soc_dpm = &dpmm_dcn4_map_mode_to_soc_dpm; + out->map_watermarks = &dpmm_dcn4_map_watermarks; + result = true; + break; + case dml2_project_invalid: + default: + break; + } + + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h new file mode 100644 index 000000000000..20ba2e446f1d --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_DPMM_FACTORY_H__ +#define __DML2_DPMM_FACTORY_H__ + +#include "dml2_internal_shared_types.h" +#include "dml_top_types.h" + +bool dml2_dpmm_create(enum dml2_project_id project_id, struct dml2_dpmm_instance *out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c new file mode 100644 index 000000000000..a265f254152c --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c @@ -0,0 +1,198 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_mcg_dcn4.h" +#include "dml_top_soc_parameter_types.h" + +static bool build_min_clock_table(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table); + +bool mcg_dcn4_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out) +{ + return build_min_clock_table(in_out->soc_bb, in_out->min_clk_table); +} + +static unsigned long long uclk_to_dram_bw_kbps(unsigned long uclk_khz, const struct dml2_dram_params *dram_config) +{ + unsigned long long bw_kbps = 0; + + bw_kbps = (unsigned long long) uclk_khz * dram_config->channel_count * dram_config->channel_width_bytes * dram_config->transactions_per_clock; + + return bw_kbps; +} + +static unsigned long round_up_to_quantized_values(unsigned long value, const unsigned long *quantized_values, int num_quantized_values) +{ + int i; + + if (!quantized_values) + return 0; + + for (i = 0; i < num_quantized_values; i++) { + if (quantized_values[i] > value) + return quantized_values[i]; + } + + return 0; +} + +static bool build_min_clk_table_fine_grained(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) +{ + bool dcfclk_fine_grained = false, fclk_fine_grained = false; + + int i; + unsigned int j; + + unsigned long min_dcfclk_khz = 0; + unsigned long min_fclk_khz = 0; + unsigned long prev_100, cur_50; + + if (soc_bb->clk_table.dcfclk.num_clk_values == 2) { + dcfclk_fine_grained = true; + } + + if (soc_bb->clk_table.fclk.num_clk_values == 2) { + fclk_fine_grained = true; + } + + min_dcfclk_khz = soc_bb->clk_table.dcfclk.clk_values_khz[0]; + min_fclk_khz = soc_bb->clk_table.fclk.clk_values_khz[0]; + + // First calculate the table for "balanced" bandwidths across UCLK/FCLK + for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { + min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps = uclk_to_dram_bw_kbps(soc_bb->clk_table.uclk.clk_values_khz[i], &soc_bb->clk_table.dram_config); + + min_table->dram_bw_table.entries[i].min_fclk_khz = (unsigned long)((((double)min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps * soc_bb->qos_parameters.derate_table.system_active_urgent.dram_derate_percent_pixel / 100) / ((double)soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / 100)) / soc_bb->fabric_datapath_to_dcn_data_return_bytes); + } + min_table->dram_bw_table.num_entries = soc_bb->clk_table.uclk.num_clk_values; + + // To create the minium table, effectively shift "up" all the dcfclk/fclk entries by 1, and then replace the lowest entry with min fclk/dcfclk + for (i = min_table->dram_bw_table.num_entries - 1; i > 0; i--) { + prev_100 = min_table->dram_bw_table.entries[i - 1].min_fclk_khz; + cur_50 = min_table->dram_bw_table.entries[i].min_fclk_khz / 2; + min_table->dram_bw_table.entries[i].min_fclk_khz = prev_100 > cur_50 ? prev_100 : cur_50; + + if (!fclk_fine_grained) { + min_table->dram_bw_table.entries[i].min_fclk_khz = round_up_to_quantized_values(min_table->dram_bw_table.entries[i].min_fclk_khz, soc_bb->clk_table.fclk.clk_values_khz, soc_bb->clk_table.fclk.num_clk_values); + } + } + min_table->dram_bw_table.entries[0].min_fclk_khz /= 2; + + // Clamp to minimums and maximums + for (i = 0; i < (int)min_table->dram_bw_table.num_entries; i++) { + if (min_table->dram_bw_table.entries[i].min_dcfclk_khz < min_dcfclk_khz) + min_table->dram_bw_table.entries[i].min_dcfclk_khz = min_dcfclk_khz; + + if (min_table->dram_bw_table.entries[i].min_fclk_khz < min_fclk_khz) + min_table->dram_bw_table.entries[i].min_fclk_khz = min_fclk_khz; + + if (soc_bb->max_fclk_for_uclk_dpm_khz > 0 && + min_table->dram_bw_table.entries[i].min_fclk_khz > soc_bb->max_fclk_for_uclk_dpm_khz) + min_table->dram_bw_table.entries[i].min_fclk_khz = soc_bb->max_fclk_for_uclk_dpm_khz; + + min_table->dram_bw_table.entries[i].min_dcfclk_khz = + min_table->dram_bw_table.entries[i].min_fclk_khz * + soc_bb->qos_parameters.derate_table.system_active_urgent.fclk_derate_percent / soc_bb->qos_parameters.derate_table.system_active_urgent.dcfclk_derate_percent; + + min_table->dram_bw_table.entries[i].min_dcfclk_khz = + min_table->dram_bw_table.entries[i].min_dcfclk_khz * soc_bb->fabric_datapath_to_dcn_data_return_bytes / soc_bb->return_bus_width_bytes; + + if (!dcfclk_fine_grained) { + min_table->dram_bw_table.entries[i].min_dcfclk_khz = round_up_to_quantized_values(min_table->dram_bw_table.entries[i].min_dcfclk_khz, soc_bb->clk_table.dcfclk.clk_values_khz, soc_bb->clk_table.dcfclk.num_clk_values); + } + } + + // Prune states which are invalid (some clocks exceed maximum) + for (i = 0; i < (int)min_table->dram_bw_table.num_entries; i++) { + if (min_table->dram_bw_table.entries[i].min_dcfclk_khz > min_table->max_clocks_khz.dcfclk || + min_table->dram_bw_table.entries[i].min_fclk_khz > min_table->max_clocks_khz.fclk) { + min_table->dram_bw_table.num_entries = i; + break; + } + } + + // Prune duplicate states + for (i = 0; i < (int)min_table->dram_bw_table.num_entries - 1; i++) { + if (min_table->dram_bw_table.entries[i].min_dcfclk_khz == min_table->dram_bw_table.entries[i + 1].min_dcfclk_khz && + min_table->dram_bw_table.entries[i].min_fclk_khz == min_table->dram_bw_table.entries[i + 1].min_fclk_khz && + min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps == min_table->dram_bw_table.entries[i + 1].pre_derate_dram_bw_kbps) { + + // i + 1 is the same state as i, so shift everything + for (j = i + 1; j < min_table->dram_bw_table.num_entries; j++) { + min_table->dram_bw_table.entries[j].min_dcfclk_khz = min_table->dram_bw_table.entries[j + 1].min_dcfclk_khz; + min_table->dram_bw_table.entries[j].min_fclk_khz = min_table->dram_bw_table.entries[j + 1].min_fclk_khz; + min_table->dram_bw_table.entries[j].pre_derate_dram_bw_kbps = min_table->dram_bw_table.entries[j + 1].pre_derate_dram_bw_kbps; + } + min_table->dram_bw_table.num_entries--; + } + } + + return true; +} + +static bool build_min_clk_table_coarse_grained(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) +{ + int i; + + for (i = 0; i < soc_bb->clk_table.uclk.num_clk_values; i++) { + min_table->dram_bw_table.entries[i].pre_derate_dram_bw_kbps = uclk_to_dram_bw_kbps(soc_bb->clk_table.uclk.clk_values_khz[i], &soc_bb->clk_table.dram_config); + min_table->dram_bw_table.entries[i].min_dcfclk_khz = soc_bb->clk_table.dcfclk.clk_values_khz[i]; + min_table->dram_bw_table.entries[i].min_fclk_khz = soc_bb->clk_table.fclk.clk_values_khz[i]; + } + min_table->dram_bw_table.num_entries = soc_bb->clk_table.uclk.num_clk_values; + + return true; +} + +static bool build_min_clock_table(const struct dml2_soc_bb *soc_bb, struct dml2_mcg_min_clock_table *min_table) +{ + bool result; + bool dcfclk_fine_grained = false, fclk_fine_grained = false, clock_state_count_equal = false; + + if (!soc_bb || !min_table) + return false; + + if (soc_bb->clk_table.dcfclk.num_clk_values < 2 || soc_bb->clk_table.fclk.num_clk_values < 2) + return false; + + if (soc_bb->clk_table.uclk.num_clk_values > DML_MCG_MAX_CLK_TABLE_SIZE) + return false; + + if (soc_bb->clk_table.dcfclk.num_clk_values == 2) { + dcfclk_fine_grained = true; + } + + if (soc_bb->clk_table.fclk.num_clk_values == 2) { + fclk_fine_grained = true; + } + + if (soc_bb->clk_table.fclk.num_clk_values == soc_bb->clk_table.dcfclk.num_clk_values && + soc_bb->clk_table.fclk.num_clk_values == soc_bb->clk_table.uclk.num_clk_values) + clock_state_count_equal = true; + + min_table->fixed_clocks_khz.amclk = 0; + min_table->fixed_clocks_khz.dprefclk = soc_bb->dprefclk_mhz * 1000; + min_table->fixed_clocks_khz.pcierefclk = soc_bb->pcie_refclk_mhz * 1000; + min_table->fixed_clocks_khz.dchubrefclk = soc_bb->dchub_refclk_mhz * 1000; + min_table->fixed_clocks_khz.xtalclk = soc_bb->xtalclk_mhz * 1000; + + min_table->max_clocks_khz.dispclk = soc_bb->clk_table.dispclk.clk_values_khz[soc_bb->clk_table.dispclk.num_clk_values - 1]; + min_table->max_clocks_khz.dppclk = soc_bb->clk_table.dppclk.clk_values_khz[soc_bb->clk_table.dppclk.num_clk_values - 1]; + min_table->max_clocks_khz.dscclk = soc_bb->clk_table.dscclk.clk_values_khz[soc_bb->clk_table.dscclk.num_clk_values - 1]; + min_table->max_clocks_khz.dtbclk = soc_bb->clk_table.dtbclk.clk_values_khz[soc_bb->clk_table.dtbclk.num_clk_values - 1]; + min_table->max_clocks_khz.phyclk = soc_bb->clk_table.phyclk.clk_values_khz[soc_bb->clk_table.phyclk.num_clk_values - 1]; + + min_table->max_ss_clocks_khz.dispclk = (unsigned int)((double)min_table->max_clocks_khz.dispclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + min_table->max_ss_clocks_khz.dppclk = (unsigned int)((double)min_table->max_clocks_khz.dppclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + min_table->max_ss_clocks_khz.dtbclk = (unsigned int)((double)min_table->max_clocks_khz.dtbclk / (1.0 + soc_bb->dcn_downspread_percent / 100.0)); + + min_table->max_clocks_khz.dcfclk = soc_bb->clk_table.dcfclk.clk_values_khz[soc_bb->clk_table.dcfclk.num_clk_values - 1]; + min_table->max_clocks_khz.fclk = soc_bb->clk_table.fclk.clk_values_khz[soc_bb->clk_table.fclk.num_clk_values - 1]; + + if (dcfclk_fine_grained || fclk_fine_grained || !clock_state_count_equal) + result = build_min_clk_table_fine_grained(soc_bb, min_table); + else + result = build_min_clk_table_coarse_grained(soc_bb, min_table); + + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h new file mode 100644 index 000000000000..f54fde8fba90 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_MCG_DCN4_H__ +#define __DML2_MCG_DCN4_H__ + +#include "dml2_internal_shared_types.h" + +bool mcg_dcn4_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out); +bool mcg_dcn4_unit_test(void); + +#endif \ No newline at end of file diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c new file mode 100644 index 000000000000..c60b8fe90819 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c @@ -0,0 +1,39 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_mcg_factory.h" +#include "dml2_mcg_dcn4.h" +#include "dml2_external_lib_deps.h" + +static bool dummy_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out) +{ + return true; +} + +bool dml2_mcg_create(enum dml2_project_id project_id, struct dml2_mcg_instance *out) +{ + bool result = false; + + if (!out) + return false; + + memset(out, 0, sizeof(struct dml2_mcg_instance)); + + switch (project_id) { + case dml2_project_dcn4x_stage1: + out->build_min_clock_table = &dummy_build_min_clock_table; + result = true; + break; + case dml2_project_dcn4x_stage2: + case dml2_project_dcn4x_stage2_auto_drr_svp: + out->build_min_clock_table = &mcg_dcn4_build_min_clock_table; + result = true; + break; + case dml2_project_invalid: + default: + break; + } + + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h new file mode 100644 index 000000000000..ad307deca3b0 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_MCG_FACTORY_H__ +#define __DML2_MCG_FACTORY_H__ + +#include "dml2_internal_shared_types.h" +#include "dml_top_types.h" + +bool dml2_mcg_create(enum dml2_project_id project_id, struct dml2_mcg_instance *out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c new file mode 100644 index 000000000000..1b9579a32ff2 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c @@ -0,0 +1,706 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_pmo_factory.h" +#include "dml2_pmo_dcn3.h" + +static void sort(double *list_a, int list_a_size) +{ + // For all elements b[i] in list_b[] + for (int i = 0; i < list_a_size - 1; i++) { + // Find the first element of list_a that's larger than b[i] + for (int j = i; j < list_a_size - 1; j++) { + if (list_a[j] > list_a[j + 1]) + swap(list_a[j], list_a[j + 1]); + } + } +} + +static double get_max_reserved_time_on_all_planes_with_stream_index(struct display_configuation_with_meta *config, unsigned int stream_index) +{ + struct dml2_plane_parameters *plane_descriptor; + long max_reserved_time_ns = 0; + + for (unsigned int i = 0; i < config->display_config.num_planes; i++) { + plane_descriptor = &config->display_config.plane_descriptors[i]; + + if (plane_descriptor->stream_index == stream_index) + if (plane_descriptor->overrides.reserved_vblank_time_ns > max_reserved_time_ns) + max_reserved_time_ns = plane_descriptor->overrides.reserved_vblank_time_ns; + } + + return (max_reserved_time_ns / 1000.0); +} + + +static void set_reserved_time_on_all_planes_with_stream_index(struct display_configuation_with_meta *config, unsigned int stream_index, double reserved_time_us) +{ + struct dml2_plane_parameters *plane_descriptor; + + for (unsigned int i = 0; i < config->display_config.num_planes; i++) { + plane_descriptor = &config->display_config.plane_descriptors[i]; + + if (plane_descriptor->stream_index == stream_index) + plane_descriptor->overrides.reserved_vblank_time_ns = (long int)(reserved_time_us * 1000); + } +} + +static void remove_duplicates(double *list_a, int *list_a_size) +{ + int j = 0; + + if (*list_a_size == 0) + return; + + for (int i = 1; i < *list_a_size; i++) { + if (list_a[j] != list_a[i]) { + j++; + list_a[j] = list_a[i]; + } + } + + *list_a_size = j + 1; +} + +static bool increase_mpc_combine_factor(unsigned int *mpc_combine_factor, unsigned int limit) +{ + if (*mpc_combine_factor < limit) { + (*mpc_combine_factor)++; + return true; + } + + return false; +} + +static bool optimize_dcc_mcache_no_odm(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out, + int free_pipes) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i; + bool result = true; + + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + // For pipes that failed dcc mcache check, we want to increase the pipe count. + // The logic for doing this depends on how many pipes is already being used, + // and whether it's mpcc or odm combine. + if (!in_out->dcc_mcache_supported[i]) { + // For the general case of "n displays", we can only optimize streams with an ODM combine factor of 1 + if (in_out->cfg_support_info->stream_support_info[in_out->optimized_display_cfg->plane_descriptors[i].stream_index].odms_used == 1) { + in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor = + in_out->cfg_support_info->plane_support_info[i].dpps_used; + // For each plane that is not passing mcache validation, just add another pipe to it, up to the limit. + if (free_pipes > 0) { + if (!increase_mpc_combine_factor(&in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor, + pmo->mpc_combine_limit)) { + // We've reached max pipes allocatable to a single plane, so we fail. + result = false; + break; + } else { + // Successfully added another pipe to this failing plane. + free_pipes--; + } + } else { + // No free pipes to add. + result = false; + break; + } + } else { + // If the stream of this plane needs ODM combine, no further optimization can be done. + result = false; + break; + } + } + } + + return result; +} + +static bool iterate_to_next_candidiate(struct dml2_pmo_instance *pmo, int size) +{ + int borrow_from, i; + bool success = false; + + if (pmo->scratch.pmo_dcn3.current_candidate[0] > 0) { + pmo->scratch.pmo_dcn3.current_candidate[0]--; + success = true; + } else { + for (borrow_from = 1; borrow_from < size && pmo->scratch.pmo_dcn3.current_candidate[borrow_from] == 0; borrow_from++) + ; + + if (borrow_from < size) { + pmo->scratch.pmo_dcn3.current_candidate[borrow_from]--; + for (i = 0; i < borrow_from; i++) { + pmo->scratch.pmo_dcn3.current_candidate[i] = pmo->scratch.pmo_dcn3.reserved_time_candidates_count[i] - 1; + } + + success = true; + } + } + + return success; +} + +static bool increase_odm_combine_factor(enum dml2_odm_mode *odm_mode, int odms_calculated) +{ + bool result = true; + + if (*odm_mode == dml2_odm_mode_auto) { + switch (odms_calculated) { + case 1: + *odm_mode = dml2_odm_mode_bypass; + break; + case 2: + *odm_mode = dml2_odm_mode_combine_2to1; + break; + case 3: + *odm_mode = dml2_odm_mode_combine_3to1; + break; + case 4: + *odm_mode = dml2_odm_mode_combine_4to1; + break; + default: + result = false; + break; + } + } + + if (result) { + if (*odm_mode == dml2_odm_mode_bypass) { + *odm_mode = dml2_odm_mode_combine_2to1; + } else if (*odm_mode == dml2_odm_mode_combine_2to1) { + *odm_mode = dml2_odm_mode_combine_3to1; + } else if (*odm_mode == dml2_odm_mode_combine_3to1) { + *odm_mode = dml2_odm_mode_combine_4to1; + } else { + result = false; + } + } + + return result; +} + +static int count_planes_with_stream_index(const struct dml2_display_cfg *display_cfg, unsigned int stream_index) +{ + unsigned int i, count; + + count = 0; + for (i = 0; i < display_cfg->num_planes; i++) { + if (display_cfg->plane_descriptors[i].stream_index == stream_index) + count++; + } + + return count; +} + +static bool are_timings_trivially_synchronizable(struct display_configuation_with_meta *display_config, int mask) +{ + unsigned int i; + bool identical = true; + bool contains_drr = false; + unsigned int remap_array[DML2_MAX_PLANES]; + unsigned int remap_array_size = 0; + + // Create a remap array to enable simple iteration through only masked stream indicies + for (i = 0; i < display_config->display_config.num_streams; i++) { + if (mask & (0x1 << i)) { + remap_array[remap_array_size++] = i; + } + } + + // 0 or 1 display is always trivially synchronizable + if (remap_array_size <= 1) + return true; + + for (i = 1; i < remap_array_size; i++) { + if (memcmp(&display_config->display_config.stream_descriptors[remap_array[i - 1]].timing, + &display_config->display_config.stream_descriptors[remap_array[i]].timing, + sizeof(struct dml2_timing_cfg))) { + identical = false; + break; + } + } + + for (i = 0; i < remap_array_size; i++) { + if (display_config->display_config.stream_descriptors[remap_array[i]].timing.drr_config.enabled) { + contains_drr = true; + break; + } + } + + return !contains_drr && identical; +} + +bool pmo_dcn3_initialize(struct dml2_pmo_initialize_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + pmo->soc_bb = in_out->soc_bb; + pmo->ip_caps = in_out->ip_caps; + pmo->mpc_combine_limit = 2; + pmo->odm_combine_limit = 4; + pmo->mcg_clock_table_size = in_out->mcg_clock_table_size; + + pmo->options = in_out->options; + + return true; +} + +static bool is_h_timing_divisible_by(const struct dml2_timing_cfg *timing, unsigned char denominator) +{ + /* + * Htotal, Hblank start/end, and Hsync start/end all must be divisible + * in order for the horizontal timing params to be considered divisible + * by 2. Hsync start is always 0. + */ + unsigned long h_blank_start = timing->h_total - timing->h_front_porch; + + return (timing->h_total % denominator == 0) && + (h_blank_start % denominator == 0) && + (timing->h_blank_end % denominator == 0) && + (timing->h_sync_width % denominator == 0); +} + +static bool is_dp_encoder(enum dml2_output_encoder_class encoder_type) +{ + switch (encoder_type) { + case dml2_dp: + case dml2_edp: + case dml2_dp2p0: + case dml2_none: + return true; + case dml2_hdmi: + case dml2_hdmifrl: + default: + return false; + } +} + +bool pmo_dcn3_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out) +{ + unsigned int i; + const struct dml2_display_cfg *display_config = + &in_out->base_display_config->display_config; + const struct dml2_core_mode_support_result *mode_support_result = + &in_out->base_display_config->mode_support_result; + + if (in_out->instance->options->disable_dyn_odm || + (in_out->instance->options->disable_dyn_odm_for_multi_stream && display_config->num_streams > 1)) + return false; + + for (i = 0; i < display_config->num_planes; i++) + /* + * vmin optimization is required to be seamlessly switched off + * at any time when the new configuration is no longer + * supported. However switching from ODM combine to MPC combine + * is not always seamless. When there not enough free pipes, we + * will have to use the same secondary OPP heads as secondary + * DPP pipes in MPC combine in new state. This transition is + * expected to cause glitches. To avoid the transition, we only + * allow vmin optimization if the stream's base configuration + * doesn't require MPC combine. This condition checks if MPC + * combine is enabled. If so do not optimize the stream. + */ + if (mode_support_result->cfg_support_info.plane_support_info[i].dpps_used > 1 && + mode_support_result->cfg_support_info.stream_support_info[display_config->plane_descriptors[i].stream_index].odms_used == 1) + in_out->base_display_config->stage4.unoptimizable_streams[display_config->plane_descriptors[i].stream_index] = true; + + for (i = 0; i < display_config->num_streams; i++) { + if (display_config->stream_descriptors[i].overrides.disable_dynamic_odm) + in_out->base_display_config->stage4.unoptimizable_streams[i] = true; + else if (in_out->base_display_config->stage3.stream_svp_meta[i].valid && + in_out->instance->options->disable_dyn_odm_for_stream_with_svp) + in_out->base_display_config->stage4.unoptimizable_streams[i] = true; + /* + * ODM Combine requires horizontal timing divisible by 2 so each + * ODM segment has the same size. + */ + else if (!is_h_timing_divisible_by(&display_config->stream_descriptors[i].timing, 2)) + in_out->base_display_config->stage4.unoptimizable_streams[i] = true; + /* + * Our hardware support seamless ODM transitions for DP encoders + * only. + */ + else if (!is_dp_encoder(display_config->stream_descriptors[i].output.output_encoder)) + in_out->base_display_config->stage4.unoptimizable_streams[i] = true; + } + + return true; +} + +bool pmo_dcn3_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out) +{ + bool is_vmin = true; + + if (in_out->vmin_limits->dispclk_khz > 0 && + in_out->display_config->mode_support_result.global.dispclk_khz > in_out->vmin_limits->dispclk_khz) + is_vmin = false; + + return is_vmin; +} + +static int find_highest_odm_load_stream_index( + const struct dml2_display_cfg *display_config, + const struct dml2_core_mode_support_result *mode_support_result) +{ + unsigned int i; + int odm_load, highest_odm_load = -1, highest_odm_load_index = -1; + + for (i = 0; i < display_config->num_streams; i++) { + if (mode_support_result->cfg_support_info.stream_support_info[i].odms_used > 0) + odm_load = display_config->stream_descriptors[i].timing.pixel_clock_khz + / mode_support_result->cfg_support_info.stream_support_info[i].odms_used; + else + odm_load = 0; + + if (odm_load > highest_odm_load) { + highest_odm_load_index = i; + highest_odm_load = odm_load; + } + } + + return highest_odm_load_index; +} + +bool pmo_dcn3_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out) +{ + int stream_index; + const struct dml2_display_cfg *display_config = + &in_out->base_display_config->display_config; + const struct dml2_core_mode_support_result *mode_support_result = + &in_out->base_display_config->mode_support_result; + unsigned int odms_used; + struct dml2_stream_parameters *stream_descriptor; + bool optimizable = false; + + /* + * highest odm load stream must be optimizable to continue as dispclk is + * bounded by it. + */ + stream_index = find_highest_odm_load_stream_index(display_config, + mode_support_result); + + if (stream_index < 0 || + in_out->base_display_config->stage4.unoptimizable_streams[stream_index]) + return false; + + odms_used = mode_support_result->cfg_support_info.stream_support_info[stream_index].odms_used; + if ((int)odms_used >= in_out->instance->odm_combine_limit) + return false; + + memcpy(in_out->optimized_display_config, + in_out->base_display_config, + sizeof(struct display_configuation_with_meta)); + + stream_descriptor = &in_out->optimized_display_config->display_config.stream_descriptors[stream_index]; + while (!optimizable && increase_odm_combine_factor( + &stream_descriptor->overrides.odm_mode, + odms_used)) { + switch (stream_descriptor->overrides.odm_mode) { + case dml2_odm_mode_combine_2to1: + optimizable = true; + break; + case dml2_odm_mode_combine_3to1: + /* + * In ODM Combine 3:1 OTG_valid_pixel rate is 1/4 of + * actual pixel rate. Therefore horizontal timing must + * be divisible by 4. + */ + if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { + /* + * DSC h slice count must be divisible + * by 3. + */ + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 3 == 0) + optimizable = true; + } else { + optimizable = true; + } + } + break; + case dml2_odm_mode_combine_4to1: + /* + * In ODM Combine 4:1 OTG_valid_pixel rate is 1/4 of + * actual pixel rate. Therefore horizontal timing must + * be divisible by 4. + */ + if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { + /* + * DSC h slice count must be divisible + * by 4. + */ + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 4 == 0) + optimizable = true; + } else { + optimizable = true; + } + } + break; + case dml2_odm_mode_auto: + case dml2_odm_mode_bypass: + case dml2_odm_mode_split_1to2: + case dml2_odm_mode_mso_1to2: + case dml2_odm_mode_mso_1to4: + default: + break; + } + } + + return optimizable; +} + +bool pmo_dcn3_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i, used_pipes, free_pipes, planes_on_stream; + bool result; + + if (in_out->display_config != in_out->optimized_display_cfg) { + memcpy(in_out->optimized_display_cfg, in_out->display_config, sizeof(struct dml2_display_cfg)); + } + + //Count number of free pipes, and check if any odm combine is in use. + used_pipes = 0; + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + used_pipes += in_out->cfg_support_info->plane_support_info[i].dpps_used; + } + free_pipes = pmo->ip_caps->pipe_count - used_pipes; + + // Optimization loop + // The goal here is to add more pipes to any planes + // which are failing mcache admissibility + result = true; + + // The optimization logic depends on whether ODM combine is enabled, and the stream count. + if (in_out->optimized_display_cfg->num_streams > 1) { + // If there are multiple streams, we are limited to only be able to optimize mcache failures on planes + // which are not ODM combined. + + result = optimize_dcc_mcache_no_odm(in_out, free_pipes); + } else if (in_out->optimized_display_cfg->num_streams == 1) { + // In single stream cases, we still optimize mcache failures when there's ODM combine with some + // additional logic. + + if (in_out->cfg_support_info->stream_support_info[0].odms_used > 1) { + // If ODM combine is enabled, then the logic is to increase ODM combine factor. + + // Optimization for streams with > 1 ODM combine factor is only supported for single display. + planes_on_stream = count_planes_with_stream_index(in_out->optimized_display_cfg, 0); + + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + // For pipes that failed dcc mcache check, we want to increase the pipe count. + // The logic for doing this depends on how many pipes is already being used, + // and whether it's mpcc or odm combine. + if (!in_out->dcc_mcache_supported[i]) { + // Increasing ODM combine factor on a stream requires a free pipe for each plane on the stream. + if (free_pipes >= planes_on_stream) { + if (!increase_odm_combine_factor(&in_out->optimized_display_cfg->stream_descriptors[i].overrides.odm_mode, + in_out->cfg_support_info->plane_support_info[i].dpps_used)) { + result = false; + } else { + break; + } + } else { + result = false; + break; + } + } + } + } else { + // If ODM combine is not enabled, then we can actually use the same logic as before. + + result = optimize_dcc_mcache_no_odm(in_out, free_pipes); + } + } else { + result = true; + } + + return result; +} + +bool pmo_dcn3_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + struct dml2_optimization_stage3_state *state = &in_out->base_display_config->stage3; + const struct dml2_stream_parameters *stream_descriptor; + const struct dml2_plane_parameters *plane_descriptor; + unsigned int stream_index, plane_index, candidate_count; + double min_reserved_vblank_time = 0; + int fclk_twait_needed_mask = 0x0; + int uclk_twait_needed_mask = 0x0; + + state->performed = true; + state->min_clk_index_for_latency = in_out->base_display_config->stage1.min_clk_index_for_latency; + pmo->scratch.pmo_dcn3.min_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; + pmo->scratch.pmo_dcn3.max_latency_index = pmo->mcg_clock_table_size - 1; + pmo->scratch.pmo_dcn3.cur_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; + + pmo->scratch.pmo_dcn3.stream_mask = 0xF; + + for (plane_index = 0; plane_index < in_out->base_display_config->display_config.num_planes; plane_index++) { + plane_descriptor = &in_out->base_display_config->display_config.plane_descriptors[plane_index]; + stream_descriptor = &in_out->base_display_config->display_config.stream_descriptors[plane_descriptor->stream_index]; + + if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[plane_index].active_latency_hiding_us < + in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us && + stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate == dml2_twait_budgeting_setting_if_needed) + uclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); + + if (stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate == dml2_twait_budgeting_setting_try) + uclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); + + if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[plane_index].active_latency_hiding_us < + in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us && + stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate == dml2_twait_budgeting_setting_if_needed) + fclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); + + if (stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate == dml2_twait_budgeting_setting_try) + fclk_twait_needed_mask |= (0x1 << plane_descriptor->stream_index); + + if (plane_descriptor->overrides.legacy_svp_config != dml2_svp_mode_override_auto) { + pmo->scratch.pmo_dcn3.stream_mask &= ~(0x1 << plane_descriptor->stream_index); + } + } + + for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) { + stream_descriptor = &in_out->base_display_config->display_config.stream_descriptors[stream_index]; + + // The absolute minimum required time is the minimum of all the required budgets + /* + if (stream_descriptor->overrides.hw.twait_budgeting.fclk_pstate + == dml2_twait_budgeting_setting_require) + + if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { + min_reserved_vblank_time = max_double2(min_reserved_vblank_time, + in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us); + } + + if (stream_descriptor->overrides.hw.twait_budgeting.uclk_pstate + == dml2_twait_budgeting_setting_require) { + + if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { + min_reserved_vblank_time = max_double2(min_reserved_vblank_time, + in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us); + } + } + + if (stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit + == dml2_twait_budgeting_setting_require) + min_reserved_vblank_time = max_double2(min_reserved_vblank_time, + in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us); + */ + + min_reserved_vblank_time = get_max_reserved_time_on_all_planes_with_stream_index(in_out->base_display_config, stream_index); + + // Insert the absolute minimum into the array + candidate_count = 1; + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][0] = min_reserved_vblank_time; + pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] = candidate_count; + + if (!(pmo->scratch.pmo_dcn3.stream_mask & (0x1 << stream_index))) + continue; + + // For every optional feature, we create a candidate for it only if it's larger minimum. + if ((fclk_twait_needed_mask & (0x1 << stream_index)) && + in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us > min_reserved_vblank_time) { + + if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = + in_out->instance->soc_bb->power_management_parameters.fclk_change_blackout_us; + } + } + + if ((uclk_twait_needed_mask & (0x1 << stream_index)) && + in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us > min_reserved_vblank_time) { + + if (are_timings_trivially_synchronizable(in_out->base_display_config, pmo->scratch.pmo_dcn3.stream_mask)) { + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = + in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us; + } + } + + if ((stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit == dml2_twait_budgeting_setting_try || + stream_descriptor->overrides.hw.twait_budgeting.stutter_enter_exit == dml2_twait_budgeting_setting_if_needed) && + in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > min_reserved_vblank_time) { + + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][candidate_count++] = + in_out->instance->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us; + } + + pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] = candidate_count; + + // Finally sort the array of candidates + sort(pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index], + pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index]); + + remove_duplicates(pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index], + &pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index]); + + pmo->scratch.pmo_dcn3.current_candidate[stream_index] = + pmo->scratch.pmo_dcn3.reserved_time_candidates_count[stream_index] - 1; + } + + return true; +} + +bool pmo_dcn3_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i, stream_index; + + for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { + stream_index = in_out->base_display_config->display_config.plane_descriptors[i].stream_index; + + if (in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][pmo->scratch.pmo_dcn3.current_candidate[stream_index]] * 1000) { + return false; + } + } + + return true; +} + +bool pmo_dcn3_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + unsigned int stream_index; + bool success = false; + bool reached_end; + + memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); + + if (in_out->last_candidate_failed) { + if (pmo->scratch.pmo_dcn3.cur_latency_index < pmo->scratch.pmo_dcn3.max_latency_index) { + // If we haven't tried all the clock bounds to support this state, try a higher one + pmo->scratch.pmo_dcn3.cur_latency_index++; + + success = true; + } else { + // If there's nothing higher to try, then we have to have a smaller canadidate + reached_end = !iterate_to_next_candidiate(pmo, in_out->optimized_display_config->display_config.num_streams); + + if (!reached_end) { + pmo->scratch.pmo_dcn3.cur_latency_index = pmo->scratch.pmo_dcn3.min_latency_index; + success = true; + } + } + } else { + success = true; + } + + if (success) { + in_out->optimized_display_config->stage3.min_clk_index_for_latency = pmo->scratch.pmo_dcn3.cur_latency_index; + + for (stream_index = 0; stream_index < in_out->optimized_display_config->display_config.num_streams; stream_index++) { + set_reserved_time_on_all_planes_with_stream_index(in_out->optimized_display_config, stream_index, + pmo->scratch.pmo_dcn3.reserved_time_candidates[stream_index][pmo->scratch.pmo_dcn3.current_candidate[stream_index]]); + } + } + + return success; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h new file mode 100644 index 000000000000..f00bd9e72a86 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h @@ -0,0 +1,22 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_PMO_DCN3_H__ +#define __DML2_PMO_DCN3_H__ + +#include "dml2_internal_shared_types.h" + +bool pmo_dcn3_initialize(struct dml2_pmo_initialize_in_out *in_out); + +bool pmo_dcn3_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); + +bool pmo_dcn3_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out); +bool pmo_dcn3_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out); +bool pmo_dcn3_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out); + +bool pmo_dcn3_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out); +bool pmo_dcn3_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out); +bool pmo_dcn3_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c new file mode 100644 index 000000000000..5769c2638f9a --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c @@ -0,0 +1,2394 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_pmo_factory.h" +#include "dml2_debug.h" +#include "lib_float_math.h" +#include "dml2_pmo_dcn4_fams2.h" + +static const double MIN_VACTIVE_MARGIN_PCT = 0.25; // We need more than non-zero margin because DET buffer granularity can alter vactive latency hiding +static const double MIN_BLANK_STUTTER_FACTOR = 3.0; + +static const struct dml2_pmo_pstate_strategy base_strategy_list_1_display[] = { + // VActive Preferred + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then SVP + { + .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // Then DRR + { + .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Finally VBlank, but allow base clocks for latency to increase + /* + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + */ +}; + +static const int base_strategy_list_1_display_size = sizeof(base_strategy_list_1_display) / sizeof(struct dml2_pmo_pstate_strategy); + +static const struct dml2_pmo_pstate_strategy base_strategy_list_2_display[] = { + // VActive only is preferred + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then VActive + VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // Then VBlank only + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // Then SVP + VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // Then SVP + DRR + { + .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then SVP + SVP + { + .per_stream_pstate_method = { dml2_pstate_method_fw_svp, dml2_pstate_method_fw_svp, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then DRR + VActive + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Then DRR + DRR + { + .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // Finally VBlank, but allow base clocks for latency to increase + /* + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + */ +}; + +static const int base_strategy_list_2_display_size = sizeof(base_strategy_list_2_display) / sizeof(struct dml2_pmo_pstate_strategy); + +static const struct dml2_pmo_pstate_strategy base_strategy_list_3_display[] = { + // All VActive + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // VActive + 1 VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vblank, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // All VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na }, + .allow_state_increase = false, + }, + + // All DRR + { + .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + + // All VBlank, with state increase allowed + /* + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_na }, + .allow_state_increase = true, + }, + */ +}; + +static const int base_strategy_list_3_display_size = sizeof(base_strategy_list_3_display) / sizeof(struct dml2_pmo_pstate_strategy); + +static const struct dml2_pmo_pstate_strategy base_strategy_list_4_display[] = { + // All VActive + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive }, + .allow_state_increase = true, + }, + + // VActive + 1 VBlank + { + .per_stream_pstate_method = { dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vactive, dml2_pstate_method_vblank }, + .allow_state_increase = false, + }, + + // All Vblank + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank }, + .allow_state_increase = false, + }, + + // All DRR + { + .per_stream_pstate_method = { dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr, dml2_pstate_method_fw_drr }, + .allow_state_increase = true, + }, + + // All VBlank, with state increase allowed + /* + { + .per_stream_pstate_method = { dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank, dml2_pstate_method_vblank }, + .allow_state_increase = true, + }, + */ +}; + +static const int base_strategy_list_4_display_size = sizeof(base_strategy_list_4_display) / sizeof(struct dml2_pmo_pstate_strategy); + + +static bool increase_odm_combine_factor(enum dml2_odm_mode *odm_mode, int odms_calculated) +{ + bool result = true; + + if (*odm_mode == dml2_odm_mode_auto) { + switch (odms_calculated) { + case 1: + *odm_mode = dml2_odm_mode_bypass; + break; + case 2: + *odm_mode = dml2_odm_mode_combine_2to1; + break; + case 3: + *odm_mode = dml2_odm_mode_combine_3to1; + break; + case 4: + *odm_mode = dml2_odm_mode_combine_4to1; + break; + default: + result = false; + break; + } + } + + if (result) { + if (*odm_mode == dml2_odm_mode_bypass) { + *odm_mode = dml2_odm_mode_combine_2to1; + } else if (*odm_mode == dml2_odm_mode_combine_2to1) { + *odm_mode = dml2_odm_mode_combine_3to1; + } else if (*odm_mode == dml2_odm_mode_combine_3to1) { + *odm_mode = dml2_odm_mode_combine_4to1; + } else { + result = false; + } + } + + return result; +} + +static bool increase_mpc_combine_factor(unsigned int *mpc_combine_factor, unsigned int limit) +{ + if (*mpc_combine_factor < limit) { + (*mpc_combine_factor)++; + return true; + } + + return false; +} + +static int count_planes_with_stream_index(const struct dml2_display_cfg *display_cfg, unsigned int stream_index) +{ + unsigned int i, count; + + count = 0; + for (i = 0; i < display_cfg->num_planes; i++) { + if (display_cfg->plane_descriptors[i].stream_index == stream_index) + count++; + } + + return count; +} + +static bool optimize_dcc_mcache_no_odm(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out, + int free_pipes) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i; + bool result = true; + + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + // For pipes that failed dcc mcache check, we want to increase the pipe count. + // The logic for doing this depends on how many pipes is already being used, + // and whether it's mpcc or odm combine. + if (!in_out->dcc_mcache_supported[i]) { + // For the general case of "n displays", we can only optimize streams with an ODM combine factor of 1 + if (in_out->cfg_support_info->stream_support_info[in_out->optimized_display_cfg->plane_descriptors[i].stream_index].odms_used == 1) { + in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor = + in_out->cfg_support_info->plane_support_info[i].dpps_used; + // For each plane that is not passing mcache validation, just add another pipe to it, up to the limit. + if (free_pipes > 0) { + if (!increase_mpc_combine_factor(&in_out->optimized_display_cfg->plane_descriptors[i].overrides.mpcc_combine_factor, + pmo->mpc_combine_limit)) { + // We've reached max pipes allocatable to a single plane, so we fail. + result = false; + break; + } else { + // Successfully added another pipe to this failing plane. + free_pipes--; + } + } else { + // No free pipes to add. + result = false; + break; + } + } else { + // If the stream of this plane needs ODM combine, no further optimization can be done. + result = false; + break; + } + } + } + + return result; +} + +bool pmo_dcn4_fams2_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i, used_pipes, free_pipes, planes_on_stream; + bool result; + + if (in_out->display_config != in_out->optimized_display_cfg) { + memcpy(in_out->optimized_display_cfg, in_out->display_config, sizeof(struct dml2_display_cfg)); + } + + //Count number of free pipes, and check if any odm combine is in use. + used_pipes = 0; + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + used_pipes += in_out->cfg_support_info->plane_support_info[i].dpps_used; + } + free_pipes = pmo->ip_caps->pipe_count - used_pipes; + + // Optimization loop + // The goal here is to add more pipes to any planes + // which are failing mcache admissibility + result = true; + + // The optimization logic depends on whether ODM combine is enabled, and the stream count. + if (in_out->optimized_display_cfg->num_streams > 1 || in_out->instance->options->disable_dyn_odm) { + // If there are multiple streams, we are limited to only be able to optimize mcache failures on planes + // which are not ODM combined. + + result = optimize_dcc_mcache_no_odm(in_out, free_pipes); + } else if (in_out->optimized_display_cfg->num_streams == 1) { + // In single stream cases, we still optimize mcache failures when there's ODM combine with some + // additional logic. + + if (in_out->cfg_support_info->stream_support_info[0].odms_used > 1) { + // If ODM combine is enabled, then the logic is to increase ODM combine factor. + + // Optimization for streams with > 1 ODM combine factor is only supported for single display. + planes_on_stream = count_planes_with_stream_index(in_out->optimized_display_cfg, 0); + + for (i = 0; i < in_out->optimized_display_cfg->num_planes; i++) { + // For pipes that failed dcc mcache check, we want to increase the pipe count. + // The logic for doing this depends on how many pipes is already being used, + // and whether it's mpcc or odm combine. + if (!in_out->dcc_mcache_supported[i]) { + // Increasing ODM combine factor on a stream requires a free pipe for each plane on the stream. + if (free_pipes >= planes_on_stream) { + if (!increase_odm_combine_factor(&in_out->optimized_display_cfg->stream_descriptors[i].overrides.odm_mode, + in_out->cfg_support_info->plane_support_info[i].dpps_used)) { + result = false; + } else { + break; + } + } else { + result = false; + break; + } + } + } + } else { + // If ODM combine is not enabled, then we can actually use the same logic as before. + + result = optimize_dcc_mcache_no_odm(in_out, free_pipes); + } + } else { + result = true; + } + + return result; +} + +static enum dml2_pstate_method convert_strategy_to_drr_variant(const enum dml2_pstate_method base_strategy) +{ + enum dml2_pstate_method variant_strategy = 0; + + switch (base_strategy) { + case dml2_pstate_method_vactive: + variant_strategy = dml2_pstate_method_fw_vactive_drr; + break; + case dml2_pstate_method_vblank: + variant_strategy = dml2_pstate_method_fw_vblank_drr; + break; + case dml2_pstate_method_fw_svp: + variant_strategy = dml2_pstate_method_fw_svp_drr; + break; + case dml2_pstate_method_fw_vactive_drr: + case dml2_pstate_method_fw_vblank_drr: + case dml2_pstate_method_fw_svp_drr: + case dml2_pstate_method_fw_drr: + case dml2_pstate_method_reserved_hw: + case dml2_pstate_method_reserved_fw: + case dml2_pstate_method_reserved_fw_drr_clamped: + case dml2_pstate_method_reserved_fw_drr_var: + case dml2_pstate_method_count: + case dml2_pstate_method_na: + default: + /* no variant for this mode */ + variant_strategy = base_strategy; + } + + return variant_strategy; +} + +static struct dml2_pmo_pstate_strategy *get_expanded_strategy_list(struct dml2_pmo_init_data *init_data, int stream_count) +{ + struct dml2_pmo_pstate_strategy *expanded_strategy_list = NULL; + + switch (stream_count) { + case 1: + expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_1_display; + break; + case 2: + expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_2_display; + break; + case 3: + expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_3_display; + break; + case 4: + expanded_strategy_list = init_data->pmo_dcn4.expanded_strategy_list_4_display; + break; + default: + break; + } + + return expanded_strategy_list; +} + +static unsigned int get_num_expanded_strategies( + struct dml2_pmo_init_data *init_data, + int stream_count) +{ + return init_data->pmo_dcn4.num_expanded_strategies_per_list[stream_count - 1]; +} + +static void insert_strategy_into_expanded_list( + const struct dml2_pmo_pstate_strategy *per_stream_pstate_strategy, + const int stream_count, + struct dml2_pmo_pstate_strategy *expanded_strategy_list, + unsigned int *num_expanded_strategies) +{ + if (expanded_strategy_list && num_expanded_strategies) { + memcpy(&expanded_strategy_list[*num_expanded_strategies], per_stream_pstate_strategy, sizeof(struct dml2_pmo_pstate_strategy)); + + (*num_expanded_strategies)++; + } +} + +static void expand_base_strategy( + const struct dml2_pmo_pstate_strategy *base_strategy, + const unsigned int stream_count, + struct dml2_pmo_pstate_strategy *expanded_strategy_list, + unsigned int *num_expanded_strategies) +{ + bool skip_to_next_stream; + bool expanded_strategy_added; + bool skip_iteration; + unsigned int i, j; + unsigned int num_streams_per_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; + unsigned int stream_iteration_indices[PMO_DCN4_MAX_DISPLAYS] = { 0 }; + struct dml2_pmo_pstate_strategy cur_strategy_list = { 0 }; + + /* determine number of displays per method */ + for (i = 0; i < stream_count; i++) { + /* increment the count of the earliest index with the same method */ + for (j = 0; j < stream_count; j++) { + if (base_strategy->per_stream_pstate_method[i] == base_strategy->per_stream_pstate_method[j]) { + num_streams_per_method[j] = num_streams_per_method[j] + 1; + break; + } + } + } + + cur_strategy_list.allow_state_increase = base_strategy->allow_state_increase; + + i = 0; + /* uses a while loop instead of recursion to build permutations of base strategy */ + while (stream_iteration_indices[0] < stream_count) { + skip_to_next_stream = false; + expanded_strategy_added = false; + skip_iteration = false; + + /* determine what to do for this iteration */ + if (stream_iteration_indices[i] < stream_count && num_streams_per_method[stream_iteration_indices[i]] != 0) { + /* decrement count and assign method */ + cur_strategy_list.per_stream_pstate_method[i] = base_strategy->per_stream_pstate_method[stream_iteration_indices[i]]; + num_streams_per_method[stream_iteration_indices[i]] -= 1; + + if (i >= stream_count - 1) { + /* insert into strategy list */ + insert_strategy_into_expanded_list(&cur_strategy_list, stream_count, expanded_strategy_list, num_expanded_strategies); + expanded_strategy_added = true; + } else { + /* skip to next stream */ + skip_to_next_stream = true; + } + } else { + skip_iteration = true; + } + + /* prepare for next iteration */ + if (skip_to_next_stream) { + i++; + } else { + /* restore count */ + if (!skip_iteration) { + num_streams_per_method[stream_iteration_indices[i]] += 1; + } + + /* increment iteration count */ + stream_iteration_indices[i]++; + + /* if iterations are complete, or last stream was reached */ + if ((stream_iteration_indices[i] >= stream_count || expanded_strategy_added) && i > 0) { + /* reset per stream index, decrement i */ + stream_iteration_indices[i] = 0; + i--; + + /* restore previous stream's count and increment index */ + num_streams_per_method[stream_iteration_indices[i]] += 1; + stream_iteration_indices[i]++; + } + } + } +} + + +static bool is_variant_method_valid(const struct dml2_pmo_pstate_strategy *base_strategy, + const struct dml2_pmo_pstate_strategy *variant_strategy, + const unsigned int num_streams_per_base_method[PMO_DCN4_MAX_DISPLAYS], + const unsigned int num_streams_per_variant_method[PMO_DCN4_MAX_DISPLAYS], + const unsigned int stream_count) +{ + bool valid = true; + unsigned int i; + + /* check all restrictions are met */ + for (i = 0; i < stream_count; i++) { + /* vblank + vblank_drr variants are invalid */ + if (base_strategy->per_stream_pstate_method[i] == dml2_pstate_method_vblank && + ((num_streams_per_base_method[i] > 0 && num_streams_per_variant_method[i] > 0) || + num_streams_per_variant_method[i] > 1)) { + valid = false; + break; + } + } + + return valid; +} + +static void expand_variant_strategy( + const struct dml2_pmo_pstate_strategy *base_strategy, + const unsigned int stream_count, + const bool should_permute, + struct dml2_pmo_pstate_strategy *expanded_strategy_list, + unsigned int *num_expanded_strategies) +{ + bool variant_found; + unsigned int i, j; + unsigned int method_index; + unsigned int stream_index; + unsigned int num_streams_per_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; + unsigned int num_streams_per_base_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; + unsigned int num_streams_per_variant_method[PMO_DCN4_MAX_DISPLAYS] = { 0 }; + enum dml2_pstate_method per_stream_variant_method[DML2_MAX_PLANES]; + struct dml2_pmo_pstate_strategy variant_strategy = { 0 }; + + /* determine number of displays per method */ + for (i = 0; i < stream_count; i++) { + /* increment the count of the earliest index with the same method */ + for (j = 0; j < stream_count; j++) { + if (base_strategy->per_stream_pstate_method[i] == base_strategy->per_stream_pstate_method[j]) { + num_streams_per_method[j] = num_streams_per_method[j] + 1; + break; + } + } + + per_stream_variant_method[i] = convert_strategy_to_drr_variant(base_strategy->per_stream_pstate_method[i]); + } + memcpy(num_streams_per_base_method, num_streams_per_method, sizeof(unsigned int) * PMO_DCN4_MAX_DISPLAYS); + + memcpy(&variant_strategy, base_strategy, sizeof(struct dml2_pmo_pstate_strategy)); + + method_index = 0; + /* uses a while loop instead of recursion to build permutations of base strategy */ + while (num_streams_per_base_method[0] > 0 || method_index != 0) { + if (method_index == stream_count) { + /* construct variant strategy */ + variant_found = false; + stream_index = 0; + + for (i = 0; i < stream_count; i++) { + for (j = 0; j < num_streams_per_base_method[i]; j++) { + variant_strategy.per_stream_pstate_method[stream_index++] = base_strategy->per_stream_pstate_method[i]; + } + + for (j = 0; j < num_streams_per_variant_method[i]; j++) { + variant_strategy.per_stream_pstate_method[stream_index++] = per_stream_variant_method[i]; + if (base_strategy->per_stream_pstate_method[i] != per_stream_variant_method[i]) { + variant_found = true; + } + } + } + + if (variant_found && is_variant_method_valid(base_strategy, &variant_strategy, num_streams_per_base_method, num_streams_per_variant_method, stream_count)) { + if (should_permute) { + /* permutations are permitted, proceed to expand */ + expand_base_strategy(&variant_strategy, stream_count, expanded_strategy_list, num_expanded_strategies); + } else { + /* no permutations allowed, so add to list now */ + insert_strategy_into_expanded_list(&variant_strategy, stream_count, expanded_strategy_list, num_expanded_strategies); + } + } + + /* rollback to earliest method with bases remaining */ + for (method_index = stream_count - 1; method_index > 0; method_index--) { + if (num_streams_per_base_method[method_index]) { + /* bases remaining */ + break; + } else { + /* reset counters */ + num_streams_per_base_method[method_index] = num_streams_per_method[method_index]; + num_streams_per_variant_method[method_index] = 0; + } + } + } + + if (num_streams_per_base_method[method_index]) { + num_streams_per_base_method[method_index]--; + num_streams_per_variant_method[method_index]++; + + method_index++; + } else if (method_index != 0) { + method_index++; + } + } +} + +void pmo_dcn4_fams2_expand_base_pstate_strategies( + const struct dml2_pmo_pstate_strategy *base_strategies_list, + const unsigned int num_base_strategies, + const unsigned int stream_count, + struct dml2_pmo_pstate_strategy *expanded_strategy_list, + unsigned int *num_expanded_strategies) +{ + unsigned int i; + + /* expand every explicit base strategy (except all DRR) */ + for (i = 0; i < num_base_strategies; i++) { + expand_base_strategy(&base_strategies_list[i], stream_count, expanded_strategy_list, num_expanded_strategies); + expand_variant_strategy(&base_strategies_list[i], stream_count, true, expanded_strategy_list, num_expanded_strategies); + } +} + +bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) +{ + int i = 0; + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int base_list_size = 0; + const struct dml2_pmo_pstate_strategy *base_list = NULL; + unsigned int *expanded_list_size = NULL; + struct dml2_pmo_pstate_strategy *expanded_list = NULL; + + pmo->soc_bb = in_out->soc_bb; + pmo->ip_caps = in_out->ip_caps; + pmo->mpc_combine_limit = 2; + pmo->odm_combine_limit = 4; + pmo->mcg_clock_table_size = in_out->mcg_clock_table_size; + + pmo->fams_params.v2.subvp.refresh_rate_limit_max = 175; + pmo->fams_params.v2.subvp.refresh_rate_limit_min = 0; + pmo->fams_params.v2.drr.refresh_rate_limit_max = 1000; + pmo->fams_params.v2.drr.refresh_rate_limit_min = 119; + + pmo->options = in_out->options; + + /* generate permutations of p-state configs from base strategy list */ + for (i = 0; i < PMO_DCN4_MAX_DISPLAYS; i++) { + switch (i+1) { + case 1: + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_1_display; + base_list_size = base_strategy_list_1_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_1_display; + + break; + case 2: + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_2_display; + base_list_size = base_strategy_list_2_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_2_display; + + break; + case 3: + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_3_display; + base_list_size = base_strategy_list_3_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_3_display; + + break; + case 4: + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_4_display; + base_list_size = base_strategy_list_4_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_4_display; + + break; + } + + DML_ASSERT(base_list_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + + /* populate list */ + pmo_dcn4_fams2_expand_base_pstate_strategies( + base_list, + base_list_size, + i + 1, + expanded_list, + expanded_list_size); + } + + return true; +} + +static bool is_h_timing_divisible_by(const struct dml2_timing_cfg *timing, unsigned char denominator) +{ + /* + * Htotal, Hblank start/end, and Hsync start/end all must be divisible + * in order for the horizontal timing params to be considered divisible + * by 2. Hsync start is always 0. + */ + unsigned long h_blank_start = timing->h_total - timing->h_front_porch; + + return (timing->h_total % denominator == 0) && + (h_blank_start % denominator == 0) && + (timing->h_blank_end % denominator == 0) && + (timing->h_sync_width % denominator == 0); +} + +static bool is_dp_encoder(enum dml2_output_encoder_class encoder_type) +{ + switch (encoder_type) { + case dml2_dp: + case dml2_edp: + case dml2_dp2p0: + case dml2_none: + return true; + case dml2_hdmi: + case dml2_hdmifrl: + default: + return false; + } +} + +bool pmo_dcn4_fams2_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out) +{ + unsigned int i; + const struct dml2_display_cfg *display_config = + &in_out->base_display_config->display_config; + const struct dml2_core_mode_support_result *mode_support_result = + &in_out->base_display_config->mode_support_result; + struct dml2_optimization_stage4_state *state = + &in_out->base_display_config->stage4; + + if (in_out->instance->options->disable_dyn_odm || + (in_out->instance->options->disable_dyn_odm_for_multi_stream && display_config->num_streams > 1)) + return false; + + for (i = 0; i < display_config->num_planes; i++) + /* + * vmin optimization is required to be seamlessly switched off + * at any time when the new configuration is no longer + * supported. However switching from ODM combine to MPC combine + * is not always seamless. When there not enough free pipes, we + * will have to use the same secondary OPP heads as secondary + * DPP pipes in MPC combine in new state. This transition is + * expected to cause glitches. To avoid the transition, we only + * allow vmin optimization if the stream's base configuration + * doesn't require MPC combine. This condition checks if MPC + * combine is enabled. If so do not optimize the stream. + */ + if (mode_support_result->cfg_support_info.plane_support_info[i].dpps_used > 1 && + mode_support_result->cfg_support_info.stream_support_info[display_config->plane_descriptors[i].stream_index].odms_used == 1) + state->unoptimizable_streams[display_config->plane_descriptors[i].stream_index] = true; + + for (i = 0; i < display_config->num_streams; i++) { + if (display_config->stream_descriptors[i].overrides.disable_dynamic_odm) + state->unoptimizable_streams[i] = true; + else if (in_out->base_display_config->stage3.stream_svp_meta[i].valid && + in_out->instance->options->disable_dyn_odm_for_stream_with_svp) + state->unoptimizable_streams[i] = true; + /* + * ODM Combine requires horizontal timing divisible by 2 so each + * ODM segment has the same size. + */ + else if (!is_h_timing_divisible_by(&display_config->stream_descriptors[i].timing, 2)) + state->unoptimizable_streams[i] = true; + /* + * Our hardware support seamless ODM transitions for DP encoders + * only. + */ + else if (!is_dp_encoder(display_config->stream_descriptors[i].output.output_encoder)) + state->unoptimizable_streams[i] = true; + } + + state->performed = true; + + return true; +} + +bool pmo_dcn4_fams2_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out) +{ + bool is_vmin = true; + + if (in_out->vmin_limits->dispclk_khz > 0 && + in_out->display_config->mode_support_result.global.dispclk_khz > in_out->vmin_limits->dispclk_khz) + is_vmin = false; + + return is_vmin; +} + +static int find_highest_odm_load_stream_index( + const struct dml2_display_cfg *display_config, + const struct dml2_core_mode_support_result *mode_support_result) +{ + unsigned int i; + int odm_load, highest_odm_load = -1, highest_odm_load_index = -1; + + for (i = 0; i < display_config->num_streams; i++) { + if (mode_support_result->cfg_support_info.stream_support_info[i].odms_used > 0) + odm_load = display_config->stream_descriptors[i].timing.pixel_clock_khz + / mode_support_result->cfg_support_info.stream_support_info[i].odms_used; + else + odm_load = 0; + + if (odm_load > highest_odm_load) { + highest_odm_load_index = i; + highest_odm_load = odm_load; + } + } + + return highest_odm_load_index; +} + +bool pmo_dcn4_fams2_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out) +{ + int stream_index; + const struct dml2_display_cfg *display_config = + &in_out->base_display_config->display_config; + const struct dml2_core_mode_support_result *mode_support_result = + &in_out->base_display_config->mode_support_result; + unsigned int odms_used; + struct dml2_stream_parameters *stream_descriptor; + bool optimizable = false; + + /* + * highest odm load stream must be optimizable to continue as dispclk is + * bounded by it. + */ + stream_index = find_highest_odm_load_stream_index(display_config, + mode_support_result); + + if (stream_index < 0 || + in_out->base_display_config->stage4.unoptimizable_streams[stream_index]) + return false; + + odms_used = mode_support_result->cfg_support_info.stream_support_info[stream_index].odms_used; + if ((int)odms_used >= in_out->instance->odm_combine_limit) + return false; + + memcpy(in_out->optimized_display_config, + in_out->base_display_config, + sizeof(struct display_configuation_with_meta)); + + stream_descriptor = &in_out->optimized_display_config->display_config.stream_descriptors[stream_index]; + while (!optimizable && increase_odm_combine_factor( + &stream_descriptor->overrides.odm_mode, + odms_used)) { + switch (stream_descriptor->overrides.odm_mode) { + case dml2_odm_mode_combine_2to1: + optimizable = true; + break; + case dml2_odm_mode_combine_3to1: + /* + * In ODM Combine 3:1 OTG_valid_pixel rate is 1/4 of + * actual pixel rate. Therefore horizontal timing must + * be divisible by 4. + */ + if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { + /* + * DSC h slice count must be divisible + * by 3. + */ + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 3 == 0) + optimizable = true; + } else { + optimizable = true; + } + } + break; + case dml2_odm_mode_combine_4to1: + /* + * In ODM Combine 4:1 OTG_valid_pixel rate is 1/4 of + * actual pixel rate. Therefore horizontal timing must + * be divisible by 4. + */ + if (is_h_timing_divisible_by(&display_config->stream_descriptors[stream_index].timing, 4)) { + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].dsc_enable) { + /* + * DSC h slice count must be divisible + * by 4. + */ + if (mode_support_result->cfg_support_info.stream_support_info[stream_index].num_dsc_slices % 4 == 0) + optimizable = true; + } else { + optimizable = true; + } + } + break; + case dml2_odm_mode_auto: + case dml2_odm_mode_bypass: + case dml2_odm_mode_split_1to2: + case dml2_odm_mode_mso_1to2: + case dml2_odm_mode_mso_1to4: + default: + break; + } + } + + return optimizable; +} + +static void set_bit_in_bitfield(unsigned int *bit_field, unsigned int bit_offset) +{ + *bit_field = *bit_field | (0x1 << bit_offset); +} + +static bool is_bit_set_in_bitfield(unsigned int bit_field, unsigned int bit_offset) +{ + if (bit_field & (0x1 << bit_offset)) + return true; + + return false; +} + +static void build_synchronized_timing_groups( + struct dml2_pmo_instance *pmo, + struct display_configuation_with_meta *display_config) +{ + unsigned int i, j; + struct dml2_timing_cfg *master_timing; + + unsigned int stream_mapped_mask = 0; + unsigned int num_timing_groups = 0; + unsigned int timing_group_idx = 0; + struct dml2_pmo_scratch *s = &pmo->scratch; + + /* clear all group masks */ + memset(s->pmo_dcn4.synchronized_timing_group_masks, 0, sizeof(s->pmo_dcn4.synchronized_timing_group_masks)); + memset(s->pmo_dcn4.group_is_drr_enabled, 0, sizeof(s->pmo_dcn4.group_is_drr_enabled)); + memset(s->pmo_dcn4.group_is_drr_active, 0, sizeof(s->pmo_dcn4.group_is_drr_active)); + memset(s->pmo_dcn4.group_line_time_us, 0, sizeof(s->pmo_dcn4.group_line_time_us)); + s->pmo_dcn4.num_timing_groups = 0; + + for (i = 0; i < display_config->display_config.num_streams; i++) { + master_timing = &display_config->display_config.stream_descriptors[i].timing; + + /* only need to build group of this stream is not in a group already */ + if (is_bit_set_in_bitfield(stream_mapped_mask, i)) { + continue; + } + set_bit_in_bitfield(&stream_mapped_mask, i); + timing_group_idx = num_timing_groups; + num_timing_groups++; + + /* trivially set default timing group to itself */ + set_bit_in_bitfield(&s->pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], i); + s->pmo_dcn4.group_line_time_us[timing_group_idx] = (double)master_timing->h_total / master_timing->pixel_clock_khz * 1000.0; + + /* if drr is in use, timing is not sychnronizable */ + if (master_timing->drr_config.enabled) { + s->pmo_dcn4.group_is_drr_enabled[timing_group_idx] = true; + s->pmo_dcn4.group_is_drr_active[timing_group_idx] = !master_timing->drr_config.disallowed && + (master_timing->drr_config.drr_active_fixed || master_timing->drr_config.drr_active_variable); + continue; + } + + /* find synchronizable timing groups */ + for (j = i + 1; j < display_config->display_config.num_streams; j++) { + if (memcmp(master_timing, + &display_config->display_config.stream_descriptors[j].timing, + sizeof(struct dml2_timing_cfg)) == 0) { + set_bit_in_bitfield(&pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], j); + set_bit_in_bitfield(&stream_mapped_mask, j); + } + } + } + + s->pmo_dcn4.num_timing_groups = num_timing_groups; +} + +static bool all_timings_support_vactive(const struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_config, + unsigned int mask) +{ + unsigned int i; + bool valid = true; + + // Create a remap array to enable simple iteration through only masked stream indicies + for (i = 0; i < display_config->display_config.num_streams; i++) { + if (is_bit_set_in_bitfield(mask, i)) { + /* check if stream has enough vactive margin */ + valid &= is_bit_set_in_bitfield(pmo->scratch.pmo_dcn4.stream_vactive_capability_mask, i); + } + } + + return valid; +} + +static bool all_timings_support_vblank(const struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_config, + unsigned int mask) +{ + unsigned int i; + + bool synchronizable = true; + + /* find first vblank stream index and compare the timing group mask */ + for (i = 0; i < display_config->display_config.num_streams; i++) { + if (is_bit_set_in_bitfield(mask, i)) { + if (mask != pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[i]) { + /* vblank streams are not synchronizable */ + synchronizable = false; + } + break; + } + } + + return synchronizable; +} + +static unsigned int calc_svp_microschedule(const struct dml2_pstate_meta *pstate_meta) +{ + return pstate_meta->contention_delay_otg_vlines + + pstate_meta->method_subvp.programming_delay_otg_vlines + + pstate_meta->method_subvp.phantom_vtotal + + pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + + pstate_meta->blackout_otg_vlines; +} + +static bool all_timings_support_drr(const struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_config, + unsigned int mask) +{ + unsigned int i; + for (i = 0; i < DML2_MAX_PLANES; i++) { + const struct dml2_stream_parameters *stream_descriptor; + const struct dml2_pstate_meta *stream_pstate_meta; + + if (is_bit_set_in_bitfield(mask, i)) { + stream_descriptor = &display_config->display_config.stream_descriptors[i]; + stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; + + if (!stream_descriptor->timing.drr_config.enabled) + return false; + + /* cannot support required vtotal */ + if (stream_pstate_meta->method_drr.stretched_vtotal > stream_pstate_meta->max_vtotal) { + return false; + } + + /* check rr is within bounds */ + if (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.drr.refresh_rate_limit_min || + stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.drr.refresh_rate_limit_max) { + return false; + } + + /* check required stretch is allowed */ + if (stream_descriptor->timing.drr_config.max_instant_vtotal_delta > 0 && + stream_pstate_meta->method_drr.stretched_vtotal - stream_pstate_meta->nom_vtotal > stream_descriptor->timing.drr_config.max_instant_vtotal_delta) { + return false; + } + } + } + + return true; +} + +static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_config, + unsigned int mask) +{ + const struct dml2_stream_parameters *stream_descriptor; + const struct dml2_plane_parameters *plane_descriptor; + const struct dml2_pstate_meta *stream_pstate_meta; + unsigned int microschedule_vlines; + unsigned int i; + unsigned int mcaches_per_plane; + unsigned int total_mcaches_required = 0; + + unsigned int num_planes_per_stream[DML2_MAX_PLANES] = { 0 }; + + /* confirm timing it is not a centered timing */ + for (i = 0; i < display_config->display_config.num_planes; i++) { + plane_descriptor = &display_config->display_config.plane_descriptors[i]; + mcaches_per_plane = 0; + + if (plane_descriptor->surface.dcc.enable) { + mcaches_per_plane += display_config->stage2.mcache_allocations[i].num_mcaches_plane0 + + display_config->stage2.mcache_allocations[i].num_mcaches_plane1 - + (display_config->stage2.mcache_allocations[i].last_slice_sharing.plane0_plane1 ? 1 : 0); + } + + if (is_bit_set_in_bitfield(mask, (unsigned char)plane_descriptor->stream_index)) { + num_planes_per_stream[plane_descriptor->stream_index]++; + + /* check recout height covers entire otg vactive, and single plane */ + if (num_planes_per_stream[plane_descriptor->stream_index] > 1 || + !plane_descriptor->composition.rect_out_height_spans_vactive || + plane_descriptor->composition.rotation_angle != dml2_rotation_0) { + return false; + } + + /* phantom requires same number of mcaches as main */ + if (plane_descriptor->surface.dcc.enable) { + mcaches_per_plane *= 2; + } + } + total_mcaches_required += mcaches_per_plane; + } + + if (total_mcaches_required > pmo->soc_bb->num_dcc_mcaches) { + /* too many mcaches required */ + return false; + } + + for (i = 0; i < DML2_MAX_PLANES; i++) { + if (is_bit_set_in_bitfield(mask, i)) { + stream_descriptor = &display_config->display_config.stream_descriptors[i]; + stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; + + if (stream_descriptor->overrides.disable_subvp) { + return false; + } + + microschedule_vlines = calc_svp_microschedule(&pmo->scratch.pmo_dcn4.stream_pstate_meta[i]); + + /* block if using an interlaced timing */ + if (stream_descriptor->timing.interlaced) { + return false; + } + + /* 1) svp main stream's vactive must be able to fit the microschedule + * 2) refresh rate must be within the allowed bounds + */ + if (microschedule_vlines >= stream_descriptor->timing.v_active || + (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.subvp.refresh_rate_limit_min || + stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.subvp.refresh_rate_limit_max)) { + return false; + } + } + } + + return true; +} + +static void insert_into_candidate_list(const struct dml2_pmo_pstate_strategy *pstate_strategy, int stream_count, struct dml2_pmo_scratch *scratch) +{ + scratch->pmo_dcn4.pstate_strategy_candidates[scratch->pmo_dcn4.num_pstate_candidates] = *pstate_strategy; + scratch->pmo_dcn4.num_pstate_candidates++; +} + +static enum dml2_pstate_method uclk_pstate_strategy_override_to_pstate_method(const enum dml2_uclk_pstate_change_strategy override_strategy) +{ + enum dml2_pstate_method method = dml2_pstate_method_na; + + switch (override_strategy) { + case dml2_uclk_pstate_change_strategy_force_vactive: + method = dml2_pstate_method_vactive; + break; + case dml2_uclk_pstate_change_strategy_force_vblank: + method = dml2_pstate_method_vblank; + break; + case dml2_uclk_pstate_change_strategy_force_drr: + method = dml2_pstate_method_fw_drr; + break; + case dml2_uclk_pstate_change_strategy_force_mall_svp: + method = dml2_pstate_method_fw_svp; + break; + case dml2_uclk_pstate_change_strategy_force_mall_full_frame: + case dml2_uclk_pstate_change_strategy_auto: + default: + method = dml2_pstate_method_na; + } + + return method; +} + +static enum dml2_uclk_pstate_change_strategy pstate_method_to_uclk_pstate_strategy_override(const enum dml2_pstate_method method) +{ + enum dml2_uclk_pstate_change_strategy override_strategy = dml2_uclk_pstate_change_strategy_auto; + + switch (method) { + case dml2_pstate_method_vactive: + case dml2_pstate_method_fw_vactive_drr: + override_strategy = dml2_uclk_pstate_change_strategy_force_vactive; + break; + case dml2_pstate_method_vblank: + case dml2_pstate_method_fw_vblank_drr: + override_strategy = dml2_uclk_pstate_change_strategy_force_vblank; + break; + case dml2_pstate_method_fw_svp: + case dml2_pstate_method_fw_svp_drr: + override_strategy = dml2_uclk_pstate_change_strategy_force_mall_svp; + break; + case dml2_pstate_method_fw_drr: + override_strategy = dml2_uclk_pstate_change_strategy_force_drr; + break; + case dml2_pstate_method_reserved_hw: + case dml2_pstate_method_reserved_fw: + case dml2_pstate_method_reserved_fw_drr_clamped: + case dml2_pstate_method_reserved_fw_drr_var: + case dml2_pstate_method_count: + case dml2_pstate_method_na: + default: + override_strategy = dml2_uclk_pstate_change_strategy_auto; + } + + return override_strategy; +} + +static bool all_planes_match_method(const struct display_configuation_with_meta *display_cfg, int plane_mask, enum dml2_pstate_method method) +{ + unsigned int i; + + for (i = 0; i < DML2_MAX_PLANES; i++) { + if (is_bit_set_in_bitfield(plane_mask, i)) { + if (display_cfg->display_config.plane_descriptors[i].overrides.uclk_pstate_change_strategy != dml2_uclk_pstate_change_strategy_auto && + display_cfg->display_config.plane_descriptors[i].overrides.uclk_pstate_change_strategy != pstate_method_to_uclk_pstate_strategy_override(method)) + return false; + } + } + + return true; +} + +static void build_method_scheduling_params( + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta, + struct dml2_pstate_meta *stream_pstate_meta) +{ + stream_method_pstate_meta->allow_time_us = + (double)((int)stream_method_pstate_meta->allow_end_otg_vline - (int)stream_method_pstate_meta->allow_start_otg_vline) * + stream_pstate_meta->otg_vline_time_us; + if (stream_method_pstate_meta->allow_time_us >= stream_method_pstate_meta->period_us) { + /* when allow wave overlaps an entire frame, it is always schedulable (DRR can do this)*/ + stream_method_pstate_meta->disallow_time_us = 0.0; + } else { + stream_method_pstate_meta->disallow_time_us = + stream_method_pstate_meta->period_us - stream_method_pstate_meta->allow_time_us; + } +} + +static struct dml2_pstate_per_method_common_meta *get_per_method_common_meta( + struct dml2_pmo_instance *pmo, + enum dml2_pstate_method stream_pstate_method, + int stream_idx) +{ + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta = NULL; + + switch (stream_pstate_method) { + case dml2_pstate_method_vactive: + case dml2_pstate_method_fw_vactive_drr: + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vactive.common; + break; + case dml2_pstate_method_vblank: + case dml2_pstate_method_fw_vblank_drr: + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vblank.common; + break; + case dml2_pstate_method_fw_svp: + case dml2_pstate_method_fw_svp_drr: + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_subvp.common; + break; + case dml2_pstate_method_fw_drr: + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_drr.common; + break; + case dml2_pstate_method_reserved_hw: + case dml2_pstate_method_reserved_fw: + case dml2_pstate_method_reserved_fw_drr_clamped: + case dml2_pstate_method_reserved_fw_drr_var: + case dml2_pstate_method_count: + case dml2_pstate_method_na: + default: + stream_method_pstate_meta = NULL; + } + + return stream_method_pstate_meta; +} + +static bool is_timing_group_schedulable( + struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_cfg, + const struct dml2_pmo_pstate_strategy *pstate_strategy, + const unsigned int timing_group_idx, + struct dml2_pstate_per_method_common_meta *group_pstate_meta) +{ + unsigned int i; + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta; + + unsigned int base_stream_idx = 0; + struct dml2_pmo_scratch *s = &pmo->scratch; + + /* find base stream idx */ + for (base_stream_idx = 0; base_stream_idx < display_cfg->display_config.num_streams; base_stream_idx++) { + if (is_bit_set_in_bitfield(s->pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], base_stream_idx)) { + /* master stream found */ + break; + } + } + + /* init allow start and end lines for timing group */ + stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[base_stream_idx], base_stream_idx); + if (!stream_method_pstate_meta) + return false; + + group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; + group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; + group_pstate_meta->period_us = stream_method_pstate_meta->period_us; + for (i = base_stream_idx + 1; i < display_cfg->display_config.num_streams; i++) { + if (is_bit_set_in_bitfield(pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], i)) { + stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[i], i); + if (!stream_method_pstate_meta) + continue; + + if (group_pstate_meta->allow_start_otg_vline < stream_method_pstate_meta->allow_start_otg_vline) { + /* set group allow start to larger otg vline */ + group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; + } + + if (group_pstate_meta->allow_end_otg_vline > stream_method_pstate_meta->allow_end_otg_vline) { + /* set group allow end to smaller otg vline */ + group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; + } + + /* check waveform still has positive width */ + if (group_pstate_meta->allow_start_otg_vline >= group_pstate_meta->allow_end_otg_vline) { + /* timing group is not schedulable */ + return false; + } + } + } + + /* calculate the rest of the meta */ + build_method_scheduling_params(group_pstate_meta, &pmo->scratch.pmo_dcn4.stream_pstate_meta[base_stream_idx]); + + return group_pstate_meta->allow_time_us > 0.0 && + group_pstate_meta->disallow_time_us < pmo->ip_caps->fams2.max_allow_delay_us; +} + +static bool is_config_schedulable( + struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_cfg, + const struct dml2_pmo_pstate_strategy *pstate_strategy) +{ + unsigned int i, j; + bool schedulable; + struct dml2_pmo_scratch *s = &pmo->scratch; + + double max_allow_delay_us = 0.0; + + memset(s->pmo_dcn4.group_common_pstate_meta, 0, sizeof(s->pmo_dcn4.group_common_pstate_meta)); + memset(s->pmo_dcn4.sorted_group_gtl_disallow_index, 0, sizeof(unsigned int) * DML2_MAX_PLANES); + + /* search for a general solution to the schedule */ + + /* STAGE 0: Early return for special cases */ + if (display_cfg->display_config.num_streams == 0) { + return true; + } + + /* STAGE 1: confirm allow waves overlap for synchronizable streams */ + schedulable = true; + for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { + s->pmo_dcn4.sorted_group_gtl_disallow_index[i] = i; + s->pmo_dcn4.sorted_group_gtl_period_index[i] = i; + if (!is_timing_group_schedulable(pmo, display_cfg, pstate_strategy, i, &s->pmo_dcn4.group_common_pstate_meta[i])) { + /* synchronized timing group was not schedulable */ + schedulable = false; + break; + } + max_allow_delay_us += s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us; + } + + if ((schedulable && s->pmo_dcn4.num_timing_groups <= 1) || !schedulable) { + /* 1. the only timing group was schedulable, so early pass + * 2. one of the timing groups was not schedulable, so early fail */ + return schedulable; + } + + /* STAGE 2: Check allow can't be masked entirely by other disallows */ + schedulable = true; + + /* sort disallow times from greatest to least */ + for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { + bool swapped = false; + + for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { + double j_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j]].disallow_time_us; + double jp1_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]].disallow_time_us; + if (j_disallow_us < jp1_disallow_us) { + /* swap as A < B */ + swap(s->pmo_dcn4.sorted_group_gtl_disallow_index[j], + s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]); + swapped = true; + } + } + + /* sorted, exit early */ + if (!swapped) + break; + } + + /* Check worst case disallow region occurs in the middle of allow for the + * other display, or when >2 streams continue to halve the remaining allow time. + */ + for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { + if (s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us <= 0.0) { + /* this timing group always allows */ + continue; + } + + double max_allow_time_us = s->pmo_dcn4.group_common_pstate_meta[i].allow_time_us; + for (j = 0; j < s->pmo_dcn4.num_timing_groups; j++) { + unsigned int sorted_j = s->pmo_dcn4.sorted_group_gtl_disallow_index[j]; + /* stream can't overlap itself */ + if (i != sorted_j && s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us > 0.0) { + max_allow_time_us = math_min2( + s->pmo_dcn4.group_common_pstate_meta[sorted_j].allow_time_us, + (max_allow_time_us - s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us) / 2); + + if (max_allow_time_us < 0.0) { + /* failed exit early */ + break; + } + } + } + + if (max_allow_time_us <= 0.0) { + /* not enough time for microschedule in the worst case */ + schedulable = false; + break; + } + } + + if (schedulable && max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { + return true; + } + + /* STAGE 3: check larger allow can fit period of all other streams */ + schedulable = true; + + /* sort periods from greatest to least */ + for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { + bool swapped = false; + + for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { + double j_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j]].period_us; + double jp1_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]].period_us; + if (j_period_us < jp1_period_us) { + /* swap as A < B */ + swap(s->pmo_dcn4.sorted_group_gtl_period_index[j], + s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]); + swapped = true; + } + } + + /* sorted, exit early */ + if (!swapped) + break; + } + + /* check larger allow can fit period of all other streams */ + for (i = 0; i < s->pmo_dcn4.num_timing_groups - 1; i++) { + unsigned int sorted_i = s->pmo_dcn4.sorted_group_gtl_period_index[i]; + unsigned int sorted_ip1 = s->pmo_dcn4.sorted_group_gtl_period_index[i + 1]; + + if (s->pmo_dcn4.group_common_pstate_meta[sorted_i].allow_time_us < s->pmo_dcn4.group_common_pstate_meta[sorted_ip1].period_us || + (s->pmo_dcn4.group_is_drr_enabled[sorted_ip1] && s->pmo_dcn4.group_is_drr_active[sorted_ip1])) { + schedulable = false; + break; + } + } + + if (schedulable && max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { + return true; + } + + /* STAGE 4: When using HW exclusive modes, check disallow alignments are within allowed threshold */ + if (s->pmo_dcn4.num_timing_groups == 2 && + !is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, pstate_strategy->per_stream_pstate_method[0]) && + !is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, pstate_strategy->per_stream_pstate_method[1])) { + double period_ratio; + double max_shift_us; + double shift_per_period; + + /* default period_0 > period_1 */ + unsigned int lrg_idx = 0; + unsigned int sml_idx = 1; + if (s->pmo_dcn4.group_common_pstate_meta[0].period_us < s->pmo_dcn4.group_common_pstate_meta[1].period_us) { + /* period_0 < period_1 */ + lrg_idx = 1; + sml_idx = 0; + } + period_ratio = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us / s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us; + shift_per_period = s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us * (period_ratio - math_floor(period_ratio)); + max_shift_us = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].disallow_time_us - s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us; + max_allow_delay_us = max_shift_us / shift_per_period * s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us; + + if (shift_per_period > 0.0 && + shift_per_period < s->pmo_dcn4.group_common_pstate_meta[lrg_idx].allow_time_us + s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us && + max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { + schedulable = true; + } + } + + return schedulable; +} + +static bool stream_matches_drr_policy(struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_cfg, + const enum dml2_pstate_method stream_pstate_method, + unsigned int stream_index) +{ + const struct dml2_stream_parameters *stream_descriptor = &display_cfg->display_config.stream_descriptors[stream_index]; + bool strategy_matches_drr_requirements = true; + + /* check if strategy is compatible with stream drr capability and strategy */ + if (is_bit_set_in_bitfield(PMO_NO_DRR_STRATEGY_MASK, stream_pstate_method) && + display_cfg->display_config.num_streams > 1 && + stream_descriptor->timing.drr_config.enabled && + (stream_descriptor->timing.drr_config.drr_active_fixed || stream_descriptor->timing.drr_config.drr_active_variable)) { + /* DRR is active, so config may become unschedulable */ + strategy_matches_drr_requirements = false; + } else if (is_bit_set_in_bitfield(PMO_NO_DRR_STRATEGY_MASK, stream_pstate_method) && + is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, stream_pstate_method) && + stream_descriptor->timing.drr_config.enabled && + stream_descriptor->timing.drr_config.drr_active_variable) { + /* DRR is variable, fw exclusive methods require DRR to be clamped */ + strategy_matches_drr_requirements = false; + } else if (is_bit_set_in_bitfield(PMO_DRR_VAR_STRATEGY_MASK, stream_pstate_method) && + pmo->options->disable_drr_var_when_var_active && + stream_descriptor->timing.drr_config.enabled && + stream_descriptor->timing.drr_config.drr_active_variable) { + /* DRR variable is active, but policy blocks DRR for p-state when this happens */ + strategy_matches_drr_requirements = false; + } else if (is_bit_set_in_bitfield(PMO_DRR_VAR_STRATEGY_MASK, stream_pstate_method) && + (pmo->options->disable_drr_var || + !stream_descriptor->timing.drr_config.enabled || + stream_descriptor->timing.drr_config.disallowed)) { + /* DRR variable strategies are disallowed due to settings or policy */ + strategy_matches_drr_requirements = false; + } else if (is_bit_set_in_bitfield(PMO_DRR_CLAMPED_STRATEGY_MASK, stream_pstate_method) && + (pmo->options->disable_drr_clamped || + (!stream_descriptor->timing.drr_config.enabled || + (!stream_descriptor->timing.drr_config.drr_active_fixed && !stream_descriptor->timing.drr_config.drr_active_variable)) || + (pmo->options->disable_drr_clamped_when_var_active && + stream_descriptor->timing.drr_config.enabled && + stream_descriptor->timing.drr_config.drr_active_variable))) { + /* DRR fixed strategies are disallowed due to settings or policy */ + strategy_matches_drr_requirements = false; + } else if (is_bit_set_in_bitfield(PMO_FW_STRATEGY_MASK, stream_pstate_method) && + pmo->options->disable_fams2) { + /* FW modes require FAMS2 */ + strategy_matches_drr_requirements = false; + } + + return strategy_matches_drr_requirements; +} + +static bool validate_pstate_support_strategy_cofunctionality(struct dml2_pmo_instance *pmo, + const struct display_configuation_with_meta *display_cfg, + const struct dml2_pmo_pstate_strategy *pstate_strategy) +{ + struct dml2_pmo_scratch *s = &pmo->scratch; + + unsigned int stream_index = 0; + + unsigned int svp_count = 0; + unsigned int svp_stream_mask = 0; + unsigned int drr_count = 0; + unsigned int drr_stream_mask = 0; + unsigned int vactive_count = 0; + unsigned int vactive_stream_mask = 0; + unsigned int vblank_count = 0; + unsigned int vblank_stream_mask = 0; + + bool strategy_matches_forced_requirements = true; + bool strategy_matches_drr_requirements = true; + + // Tabulate everything + for (stream_index = 0; stream_index < display_cfg->display_config.num_streams; stream_index++) { + + if (!all_planes_match_method(display_cfg, s->pmo_dcn4.stream_plane_mask[stream_index], + pstate_strategy->per_stream_pstate_method[stream_index])) { + strategy_matches_forced_requirements = false; + break; + } + + strategy_matches_drr_requirements &= + stream_matches_drr_policy(pmo, display_cfg, pstate_strategy->per_stream_pstate_method[stream_index], stream_index); + + if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp || + pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { + svp_count++; + set_bit_in_bitfield(&svp_stream_mask, stream_index); + } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { + drr_count++; + set_bit_in_bitfield(&drr_stream_mask, stream_index); + } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive || + pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { + vactive_count++; + set_bit_in_bitfield(&vactive_stream_mask, stream_index); + } else if (pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank || + pstate_strategy->per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { + vblank_count++; + set_bit_in_bitfield(&vblank_stream_mask, stream_index); + } + } + + if (!strategy_matches_forced_requirements || !strategy_matches_drr_requirements) + return false; + + if (vactive_count > 0 && !all_timings_support_vactive(pmo, display_cfg, vactive_stream_mask)) + return false; + + if (vblank_count > 0 && (pmo->options->disable_vblank || !all_timings_support_vblank(pmo, display_cfg, vblank_stream_mask))) + return false; + + if (drr_count > 0 && (pmo->options->disable_drr_var || !all_timings_support_drr(pmo, display_cfg, drr_stream_mask))) + return false; + + if (svp_count > 0 && (pmo->options->disable_svp || !all_timings_support_svp(pmo, display_cfg, svp_stream_mask))) + return false; + + return is_config_schedulable(pmo, display_cfg, pstate_strategy); +} + +static int get_vactive_pstate_margin(const struct display_configuation_with_meta *display_cfg, int plane_mask) +{ + unsigned int i; + int min_vactive_margin_us = 0xFFFFFFF; + + for (i = 0; i < DML2_MAX_PLANES; i++) { + if (is_bit_set_in_bitfield(plane_mask, i)) { + if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_latency_hiding_margin_in_active < min_vactive_margin_us) + min_vactive_margin_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_latency_hiding_margin_in_active; + } + } + + return min_vactive_margin_us; +} + +static unsigned int get_vactive_det_fill_latency_delay_us(const struct display_configuation_with_meta *display_cfg, int plane_mask) +{ + unsigned char i; + unsigned int max_vactive_fill_us = 0; + + for (i = 0; i < DML2_MAX_PLANES; i++) { + if (is_bit_set_in_bitfield(plane_mask, i)) { + if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us > max_vactive_fill_us) + max_vactive_fill_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us; + } + } + + return max_vactive_fill_us; +} + +static void build_pstate_meta_per_stream(struct dml2_pmo_instance *pmo, + struct display_configuation_with_meta *display_config, + int stream_index) +{ + const struct dml2_ip_capabilities *ip_caps = pmo->ip_caps; + const struct dml2_stream_parameters *stream_descriptor = &display_config->display_config.stream_descriptors[stream_index]; + const struct core_stream_support_info *stream_info = &display_config->mode_support_result.cfg_support_info.stream_support_info[stream_index]; + const struct dml2_timing_cfg *timing = &stream_descriptor->timing; + struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; + + /* worst case all other streams require some programming at the same time, 0 if only 1 stream */ + unsigned int contention_delay_us = (ip_caps->fams2.vertical_interrupt_ack_delay_us + + (unsigned int)math_max3(ip_caps->fams2.subvp_programming_delay_us, ip_caps->fams2.drr_programming_delay_us, ip_caps->fams2.allow_programming_delay_us)) * + (display_config->display_config.num_streams - 1); + + /* common */ + stream_pstate_meta->valid = true; + stream_pstate_meta->otg_vline_time_us = (double)timing->h_total / timing->pixel_clock_khz * 1000.0; + stream_pstate_meta->nom_vtotal = stream_descriptor->timing.vblank_nom + stream_descriptor->timing.v_active; + stream_pstate_meta->nom_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / + (stream_pstate_meta->nom_vtotal * timing->h_total); + stream_pstate_meta->nom_frame_time_us = + (double)stream_pstate_meta->nom_vtotal * stream_pstate_meta->otg_vline_time_us; + stream_pstate_meta->vblank_start = timing->v_blank_end + timing->v_active; + + if (stream_descriptor->timing.drr_config.enabled == true) { + if (stream_descriptor->timing.drr_config.min_refresh_uhz != 0.0) { + stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / + ((double)stream_descriptor->timing.drr_config.min_refresh_uhz * stream_descriptor->timing.h_total) * 1e9); + } else { + /* assume min of 48Hz */ + stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / + (48000000.0 * stream_descriptor->timing.h_total) * 1e9); + } + } else { + stream_pstate_meta->max_vtotal = stream_pstate_meta->nom_vtotal; + } + stream_pstate_meta->min_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / + (stream_pstate_meta->max_vtotal * timing->h_total); + stream_pstate_meta->max_frame_time_us = + (double)stream_pstate_meta->max_vtotal * stream_pstate_meta->otg_vline_time_us; + + stream_pstate_meta->scheduling_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.scheduling_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.vertical_interrupt_ack_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->contention_delay_otg_vlines = + (unsigned int)math_ceil(contention_delay_us / stream_pstate_meta->otg_vline_time_us); + /* worst case allow to target needs to account for all streams' allow events overlapping, and 1 line for error */ + stream_pstate_meta->allow_to_target_delay_otg_vlines = + (unsigned int)(math_ceil((ip_caps->fams2.vertical_interrupt_ack_delay_us + contention_delay_us + ip_caps->fams2.allow_programming_delay_us) / stream_pstate_meta->otg_vline_time_us)) + 1; + stream_pstate_meta->min_allow_width_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.min_allow_width_us / stream_pstate_meta->otg_vline_time_us); + /* this value should account for urgent latency */ + stream_pstate_meta->blackout_otg_vlines = + (unsigned int)math_ceil(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us / + stream_pstate_meta->otg_vline_time_us); + + /* scheduling params should be built based on the worst case for allow_time:disallow_time */ + + /* vactive */ + if (display_config->display_config.num_streams == 1) { + /* for single stream, guarantee at least an instant of allow */ + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = (unsigned int)math_floor( + math_max2(0.0, + timing->v_active - math_max2(1.0, stream_pstate_meta->min_allow_width_otg_vlines) - stream_pstate_meta->blackout_otg_vlines)); + } else { + /* for multi stream, bound to a max fill time defined by IP caps */ + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = + (unsigned int)math_floor((double)ip_caps->max_vactive_det_fill_delay_us / stream_pstate_meta->otg_vline_time_us); + } + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us = stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines * stream_pstate_meta->otg_vline_time_us; + + if (stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us > 0.0) { + stream_pstate_meta->method_vactive.common.allow_start_otg_vline = + timing->v_blank_end + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; + stream_pstate_meta->method_vactive.common.allow_end_otg_vline = + stream_pstate_meta->vblank_start - + stream_pstate_meta->blackout_otg_vlines; + } else { + stream_pstate_meta->method_vactive.common.allow_start_otg_vline = 0; + stream_pstate_meta->method_vactive.common.allow_end_otg_vline = 0; + } + stream_pstate_meta->method_vactive.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_vactive.common, stream_pstate_meta); + + /* vblank */ + stream_pstate_meta->method_vblank.common.allow_start_otg_vline = stream_pstate_meta->vblank_start; + stream_pstate_meta->method_vblank.common.allow_end_otg_vline = + stream_pstate_meta->method_vblank.common.allow_start_otg_vline + 1; + stream_pstate_meta->method_vblank.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_vblank.common, stream_pstate_meta); + + /* subvp */ + stream_pstate_meta->method_subvp.programming_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_programming_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_df_throttle_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_prefetch_to_mall_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.phantom_vactive = + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + + stream_info->phantom_min_v_active; + stream_pstate_meta->method_subvp.phantom_vfp = + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines; + /* phantom vtotal = v_bp(vstartup) + v_sync(1) + v_fp(throttle_delay) + v_active(allow_to_target + min_allow + min_vactive)*/ + stream_pstate_meta->method_subvp.phantom_vtotal = + stream_info->phantom_v_startup + + stream_pstate_meta->method_subvp.phantom_vfp + + 1 + + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines + + stream_pstate_meta->method_subvp.phantom_vactive; + stream_pstate_meta->method_subvp.common.allow_start_otg_vline = + stream_descriptor->timing.v_blank_end + + stream_pstate_meta->contention_delay_otg_vlines + + stream_pstate_meta->method_subvp.programming_delay_otg_vlines + + stream_pstate_meta->method_subvp.phantom_vtotal + + stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + + stream_pstate_meta->allow_to_target_delay_otg_vlines; + stream_pstate_meta->method_subvp.common.allow_end_otg_vline = + stream_pstate_meta->vblank_start - + stream_pstate_meta->blackout_otg_vlines; + stream_pstate_meta->method_subvp.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_subvp.common, stream_pstate_meta); + + /* drr */ + stream_pstate_meta->method_drr.programming_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.drr_programming_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_drr.common.allow_start_otg_vline = + stream_pstate_meta->vblank_start + + stream_pstate_meta->allow_to_target_delay_otg_vlines; + stream_pstate_meta->method_drr.common.period_us = stream_pstate_meta->nom_frame_time_us; + if (display_config->display_config.num_streams <= 1) { + /* only need to stretch vblank for blackout time */ + stream_pstate_meta->method_drr.stretched_vtotal = + stream_pstate_meta->nom_vtotal + + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + + stream_pstate_meta->blackout_otg_vlines; + } else { + /* multi display needs to always be schedulable */ + stream_pstate_meta->method_drr.stretched_vtotal = + stream_pstate_meta->nom_vtotal * 2 + + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + + stream_pstate_meta->blackout_otg_vlines; + } + stream_pstate_meta->method_drr.common.allow_end_otg_vline = + stream_pstate_meta->method_drr.stretched_vtotal - + stream_pstate_meta->blackout_otg_vlines; + build_method_scheduling_params(&stream_pstate_meta->method_drr.common, stream_pstate_meta); +} + +static void build_subvp_meta_per_stream(struct dml2_pmo_instance *pmo, + struct display_configuation_with_meta *display_config, + int stream_index) +{ + struct dml2_implicit_svp_meta *stream_svp_meta = &pmo->scratch.pmo_dcn4.stream_svp_meta[stream_index]; + struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; + + stream_svp_meta->valid = true; + + /* PMO FAMS2 precaulcates these values */ + stream_svp_meta->v_active = stream_pstate_meta->method_subvp.phantom_vactive; + stream_svp_meta->v_front_porch = stream_pstate_meta->method_subvp.phantom_vfp; + stream_svp_meta->v_total = stream_pstate_meta->method_subvp.phantom_vtotal; +} + +bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out) +{ + struct dml2_pmo_instance *pmo = in_out->instance; + struct dml2_optimization_stage3_state *state = &in_out->base_display_config->stage3; + struct dml2_pmo_scratch *s = &pmo->scratch; + + struct display_configuation_with_meta *display_config; + const struct dml2_plane_parameters *plane_descriptor; + const struct dml2_pmo_pstate_strategy *strategy_list = NULL; + struct dml2_pmo_pstate_strategy override_base_strategy = { 0 }; + unsigned int strategy_list_size = 0; + unsigned int plane_index, stream_index, i; + bool build_override_strategy = true; + + state->performed = true; + in_out->base_display_config->stage3.min_clk_index_for_latency = in_out->base_display_config->stage1.min_clk_index_for_latency; + + display_config = in_out->base_display_config; + display_config->display_config.overrides.enable_subvp_implicit_pmo = true; + + memset(s, 0, sizeof(struct dml2_pmo_scratch)); + + if (display_config->display_config.overrides.all_streams_blanked) { + return true; + } + + pmo->scratch.pmo_dcn4.min_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; + pmo->scratch.pmo_dcn4.max_latency_index = pmo->mcg_clock_table_size; + pmo->scratch.pmo_dcn4.cur_latency_index = in_out->base_display_config->stage1.min_clk_index_for_latency; + + // First build the stream plane mask (array of bitfields indexed by stream, indicating plane mapping) + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + plane_descriptor = &display_config->display_config.plane_descriptors[plane_index]; + + set_bit_in_bitfield(&s->pmo_dcn4.stream_plane_mask[plane_descriptor->stream_index], plane_index); + + state->pstate_switch_modes[plane_index] = dml2_pstate_method_vactive; + + build_override_strategy &= plane_descriptor->overrides.uclk_pstate_change_strategy != dml2_uclk_pstate_change_strategy_auto; + override_base_strategy.per_stream_pstate_method[plane_descriptor->stream_index] = + uclk_pstate_strategy_override_to_pstate_method(plane_descriptor->overrides.uclk_pstate_change_strategy); + } + + // Figure out which streams can do vactive, and also build up implicit SVP and FAMS2 meta + for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { + if (get_vactive_pstate_margin(display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) >= (int)(MIN_VACTIVE_MARGIN_PCT * pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us)) + set_bit_in_bitfield(&s->pmo_dcn4.stream_vactive_capability_mask, stream_index); + + /* FAMS2 meta */ + build_pstate_meta_per_stream(pmo, display_config, stream_index); + + /* SVP meta */ + build_subvp_meta_per_stream(pmo, display_config, stream_index); + } + + /* get synchronized timing groups */ + build_synchronized_timing_groups(pmo, display_config); + + if (build_override_strategy) { + /* build expanded override strategy list (no permutations) */ + override_base_strategy.allow_state_increase = true; + s->pmo_dcn4.num_expanded_override_strategies = 0; + insert_strategy_into_expanded_list(&override_base_strategy, + display_config->display_config.num_streams, + s->pmo_dcn4.expanded_override_strategy_list, + &s->pmo_dcn4.num_expanded_override_strategies); + expand_variant_strategy(&override_base_strategy, + display_config->display_config.num_streams, + false, + s->pmo_dcn4.expanded_override_strategy_list, + &s->pmo_dcn4.num_expanded_override_strategies); + + /* use override strategy list */ + strategy_list = s->pmo_dcn4.expanded_override_strategy_list; + strategy_list_size = s->pmo_dcn4.num_expanded_override_strategies; + } else { + /* use predefined strategy list */ + strategy_list = get_expanded_strategy_list(&pmo->init_data, display_config->display_config.num_streams); + strategy_list_size = get_num_expanded_strategies(&pmo->init_data, display_config->display_config.num_streams); + } + + if (!strategy_list || strategy_list_size == 0) + return false; + + s->pmo_dcn4.num_pstate_candidates = 0; + + for (i = 0; i < strategy_list_size && s->pmo_dcn4.num_pstate_candidates < DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE; i++) { + if (validate_pstate_support_strategy_cofunctionality(pmo, display_config, &strategy_list[i])) { + insert_into_candidate_list(&strategy_list[i], display_config->display_config.num_streams, s); + } + } + + if (s->pmo_dcn4.num_pstate_candidates > 0) { + s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.num_pstate_candidates-1].allow_state_increase = true; + s->pmo_dcn4.cur_pstate_candidate = -1; + return true; + } else { + return false; + } +} + +static void reset_display_configuration(struct display_configuation_with_meta *display_config) +{ + unsigned int plane_index; + unsigned int stream_index; + struct dml2_plane_parameters *plane; + + for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { + display_config->stage3.stream_svp_meta[stream_index].valid = false; + + display_config->display_config.stream_descriptors[stream_index].overrides.minimize_active_latency_hiding = false; + display_config->display_config.overrides.best_effort_min_active_latency_hiding_us = 0; + } + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + plane = &display_config->display_config.plane_descriptors[plane_index]; + + // Unset SubVP + plane->overrides.legacy_svp_config = dml2_svp_mode_override_auto; + + // Remove reserve time + plane->overrides.reserved_vblank_time_ns = 0; + + // Reset strategy to auto + plane->overrides.uclk_pstate_change_strategy = dml2_uclk_pstate_change_strategy_auto; + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_na; + } +} + +static void setup_planes_for_drr_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + unsigned int plane_index; + struct dml2_plane_parameters *plane; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + plane = &display_config->display_config.plane_descriptors[plane_index]; + + plane->overrides.uclk_pstate_change_strategy = dml2_uclk_pstate_change_strategy_force_drr; + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_drr; + + } + } +} + +static void setup_planes_for_svp_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + struct dml2_pmo_scratch *scratch = &pmo->scratch; + + unsigned int plane_index; + int stream_index = -1; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + stream_index = (char)display_config->display_config.plane_descriptors[plane_index].stream_index; + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_svp; + } + } + + if (stream_index >= 0) { + memcpy(&display_config->stage3.stream_svp_meta[stream_index], + &scratch->pmo_dcn4.stream_svp_meta[stream_index], + sizeof(struct dml2_implicit_svp_meta)); + } +} + +static void setup_planes_for_svp_drr_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + struct dml2_pmo_scratch *scratch = &pmo->scratch; + + unsigned int plane_index; + int stream_index = -1; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + stream_index = (char)display_config->display_config.plane_descriptors[plane_index].stream_index; + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_svp_drr; + } + } + + if (stream_index >= 0) { + memcpy(&display_config->stage3.stream_svp_meta[stream_index], + &scratch->pmo_dcn4.stream_svp_meta[stream_index], + sizeof(struct dml2_implicit_svp_meta)); + } +} + +static void setup_planes_for_vblank_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + unsigned int plane_index; + struct dml2_plane_parameters *plane; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + plane = &display_config->display_config.plane_descriptors[plane_index]; + + plane->overrides.reserved_vblank_time_ns = (long)math_max2(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us * 1000.0, + plane->overrides.reserved_vblank_time_ns); + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vblank; + + } + } +} + +static void setup_planes_for_vblank_drr_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + unsigned int plane_index; + struct dml2_plane_parameters *plane; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + plane = &display_config->display_config.plane_descriptors[plane_index]; + plane->overrides.reserved_vblank_time_ns = (long)(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us * 1000); + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vblank_drr; + } + } +} + +static void setup_planes_for_vactive_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + unsigned int plane_index; + unsigned int stream_index; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + stream_index = display_config->display_config.plane_descriptors[plane_index].stream_index; + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vactive; + + if (!pmo->options->disable_vactive_det_fill_bw_pad) { + display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = + (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); + } + } + } +} + +static void setup_planes_for_vactive_drr_by_mask(struct display_configuation_with_meta *display_config, + struct dml2_pmo_instance *pmo, + int plane_mask) +{ + unsigned int plane_index; + unsigned int stream_index; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + stream_index = display_config->display_config.plane_descriptors[plane_index].stream_index; + + display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vactive_drr; + + if (!pmo->options->disable_vactive_det_fill_bw_pad) { + display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = + (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); + } + } + } +} + +static bool setup_display_config(struct display_configuation_with_meta *display_config, struct dml2_pmo_instance *pmo, int strategy_index) +{ + struct dml2_pmo_scratch *scratch = &pmo->scratch; + + bool fams2_required = false; + bool success = true; + unsigned int stream_index; + + reset_display_configuration(display_config); + + for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { + + if (pmo->scratch.pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_na) { + success = false; + break; + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive) { + setup_planes_for_vactive_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank) { + setup_planes_for_vblank_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp) { + fams2_required = true; + setup_planes_for_svp_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { + fams2_required = true; + setup_planes_for_vactive_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { + fams2_required = true; + setup_planes_for_vblank_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { + fams2_required = true; + setup_planes_for_svp_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } else if (scratch->pmo_dcn4.pstate_strategy_candidates[strategy_index].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { + fams2_required = true; + setup_planes_for_drr_by_mask(display_config, pmo, scratch->pmo_dcn4.stream_plane_mask[stream_index]); + } + } + + /* copy FAMS2 meta */ + if (success) { + display_config->stage3.fams2_required = fams2_required; + memcpy(&display_config->stage3.stream_pstate_meta, + &scratch->pmo_dcn4.stream_pstate_meta, + sizeof(struct dml2_pstate_meta) * DML2_MAX_PLANES); + } + + return success; +} + +static int get_minimum_reserved_time_us_for_planes(struct display_configuation_with_meta *display_config, int plane_mask) +{ + int min_time_us = 0xFFFFFF; + unsigned int plane_index = 0; + + for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { + if (is_bit_set_in_bitfield(plane_mask, plane_index)) { + if (min_time_us > (display_config->display_config.plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000)) + min_time_us = display_config->display_config.plane_descriptors[plane_index].overrides.reserved_vblank_time_ns / 1000; + } + } + return min_time_us; +} + +bool pmo_dcn4_fams2_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out) +{ + bool p_state_supported = true; + unsigned int stream_index; + struct dml2_pmo_scratch *s = &in_out->instance->scratch; + + int MIN_VACTIVE_MARGIN_VBLANK = 0; + int MIN_VACTIVE_MARGIN_DRR = 0; + int REQUIRED_RESERVED_TIME = 0; + + if (in_out->base_display_config->display_config.overrides.all_streams_blanked) { + return true; + } + + MIN_VACTIVE_MARGIN_VBLANK = INT_MIN; + MIN_VACTIVE_MARGIN_DRR = INT_MIN; + REQUIRED_RESERVED_TIME = (int)in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us; + + if (s->pmo_dcn4.cur_pstate_candidate < 0) + return false; + + for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) { + struct dml2_pstate_meta *stream_pstate_meta = &s->pmo_dcn4.stream_pstate_meta[stream_index]; + + if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive || + s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { + if (get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < (MIN_VACTIVE_MARGIN_PCT * in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us) || + get_vactive_det_fill_latency_delay_us(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) > stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us) { + p_state_supported = false; + break; + } + } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vblank || + s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vblank_drr) { + if (get_minimum_reserved_time_us_for_planes(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < + REQUIRED_RESERVED_TIME || + get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < MIN_VACTIVE_MARGIN_VBLANK) { + p_state_supported = false; + break; + } + } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp || + s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_svp_drr) { + if (in_out->base_display_config->stage3.stream_svp_meta[stream_index].valid == false) { + p_state_supported = false; + break; + } + } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_drr) { + if (!all_planes_match_method(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index], dml2_pstate_method_fw_drr) || + get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < MIN_VACTIVE_MARGIN_DRR) { + p_state_supported = false; + break; + } + } else if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_na) { + p_state_supported = false; + break; + } + } + + return p_state_supported; +} + +bool pmo_dcn4_fams2_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out) +{ + bool success = false; + struct dml2_pmo_scratch *s = &in_out->instance->scratch; + + memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); + + if (in_out->last_candidate_failed) { + if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].allow_state_increase && + s->pmo_dcn4.cur_latency_index < s->pmo_dcn4.max_latency_index - 1) { + s->pmo_dcn4.cur_latency_index++; + + success = true; + } + } + + if (!success) { + s->pmo_dcn4.cur_latency_index = s->pmo_dcn4.min_latency_index; + s->pmo_dcn4.cur_pstate_candidate++; + + if (s->pmo_dcn4.cur_pstate_candidate < s->pmo_dcn4.num_pstate_candidates) { + success = true; + } + } + + if (success) { + in_out->optimized_display_config->stage3.min_clk_index_for_latency = s->pmo_dcn4.cur_latency_index; + setup_display_config(in_out->optimized_display_config, in_out->instance, in_out->instance->scratch.pmo_dcn4.cur_pstate_candidate); + } + + return success; +} + +bool pmo_dcn4_fams2_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out) +{ + bool success = true; + struct dml2_pmo_instance *pmo = in_out->instance; + bool stutter_period_meets_z8_eco = true; + bool z8_stutter_optimization_too_expensive = false; + bool stutter_optimization_too_expensive = false; + double line_time_us, vblank_nom_time_us; + + unsigned int i; + + if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && + pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && + pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us < pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us) + return false; // Unexpected SoCBB setup + + for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { + if (in_out->base_display_config->mode_support_result.cfg_support_info.plane_support_info[i].active_latency_hiding_us < + pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us + pmo->soc_bb->power_management_parameters.z8_min_idle_time) { + stutter_period_meets_z8_eco = false; + break; + } + } + + for (i = 0; i < in_out->base_display_config->display_config.num_streams; i++) { + line_time_us = (double)in_out->base_display_config->display_config.stream_descriptors[i].timing.h_total / (in_out->base_display_config->display_config.stream_descriptors[i].timing.pixel_clock_khz * 1000) * 1000000; + vblank_nom_time_us = line_time_us * in_out->base_display_config->display_config.stream_descriptors[i].timing.vblank_nom; + + if (vblank_nom_time_us < pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us * MIN_BLANK_STUTTER_FACTOR) { + z8_stutter_optimization_too_expensive = true; + break; + } + + if (vblank_nom_time_us < pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us * MIN_BLANK_STUTTER_FACTOR) { + stutter_optimization_too_expensive = true; + break; + } + } + + pmo->scratch.pmo_dcn4.num_stutter_candidates = 0; + pmo->scratch.pmo_dcn4.cur_stutter_candidate = 0; + + if (stutter_period_meets_z8_eco && !z8_stutter_optimization_too_expensive) { + if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0) { + pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.num_stutter_candidates] = (unsigned int)pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us; + pmo->scratch.pmo_dcn4.num_stutter_candidates++; + pmo->scratch.pmo_dcn4.z8_vblank_optimizable = true; + } + } else { + pmo->scratch.pmo_dcn4.z8_vblank_optimizable = false; + } + + if (!stutter_optimization_too_expensive && pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0) { + pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.num_stutter_candidates] = (unsigned int)pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us; + pmo->scratch.pmo_dcn4.num_stutter_candidates++; + } + + if (pmo->scratch.pmo_dcn4.num_stutter_candidates == 0) + success = false; + + return success; +} + +bool pmo_dcn4_fams2_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out) +{ + bool success = true; + struct dml2_pmo_instance *pmo = in_out->instance; + + unsigned int i; + + for (i = 0; i < in_out->base_display_config->display_config.num_planes; i++) { + if (pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us > 0 && + pmo->scratch.pmo_dcn4.z8_vblank_optimizable && + in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < (int)pmo->soc_bb->power_management_parameters.z8_stutter_exit_latency_us * 1000) { + success = false; + break; + } + if (pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us > 0 && + in_out->base_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns < (int)pmo->soc_bb->power_management_parameters.stutter_enter_plus_exit_latency_us * 1000) { + success = false; + break; + } + } + + return success; +} + +bool pmo_dcn4_fams2_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out) +{ + bool success = false; + struct dml2_pmo_instance *pmo = in_out->instance; + unsigned int i; + + memcpy(in_out->optimized_display_config, in_out->base_display_config, sizeof(struct display_configuation_with_meta)); + + if (!in_out->last_candidate_failed) { + if (pmo->scratch.pmo_dcn4.cur_stutter_candidate < pmo->scratch.pmo_dcn4.num_stutter_candidates) { + for (i = 0; i < in_out->optimized_display_config->display_config.num_planes; i++) { + /* take the max of the current and the optimal reserved time */ + in_out->optimized_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns = + (long)math_max2(pmo->scratch.pmo_dcn4.optimal_vblank_reserved_time_for_stutter_us[pmo->scratch.pmo_dcn4.cur_stutter_candidate] * 1000, + in_out->optimized_display_config->display_config.plane_descriptors[i].overrides.reserved_vblank_time_ns); + } + + success = true; + } + } + + return success; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h new file mode 100644 index 000000000000..6baab7ad6ecc --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h @@ -0,0 +1,33 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_PMO_FAMS2_DCN4_H__ +#define __DML2_PMO_FAMS2_DCN4_H__ + +#include "dml2_internal_shared_types.h" + +bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out); + +bool pmo_dcn4_fams2_optimize_dcc_mcache(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); + +bool pmo_dcn4_fams2_init_for_vmin(struct dml2_pmo_init_for_vmin_in_out *in_out); +bool pmo_dcn4_fams2_test_for_vmin(struct dml2_pmo_test_for_vmin_in_out *in_out); +bool pmo_dcn4_fams2_optimize_for_vmin(struct dml2_pmo_optimize_for_vmin_in_out *in_out); + +bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out); +bool pmo_dcn4_fams2_test_for_pstate_support(struct dml2_pmo_test_for_pstate_support_in_out *in_out); +bool pmo_dcn4_fams2_optimize_for_pstate_support(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); + +bool pmo_dcn4_fams2_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out); +bool pmo_dcn4_fams2_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out); +bool pmo_dcn4_fams2_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out); + +void pmo_dcn4_fams2_expand_base_pstate_strategies( + const struct dml2_pmo_pstate_strategy *base_strategies_list, + const unsigned int num_base_strategies, + const unsigned int stream_count, + struct dml2_pmo_pstate_strategy *expanded_strategy_list, + unsigned int *num_expanded_strategies); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c new file mode 100644 index 000000000000..55d2464365d0 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c @@ -0,0 +1,83 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_pmo_factory.h" +#include "dml2_pmo_dcn4_fams2.h" +#include "dml2_pmo_dcn3.h" +#include "dml2_external_lib_deps.h" + +static bool dummy_init_for_stutter(struct dml2_pmo_init_for_stutter_in_out *in_out) +{ + return false; +} + +static bool dummy_test_for_stutter(struct dml2_pmo_test_for_stutter_in_out *in_out) +{ + return true; +} + +static bool dummy_optimize_for_stutter(struct dml2_pmo_optimize_for_stutter_in_out *in_out) +{ + return false; +} + +bool dml2_pmo_create(enum dml2_project_id project_id, struct dml2_pmo_instance *out) +{ + bool result = false; + + if (!out) + return false; + + memset(out, 0, sizeof(struct dml2_pmo_instance)); + + switch (project_id) { + case dml2_project_dcn4x_stage1: + out->initialize = pmo_dcn4_fams2_initialize; + out->optimize_dcc_mcache = pmo_dcn4_fams2_optimize_dcc_mcache; + result = true; + break; + case dml2_project_dcn4x_stage2: + out->initialize = pmo_dcn3_initialize; + + out->optimize_dcc_mcache = pmo_dcn3_optimize_dcc_mcache; + + out->init_for_vmin = pmo_dcn3_init_for_vmin; + out->test_for_vmin = pmo_dcn3_test_for_vmin; + out->optimize_for_vmin = pmo_dcn3_optimize_for_vmin; + + out->init_for_uclk_pstate = pmo_dcn3_init_for_pstate_support; + out->test_for_uclk_pstate = pmo_dcn3_test_for_pstate_support; + out->optimize_for_uclk_pstate = pmo_dcn3_optimize_for_pstate_support; + + out->init_for_stutter = dummy_init_for_stutter; + out->test_for_stutter = dummy_test_for_stutter; + out->optimize_for_stutter = dummy_optimize_for_stutter; + + result = true; + break; + case dml2_project_dcn4x_stage2_auto_drr_svp: + out->initialize = pmo_dcn4_fams2_initialize; + + out->optimize_dcc_mcache = pmo_dcn4_fams2_optimize_dcc_mcache; + + out->init_for_vmin = pmo_dcn4_fams2_init_for_vmin; + out->test_for_vmin = pmo_dcn4_fams2_test_for_vmin; + out->optimize_for_vmin = pmo_dcn4_fams2_optimize_for_vmin; + + out->init_for_uclk_pstate = pmo_dcn4_fams2_init_for_pstate_support; + out->test_for_uclk_pstate = pmo_dcn4_fams2_test_for_pstate_support; + out->optimize_for_uclk_pstate = pmo_dcn4_fams2_optimize_for_pstate_support; + + out->init_for_stutter = pmo_dcn4_fams2_init_for_stutter; + out->test_for_stutter = pmo_dcn4_fams2_test_for_stutter; + out->optimize_for_stutter = pmo_dcn4_fams2_optimize_for_stutter; + result = true; + break; + case dml2_project_invalid: + default: + break; + } + + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h new file mode 100644 index 000000000000..b90f6263cd85 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_PMO_FACTORY_H__ +#define __DML2_PMO_FACTORY_H__ + +#include "dml2_internal_shared_types.h" +#include "dml_top_types.h" + +bool dml2_pmo_create(enum dml2_project_id project_id, struct dml2_pmo_instance *out); + +#endif \ No newline at end of file diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c new file mode 100644 index 000000000000..e17b5ceba447 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c @@ -0,0 +1,147 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "lib_float_math.h" + +#define ASSERT(condition) + +#define isNaN(number) ((number) != (number)) + + /* + * NOTE: + * This file is gcc-parseable HW gospel, coming straight from HW engineers. + * + * It doesn't adhere to Linux kernel style and sometimes will do things in odd + * ways. Unless there is something clearly wrong with it the code should + * remain as-is as it provides us with a guarantee from HW that it is correct. + */ + +double math_mod(const double arg1, const double arg2) +{ + if (isNaN(arg1)) + return arg2; + if (isNaN(arg2)) + return arg1; + return arg1 - arg1 * ((int)(arg1 / arg2)); +} + +double math_min2(const double arg1, const double arg2) +{ + if (isNaN(arg1)) + return arg2; + if (isNaN(arg2)) + return arg1; + return arg1 < arg2 ? arg1 : arg2; +} + +double math_max2(const double arg1, const double arg2) +{ + if (isNaN(arg1)) + return arg2; + if (isNaN(arg2)) + return arg1; + return arg1 > arg2 ? arg1 : arg2; +} + +double math_floor2(const double arg, const double significance) +{ + ASSERT(significance != 0); + + return ((int)(arg / significance)) * significance; +} + +double math_floor(const double arg) +{ + return ((int)(arg)); +} + +double math_ceil(const double arg) +{ + return (int)(arg + 0.99999); +} + +double math_ceil2(const double arg, const double significance) +{ + return ((int)(arg / significance + 0.99999)) * significance; +} + +double math_max3(double v1, double v2, double v3) +{ + return v3 > math_max2(v1, v2) ? v3 : math_max2(v1, v2); +} + +double math_max4(double v1, double v2, double v3, double v4) +{ + return v4 > math_max3(v1, v2, v3) ? v4 : math_max3(v1, v2, v3); +} + +double math_max5(double v1, double v2, double v3, double v4, double v5) +{ + return math_max3(v1, v2, v3) > math_max2(v4, v5) ? math_max3(v1, v2, v3) : math_max2(v4, v5); +} + +float math_pow(float a, float exp) +{ + double temp; + if ((int)exp == 0) + return 1; + temp = math_pow(a, (float)((int)(exp / 2))); + if (((int)exp % 2) == 0) { + return (float)(temp * temp); + } else { + if ((int)exp > 0) + return (float)(a * temp * temp); + else + return (float)((temp * temp) / a); + } +} + +double math_fabs(double a) +{ + if (a > 0) + return (a); + else + return (-a); +} + +float math_log(float a, float b) +{ + int *const exp_ptr = (int *)(&a); + int x = *exp_ptr; + const int log_2 = ((x >> 23) & 255) - 128; + x &= ~(255 << 23); + x += 127 << 23; + *exp_ptr = x; + + a = ((-1.0f / 3) * a + 2) * a - 2.0f / 3; + + if (b > 2.00001 || b < 1.99999) + return (a + log_2) / math_log(b, 2); + else + return (a + log_2); +} + +float math_log2(float a) +{ + return math_log(a, 2.0); +} + +// approximate log2 value of a input +// - precise if the input pwr of 2, else the approximation will be an integer = floor(actual_log2) +unsigned int math_log2_approx(unsigned int a) +{ + unsigned int log2_val = 0; + while (a > 1) { + a = a >> 1; + log2_val++; + } + return log2_val; +} + +double math_round(double a) +{ + const double round_pt = 0.5; + + return math_floor(a + round_pt); +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h new file mode 100644 index 000000000000..e13b0c5939b0 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h @@ -0,0 +1,25 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __LIB_FLOAT_MATH_H__ +#define __LIB_FLOAT_MATH_H__ + +double math_mod(const double arg1, const double arg2); +double math_min2(const double arg1, const double arg2); +double math_max2(const double arg1, const double arg2); +double math_floor2(const double arg, const double significance); +double math_floor(const double arg); +double math_ceil(const double arg); +double math_ceil2(const double arg, const double significance); +double math_max3(double v1, double v2, double v3); +double math_max4(double v1, double v2, double v3, double v4); +double math_max5(double v1, double v2, double v3, double v4, double v5); +float math_pow(float a, float exp); +double math_fabs(double a); +float math_log(float a, float b); +float math_log2(float a); +unsigned int math_log2_approx(unsigned int a); +double math_round(double a); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c new file mode 100644 index 000000000000..5a33e2f357f4 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c @@ -0,0 +1,49 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml_top.h" +#include "dml2_internal_shared_types.h" +#include "dml2_top_soc15.h" + +unsigned int dml2_get_instance_size_bytes(void) +{ + return sizeof(struct dml2_instance); +} + +bool dml2_initialize_instance(struct dml2_initialize_instance_in_out *in_out) +{ + switch (in_out->options.project_id) { + case dml2_project_dcn4x_stage1: + case dml2_project_dcn4x_stage2: + case dml2_project_dcn4x_stage2_auto_drr_svp: + return dml2_top_soc15_initialize_instance(in_out); + case dml2_project_invalid: + default: + return false; + } +} + +bool dml2_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out) +{ + if (!in_out->dml2_instance->funcs.check_mode_supported) + return false; + + return in_out->dml2_instance->funcs.check_mode_supported(in_out); +} + +bool dml2_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out) +{ + if (!in_out->dml2_instance->funcs.build_mode_programming) + return false; + + return in_out->dml2_instance->funcs.build_mode_programming(in_out); +} + +bool dml2_build_mcache_programming(struct dml2_build_mcache_programming_in_out *in_out) +{ + if (!in_out->dml2_instance->funcs.build_mcache_programming) + return false; + + return in_out->dml2_instance->funcs.build_mcache_programming(in_out); +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c new file mode 100644 index 000000000000..5e14d85821e2 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c @@ -0,0 +1,10 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_top_legacy.h" +#include "dml2_top_soc15.h" +#include "dml2_core_factory.h" +#include "dml2_pmo_factory.h" +#include "display_mode_core_structs.h" + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h new file mode 100644 index 000000000000..14d0ae03dce6 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h @@ -0,0 +1,9 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_TOP_LEGACY_H__ +#define __DML2_TOP_LEGACY_H__ +#include "dml2_internal_shared_types.h" +bool dml2_top_legacy_initialize_instance(struct dml2_initialize_instance_in_out *in_out); +#endif /* __DML2_TOP_LEGACY_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c new file mode 100644 index 000000000000..4a7c4c62111e --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c @@ -0,0 +1,1170 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#include "dml2_top_soc15.h" +#include "dml2_mcg_factory.h" +#include "dml2_dpmm_factory.h" +#include "dml2_core_factory.h" +#include "dml2_pmo_factory.h" +#include "lib_float_math.h" +#include "dml2_debug.h" +static void setup_unoptimized_display_config_with_meta(const struct dml2_instance *dml, struct display_configuation_with_meta *out, const struct dml2_display_cfg *display_config) +{ + memcpy(&out->display_config, display_config, sizeof(struct dml2_display_cfg)); + out->stage1.min_clk_index_for_latency = dml->min_clk_table.dram_bw_table.num_entries - 1; //dml->min_clk_table.clean_me_up.soc_bb.num_states - 1; +} + +static void setup_speculative_display_config_with_meta(const struct dml2_instance *dml, struct display_configuation_with_meta *out, const struct dml2_display_cfg *display_config) +{ + memcpy(&out->display_config, display_config, sizeof(struct dml2_display_cfg)); + out->stage1.min_clk_index_for_latency = 0; +} + +static void copy_display_configuration_with_meta(struct display_configuation_with_meta *dst, const struct display_configuation_with_meta *src) +{ + memcpy(dst, src, sizeof(struct display_configuation_with_meta)); +} + +static bool dml2_top_optimization_init_function_min_clk_for_latency(const struct optimization_init_function_params *params) +{ + struct dml2_optimization_stage1_state *state = ¶ms->display_config->stage1; + + state->performed = true; + + return true; +} + +static bool dml2_top_optimization_test_function_min_clk_for_latency(const struct optimization_test_function_params *params) +{ + struct dml2_optimization_stage1_state *state = ¶ms->display_config->stage1; + + return state->min_clk_index_for_latency == 0; +} + +static bool dml2_top_optimization_optimize_function_min_clk_for_latency(const struct optimization_optimize_function_params *params) +{ + bool result = false; + + if (params->display_config->stage1.min_clk_index_for_latency > 0) { + copy_display_configuration_with_meta(params->optimized_display_config, params->display_config); + params->optimized_display_config->stage1.min_clk_index_for_latency--; + result = true; + } + + return result; +} + +static bool dml2_top_optimization_test_function_mcache(const struct optimization_test_function_params *params) +{ + struct dml2_optimization_test_function_locals *l = params->locals; + bool mcache_success = false; + bool result = false; + + memset(l, 0, sizeof(struct dml2_optimization_test_function_locals)); + + l->test_mcache.calc_mcache_count_params.dml2_instance = params->dml; + l->test_mcache.calc_mcache_count_params.display_config = ¶ms->display_config->display_config; + l->test_mcache.calc_mcache_count_params.mcache_allocations = params->display_config->stage2.mcache_allocations; + + result = dml2_top_mcache_calc_mcache_count_and_offsets(&l->test_mcache.calc_mcache_count_params); // use core to get the basic mcache_allocations + + if (result) { + l->test_mcache.assign_global_mcache_ids_params.allocations = params->display_config->stage2.mcache_allocations; + l->test_mcache.assign_global_mcache_ids_params.num_allocations = params->display_config->display_config.num_planes; + + dml2_top_mcache_assign_global_mcache_ids(&l->test_mcache.assign_global_mcache_ids_params); + + l->test_mcache.validate_admissibility_params.dml2_instance = params->dml; + l->test_mcache.validate_admissibility_params.display_cfg = ¶ms->display_config->display_config; + l->test_mcache.validate_admissibility_params.mcache_allocations = params->display_config->stage2.mcache_allocations; + l->test_mcache.validate_admissibility_params.cfg_support_info = ¶ms->display_config->mode_support_result.cfg_support_info; + + mcache_success = dml2_top_mcache_validate_admissability(&l->test_mcache.validate_admissibility_params); // also find the shift to make mcache allocation works + + memcpy(params->display_config->stage2.per_plane_mcache_support, l->test_mcache.validate_admissibility_params.per_plane_status, sizeof(bool) * DML2_MAX_PLANES); + } + + return mcache_success; +} + +static bool dml2_top_optimization_optimize_function_mcache(const struct optimization_optimize_function_params *params) +{ + struct dml2_optimization_optimize_function_locals *l = params->locals; + bool optimize_success = false; + + if (params->last_candidate_supported == false) + return false; + + copy_display_configuration_with_meta(params->optimized_display_config, params->display_config); + + l->optimize_mcache.optimize_mcache_params.instance = ¶ms->dml->pmo_instance; + l->optimize_mcache.optimize_mcache_params.dcc_mcache_supported = params->display_config->stage2.per_plane_mcache_support; + l->optimize_mcache.optimize_mcache_params.display_config = ¶ms->display_config->display_config; + l->optimize_mcache.optimize_mcache_params.optimized_display_cfg = ¶ms->optimized_display_config->display_config; + l->optimize_mcache.optimize_mcache_params.cfg_support_info = ¶ms->optimized_display_config->mode_support_result.cfg_support_info; + + optimize_success = params->dml->pmo_instance.optimize_dcc_mcache(&l->optimize_mcache.optimize_mcache_params); + + return optimize_success; +} + +static bool dml2_top_optimization_init_function_vmin(const struct optimization_init_function_params *params) +{ + struct dml2_optimization_init_function_locals *l = params->locals; + + l->vmin.init_params.instance = ¶ms->dml->pmo_instance; + l->vmin.init_params.base_display_config = params->display_config; + return params->dml->pmo_instance.init_for_vmin(&l->vmin.init_params); +} + +static bool dml2_top_optimization_test_function_vmin(const struct optimization_test_function_params *params) +{ + struct dml2_optimization_test_function_locals *l = params->locals; + + l->test_vmin.pmo_test_vmin_params.instance = ¶ms->dml->pmo_instance; + l->test_vmin.pmo_test_vmin_params.display_config = params->display_config; + l->test_vmin.pmo_test_vmin_params.vmin_limits = ¶ms->dml->soc_bbox.vmin_limit; + return params->dml->pmo_instance.test_for_vmin(&l->test_vmin.pmo_test_vmin_params); +} + +static bool dml2_top_optimization_optimize_function_vmin(const struct optimization_optimize_function_params *params) +{ + struct dml2_optimization_optimize_function_locals *l = params->locals; + + if (params->last_candidate_supported == false) + return false; + + l->optimize_vmin.pmo_optimize_vmin_params.instance = ¶ms->dml->pmo_instance; + l->optimize_vmin.pmo_optimize_vmin_params.base_display_config = params->display_config; + l->optimize_vmin.pmo_optimize_vmin_params.optimized_display_config = params->optimized_display_config; + return params->dml->pmo_instance.optimize_for_vmin(&l->optimize_vmin.pmo_optimize_vmin_params); +} + +static bool dml2_top_optimization_init_function_uclk_pstate(const struct optimization_init_function_params *params) +{ + struct dml2_optimization_init_function_locals *l = params->locals; + + l->uclk_pstate.init_params.instance = ¶ms->dml->pmo_instance; + l->uclk_pstate.init_params.base_display_config = params->display_config; + + return params->dml->pmo_instance.init_for_uclk_pstate(&l->uclk_pstate.init_params); +} + +static bool dml2_top_optimization_test_function_uclk_pstate(const struct optimization_test_function_params *params) +{ + struct dml2_optimization_test_function_locals *l = params->locals; + + l->uclk_pstate.test_params.instance = ¶ms->dml->pmo_instance; + l->uclk_pstate.test_params.base_display_config = params->display_config; + + return params->dml->pmo_instance.test_for_uclk_pstate(&l->uclk_pstate.test_params); +} + +static bool dml2_top_optimization_optimize_function_uclk_pstate(const struct optimization_optimize_function_params *params) +{ + struct dml2_optimization_optimize_function_locals *l = params->locals; + + l->uclk_pstate.optimize_params.instance = ¶ms->dml->pmo_instance; + l->uclk_pstate.optimize_params.base_display_config = params->display_config; + l->uclk_pstate.optimize_params.optimized_display_config = params->optimized_display_config; + l->uclk_pstate.optimize_params.last_candidate_failed = !params->last_candidate_supported; + + return params->dml->pmo_instance.optimize_for_uclk_pstate(&l->uclk_pstate.optimize_params); +} + +static bool dml2_top_optimization_init_function_stutter(const struct optimization_init_function_params *params) +{ + struct dml2_optimization_init_function_locals *l = params->locals; + + l->uclk_pstate.init_params.instance = ¶ms->dml->pmo_instance; + l->uclk_pstate.init_params.base_display_config = params->display_config; + + return params->dml->pmo_instance.init_for_stutter(&l->stutter.stutter_params); +} + +static bool dml2_top_optimization_test_function_stutter(const struct optimization_test_function_params *params) +{ + struct dml2_optimization_test_function_locals *l = params->locals; + + l->stutter.stutter_params.instance = ¶ms->dml->pmo_instance; + l->stutter.stutter_params.base_display_config = params->display_config; + return params->dml->pmo_instance.test_for_stutter(&l->stutter.stutter_params); +} + +static bool dml2_top_optimization_optimize_function_stutter(const struct optimization_optimize_function_params *params) +{ + struct dml2_optimization_optimize_function_locals *l = params->locals; + + l->stutter.stutter_params.instance = ¶ms->dml->pmo_instance; + l->stutter.stutter_params.base_display_config = params->display_config; + l->stutter.stutter_params.optimized_display_config = params->optimized_display_config; + l->stutter.stutter_params.last_candidate_failed = !params->last_candidate_supported; + return params->dml->pmo_instance.optimize_for_stutter(&l->stutter.stutter_params); +} + +static bool dml2_top_optimization_perform_optimization_phase(struct dml2_optimization_phase_locals *l, const struct optimization_phase_params *params) +{ + bool test_passed = false; + bool optimize_succeeded = true; + bool candidate_validation_passed = true; + struct optimization_init_function_params init_params = { 0 }; + struct optimization_test_function_params test_params = { 0 }; + struct optimization_optimize_function_params optimize_params = { 0 }; + + if (!params->dml || + !params->optimize_function || + !params->test_function || + !params->display_config || + !params->optimized_display_config) + return false; + + copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, params->display_config); + + init_params.locals = &l->init_function_locals; + init_params.dml = params->dml; + init_params.display_config = &l->cur_candidate_display_cfg; + + if (params->init_function && !params->init_function(&init_params)) + return false; + + test_params.locals = &l->test_function_locals; + test_params.dml = params->dml; + test_params.display_config = &l->cur_candidate_display_cfg; + + test_passed = params->test_function(&test_params); + + while (!test_passed && optimize_succeeded) { + memset(&optimize_params, 0, sizeof(struct optimization_optimize_function_params)); + + optimize_params.locals = &l->optimize_function_locals; + optimize_params.dml = params->dml; + optimize_params.display_config = &l->cur_candidate_display_cfg; + optimize_params.optimized_display_config = &l->next_candidate_display_cfg; + optimize_params.last_candidate_supported = candidate_validation_passed; + + optimize_succeeded = params->optimize_function(&optimize_params); + + if (optimize_succeeded) { + l->mode_support_params.instance = ¶ms->dml->core_instance; + l->mode_support_params.display_cfg = &l->next_candidate_display_cfg; + l->mode_support_params.min_clk_table = ¶ms->dml->min_clk_table; + + if (l->next_candidate_display_cfg.stage3.performed) + l->mode_support_params.min_clk_index = l->next_candidate_display_cfg.stage3.min_clk_index_for_latency; + else + l->mode_support_params.min_clk_index = l->next_candidate_display_cfg.stage1.min_clk_index_for_latency; + candidate_validation_passed = params->dml->core_instance.mode_support(&l->mode_support_params); + l->next_candidate_display_cfg.mode_support_result = l->mode_support_params.mode_support_result; + } + + if (optimize_succeeded && candidate_validation_passed) { + memset(&test_params, 0, sizeof(struct optimization_test_function_params)); + test_params.locals = &l->test_function_locals; + test_params.dml = params->dml; + test_params.display_config = &l->next_candidate_display_cfg; + test_passed = params->test_function(&test_params); + + copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, &l->next_candidate_display_cfg); + + // If optimization is not all or nothing, then store partial progress in output + if (!params->all_or_nothing) + copy_display_configuration_with_meta(params->optimized_display_config, &l->next_candidate_display_cfg); + } + } + + if (test_passed) + copy_display_configuration_with_meta(params->optimized_display_config, &l->cur_candidate_display_cfg); + + return test_passed; +} + +static bool dml2_top_optimization_perform_optimization_phase_1(struct dml2_optimization_phase_locals *l, const struct optimization_phase_params *params) +{ + int highest_state, lowest_state, cur_state; + bool supported = false; + + if (!params->dml || + !params->optimize_function || + !params->test_function || + !params->display_config || + !params->optimized_display_config) + return false; + + copy_display_configuration_with_meta(&l->cur_candidate_display_cfg, params->display_config); + highest_state = l->cur_candidate_display_cfg.stage1.min_clk_index_for_latency; + lowest_state = 0; + + while (highest_state > lowest_state) { + cur_state = (highest_state + lowest_state) / 2; + + l->mode_support_params.instance = ¶ms->dml->core_instance; + l->mode_support_params.display_cfg = &l->cur_candidate_display_cfg; + l->mode_support_params.min_clk_table = ¶ms->dml->min_clk_table; + l->mode_support_params.min_clk_index = cur_state; + supported = params->dml->core_instance.mode_support(&l->mode_support_params); + + if (supported) { + l->cur_candidate_display_cfg.mode_support_result = l->mode_support_params.mode_support_result; + highest_state = cur_state; + } else { + lowest_state = cur_state + 1; + } + } + l->cur_candidate_display_cfg.stage1.min_clk_index_for_latency = lowest_state; + + copy_display_configuration_with_meta(params->optimized_display_config, &l->cur_candidate_display_cfg); + + return true; +} + +/* +* Takes an input set of mcache boundaries and finds the appropriate setting of cache programming. +* Returns true if a valid set of programming can be made, and false otherwise. "Valid" means +* that the horizontal viewport does not span more than 2 cache slices. +* +* It optionally also can apply a constant shift to all the cache boundaries. +*/ +static const uint32_t MCACHE_ID_UNASSIGNED = 0xF; +static const uint32_t SPLIT_LOCATION_UNDEFINED = 0xFFFF; + +static bool calculate_first_second_splitting(const int *mcache_boundaries, int num_boundaries, int shift, + int pipe_h_vp_start, int pipe_h_vp_end, int *first_offset, int *second_offset) +{ + const int MAX_VP = 0xFFFFFF; + int left_cache_id; + int right_cache_id; + int range_start; + int range_end; + bool success = false; + + if (num_boundaries <= 1) { + if (first_offset && second_offset) { + *first_offset = 0; + *second_offset = -1; + } + success = true; + return success; + } else { + range_start = 0; + for (left_cache_id = 0; left_cache_id < num_boundaries; left_cache_id++) { + range_end = mcache_boundaries[left_cache_id] - shift - 1; + + if (range_start <= pipe_h_vp_start && pipe_h_vp_start <= range_end) + break; + + range_start = range_end + 1; + } + + range_end = MAX_VP; + for (right_cache_id = num_boundaries - 1; right_cache_id >= -1; right_cache_id--) { + if (right_cache_id >= 0) + range_start = mcache_boundaries[right_cache_id] - shift; + else + range_start = 0; + + if (range_start <= pipe_h_vp_end && pipe_h_vp_end <= range_end) { + break; + } + range_end = range_start - 1; + } + right_cache_id = (right_cache_id + 1) % num_boundaries; + + if (right_cache_id == left_cache_id) { + if (first_offset && second_offset) { + *first_offset = left_cache_id; + *second_offset = -1; + } + success = true; + } else if (right_cache_id == (left_cache_id + 1) % num_boundaries) { + if (first_offset && second_offset) { + *first_offset = left_cache_id; + *second_offset = right_cache_id; + } + success = true; + } + } + + return success; +} + +/* +* For a given set of pipe start/end x positions, checks to see it can support the input mcache splitting. +* It also attempts to "optimize" by finding a shift if the default 0 shift does not work. +*/ +static bool find_shift_for_valid_cache_id_assignment(int *mcache_boundaries, unsigned int num_boundaries, + int *pipe_vp_startx, int *pipe_vp_endx, unsigned int pipe_count, int shift_granularity, int *shift) +{ + int max_shift = 0xFFFF; + unsigned int pipe_index; + unsigned int i, slice_width; + bool success = false; + + for (i = 0; i < num_boundaries; i++) { + if (i == 0) + slice_width = mcache_boundaries[i]; + else + slice_width = mcache_boundaries[i] - mcache_boundaries[i - 1]; + + if (max_shift > (int)slice_width) { + max_shift = slice_width; + } + } + + for (*shift = 0; *shift <= max_shift; *shift += shift_granularity) { + success = true; + for (pipe_index = 0; pipe_index < pipe_count; pipe_index++) { + if (!calculate_first_second_splitting(mcache_boundaries, num_boundaries, *shift, + pipe_vp_startx[pipe_index], pipe_vp_endx[pipe_index], 0, 0)) { + success = false; + break; + } + } + if (success) + break; + } + + return success; +} + +/* +* Counts the number of elements inside input array within the given span length. +* Formally, what is the size of the largest subset of the array where the largest and smallest element +* differ no more than the span. +*/ +static unsigned int count_elements_in_span(int *array, unsigned int array_size, unsigned int span) +{ + unsigned int i; + unsigned int span_start_value; + unsigned int span_start_index; + unsigned int greatest_element_count; + + if (array_size == 0) + return 1; + + if (span == 0) + return array_size > 0 ? 1 : 0; + + span_start_value = 0; + span_start_index = 0; + greatest_element_count = 0; + + while (span_start_index < array_size) { + for (i = span_start_index; i < array_size; i++) { + if (array[i] - span_start_value <= span) { + if (i - span_start_index + 1 > greatest_element_count) { + greatest_element_count = i - span_start_index + 1; + } + } else + break; + } + + span_start_index++; + + if (span_start_index < array_size) { + span_start_value = array[span_start_index - 1] + 1; + } + } + + return greatest_element_count; +} + +static bool calculate_h_split_for_scaling_transform(int full_vp_width, int h_active, int num_pipes, + enum dml2_scaling_transform scaling_transform, int *pipe_vp_x_start, int *pipe_vp_x_end) +{ + int i, slice_width; + const char MAX_SCL_VP_OVERLAP = 3; + bool success = false; + + switch (scaling_transform) { + case dml2_scaling_transform_centered: + case dml2_scaling_transform_aspect_ratio: + case dml2_scaling_transform_fullscreen: + slice_width = full_vp_width / num_pipes; + for (i = 0; i < num_pipes; i++) { + pipe_vp_x_start[i] = i * slice_width; + pipe_vp_x_end[i] = (i + 1) * slice_width - 1; + + if (pipe_vp_x_start[i] < MAX_SCL_VP_OVERLAP) + pipe_vp_x_start[i] = 0; + else + pipe_vp_x_start[i] -= MAX_SCL_VP_OVERLAP; + + if (pipe_vp_x_end[i] > full_vp_width - MAX_SCL_VP_OVERLAP - 1) + pipe_vp_x_end[i] = full_vp_width - 1; + else + pipe_vp_x_end[i] += MAX_SCL_VP_OVERLAP; + } + break; + case dml2_scaling_transform_explicit: + default: + success = false; + break; + } + + return success; +} + +bool dml2_top_mcache_validate_admissability(struct top_mcache_validate_admissability_in_out *params) +{ + struct dml2_instance *dml = (struct dml2_instance *)params->dml2_instance; + struct dml2_top_mcache_validate_admissability_locals *l = &dml->scratch.mcache_validate_admissability_locals; + + const int MAX_PIXEL_OVERLAP = 6; + int max_per_pipe_vp_p0 = 0; + int max_per_pipe_vp_p1 = 0; + int temp, p0shift, p1shift; + unsigned int plane_index = 0; + unsigned int i; + unsigned int odm_combine_factor; + unsigned int mpc_combine_factor; + unsigned int num_dpps; + unsigned int num_boundaries; + enum dml2_scaling_transform scaling_transform; + const struct dml2_plane_parameters *plane; + const struct dml2_stream_parameters *stream; + + bool p0pass = false; + bool p1pass = false; + bool all_pass = true; + + for (plane_index = 0; plane_index < params->display_cfg->num_planes; plane_index++) { + if (!params->display_cfg->plane_descriptors[plane_index].surface.dcc.enable) + continue; + + plane = ¶ms->display_cfg->plane_descriptors[plane_index]; + stream = ¶ms->display_cfg->stream_descriptors[plane->stream_index]; + + num_dpps = odm_combine_factor = params->cfg_support_info->stream_support_info[plane->stream_index].odms_used; + + if (odm_combine_factor == 1) + num_dpps = mpc_combine_factor = (unsigned int)params->cfg_support_info->plane_support_info[plane_index].dpps_used; + else + mpc_combine_factor = 1; + + if (odm_combine_factor > 1) { + max_per_pipe_vp_p0 = plane->surface.plane0.width; + temp = (unsigned int)math_ceil(plane->composition.scaler_info.plane0.h_ratio * stream->timing.h_active / odm_combine_factor); + + if (temp < max_per_pipe_vp_p0) + max_per_pipe_vp_p0 = temp; + + max_per_pipe_vp_p1 = plane->surface.plane1.width; + temp = (unsigned int)math_ceil(plane->composition.scaler_info.plane1.h_ratio * stream->timing.h_active / odm_combine_factor); + + if (temp < max_per_pipe_vp_p1) + max_per_pipe_vp_p1 = temp; + } else { + max_per_pipe_vp_p0 = plane->surface.plane0.width / mpc_combine_factor; + max_per_pipe_vp_p1 = plane->surface.plane1.width / mpc_combine_factor; + } + + max_per_pipe_vp_p0 += 2 * MAX_PIXEL_OVERLAP; + max_per_pipe_vp_p1 += MAX_PIXEL_OVERLAP; + + p0shift = 0; + p1shift = 0; + + // The last element in the unshifted boundary array will always be the first pixel outside the + // plane, which means theres no mcache associated with it, so -1 + num_boundaries = params->mcache_allocations[plane_index].num_mcaches_plane0 == 0 ? 0 : params->mcache_allocations[plane_index].num_mcaches_plane0 - 1; + if ((count_elements_in_span(params->mcache_allocations[plane_index].mcache_x_offsets_plane0, + num_boundaries, max_per_pipe_vp_p0) <= 1) && (num_boundaries <= num_dpps)) { + p0pass = true; + } + num_boundaries = params->mcache_allocations[plane_index].num_mcaches_plane1 == 0 ? 0 : params->mcache_allocations[plane_index].num_mcaches_plane1 - 1; + if ((count_elements_in_span(params->mcache_allocations[plane_index].mcache_x_offsets_plane1, + num_boundaries, max_per_pipe_vp_p1) <= 1) && (num_boundaries <= num_dpps)) { + p1pass = true; + } + + if (!p0pass || !p1pass) { + if (odm_combine_factor > 1) { + num_dpps = odm_combine_factor; + scaling_transform = plane->composition.scaling_transform; + } else { + num_dpps = mpc_combine_factor; + scaling_transform = dml2_scaling_transform_fullscreen; + } + + if (!p0pass) { + if (plane->composition.viewport.stationary) { + calculate_h_split_for_scaling_transform(plane->surface.plane0.width, + stream->timing.h_active, num_dpps, scaling_transform, + &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index]); + p0pass = find_shift_for_valid_cache_id_assignment(params->mcache_allocations[plane_index].mcache_x_offsets_plane0, + params->mcache_allocations[plane_index].num_mcaches_plane0, + &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index], num_dpps, + params->mcache_allocations[plane_index].shift_granularity.p0, &p0shift); + } + } + if (!p1pass) { + if (plane->composition.viewport.stationary) { + calculate_h_split_for_scaling_transform(plane->surface.plane1.width, + stream->timing.h_active, num_dpps, scaling_transform, + &l->plane0.pipe_vp_startx[plane_index], &l->plane0.pipe_vp_endx[plane_index]); + p1pass = find_shift_for_valid_cache_id_assignment(params->mcache_allocations[plane_index].mcache_x_offsets_plane1, + params->mcache_allocations[plane_index].num_mcaches_plane1, + &l->plane1.pipe_vp_startx[plane_index], &l->plane1.pipe_vp_endx[plane_index], num_dpps, + params->mcache_allocations[plane_index].shift_granularity.p1, &p1shift); + } + } + } + + if (p0pass && p1pass) { + for (i = 0; i < params->mcache_allocations[plane_index].num_mcaches_plane0; i++) { + params->mcache_allocations[plane_index].mcache_x_offsets_plane0[i] -= p0shift; + } + for (i = 0; i < params->mcache_allocations[plane_index].num_mcaches_plane1; i++) { + params->mcache_allocations[plane_index].mcache_x_offsets_plane1[i] -= p1shift; + } + } + + params->per_plane_status[plane_index] = p0pass && p1pass; + all_pass &= p0pass && p1pass; + } + + return all_pass; +} + +static void reset_mcache_allocations(struct dml2_hubp_pipe_mcache_regs *per_plane_pipe_mcache_regs) +{ + // Initialize all entries to special valid MCache ID and special valid split coordinate + per_plane_pipe_mcache_regs->main.p0.mcache_id_first = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->main.p0.mcache_id_second = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->main.p0.split_location = SPLIT_LOCATION_UNDEFINED; + + per_plane_pipe_mcache_regs->mall.p0.mcache_id_first = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->mall.p0.mcache_id_second = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->mall.p0.split_location = SPLIT_LOCATION_UNDEFINED; + + per_plane_pipe_mcache_regs->main.p1.mcache_id_first = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->main.p1.mcache_id_second = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->main.p1.split_location = SPLIT_LOCATION_UNDEFINED; + + per_plane_pipe_mcache_regs->mall.p1.mcache_id_first = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->mall.p1.mcache_id_second = MCACHE_ID_UNASSIGNED; + per_plane_pipe_mcache_regs->mall.p1.split_location = SPLIT_LOCATION_UNDEFINED; +} + +void dml2_top_mcache_assign_global_mcache_ids(struct top_mcache_assign_global_mcache_ids_in_out *params) +{ + int i; + unsigned int j; + int next_unused_cache_id = 0; + + for (i = 0; i < params->num_allocations; i++) { + if (!params->allocations[i].valid) + continue; + + for (j = 0; j < params->allocations[i].num_mcaches_plane0; j++) { + params->allocations[i].global_mcache_ids_plane0[j] = next_unused_cache_id++; + } + for (j = 0; j < params->allocations[i].num_mcaches_plane1; j++) { + params->allocations[i].global_mcache_ids_plane1[j] = next_unused_cache_id++; + } + + // The "psuedo-last" slice is always wrapped around + params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0] = + params->allocations[i].global_mcache_ids_plane0[0]; + params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1] = + params->allocations[i].global_mcache_ids_plane1[0]; + + // If we need dedicated caches for mall requesting, then we assign them here. + if (params->allocations[i].requires_dedicated_mall_mcache) { + for (j = 0; j < params->allocations[i].num_mcaches_plane0; j++) { + params->allocations[i].global_mcache_ids_mall_plane0[j] = next_unused_cache_id++; + } + for (j = 0; j < params->allocations[i].num_mcaches_plane1; j++) { + params->allocations[i].global_mcache_ids_mall_plane1[j] = next_unused_cache_id++; + } + + // The "psuedo-last" slice is always wrapped around + params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0] = + params->allocations[i].global_mcache_ids_mall_plane0[0]; + params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1] = + params->allocations[i].global_mcache_ids_mall_plane1[0]; + } + + // If P0 and P1 are sharing caches, then it means the largest mcache IDs for p0 and p1 can be the same + // since mcache IDs are always ascending, then it means the largest mcacheID of p1 should be the + // largest mcacheID of P0 + if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].num_mcaches_plane1 > 0 && + params->allocations[i].last_slice_sharing.plane0_plane1) { + params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1 - 1] = + params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0 - 1]; + } + + // If we need dedicated caches handle last slice sharing + if (params->allocations[i].requires_dedicated_mall_mcache) { + if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].num_mcaches_plane1 > 0 && + params->allocations[i].last_slice_sharing.plane0_plane1) { + params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1 - 1] = + params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0 - 1]; + } + // If mall_comb_mcache_l is set then it means that largest mcache ID for MALL p0 can be same as regular read p0 + if (params->allocations[i].num_mcaches_plane0 > 0 && params->allocations[i].last_slice_sharing.mall_comb_mcache_p0) { + params->allocations[i].global_mcache_ids_mall_plane0[params->allocations[i].num_mcaches_plane0 - 1] = + params->allocations[i].global_mcache_ids_plane0[params->allocations[i].num_mcaches_plane0 - 1]; + } + // If mall_comb_mcache_c is set then it means that largest mcache ID for MALL p1 can be same as regular + // read p1 (which can be same as regular read p0 if plane0_plane1 is also set) + if (params->allocations[i].num_mcaches_plane1 > 0 && params->allocations[i].last_slice_sharing.mall_comb_mcache_p1) { + params->allocations[i].global_mcache_ids_mall_plane1[params->allocations[i].num_mcaches_plane1 - 1] = + params->allocations[i].global_mcache_ids_plane1[params->allocations[i].num_mcaches_plane1 - 1]; + } + } + + // If you don't need dedicated mall mcaches, the mall mcache assignments are identical to the normal requesting + if (!params->allocations[i].requires_dedicated_mall_mcache) { + memcpy(params->allocations[i].global_mcache_ids_mall_plane0, params->allocations[i].global_mcache_ids_plane0, + sizeof(params->allocations[i].global_mcache_ids_mall_plane0)); + memcpy(params->allocations[i].global_mcache_ids_mall_plane1, params->allocations[i].global_mcache_ids_plane1, + sizeof(params->allocations[i].global_mcache_ids_mall_plane1)); + } + } +} + +bool dml2_top_mcache_calc_mcache_count_and_offsets(struct top_mcache_calc_mcache_count_and_offsets_in_out *params) +{ + struct dml2_instance *dml = (struct dml2_instance *)params->dml2_instance; + struct dml2_top_mcache_verify_mcache_size_locals *l = &dml->scratch.mcache_verify_mcache_size_locals; + + unsigned int total_mcaches_required; + unsigned int i; + bool result = false; + + if (dml->soc_bbox.num_dcc_mcaches == 0) { + return true; + } + + total_mcaches_required = 0; + l->calc_mcache_params.instance = &dml->core_instance; + for (i = 0; i < params->display_config->num_planes; i++) { + if (!params->display_config->plane_descriptors[i].surface.dcc.enable) { + memset(¶ms->mcache_allocations[i], 0, sizeof(struct dml2_mcache_surface_allocation)); + continue; + } + + l->calc_mcache_params.plane_descriptor = ¶ms->display_config->plane_descriptors[i]; + l->calc_mcache_params.mcache_allocation = ¶ms->mcache_allocations[i]; + l->calc_mcache_params.plane_index = i; + + if (!dml->core_instance.calculate_mcache_allocation(&l->calc_mcache_params)) { + result = false; + break; + } + + if (params->mcache_allocations[i].valid) { + total_mcaches_required += params->mcache_allocations[i].num_mcaches_plane0 + params->mcache_allocations[i].num_mcaches_plane1; + if (params->mcache_allocations[i].last_slice_sharing.plane0_plane1) + total_mcaches_required--; + } + } + DML_LOG_VERBOSE("DML_CORE_DCN3::%s: plane_%d, total_mcaches_required=%d\n", __func__, i, total_mcaches_required); + + if (total_mcaches_required > dml->soc_bbox.num_dcc_mcaches) { + result = false; + } else { + result = true; + } + + return result; +} + +static bool dml2_top_soc15_check_mode_supported(struct dml2_check_mode_supported_in_out *in_out) +{ + struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; + struct dml2_check_mode_supported_locals *l = &dml->scratch.check_mode_supported_locals; + struct dml2_display_cfg_programming *dpmm_programming = &dml->dpmm_instance.dpmm_scratch.programming; + + bool result = false; + bool mcache_success = false; + memset(dpmm_programming, 0, sizeof(struct dml2_display_cfg_programming)); + + setup_unoptimized_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); + + l->mode_support_params.instance = &dml->core_instance; + l->mode_support_params.display_cfg = &l->base_display_config_with_meta; + l->mode_support_params.min_clk_table = &dml->min_clk_table; + l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; + result = dml->core_instance.mode_support(&l->mode_support_params); + l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; + + if (result) { + struct optimization_phase_params mcache_phase = { + .dml = dml, + .display_config = &l->base_display_config_with_meta, + .test_function = dml2_top_optimization_test_function_mcache, + .optimize_function = dml2_top_optimization_optimize_function_mcache, + .optimized_display_config = &l->optimized_display_config_with_meta, + .all_or_nothing = false, + }; + mcache_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &mcache_phase); + } + + /* + * Call DPMM to map all requirements to minimum clock state + */ + if (result) { + l->dppm_map_mode_params.min_clk_table = &dml->min_clk_table; + l->dppm_map_mode_params.display_cfg = &l->base_display_config_with_meta; + l->dppm_map_mode_params.programming = dpmm_programming; + l->dppm_map_mode_params.soc_bb = &dml->soc_bbox; + l->dppm_map_mode_params.ip = &dml->core_instance.clean_me_up.mode_lib.ip; + result = dml->dpmm_instance.map_mode_to_soc_dpm(&l->dppm_map_mode_params); + } + + in_out->is_supported = mcache_success; + result = result && in_out->is_supported; + + return result; +} + +static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_programming_in_out *in_out) +{ + struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; + struct dml2_build_mode_programming_locals *l = &dml->scratch.build_mode_programming_locals; + + bool result = false; + bool mcache_success = false; + bool uclk_pstate_success = false; + bool vmin_success = false; + bool stutter_success = false; + + memset(l, 0, sizeof(struct dml2_build_mode_programming_locals)); + memset(in_out->programming, 0, sizeof(struct dml2_display_cfg_programming)); + + memcpy(&in_out->programming->display_config, in_out->display_config, sizeof(struct dml2_display_cfg)); + + setup_speculative_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); + + l->mode_support_params.instance = &dml->core_instance; + l->mode_support_params.display_cfg = &l->base_display_config_with_meta; + l->mode_support_params.min_clk_table = &dml->min_clk_table; + l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; + result = dml->core_instance.mode_support(&l->mode_support_params); + + l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; + + if (!result) { + setup_unoptimized_display_config_with_meta(dml, &l->base_display_config_with_meta, in_out->display_config); + + l->mode_support_params.instance = &dml->core_instance; + l->mode_support_params.display_cfg = &l->base_display_config_with_meta; + l->mode_support_params.min_clk_table = &dml->min_clk_table; + l->mode_support_params.min_clk_index = l->base_display_config_with_meta.stage1.min_clk_index_for_latency; + result = dml->core_instance.mode_support(&l->mode_support_params); + l->base_display_config_with_meta.mode_support_result = l->mode_support_params.mode_support_result; + + if (!result) { + l->informative_params.instance = &dml->core_instance; + l->informative_params.programming = in_out->programming; + l->informative_params.mode_is_supported = false; + dml->core_instance.populate_informative(&l->informative_params); + + return false; + } + + /* + * Phase 1: Determine minimum clocks to satisfy latency requirements for this mode + */ + memset(&l->min_clock_for_latency_phase, 0, sizeof(struct optimization_phase_params)); + l->min_clock_for_latency_phase.dml = dml; + l->min_clock_for_latency_phase.display_config = &l->base_display_config_with_meta; + l->min_clock_for_latency_phase.init_function = dml2_top_optimization_init_function_min_clk_for_latency; + l->min_clock_for_latency_phase.test_function = dml2_top_optimization_test_function_min_clk_for_latency; + l->min_clock_for_latency_phase.optimize_function = dml2_top_optimization_optimize_function_min_clk_for_latency; + l->min_clock_for_latency_phase.optimized_display_config = &l->optimized_display_config_with_meta; + l->min_clock_for_latency_phase.all_or_nothing = false; + + dml2_top_optimization_perform_optimization_phase_1(&l->optimization_phase_locals, &l->min_clock_for_latency_phase); + + memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); + } + + /* + * Phase 2: Satisfy DCC mcache requirements + */ + memset(&l->mcache_phase, 0, sizeof(struct optimization_phase_params)); + l->mcache_phase.dml = dml; + l->mcache_phase.display_config = &l->base_display_config_with_meta; + l->mcache_phase.test_function = dml2_top_optimization_test_function_mcache; + l->mcache_phase.optimize_function = dml2_top_optimization_optimize_function_mcache; + l->mcache_phase.optimized_display_config = &l->optimized_display_config_with_meta; + l->mcache_phase.all_or_nothing = true; + + mcache_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->mcache_phase); + + if (!mcache_success) { + l->informative_params.instance = &dml->core_instance; + l->informative_params.programming = in_out->programming; + l->informative_params.mode_is_supported = false; + + dml->core_instance.populate_informative(&l->informative_params); + + in_out->programming->informative.failed_mcache_validation = true; + return false; + } + + memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); + + /* + * Phase 3: Optimize for Pstate + */ + memset(&l->uclk_pstate_phase, 0, sizeof(struct optimization_phase_params)); + l->uclk_pstate_phase.dml = dml; + l->uclk_pstate_phase.display_config = &l->base_display_config_with_meta; + l->uclk_pstate_phase.init_function = dml2_top_optimization_init_function_uclk_pstate; + l->uclk_pstate_phase.test_function = dml2_top_optimization_test_function_uclk_pstate; + l->uclk_pstate_phase.optimize_function = dml2_top_optimization_optimize_function_uclk_pstate; + l->uclk_pstate_phase.optimized_display_config = &l->optimized_display_config_with_meta; + l->uclk_pstate_phase.all_or_nothing = true; + + uclk_pstate_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->uclk_pstate_phase); + + if (uclk_pstate_success) { + memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); + l->base_display_config_with_meta.stage3.success = true; + } + + /* + * Phase 4: Optimize for Vmin + */ + memset(&l->vmin_phase, 0, sizeof(struct optimization_phase_params)); + l->vmin_phase.dml = dml; + l->vmin_phase.display_config = &l->base_display_config_with_meta; + l->vmin_phase.init_function = dml2_top_optimization_init_function_vmin; + l->vmin_phase.test_function = dml2_top_optimization_test_function_vmin; + l->vmin_phase.optimize_function = dml2_top_optimization_optimize_function_vmin; + l->vmin_phase.optimized_display_config = &l->optimized_display_config_with_meta; + l->vmin_phase.all_or_nothing = false; + + vmin_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->vmin_phase); + + if (l->optimized_display_config_with_meta.stage4.performed) { + /* + * when performed is true, optimization has applied to + * optimized_display_config_with_meta and it has passed mode + * support. However it may or may not pass the test function to + * reach actual Vmin. As long as voltage is optimized even if it + * doesn't reach Vmin level, there is still power benefit so in + * this case we will still copy this optimization into base + * display config. + */ + memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); + l->base_display_config_with_meta.stage4.success = vmin_success; + } + + /* + * Phase 5: Optimize for Stutter + */ + memset(&l->stutter_phase, 0, sizeof(struct optimization_phase_params)); + l->stutter_phase.dml = dml; + l->stutter_phase.display_config = &l->base_display_config_with_meta; + l->stutter_phase.init_function = dml2_top_optimization_init_function_stutter; + l->stutter_phase.test_function = dml2_top_optimization_test_function_stutter; + l->stutter_phase.optimize_function = dml2_top_optimization_optimize_function_stutter; + l->stutter_phase.optimized_display_config = &l->optimized_display_config_with_meta; + l->stutter_phase.all_or_nothing = true; + + stutter_success = dml2_top_optimization_perform_optimization_phase(&l->optimization_phase_locals, &l->stutter_phase); + + if (stutter_success) { + memcpy(&l->base_display_config_with_meta, &l->optimized_display_config_with_meta, sizeof(struct display_configuation_with_meta)); + l->base_display_config_with_meta.stage5.success = true; + } + + /* + * Call DPMM to map all requirements to minimum clock state + */ + if (result) { + l->dppm_map_mode_params.min_clk_table = &dml->min_clk_table; + l->dppm_map_mode_params.display_cfg = &l->base_display_config_with_meta; + l->dppm_map_mode_params.programming = in_out->programming; + l->dppm_map_mode_params.soc_bb = &dml->soc_bbox; + l->dppm_map_mode_params.ip = &dml->core_instance.clean_me_up.mode_lib.ip; + result = dml->dpmm_instance.map_mode_to_soc_dpm(&l->dppm_map_mode_params); + if (!result) + in_out->programming->informative.failed_dpmm = true; + } + + if (result) { + l->mode_programming_params.instance = &dml->core_instance; + l->mode_programming_params.display_cfg = &l->base_display_config_with_meta; + l->mode_programming_params.cfg_support_info = &l->base_display_config_with_meta.mode_support_result.cfg_support_info; + l->mode_programming_params.programming = in_out->programming; + result = dml->core_instance.mode_programming(&l->mode_programming_params); + if (!result) + in_out->programming->informative.failed_mode_programming = true; + } + + if (result) { + l->dppm_map_watermarks_params.core = &dml->core_instance; + l->dppm_map_watermarks_params.display_cfg = &l->base_display_config_with_meta; + l->dppm_map_watermarks_params.programming = in_out->programming; + result = dml->dpmm_instance.map_watermarks(&l->dppm_map_watermarks_params); + } + + l->informative_params.instance = &dml->core_instance; + l->informative_params.programming = in_out->programming; + l->informative_params.mode_is_supported = result; + + dml->core_instance.populate_informative(&l->informative_params); + + return result; +} + +bool dml2_top_soc15_build_mcache_programming(struct dml2_build_mcache_programming_in_out *params) +{ + bool success = true; + int config_index, pipe_index; + int first_offset, second_offset; + int free_per_plane_reg_index = 0; + + memset(params->per_plane_pipe_mcache_regs, 0, DML2_MAX_PLANES * DML2_MAX_DCN_PIPES * sizeof(struct dml2_hubp_pipe_mcache_regs *)); + + for (config_index = 0; config_index < params->num_configurations; config_index++) { + for (pipe_index = 0; pipe_index < params->mcache_configurations[config_index].num_pipes; pipe_index++) { + // Allocate storage for the mcache regs + params->per_plane_pipe_mcache_regs[config_index][pipe_index] = ¶ms->mcache_regs_set[free_per_plane_reg_index++]; + + reset_mcache_allocations(params->per_plane_pipe_mcache_regs[config_index][pipe_index]); + + if (params->mcache_configurations[config_index].plane_descriptor->surface.dcc.enable) { + // P0 always enabled + if (!calculate_first_second_splitting(params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0, + params->mcache_configurations[config_index].mcache_allocation->num_mcaches_plane0, + 0, + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_x_start, + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_x_start + + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane0.viewport_width - 1, + &first_offset, &second_offset)) { + success = false; + break; + } + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.mcache_id_first = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane0[first_offset]; + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.mcache_id_first = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane0[first_offset]; + + if (second_offset >= 0) { + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.mcache_id_second = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane0[second_offset]; + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p0.split_location = + params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0[first_offset] - 1; + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.mcache_id_second = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane0[second_offset]; + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p0.split_location = + params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane0[first_offset] - 1; + } + + // Populate P1 if enabled + if (params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1_enabled) { + if (!calculate_first_second_splitting(params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1, + params->mcache_configurations[config_index].mcache_allocation->num_mcaches_plane1, + 0, + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_x_start, + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_x_start + + params->mcache_configurations[config_index].pipe_configurations[pipe_index].plane1.viewport_width - 1, + &first_offset, &second_offset)) { + success = false; + break; + } + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.mcache_id_first = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane1[first_offset]; + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.mcache_id_first = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane1[first_offset]; + + if (second_offset >= 0) { + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.mcache_id_second = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_plane1[second_offset]; + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->main.p1.split_location = + params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1[first_offset] - 1; + + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.mcache_id_second = + params->mcache_configurations[config_index].mcache_allocation->global_mcache_ids_mall_plane1[second_offset]; + params->per_plane_pipe_mcache_regs[config_index][pipe_index]->mall.p1.split_location = + params->mcache_configurations[config_index].mcache_allocation->mcache_x_offsets_plane1[first_offset] - 1; + } + } + } + } + } + + return success; +} + +static const struct dml2_top_funcs soc15_funcs = { + .check_mode_supported = dml2_top_soc15_check_mode_supported, + .build_mode_programming = dml2_top_soc15_build_mode_programming, + .build_mcache_programming = dml2_top_soc15_build_mcache_programming, +}; + +bool dml2_top_soc15_initialize_instance(struct dml2_initialize_instance_in_out *in_out) +{ + struct dml2_instance *dml = (struct dml2_instance *)in_out->dml2_instance; + struct dml2_initialize_instance_locals *l = &dml->scratch.initialize_instance_locals; + struct dml2_core_initialize_in_out core_init_params = { 0 }; + struct dml2_mcg_build_min_clock_table_params_in_out mcg_build_min_clk_params = { 0 }; + struct dml2_pmo_initialize_in_out pmo_init_params = { 0 }; + bool result = false; + + memset(l, 0, sizeof(struct dml2_initialize_instance_locals)); + memset(dml, 0, sizeof(struct dml2_instance)); + + memcpy(&dml->ip_caps, &in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); + memcpy(&dml->soc_bbox, &in_out->soc_bb, sizeof(struct dml2_soc_bb)); + + dml->project_id = in_out->options.project_id; + dml->pmo_options = in_out->options.pmo_options; + + // Initialize All Components + result = dml2_mcg_create(in_out->options.project_id, &dml->mcg_instance); + + if (result) + result = dml2_dpmm_create(in_out->options.project_id, &dml->dpmm_instance); + + if (result) + result = dml2_core_create(in_out->options.project_id, &dml->core_instance); + + if (result) { + mcg_build_min_clk_params.soc_bb = &in_out->soc_bb; + mcg_build_min_clk_params.min_clk_table = &dml->min_clk_table; + result = dml->mcg_instance.build_min_clock_table(&mcg_build_min_clk_params); + } + + if (result) { + core_init_params.project_id = in_out->options.project_id; + core_init_params.instance = &dml->core_instance; + core_init_params.minimum_clock_table = &dml->min_clk_table; + core_init_params.explicit_ip_bb = in_out->overrides.explicit_ip_bb; + core_init_params.explicit_ip_bb_size = in_out->overrides.explicit_ip_bb_size; + core_init_params.ip_caps = &in_out->ip_caps; + core_init_params.soc_bb = &in_out->soc_bb; + result = dml->core_instance.initialize(&core_init_params); + + if (core_init_params.explicit_ip_bb && core_init_params.explicit_ip_bb_size > 0) { + memcpy(&dml->ip_caps, &in_out->ip_caps, sizeof(struct dml2_ip_capabilities)); + } + } + + if (result) + result = dml2_pmo_create(in_out->options.project_id, &dml->pmo_instance); + + if (result) { + pmo_init_params.instance = &dml->pmo_instance; + pmo_init_params.soc_bb = &dml->soc_bbox; + pmo_init_params.ip_caps = &dml->ip_caps; + pmo_init_params.mcg_clock_table_size = dml->min_clk_table.dram_bw_table.num_entries; + pmo_init_params.options = &dml->pmo_options; + dml->pmo_instance.initialize(&pmo_init_params); + } + dml->funcs = soc15_funcs; + return result; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h new file mode 100644 index 000000000000..53bd8602f9ef --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h @@ -0,0 +1,14 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_TOP_SOC15_H__ +#define __DML2_TOP_SOC15_H__ +#include "dml2_internal_shared_types.h" +bool dml2_top_soc15_initialize_instance(struct dml2_initialize_instance_in_out *in_out); + +bool dml2_top_mcache_calc_mcache_count_and_offsets(struct top_mcache_calc_mcache_count_and_offsets_in_out *params); +void dml2_top_mcache_assign_global_mcache_ids(struct top_mcache_assign_global_mcache_ids_in_out *params); +bool dml2_top_mcache_validate_admissability(struct top_mcache_validate_admissability_in_out *params); +bool dml2_top_soc15_build_mcache_programming(struct dml2_build_mcache_programming_in_out *params); +#endif /* __DML2_TOP_SOC15_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h new file mode 100644 index 000000000000..611c80f4f1bf --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h @@ -0,0 +1,189 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_DEBUG_H__ +#define __DML2_DEBUG_H__ + +#include "os_types.h" +#define DML_ASSERT(condition) ASSERT(condition) +#define DML_LOG_LEVEL_DEFAULT DML_LOG_LEVEL_WARN +#define DML_LOG_INTERNAL(fmt, ...) dm_output_to_console(fmt, ## __VA_ARGS__) + +/* private helper macros */ +#define _BOOL_FORMAT(field) "%s", field ? "true" : "false" +#define _UINT_FORMAT(field) "%u", field +#define _INT_FORMAT(field) "%d", field +#define _DOUBLE_FORMAT(field) "%lf", field +#define _ELEMENT_FUNC "function" +#define _ELEMENT_COMP_IF "component_interface" +#define _ELEMENT_TOP_IF "top_interface" +#define _LOG_ENTRY(element) do { \ + DML_LOG_INTERNAL("<"element" name=\""); \ + DML_LOG_INTERNAL(__func__); \ + DML_LOG_INTERNAL("\">\n"); \ +} while (0) +#define _LOG_EXIT(element) DML_LOG_INTERNAL("\n") +#define _LOG_SCALAR(field, format) do { \ + DML_LOG_INTERNAL(#field" = "format(field)); \ + DML_LOG_INTERNAL("\n"); \ +} while (0) +#define _LOG_ARRAY(field, size, format) do { \ + DML_LOG_INTERNAL(#field " = ["); \ + for (int _i = 0; _i < (int) size; _i++) { \ + DML_LOG_INTERNAL(format(field[_i])); \ + if (_i + 1 == (int) size) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ +}} while (0) +#define _LOG_2D_ARRAY(field, size0, size1, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL(format(field[_i][_j])); \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) +#define _LOG_3D_ARRAY(field, size0, size1, size2, format) do { \ + DML_LOG_INTERNAL(#field" = ["); \ + for (int _i = 0; _i < (int) size0; _i++) { \ + DML_LOG_INTERNAL("\n\t["); \ + for (int _j = 0; _j < (int) size1; _j++) { \ + DML_LOG_INTERNAL("["); \ + for (int _k = 0; _k < (int) size2; _k++) { \ + DML_LOG_INTERNAL(format(field[_i][_j][_k])); \ + if (_k + 1 == (int) size2) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_j + 1 == (int) size1) \ + DML_LOG_INTERNAL("]"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ + if (_i + 1 == (int) size0) \ + DML_LOG_INTERNAL("]\n"); \ + else \ + DML_LOG_INTERNAL(", "); \ + } \ +} while (0) + +/* fatal errors for unrecoverable DML states until a full reset */ +#define DML_LOG_LEVEL_FATAL 0 +/* unexpected but recoverable failures inside DML */ +#define DML_LOG_LEVEL_ERROR 1 +/* unexpected inputs or events to DML */ +#define DML_LOG_LEVEL_WARN 2 +/* high level tracing of DML interfaces */ +#define DML_LOG_LEVEL_INFO 3 +/* tracing of DML internal executions */ +#define DML_LOG_LEVEL_DEBUG 4 +/* detailed tracing of DML calculation procedure */ +#define DML_LOG_LEVEL_VERBOSE 5 + +#ifndef DML_LOG_LEVEL +#define DML_LOG_LEVEL DML_LOG_LEVEL_DEFAULT +#endif /* #ifndef DML_LOG_LEVEL */ + +/* public macros for DML_LOG_LEVEL_FATAL and up */ +#define DML_LOG_FATAL(fmt, ...) DML_LOG_INTERNAL("[DML FATAL] " fmt, ## __VA_ARGS__) + +/* public macros for DML_LOG_LEVEL_ERROR and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_ERROR +#define DML_LOG_ERROR(fmt, ...) DML_LOG_INTERNAL("[DML ERROR] "fmt, ## __VA_ARGS__) +#define DML_ASSERT_MSG(condition, fmt, ...) \ + do { \ + if (!(condition)) { \ + DML_LOG_ERROR("ASSERT hit in %s line %d\n", __func__, __LINE__); \ + DML_LOG_ERROR(fmt, ## __VA_ARGS__); \ + DML_ASSERT(condition); \ + } \ + } while (0) +#else +#define DML_LOG_ERROR(fmt, ...) ((void)0) +#define DML_ASSERT_MSG(condition, fmt, ...) ((void)0) +#endif + +/* public macros for DML_LOG_LEVEL_WARN and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_WARN +#define DML_LOG_WARN(fmt, ...) DML_LOG_INTERNAL("[DML WARN] "fmt, ## __VA_ARGS__) +#else +#define DML_LOG_WARN(fmt, ...) ((void)0) +#endif + +/* public macros for DML_LOG_LEVEL_INFO and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_INFO +#define DML_LOG_INFO(fmt, ...) DML_LOG_INTERNAL("[DML INFO] "fmt, ## __VA_ARGS__) +#define DML_LOG_TOP_IF_ENTER() _LOG_ENTRY(_ELEMENT_TOP_IF) +#define DML_LOG_TOP_IF_EXIT() _LOG_EXIT(_ELEMENT_TOP_IF) +#else +#define DML_LOG_INFO(fmt, ...) ((void)0) +#define DML_LOG_TOP_IF_ENTER() ((void)0) +#define DML_LOG_TOP_IF_EXIT() ((void)0) +#endif + +/* public macros for DML_LOG_LEVEL_DEBUG and up */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_DEBUG +#define DML_LOG_DEBUG(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) +#define DML_LOG_COMP_IF_ENTER() _LOG_ENTRY(_ELEMENT_COMP_IF) +#define DML_LOG_COMP_IF_EXIT() _LOG_EXIT(_ELEMENT_COMP_IF) +#define DML_LOG_FUNC_ENTER() _LOG_ENTRY(_ELEMENT_FUNC) +#define DML_LOG_FUNC_EXIT() _LOG_EXIT(_ELEMENT_FUNC) +#define DML_LOG_DEBUG_BOOL(field) _LOG_SCALAR(field, _BOOL_FORMAT) +#define DML_LOG_DEBUG_UINT(field) _LOG_SCALAR(field, _UINT_FORMAT) +#define DML_LOG_DEBUG_INT(field) _LOG_SCALAR(field, _INT_FORMAT) +#define DML_LOG_DEBUG_DOUBLE(field) _LOG_SCALAR(field, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) _LOG_ARRAY(field, size, _BOOL_FORMAT) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) _LOG_ARRAY(field, size, _UINT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) _LOG_ARRAY(field, size, _INT_FORMAT) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) _LOG_ARRAY(field, size, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _BOOL_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _UINT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _INT_FORMAT) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) _LOG_2D_ARRAY(field, size0, size1, _DOUBLE_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _BOOL_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _UINT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _INT_FORMAT) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) _LOG_3D_ARRAY(field, size0, size1, size2, _DOUBLE_FORMAT) +#else +#define DML_LOG_DEBUG(fmt, ...) ((void)0) +#define DML_LOG_COMP_IF_ENTER() ((void)0) +#define DML_LOG_COMP_IF_EXIT() ((void)0) +#define DML_LOG_FUNC_ENTER() ((void)0) +#define DML_LOG_FUNC_EXIT() ((void)0) +#define DML_LOG_DEBUG_BOOL(field) ((void)0) +#define DML_LOG_DEBUG_UINT(field) ((void)0) +#define DML_LOG_DEBUG_INT(field) ((void)0) +#define DML_LOG_DEBUG_DOUBLE(field) ((void)0) +#define DML_LOG_DEBUG_ARRAY_BOOL(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_UINT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_INT(field, size) ((void)0) +#define DML_LOG_DEBUG_ARRAY_DOUBLE(field, size) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_BOOL(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_UINT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_INT(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_2D_ARRAY_DOUBLE(field, size0, size1) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_BOOL(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_UINT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_INT(field, size0, size1, size2) ((void)0) +#define DML_LOG_DEBUG_3D_ARRAY_DOUBLE(field, size0, size1, size2) ((void)0) +#endif + +/* public macros for DML_LOG_LEVEL_VERBOSE */ +#if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE +#define DML_LOG_VERBOSE(fmt, ...) DML_LOG_INTERNAL(fmt, ## __VA_ARGS__) +#else +#define DML_LOG_VERBOSE(fmt, ...) ((void)0) +#endif /* #if DML_LOG_LEVEL >= DML_LOG_LEVEL_VERBOSE */ +#endif /* __DML2_DEBUG_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h new file mode 100644 index 000000000000..9f562f0c4797 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h @@ -0,0 +1,1004 @@ +// SPDX-License-Identifier: MIT +// +// Copyright 2024 Advanced Micro Devices, Inc. + +#ifndef __DML2_INTERNAL_SHARED_TYPES_H__ +#define __DML2_INTERNAL_SHARED_TYPES_H__ + +#include "dml2_external_lib_deps.h" +#include "dml_top_types.h" +#include "dml2_core_shared_types.h" +/* +* DML2 MCG Types and Interfaces +*/ + +#define DML_MCG_MAX_CLK_TABLE_SIZE 20 + +struct dram_bw_to_min_clk_table_entry { + unsigned long long pre_derate_dram_bw_kbps; + unsigned long min_fclk_khz; + unsigned long min_dcfclk_khz; +}; + +struct dml2_mcg_dram_bw_to_min_clk_table { + struct dram_bw_to_min_clk_table_entry entries[DML_MCG_MAX_CLK_TABLE_SIZE]; + + unsigned int num_entries; +}; + +struct dml2_mcg_min_clock_table { + struct { + unsigned int dispclk; + unsigned int dppclk; + unsigned int dscclk; + unsigned int dtbclk; + unsigned int phyclk; + unsigned int fclk; + unsigned int dcfclk; + } max_clocks_khz; + + struct { + unsigned int dispclk; + unsigned int dppclk; + unsigned int dtbclk; + } max_ss_clocks_khz; + + struct { + unsigned int dprefclk; + unsigned int xtalclk; + unsigned int pcierefclk; + unsigned int dchubrefclk; + unsigned int amclk; + } fixed_clocks_khz; + + struct dml2_mcg_dram_bw_to_min_clk_table dram_bw_table; +}; + +struct dml2_mcg_build_min_clock_table_params_in_out { + /* + * Input + */ + struct dml2_soc_bb *soc_bb; + struct { + bool perform_pseudo_build; + } clean_me_up; + + /* + * Output + */ + struct dml2_mcg_min_clock_table *min_clk_table; +}; +struct dml2_mcg_instance { + bool (*build_min_clock_table)(struct dml2_mcg_build_min_clock_table_params_in_out *in_out); +}; + +/* +* DML2 DPMM Types and Interfaces +*/ + +struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out { + /* + * Input + */ + struct dml2_core_ip_params *ip; + struct dml2_soc_bb *soc_bb; + struct dml2_mcg_min_clock_table *min_clk_table; + const struct display_configuation_with_meta *display_cfg; + struct { + bool perform_pseudo_map; + struct dml2_core_internal_soc_bb *soc_bb; + } clean_me_up; + + /* + * Output + */ + struct dml2_display_cfg_programming *programming; +}; + +struct dml2_dpmm_map_watermarks_params_in_out { + /* + * Input + */ + const struct display_configuation_with_meta *display_cfg; + const struct dml2_core_instance *core; + + /* + * Output + */ + struct dml2_display_cfg_programming *programming; +}; + +struct dml2_dpmm_scratch { + struct dml2_display_cfg_programming programming; +}; + +struct dml2_dpmm_instance { + bool (*map_mode_to_soc_dpm)(struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out *in_out); + bool (*map_watermarks)(struct dml2_dpmm_map_watermarks_params_in_out *in_out); + + struct dml2_dpmm_scratch dpmm_scratch; +}; + +/* +* DML2 Core Types and Interfaces +*/ + +struct dml2_core_initialize_in_out { + enum dml2_project_id project_id; + struct dml2_core_instance *instance; + struct dml2_soc_bb *soc_bb; + struct dml2_ip_capabilities *ip_caps; + + struct dml2_mcg_min_clock_table *minimum_clock_table; + + void *explicit_ip_bb; + unsigned int explicit_ip_bb_size; + + // FIXME_STAGE2 can remove but dcn3 version still need this + struct { + struct soc_bounding_box_st *soc_bb; + struct soc_states_st *soc_states; + } legacy; +}; + +struct core_bandwidth_requirements { + int urgent_bandwidth_kbytes_per_sec; + int average_bandwidth_kbytes_per_sec; +}; + +struct core_plane_support_info { + int dpps_used; + int dram_change_latency_hiding_margin_in_active; + int active_latency_hiding_us; + int mall_svp_size_requirement_ways; + int nominal_vblank_pstate_latency_hiding_us; + unsigned int dram_change_vactive_det_fill_delay_us; +}; + +struct core_stream_support_info { + unsigned int odms_used; + unsigned int num_odm_output_segments; // for odm split mode (e.g. a value of 2 for odm_mode_mso_1to2) + + /* FAMS2 SubVP support info */ + unsigned int phantom_min_v_active; + unsigned int phantom_v_startup; + + unsigned int phantom_v_active; + unsigned int phantom_v_total; + int vblank_reserved_time_us; + int num_dsc_slices; + bool dsc_enable; +}; + +struct core_display_cfg_support_info { + bool is_supported; + + struct core_stream_support_info stream_support_info[DML2_MAX_PLANES]; + struct core_plane_support_info plane_support_info[DML2_MAX_PLANES]; + + struct { + struct dml2_core_internal_mode_support_info support_info; + } clean_me_up; +}; + +struct dml2_core_mode_support_result { + struct { + struct { + unsigned long urgent_bw_sdp_kbps; + unsigned long average_bw_sdp_kbps; + unsigned long urgent_bw_dram_kbps; + unsigned long average_bw_dram_kbps; + unsigned long dcfclk_khz; + unsigned long fclk_khz; + } svp_prefetch; + + struct { + unsigned long urgent_bw_sdp_kbps; + unsigned long average_bw_sdp_kbps; + unsigned long urgent_bw_dram_kbps; + unsigned long average_bw_dram_kbps; + unsigned long dcfclk_khz; + unsigned long fclk_khz; + } active; + + unsigned int dispclk_khz; + unsigned int dpprefclk_khz; + unsigned int dtbrefclk_khz; + unsigned int dcfclk_deepsleep_khz; + unsigned int socclk_khz; + + unsigned int uclk_pstate_supported; + unsigned int fclk_pstate_supported; + } global; + + struct { + unsigned int dscclk_khz; + unsigned int dtbclk_khz; + unsigned int phyclk_khz; + } per_stream[DML2_MAX_PLANES]; + + struct { + unsigned int dppclk_khz; + unsigned int mall_svp_allocation_mblks; + unsigned int mall_full_frame_allocation_mblks; + } per_plane[DML2_MAX_PLANES]; + + struct core_display_cfg_support_info cfg_support_info; +}; + +struct dml2_optimization_stage1_state { + bool performed; + bool success; + + int min_clk_index_for_latency; +}; + +struct dml2_optimization_stage2_state { + bool performed; + bool success; + + // Whether or not each plane supports mcache + // The number of valid elements == display_cfg.num_planes + // The indexing of pstate_switch_modes matches plane_descriptors[] + bool per_plane_mcache_support[DML2_MAX_PLANES]; + struct dml2_mcache_surface_allocation mcache_allocations[DML2_MAX_PLANES]; +}; + +#define DML2_PMO_LEGACY_PREFETCH_MAX_TWAIT_OPTIONS 8 +#define DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE 10 +#define DML2_PMO_STUTTER_CANDIDATE_LIST_SIZE 3 + +struct dml2_implicit_svp_meta { + bool valid; + unsigned long v_active; + unsigned long v_total; + unsigned long v_front_porch; +}; + +struct dml2_pstate_per_method_common_meta { + /* generic params */ + unsigned int allow_start_otg_vline; + unsigned int allow_end_otg_vline; + /* scheduling params */ + double allow_time_us; + double disallow_time_us; + double period_us; +}; + +struct dml2_pstate_meta { + bool valid; + double otg_vline_time_us; + unsigned int scheduling_delay_otg_vlines; + unsigned int vertical_interrupt_ack_delay_otg_vlines; + unsigned int allow_to_target_delay_otg_vlines; + unsigned int contention_delay_otg_vlines; + unsigned int min_allow_width_otg_vlines; + unsigned int nom_vtotal; + unsigned int vblank_start; + double nom_refresh_rate_hz; + double nom_frame_time_us; + unsigned int max_vtotal; + double min_refresh_rate_hz; + double max_frame_time_us; + unsigned int blackout_otg_vlines; + struct { + double max_vactive_det_fill_delay_us; + unsigned int max_vactive_det_fill_delay_otg_vlines; + struct dml2_pstate_per_method_common_meta common; + } method_vactive; + struct { + struct dml2_pstate_per_method_common_meta common; + } method_vblank; + struct { + unsigned int programming_delay_otg_vlines; + unsigned int df_throttle_delay_otg_vlines; + unsigned int prefetch_to_mall_delay_otg_vlines; + unsigned long phantom_vactive; + unsigned long phantom_vfp; + unsigned long phantom_vtotal; + struct dml2_pstate_per_method_common_meta common; + } method_subvp; + struct { + unsigned int programming_delay_otg_vlines; + unsigned int stretched_vtotal; + struct dml2_pstate_per_method_common_meta common; + } method_drr; +}; + +/* mask of synchronized timings by stream index */ +struct dml2_pmo_synchronized_timing_groups { + unsigned int num_timing_groups; + unsigned int synchronized_timing_group_masks[DML2_MAX_PLANES]; + bool group_is_drr_enabled[DML2_MAX_PLANES]; + bool group_is_drr_active[DML2_MAX_PLANES]; + double group_line_time_us[DML2_MAX_PLANES]; +}; + +struct dml2_optimization_stage3_state { + bool performed; + bool success; + + // The pstate support mode for each plane + // The number of valid elements == display_cfg.num_planes + // The indexing of pstate_switch_modes matches plane_descriptors[] + enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; + + // Meta-data for implicit SVP generation, indexed by stream index + struct dml2_implicit_svp_meta stream_svp_meta[DML2_MAX_PLANES]; + + // Meta-data for FAMS2 + bool fams2_required; + struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; + + int min_clk_index_for_latency; +}; + +struct dml2_optimization_stage4_state { + bool performed; + bool success; + bool unoptimizable_streams[DML2_MAX_DCN_PIPES]; +}; + +struct dml2_optimization_stage5_state { + bool performed; + bool success; + + bool optimal_reserved_time_in_vblank_us; + bool vblank_includes_z8_optimization; +}; + +struct display_configuation_with_meta { + struct dml2_display_cfg display_config; + + struct dml2_core_mode_support_result mode_support_result; + + // Stage 1 = Min Clocks for Latency + struct dml2_optimization_stage1_state stage1; + + // Stage 2 = MCache + struct dml2_optimization_stage2_state stage2; + + // Stage 3 = UCLK PState + struct dml2_optimization_stage3_state stage3; + + // Stage 4 = Vmin + struct dml2_optimization_stage4_state stage4; + + // Stage 5 = Stutter + struct dml2_optimization_stage5_state stage5; +}; + +struct dml2_pmo_pstate_strategy { + enum dml2_pstate_method per_stream_pstate_method[DML2_MAX_PLANES]; + bool allow_state_increase; +}; +struct dml2_core_mode_support_in_out { + /* + * Inputs + */ + struct dml2_core_instance *instance; + const struct display_configuation_with_meta *display_cfg; + + struct dml2_mcg_min_clock_table *min_clk_table; + int min_clk_index; + /* + * Outputs + */ + struct dml2_core_mode_support_result mode_support_result; + + struct { + // Inputs + struct dml_display_cfg_st *display_cfg; + + // Outputs + struct dml_mode_support_info_st *support_info; + unsigned int out_lowest_state_idx; + unsigned int min_fclk_khz; + unsigned int min_dcfclk_khz; + unsigned int min_dram_speed_mts; + unsigned int min_socclk_khz; + unsigned int min_dscclk_khz; + unsigned int min_dtbclk_khz; + unsigned int min_phyclk_khz; + } legacy; +}; + +struct dml2_core_mode_programming_in_out { + /* + * Inputs + */ + struct dml2_core_instance *instance; + const struct display_configuation_with_meta *display_cfg; + const struct core_display_cfg_support_info *cfg_support_info; + /* + * Outputs (also Input the clk freq are also from programming struct) + */ + struct dml2_display_cfg_programming *programming; + +}; + +struct dml2_core_populate_informative_in_out { + /* + * Inputs + */ + struct dml2_core_instance *instance; + + // If this is set, then the mode was supported, and mode programming + // was successfully run. + // Otherwise, mode programming was not run, because mode support failed. + bool mode_is_supported; + + /* + * Outputs + */ + struct dml2_display_cfg_programming *programming; +}; + +struct dml2_calculate_mcache_allocation_in_out { + /* + * Inputs + */ + struct dml2_core_instance *instance; + const struct dml2_plane_parameters *plane_descriptor; + unsigned int plane_index; + + /* + * Outputs + */ + struct dml2_mcache_surface_allocation *mcache_allocation; +}; + +struct dml2_core_internal_state_inputs { + unsigned int dummy; +}; + +struct dml2_core_internal_state_intermediates { + unsigned int dummy; +}; + +struct dml2_core_mode_support_locals { + union { + struct dml2_core_calcs_mode_support_ex mode_support_ex_params; + }; + struct dml2_display_cfg svp_expanded_display_cfg; + struct dml2_calculate_mcache_allocation_in_out calc_mcache_allocation_params; +}; + +struct dml2_core_mode_programming_locals { + union { + struct dml2_core_calcs_mode_programming_ex mode_programming_ex_params; + }; + struct dml2_display_cfg svp_expanded_display_cfg; +}; + +struct dml2_core_scratch { + struct dml2_core_mode_support_locals mode_support_locals; + struct dml2_core_mode_programming_locals mode_programming_locals; + int main_stream_index_from_svp_stream_index[DML2_MAX_PLANES]; + int svp_stream_index_from_main_stream_index[DML2_MAX_PLANES]; + int main_plane_index_to_phantom_plane_index[DML2_MAX_PLANES]; + int phantom_plane_index_to_main_plane_index[DML2_MAX_PLANES]; +}; + +struct dml2_core_instance { + enum dml2_project_id project_id; + struct dml2_mcg_min_clock_table *minimum_clock_table; + struct dml2_core_internal_state_inputs inputs; + struct dml2_core_internal_state_intermediates intermediates; + + struct dml2_core_scratch scratch; + + bool (*initialize)(struct dml2_core_initialize_in_out *in_out); + bool (*mode_support)(struct dml2_core_mode_support_in_out *in_out); + bool (*mode_programming)(struct dml2_core_mode_programming_in_out *in_out); + bool (*populate_informative)(struct dml2_core_populate_informative_in_out *in_out); + bool (*calculate_mcache_allocation)(struct dml2_calculate_mcache_allocation_in_out *in_out); + + struct { + struct dml2_core_internal_display_mode_lib mode_lib; + } clean_me_up; +}; + +/* +* DML2 PMO Types and Interfaces +*/ + +struct dml2_pmo_initialize_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct dml2_soc_bb *soc_bb; + struct dml2_ip_capabilities *ip_caps; + struct dml2_pmo_options *options; + int mcg_clock_table_size; +}; + +struct dml2_pmo_optimize_dcc_mcache_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + const struct dml2_display_cfg *display_config; + bool *dcc_mcache_supported; + struct core_display_cfg_support_info *cfg_support_info; + + /* + * Output + */ + struct dml2_display_cfg *optimized_display_cfg; +}; + +struct dml2_pmo_init_for_vmin_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; +}; + +struct dml2_pmo_test_for_vmin_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + const struct display_configuation_with_meta *display_config; + const struct dml2_soc_vmin_clock_limits *vmin_limits; +}; + +struct dml2_pmo_optimize_for_vmin_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; + + /* + * Output + */ + struct display_configuation_with_meta *optimized_display_config; +}; + +struct dml2_pmo_init_for_pstate_support_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; +}; + +struct dml2_pmo_test_for_pstate_support_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; +}; + +struct dml2_pmo_optimize_for_pstate_support_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; + bool last_candidate_failed; + + /* + * Output + */ + struct display_configuation_with_meta *optimized_display_config; +}; + +struct dml2_pmo_init_for_stutter_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; +}; + +struct dml2_pmo_test_for_stutter_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; +}; + +struct dml2_pmo_optimize_for_stutter_in_out { + /* + * Input + */ + struct dml2_pmo_instance *instance; + struct display_configuation_with_meta *base_display_config; + bool last_candidate_failed; + + /* + * Output + */ + struct display_configuation_with_meta *optimized_display_config; +}; + +#define PMO_NO_DRR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw - dml2_pstate_method_na + 1)) - 1) << dml2_pstate_method_na) +#define PMO_DRR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_vactive_drr + 1)) - 1) << dml2_pstate_method_fw_vactive_drr) +#define PMO_DRR_CLAMPED_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_clamped - dml2_pstate_method_fw_vactive_drr + 1)) - 1) << dml2_pstate_method_fw_vactive_drr) +#define PMO_DRR_VAR_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_drr + 1)) - 1) << dml2_pstate_method_fw_drr) +#define PMO_FW_STRATEGY_MASK (((1 << (dml2_pstate_method_reserved_fw_drr_var - dml2_pstate_method_fw_svp + 1)) - 1) << dml2_pstate_method_fw_svp) + +#define PMO_DCN4_MAX_DISPLAYS 4 +#define PMO_DCN4_MAX_NUM_VARIANTS 2 +#define PMO_DCN4_MAX_BASE_STRATEGIES 10 + +struct dml2_scheduling_check_locals { + struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; +}; + +struct dml2_pmo_scratch { + union { + struct { + double reserved_time_candidates[DML2_MAX_PLANES][DML2_PMO_LEGACY_PREFETCH_MAX_TWAIT_OPTIONS]; + int reserved_time_candidates_count[DML2_MAX_PLANES]; + int current_candidate[DML2_MAX_PLANES]; + int min_latency_index; + int max_latency_index; + int cur_latency_index; + int stream_mask; + } pmo_dcn3; + struct { + struct dml2_pmo_pstate_strategy expanded_override_strategy_list[2 * 2 * 2 * 2]; + unsigned int num_expanded_override_strategies; + struct dml2_pmo_pstate_strategy pstate_strategy_candidates[DML2_PMO_PSTATE_CANDIDATE_LIST_SIZE]; + int num_pstate_candidates; + int cur_pstate_candidate; + + unsigned int stream_plane_mask[DML2_MAX_PLANES]; + + unsigned int stream_vactive_capability_mask; + + int min_latency_index; + int max_latency_index; + int cur_latency_index; + + // Stores all the implicit SVP meta information indexed by stream index of the display + // configuration under inspection, built at optimization stage init + struct dml2_implicit_svp_meta stream_svp_meta[DML2_MAX_PLANES]; + struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; + + unsigned int optimal_vblank_reserved_time_for_stutter_us[DML2_PMO_STUTTER_CANDIDATE_LIST_SIZE]; + unsigned int num_stutter_candidates; + unsigned int cur_stutter_candidate; + bool z8_vblank_optimizable; + + /* mask of synchronized timings by stream index */ + unsigned int num_timing_groups; + unsigned int synchronized_timing_group_masks[DML2_MAX_PLANES]; + bool group_is_drr_enabled[DML2_MAX_PLANES]; + bool group_is_drr_active[DML2_MAX_PLANES]; + double group_line_time_us[DML2_MAX_PLANES]; + + /* scheduling check locals */ + struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; + double group_phase_offset[DML2_MAX_PLANES]; + } pmo_dcn4; + }; +}; + +struct dml2_pmo_init_data { + union { + struct { + /* populated once during initialization */ + struct dml2_pmo_pstate_strategy expanded_strategy_list_1_display[PMO_DCN4_MAX_BASE_STRATEGIES * 2]; + struct dml2_pmo_pstate_strategy expanded_strategy_list_2_display[PMO_DCN4_MAX_BASE_STRATEGIES * 4 * 4]; + struct dml2_pmo_pstate_strategy expanded_strategy_list_3_display[PMO_DCN4_MAX_BASE_STRATEGIES * 6 * 6 * 6]; + struct dml2_pmo_pstate_strategy expanded_strategy_list_4_display[PMO_DCN4_MAX_BASE_STRATEGIES * 8 * 8 * 8 * 8]; + unsigned int num_expanded_strategies_per_list[PMO_DCN4_MAX_DISPLAYS]; + } pmo_dcn4; + }; +}; + +struct dml2_pmo_instance { + struct dml2_soc_bb *soc_bb; + struct dml2_ip_capabilities *ip_caps; + + struct dml2_pmo_options *options; + + int disp_clk_vmin_threshold; + int mpc_combine_limit; + int odm_combine_limit; + int mcg_clock_table_size; + union { + struct { + struct { + int prefetch_end_to_mall_start_us; + int fw_processing_delay_us; + int refresh_rate_limit_min; + int refresh_rate_limit_max; + } subvp; + } v1; + struct { + struct { + int refresh_rate_limit_min; + int refresh_rate_limit_max; + } subvp; + struct { + int refresh_rate_limit_min; + int refresh_rate_limit_max; + } drr; + } v2; + } fams_params; + + bool (*initialize)(struct dml2_pmo_initialize_in_out *in_out); + bool (*optimize_dcc_mcache)(struct dml2_pmo_optimize_dcc_mcache_in_out *in_out); + + bool (*init_for_vmin)(struct dml2_pmo_init_for_vmin_in_out *in_out); + bool (*test_for_vmin)(struct dml2_pmo_test_for_vmin_in_out *in_out); + bool (*optimize_for_vmin)(struct dml2_pmo_optimize_for_vmin_in_out *in_out); + + bool (*init_for_uclk_pstate)(struct dml2_pmo_init_for_pstate_support_in_out *in_out); + bool (*test_for_uclk_pstate)(struct dml2_pmo_test_for_pstate_support_in_out *in_out); + bool (*optimize_for_uclk_pstate)(struct dml2_pmo_optimize_for_pstate_support_in_out *in_out); + + bool (*init_for_stutter)(struct dml2_pmo_init_for_stutter_in_out *in_out); + bool (*test_for_stutter)(struct dml2_pmo_test_for_stutter_in_out *in_out); + bool (*optimize_for_stutter)(struct dml2_pmo_optimize_for_stutter_in_out *in_out); + + struct dml2_pmo_init_data init_data; + struct dml2_pmo_scratch scratch; +}; + +/* +* DML2 MCache Types +*/ + +struct top_mcache_validate_admissability_in_out { + struct dml2_instance *dml2_instance; + + const struct dml2_display_cfg *display_cfg; + const struct core_display_cfg_support_info *cfg_support_info; + struct dml2_mcache_surface_allocation *mcache_allocations; + + bool per_plane_status[DML2_MAX_PLANES]; + + struct { + const struct dml_mode_support_info_st *mode_support_info; + } legacy; +}; + +struct top_mcache_assign_ids_in_out { + /* + * Input + */ + const struct dml2_mcache_surface_allocation *mcache_allocations; + int plane_count; + + int per_pipe_viewport_x_start[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; + int per_pipe_viewport_x_end[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; + int pipe_count_per_plane[DML2_MAX_PLANES]; + + struct dml2_display_mcache_regs *current_mcache_regs[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; //One set per pipe/hubp + + /* + * Output + */ + struct dml2_display_mcache_regs mcache_regs[DML2_MAX_PLANES][DML2_MAX_DCN_PIPES]; //One set per pipe/hubp + struct dml2_build_mcache_programming_in_out *mcache_programming; +}; + +struct top_mcache_calc_mcache_count_and_offsets_in_out { + /* + * Inputs + */ + struct dml2_instance *dml2_instance; + const struct dml2_display_cfg *display_config; + + /* + * Outputs + */ + struct dml2_mcache_surface_allocation *mcache_allocations; +}; + +struct top_mcache_assign_global_mcache_ids_in_out { + /* + * Inputs/Outputs + */ + struct dml2_mcache_surface_allocation *allocations; + int num_allocations; +}; + +/* +* DML2 Top Types +*/ + +struct dml2_initialize_instance_locals { + int dummy; +}; + +struct dml2_optimization_init_function_locals { + union { + struct { + struct dml2_pmo_init_for_pstate_support_in_out init_params; + } uclk_pstate; + struct { + struct dml2_pmo_init_for_stutter_in_out stutter_params; + } stutter; + struct { + struct dml2_pmo_init_for_vmin_in_out init_params; + } vmin; + }; +}; + +struct dml2_optimization_test_function_locals { + union { + struct { + struct top_mcache_calc_mcache_count_and_offsets_in_out calc_mcache_count_params; + struct top_mcache_assign_global_mcache_ids_in_out assign_global_mcache_ids_params; + struct top_mcache_validate_admissability_in_out validate_admissibility_params; + } test_mcache; + struct { + struct dml2_pmo_test_for_vmin_in_out pmo_test_vmin_params; + } test_vmin; + struct { + struct dml2_pmo_test_for_pstate_support_in_out test_params; + } uclk_pstate; + struct { + struct dml2_pmo_test_for_stutter_in_out stutter_params; + } stutter; + }; +}; + +struct dml2_optimization_optimize_function_locals { + union { + struct { + struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; + } optimize_mcache; + struct { + struct dml2_pmo_optimize_for_vmin_in_out pmo_optimize_vmin_params; + } optimize_vmin; + struct { + struct dml2_pmo_optimize_for_pstate_support_in_out optimize_params; + } uclk_pstate; + struct { + struct dml2_pmo_optimize_for_stutter_in_out stutter_params; + } stutter; + }; +}; + +struct dml2_optimization_phase_locals { + struct display_configuation_with_meta cur_candidate_display_cfg; + struct display_configuation_with_meta next_candidate_display_cfg; + struct dml2_core_mode_support_in_out mode_support_params; + struct dml2_optimization_init_function_locals init_function_locals; + struct dml2_optimization_test_function_locals test_function_locals; + struct dml2_optimization_optimize_function_locals optimize_function_locals; +}; + +struct dml2_check_mode_supported_locals { + struct dml2_display_cfg display_cfg_working_copy; + struct dml2_core_mode_support_in_out mode_support_params; + struct dml2_optimization_phase_locals optimization_phase_locals; + struct display_configuation_with_meta base_display_config_with_meta; + struct display_configuation_with_meta optimized_display_config_with_meta; + struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out dppm_map_mode_params; +}; + +struct optimization_init_function_params { + struct dml2_optimization_init_function_locals *locals; + struct dml2_instance *dml; + struct display_configuation_with_meta *display_config; +}; + +struct optimization_test_function_params { + struct dml2_optimization_test_function_locals *locals; + struct dml2_instance *dml; + struct display_configuation_with_meta *display_config; +}; + +struct optimization_optimize_function_params { + bool last_candidate_supported; + struct dml2_optimization_optimize_function_locals *locals; + struct dml2_instance *dml; + struct display_configuation_with_meta *display_config; + struct display_configuation_with_meta *optimized_display_config; +}; + +struct optimization_phase_params { + struct dml2_instance *dml; + const struct display_configuation_with_meta *display_config; // Initial Display Configuration + bool (*init_function)(const struct optimization_init_function_params *params); // Test function to determine optimization is complete + bool (*test_function)(const struct optimization_test_function_params *params); // Test function to determine optimization is complete + bool (*optimize_function)(const struct optimization_optimize_function_params *params); // Function which produces a more optimized display configuration + struct display_configuation_with_meta *optimized_display_config; // The optimized display configuration + + bool all_or_nothing; +}; + +struct dml2_build_mode_programming_locals { + struct dml2_core_mode_support_in_out mode_support_params; + struct dml2_core_mode_programming_in_out mode_programming_params; + struct dml2_core_populate_informative_in_out informative_params; + struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; + struct display_configuation_with_meta base_display_config_with_meta; + struct display_configuation_with_meta optimized_display_config_with_meta; + struct dml2_dpmm_map_mode_to_soc_dpm_params_in_out dppm_map_mode_params; + struct dml2_dpmm_map_watermarks_params_in_out dppm_map_watermarks_params; + struct dml2_optimization_phase_locals optimization_phase_locals; + struct optimization_phase_params min_clock_for_latency_phase; + struct optimization_phase_params mcache_phase; + struct optimization_phase_params uclk_pstate_phase; + struct optimization_phase_params vmin_phase; + struct optimization_phase_params stutter_phase; +}; + +struct dml2_legacy_core_build_mode_programming_wrapper_locals { + struct dml2_core_mode_support_in_out mode_support_params; + struct dml2_core_mode_programming_in_out mode_programming_params; + struct dml2_core_populate_informative_in_out informative_params; + struct top_mcache_calc_mcache_count_and_offsets_in_out calc_mcache_count_params; + struct top_mcache_validate_admissability_in_out validate_admissibility_params; + struct dml2_mcache_surface_allocation mcache_allocations[DML2_MAX_PLANES]; + struct top_mcache_assign_global_mcache_ids_in_out assign_global_mcache_ids_params; + struct dml2_pmo_optimize_dcc_mcache_in_out optimize_mcache_params; + struct dml2_display_cfg optimized_display_cfg; + struct core_display_cfg_support_info core_support_info; +}; + +struct dml2_top_mcache_verify_mcache_size_locals { + struct dml2_calculate_mcache_allocation_in_out calc_mcache_params; +}; + +struct dml2_top_mcache_validate_admissability_locals { + struct { + int pipe_vp_startx[DML2_MAX_DCN_PIPES]; + int pipe_vp_endx[DML2_MAX_DCN_PIPES]; + } plane0; + struct { + int pipe_vp_startx[DML2_MAX_DCN_PIPES]; + int pipe_vp_endx[DML2_MAX_DCN_PIPES]; + } plane1; +}; + +struct dml2_top_display_cfg_support_info { + const struct dml2_display_cfg *display_config; + struct core_display_cfg_support_info core_info; +}; + +struct dml2_top_funcs { + bool (*check_mode_supported)(struct dml2_check_mode_supported_in_out *in_out); + bool (*build_mode_programming)(struct dml2_build_mode_programming_in_out *in_out); + bool (*build_mcache_programming)(struct dml2_build_mcache_programming_in_out *in_out); +}; + +struct dml2_instance { + enum dml2_project_id project_id; + + struct dml2_core_instance core_instance; + struct dml2_mcg_instance mcg_instance; + struct dml2_dpmm_instance dpmm_instance; + struct dml2_pmo_instance pmo_instance; + + struct dml2_soc_bb soc_bbox; + struct dml2_ip_capabilities ip_caps; + + struct dml2_mcg_min_clock_table min_clk_table; + struct dml2_pmo_options pmo_options; + struct dml2_top_funcs funcs; + + struct { + struct dml2_initialize_instance_locals initialize_instance_locals; + struct dml2_top_mcache_verify_mcache_size_locals mcache_verify_mcache_size_locals; + struct dml2_top_mcache_validate_admissability_locals mcache_validate_admissability_locals; + struct dml2_check_mode_supported_locals check_mode_supported_locals; + struct dml2_build_mode_programming_locals build_mode_programming_locals; + } scratch; + + struct { + struct { + struct dml2_legacy_core_build_mode_programming_wrapper_locals legacy_core_build_mode_programming_wrapper_locals; + } scratch; + } legacy; +}; +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c new file mode 100644 index 000000000000..4cfe64aa8492 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c @@ -0,0 +1,1174 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "dml2_mall_phantom.h" + +#include "dml2_dc_types.h" +#include "dml2_internal_types.h" +#include "dml2_utils.h" +#include "dml2_dc_resource_mgmt.h" + +#define MAX_ODM_FACTOR 4 +#define MAX_MPCC_FACTOR 4 + +struct dc_plane_pipe_pool { + int pipes_assigned_to_plane[MAX_ODM_FACTOR][MAX_MPCC_FACTOR]; + bool pipe_used[MAX_ODM_FACTOR][MAX_MPCC_FACTOR]; + int num_pipes_assigned_to_plane_for_mpcc_combine; + int num_pipes_assigned_to_plane_for_odm_combine; +}; + +struct dc_pipe_mapping_scratch { + struct { + unsigned int odm_factor; + unsigned int odm_slice_end_x[MAX_PIPES]; + struct pipe_ctx *next_higher_pipe_for_odm_slice[MAX_PIPES]; + } odm_info; + struct { + unsigned int mpc_factor; + struct pipe_ctx *prev_odm_pipe; + } mpc_info; + + struct dc_plane_pipe_pool pipe_pool; +}; + +static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *state, const struct dc_plane_state *plane, + unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) +{ + int i, j; + bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; + + if (!plane_id) + return false; + + for (i = 0; i < state->stream_count; i++) { + if (state->streams[i]->stream_id == stream_id) { + for (j = 0; j < state->stream_status[i].plane_count; j++) { + if (state->stream_status[i].plane_states[j] == plane && + (!is_plane_duplicate || (j == plane_index))) { + *plane_id = (i << 16) | j; + return true; + } + } + } + } + + return false; +} + +static int find_disp_cfg_idx_by_plane_id(struct dml2_dml_to_dc_pipe_mapping *mapping, unsigned int plane_id) +{ + int i; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (mapping->disp_cfg_to_plane_id_valid[i] && mapping->disp_cfg_to_plane_id[i] == plane_id) + return i; + } + + ASSERT(false); + return __DML2_WRAPPER_MAX_STREAMS_PLANES__; +} + +static int find_disp_cfg_idx_by_stream_id(struct dml2_dml_to_dc_pipe_mapping *mapping, unsigned int stream_id) +{ + int i; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (mapping->disp_cfg_to_stream_id_valid[i] && mapping->disp_cfg_to_stream_id[i] == stream_id) + return i; + } + + ASSERT(false); + return __DML2_WRAPPER_MAX_STREAMS_PLANES__; +} + +// The master pipe of a stream is defined as the top pipe in odm slice 0 +static struct pipe_ctx *find_master_pipe_of_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int stream_id) +{ + int i; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + if (state->res_ctx.pipe_ctx[i].stream && state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id) { + if (!state->res_ctx.pipe_ctx[i].prev_odm_pipe && !state->res_ctx.pipe_ctx[i].top_pipe) + return &state->res_ctx.pipe_ctx[i]; + } + } + + return NULL; +} + +static struct pipe_ctx *find_master_pipe_of_plane(struct dml2_context *ctx, + struct dc_state *state, unsigned int plane_id) +{ + int i; + unsigned int plane_id_assigned_to_pipe; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + if (state->res_ctx.pipe_ctx[i].plane_state && get_plane_id(ctx, state, state->res_ctx.pipe_ctx[i].plane_state, + state->res_ctx.pipe_ctx[i].stream->stream_id, + ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[state->res_ctx.pipe_ctx[i].pipe_idx], &plane_id_assigned_to_pipe)) { + if (plane_id_assigned_to_pipe == plane_id) + return &state->res_ctx.pipe_ctx[i]; + } + } + + return NULL; +} + +static unsigned int find_pipes_assigned_to_plane(struct dml2_context *ctx, + struct dc_state *state, unsigned int plane_id, unsigned int *pipes) +{ + int i; + unsigned int num_found = 0; + unsigned int plane_id_assigned_to_pipe = -1; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + if (!pipe->plane_state || !pipe->stream) + continue; + + get_plane_id(ctx, state, pipe->plane_state, pipe->stream->stream_id, + ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[pipe->pipe_idx], + &plane_id_assigned_to_pipe); + if (plane_id_assigned_to_pipe == plane_id && !pipe->prev_odm_pipe + && (!pipe->top_pipe || pipe->top_pipe->plane_state != pipe->plane_state)) { + while (pipe) { + struct pipe_ctx *mpc_pipe = pipe; + + while (mpc_pipe) { + pipes[num_found++] = mpc_pipe->pipe_idx; + mpc_pipe = mpc_pipe->bottom_pipe; + if (!mpc_pipe) + break; + if (mpc_pipe->plane_state != pipe->plane_state) + mpc_pipe = NULL; + } + pipe = pipe->next_odm_pipe; + } + break; + } + } + + return num_found; +} + +static bool validate_pipe_assignment(const struct dml2_context *ctx, const struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, const struct dml2_dml_to_dc_pipe_mapping *mapping) +{ +// int i, j, k; +// +// unsigned int plane_id; +// +// unsigned int disp_cfg_index; +// +// unsigned int pipes_assigned_to_plane[MAX_PIPES]; +// unsigned int num_pipes_assigned_to_plane; +// +// struct pipe_ctx *top_pipe; +// +// for (i = 0; i < state->stream_count; i++) { +// for (j = 0; j < state->stream_status[i]->plane_count; j++) { +// if (get_plane_id(state, state->stream_status.plane_states[j], &plane_id)) { +// disp_cfg_index = find_disp_cfg_idx_by_plane_id(mapping, plane_id); +// num_pipes_assigned_to_plane = find_pipes_assigned_to_plane(ctx, state, plane_id, pipes_assigned_to_plane); +// +// if (disp_cfg_index >= 0 && num_pipes_assigned_to_plane > 0) { +// // Verify the number of pipes assigned matches +// if (disp_cfg->hw.DPPPerSurface != num_pipes_assigned_to_plane) +// return false; +// +// top_pipe = find_top_pipe_in_tree(state->res_ctx.pipe_ctx[pipes_assigned_to_plane[0]]); +// +// // Verify MPC and ODM combine +// if (disp_cfg->hw.ODMMode == dml_odm_mode_bypass) { +// verify_combine_tree(top_pipe, state->streams[i]->stream_id, plane_id, state, false); +// } else { +// verify_combine_tree(top_pipe, state->streams[i]->stream_id, plane_id, state, true); +// } +// +// // TODO: could also do additional verification that the pipes in tree are the same as +// // pipes_assigned_to_plane +// } else { +// ASSERT(false); +// return false; +// } +// } else { +// ASSERT(false); +// return false; +// } +// } +// } + return true; +} + +static bool is_plane_using_pipe(const struct pipe_ctx *pipe) +{ + if (pipe->plane_state) + return true; + + return false; +} + +static bool is_pipe_free(const struct pipe_ctx *pipe) +{ + if (!pipe->plane_state && !pipe->stream) + return true; + + return false; +} + +static unsigned int find_preferred_pipe_candidates(const struct dc_state *existing_state, + const int pipe_count, + const unsigned int stream_id, + unsigned int *preferred_pipe_candidates) +{ + unsigned int num_preferred_candidates = 0; + int i; + + /* There is only one case which we consider for adding a pipe to the preferred + * pipe candidate array: + * + * 1. If the existing stream id of the pipe is equivalent to the stream id + * of the stream we are trying to achieve MPC/ODM combine for. This allows + * us to minimize the changes in pipe topology during the transition. + * + * However this condition comes with a caveat. We need to ignore pipes that will + * require a change in OPP but still have the same stream id. For example during + * an MPC to ODM transiton. + * + * Adding check to avoid pipe select on the head pipe by utilizing dc resource + * helper function resource_get_primary_dpp_pipe and comparing the pipe index. + */ + if (existing_state) { + for (i = 0; i < pipe_count; i++) { + if (existing_state->res_ctx.pipe_ctx[i].stream && existing_state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id) { + struct pipe_ctx *head_pipe = + resource_is_pipe_type(&existing_state->res_ctx.pipe_ctx[i], DPP_PIPE) ? + resource_get_primary_dpp_pipe(&existing_state->res_ctx.pipe_ctx[i]) : + NULL; + + // we should always respect the head pipe from selection + if (head_pipe && head_pipe->pipe_idx == i) + continue; + if (existing_state->res_ctx.pipe_ctx[i].plane_res.hubp && + existing_state->res_ctx.pipe_ctx[i].plane_res.hubp->opp_id != i && + (existing_state->res_ctx.pipe_ctx[i].prev_odm_pipe || + existing_state->res_ctx.pipe_ctx[i].next_odm_pipe)) + continue; + + preferred_pipe_candidates[num_preferred_candidates++] = i; + } + } + } + + return num_preferred_candidates; +} + +static unsigned int find_last_resort_pipe_candidates(const struct dc_state *existing_state, + const int pipe_count, + const unsigned int stream_id, + unsigned int *last_resort_pipe_candidates) +{ + unsigned int num_last_resort_candidates = 0; + int i; + + /* There are two cases where we would like to add a given pipe into the last + * candidate array: + * + * 1. If the pipe requires a change in OPP, for example during an MPC + * to ODM transiton. + * + * 2. If the pipe already has an enabled OTG. + */ + if (existing_state) { + for (i = 0; i < pipe_count; i++) { + struct pipe_ctx *head_pipe = + resource_is_pipe_type(&existing_state->res_ctx.pipe_ctx[i], DPP_PIPE) ? + resource_get_primary_dpp_pipe(&existing_state->res_ctx.pipe_ctx[i]) : + NULL; + + // we should always respect the head pipe from selection + if (head_pipe && head_pipe->pipe_idx == i) + continue; + if ((existing_state->res_ctx.pipe_ctx[i].plane_res.hubp && + existing_state->res_ctx.pipe_ctx[i].plane_res.hubp->opp_id != i) || + existing_state->res_ctx.pipe_ctx[i].stream_res.tg) + last_resort_pipe_candidates[num_last_resort_candidates++] = i; + } + } + + return num_last_resort_candidates; +} + +static bool is_pipe_in_candidate_array(const unsigned int pipe_idx, + const unsigned int *candidate_array, + const unsigned int candidate_array_size) +{ + int i; + + for (i = 0; i < candidate_array_size; i++) { + if (candidate_array[i] == pipe_idx) + return true; + } + + return false; +} + +static bool find_more_pipes_for_stream(struct dml2_context *ctx, + struct dc_state *state, // The state we want to find a free mapping in + unsigned int stream_id, // The stream we want this pipe to drive + int *assigned_pipes, + int *assigned_pipe_count, + int pipes_needed, + const struct dc_state *existing_state) // The state (optional) that we want to minimize remapping relative to +{ + struct pipe_ctx *pipe = NULL; + unsigned int preferred_pipe_candidates[MAX_PIPES] = {0}; + unsigned int last_resort_pipe_candidates[MAX_PIPES] = {0}; + unsigned int num_preferred_candidates = 0; + unsigned int num_last_resort_candidates = 0; + int i; + + if (existing_state) { + num_preferred_candidates = + find_preferred_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, preferred_pipe_candidates); + + num_last_resort_candidates = + find_last_resort_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, last_resort_pipe_candidates); + } + + // First see if any of the preferred are unmapped, and choose those instead + for (i = 0; pipes_needed > 0 && i < num_preferred_candidates; i++) { + pipe = &state->res_ctx.pipe_ctx[preferred_pipe_candidates[i]]; + if (!is_plane_using_pipe(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = preferred_pipe_candidates[i]; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + // We like to pair pipes starting from the higher order indicies for combining + for (i = ctx->config.dcn_pipe_count - 1; pipes_needed > 0 && i >= 0; i--) { + // Ignore any pipes that are the preferred or last resort candidate + if (is_pipe_in_candidate_array(i, preferred_pipe_candidates, num_preferred_candidates) || + is_pipe_in_candidate_array(i, last_resort_pipe_candidates, num_last_resort_candidates)) + continue; + + pipe = &state->res_ctx.pipe_ctx[i]; + if (!is_plane_using_pipe(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = i; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + // Only use the last resort pipe candidates as a last resort + for (i = 0; pipes_needed > 0 && i < num_last_resort_candidates; i++) { + pipe = &state->res_ctx.pipe_ctx[last_resort_pipe_candidates[i]]; + if (!is_plane_using_pipe(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = last_resort_pipe_candidates[i]; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + ASSERT(pipes_needed <= 0); // Validation should prevent us from building a pipe context that exceeds the number of HW resoruces available + + return pipes_needed <= 0; +} + +static bool find_more_free_pipes(struct dml2_context *ctx, + struct dc_state *state, // The state we want to find a free mapping in + unsigned int stream_id, // The stream we want this pipe to drive + int *assigned_pipes, + int *assigned_pipe_count, + int pipes_needed, + const struct dc_state *existing_state) // The state (optional) that we want to minimize remapping relative to +{ + struct pipe_ctx *pipe = NULL; + unsigned int preferred_pipe_candidates[MAX_PIPES] = {0}; + unsigned int last_resort_pipe_candidates[MAX_PIPES] = {0}; + unsigned int num_preferred_candidates = 0; + unsigned int num_last_resort_candidates = 0; + int i; + + if (existing_state) { + num_preferred_candidates = + find_preferred_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, preferred_pipe_candidates); + + num_last_resort_candidates = + find_last_resort_pipe_candidates(existing_state, ctx->config.dcn_pipe_count, stream_id, last_resort_pipe_candidates); + } + + // First see if any of the preferred are unmapped, and choose those instead + for (i = 0; pipes_needed > 0 && i < num_preferred_candidates; i++) { + pipe = &state->res_ctx.pipe_ctx[preferred_pipe_candidates[i]]; + if (is_pipe_free(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = preferred_pipe_candidates[i]; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + // We like to pair pipes starting from the higher order indicies for combining + for (i = ctx->config.dcn_pipe_count - 1; pipes_needed > 0 && i >= 0; i--) { + // Ignore any pipes that are the preferred or last resort candidate + if (is_pipe_in_candidate_array(i, preferred_pipe_candidates, num_preferred_candidates) || + is_pipe_in_candidate_array(i, last_resort_pipe_candidates, num_last_resort_candidates)) + continue; + + pipe = &state->res_ctx.pipe_ctx[i]; + if (is_pipe_free(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = i; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + // Only use the last resort pipe candidates as a last resort + for (i = 0; pipes_needed > 0 && i < num_last_resort_candidates; i++) { + pipe = &state->res_ctx.pipe_ctx[last_resort_pipe_candidates[i]]; + if (is_pipe_free(pipe)) { + pipes_needed--; + // TODO: This doens't make sense really, pipe_idx should always be valid + pipe->pipe_idx = last_resort_pipe_candidates[i]; + assigned_pipes[(*assigned_pipe_count)++] = pipe->pipe_idx; + } + } + + ASSERT(pipes_needed == 0); // Validation should prevent us from building a pipe context that exceeds the number of HW resoruces available + + return pipes_needed == 0; +} + +static void sort_pipes_for_splitting(struct dc_plane_pipe_pool *pipes) +{ + bool sorted, swapped; + unsigned int cur_index; + int odm_slice_index; + + for (odm_slice_index = 0; odm_slice_index < pipes->num_pipes_assigned_to_plane_for_odm_combine; odm_slice_index++) { + // Sort each MPCC set + //Un-optimized bubble sort, but that's okay for array sizes <= 6 + + if (pipes->num_pipes_assigned_to_plane_for_mpcc_combine <= 1) + sorted = true; + else + sorted = false; + + cur_index = 0; + swapped = false; + while (!sorted) { + if (pipes->pipes_assigned_to_plane[odm_slice_index][cur_index] > pipes->pipes_assigned_to_plane[odm_slice_index][cur_index + 1]) { + swap(pipes->pipes_assigned_to_plane[odm_slice_index][cur_index + 1], + pipes->pipes_assigned_to_plane[odm_slice_index][cur_index]); + + swapped = true; + } + + cur_index++; + + if (cur_index == pipes->num_pipes_assigned_to_plane_for_mpcc_combine - 1) { + cur_index = 0; + + if (swapped) + sorted = false; + else + sorted = true; + + swapped = false; + } + + } + } +} + +// For example, 3840 x 2160, ODM2:1 has a slice array of [1919, 3839], meaning, slice0 spans h_pixels 0->1919, and slice1 spans 1920->3840 +static void calculate_odm_slices(const struct dc_stream_state *stream, unsigned int odm_factor, unsigned int *odm_slice_end_x) +{ + unsigned int slice_size = 0; + int i; + + if (odm_factor < 1 || odm_factor > 4) { + ASSERT(false); + return; + } + + slice_size = stream->src.width / odm_factor; + + for (i = 0; i < odm_factor; i++) + odm_slice_end_x[i] = (slice_size * (i + 1)) - 1; + + odm_slice_end_x[odm_factor - 1] = stream->src.width - 1; +} + +static void add_odm_slice_to_odm_tree(struct dml2_context *ctx, + struct dc_state *state, + struct dc_pipe_mapping_scratch *scratch, + unsigned int odm_slice_index) +{ + struct pipe_ctx *pipe = NULL; + int i; + + // MPCC Combine + ODM Combine is not supported, so there should never be a case where the current plane + // has more than 1 pipe mapped to it for a given slice. + ASSERT(scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine == 1 || scratch->pipe_pool.num_pipes_assigned_to_plane_for_odm_combine == 1); + + for (i = 0; i < scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine; i++) { + pipe = &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][i]]; + + if (scratch->mpc_info.prev_odm_pipe) + scratch->mpc_info.prev_odm_pipe->next_odm_pipe = pipe; + + pipe->prev_odm_pipe = scratch->mpc_info.prev_odm_pipe; + pipe->next_odm_pipe = NULL; + } + scratch->mpc_info.prev_odm_pipe = pipe; +} + +static struct pipe_ctx *add_plane_to_blend_tree(struct dml2_context *ctx, + struct dc_state *state, + const struct dc_plane_state *plane, + struct dc_plane_pipe_pool *pipe_pool, + unsigned int odm_slice, + struct pipe_ctx *top_pipe) +{ + int i; + + for (i = 0; i < pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine; i++) { + if (top_pipe) + top_pipe->bottom_pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]]; + + pipe_pool->pipe_used[odm_slice][i] = true; + + state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]].top_pipe = top_pipe; + state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]].bottom_pipe = NULL; + + top_pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][i]]; + } + + // After running the above loop, the top pipe actually ends up pointing to the bottom of this MPCC combine tree, so we are actually + // returning the bottom pipe here + return top_pipe; +} + +static unsigned int find_pipes_assigned_to_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int stream_id, unsigned int *pipes) +{ + int i; + unsigned int num_found = 0; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + if (pipe->stream && pipe->stream->stream_id == stream_id && !pipe->top_pipe && !pipe->prev_odm_pipe) { + while (pipe) { + pipes[num_found++] = pipe->pipe_idx; + pipe = pipe->next_odm_pipe; + } + break; + } + } + + return num_found; +} + +static struct pipe_ctx *assign_pipes_to_stream(struct dml2_context *ctx, struct dc_state *state, + const struct dc_stream_state *stream, + int odm_factor, + struct dc_plane_pipe_pool *pipe_pool, + const struct dc_state *existing_state) +{ + struct pipe_ctx *master_pipe; + unsigned int pipes_needed; + unsigned int pipes_assigned; + unsigned int pipes[MAX_PIPES] = {0}; + unsigned int next_pipe_to_assign; + int odm_slice; + + pipes_needed = odm_factor; + + master_pipe = find_master_pipe_of_stream(ctx, state, stream->stream_id); + ASSERT(master_pipe); + + pipes_assigned = find_pipes_assigned_to_stream(ctx, state, stream->stream_id, pipes); + + find_more_free_pipes(ctx, state, stream->stream_id, pipes, &pipes_assigned, pipes_needed - pipes_assigned, existing_state); + + ASSERT(pipes_assigned == pipes_needed); + + next_pipe_to_assign = 0; + for (odm_slice = 0; odm_slice < odm_factor; odm_slice++) + pipe_pool->pipes_assigned_to_plane[odm_slice][0] = pipes[next_pipe_to_assign++]; + + pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine = 1; + pipe_pool->num_pipes_assigned_to_plane_for_odm_combine = odm_factor; + + return master_pipe; +} + +static struct pipe_ctx *assign_pipes_to_plane(struct dml2_context *ctx, struct dc_state *state, + const struct dc_stream_state *stream, + const struct dc_plane_state *plane, + int odm_factor, + int mpc_factor, + int plane_index, + struct dc_plane_pipe_pool *pipe_pool, + const struct dc_state *existing_state) +{ + struct pipe_ctx *master_pipe = NULL; + unsigned int plane_id; + unsigned int pipes_needed; + unsigned int pipes_assigned; + unsigned int pipes[MAX_PIPES] = {0}; + unsigned int next_pipe_to_assign; + int odm_slice, mpc_slice; + + if (!get_plane_id(ctx, state, plane, stream->stream_id, plane_index, &plane_id)) { + ASSERT(false); + return master_pipe; + } + + pipes_needed = mpc_factor * odm_factor; + + master_pipe = find_master_pipe_of_plane(ctx, state, plane_id); + ASSERT(master_pipe); + + pipes_assigned = find_pipes_assigned_to_plane(ctx, state, plane_id, pipes); + + find_more_pipes_for_stream(ctx, state, stream->stream_id, pipes, &pipes_assigned, pipes_needed - pipes_assigned, existing_state); + + ASSERT(pipes_assigned >= pipes_needed); + + next_pipe_to_assign = 0; + for (odm_slice = 0; odm_slice < odm_factor; odm_slice++) + for (mpc_slice = 0; mpc_slice < mpc_factor; mpc_slice++) + pipe_pool->pipes_assigned_to_plane[odm_slice][mpc_slice] = pipes[next_pipe_to_assign++]; + + pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine = mpc_factor; + pipe_pool->num_pipes_assigned_to_plane_for_odm_combine = odm_factor; + + return master_pipe; +} + +static bool is_pipe_used(const struct dc_plane_pipe_pool *pool, unsigned int pipe_idx) +{ + int i, j; + + for (i = 0; i < pool->num_pipes_assigned_to_plane_for_odm_combine; i++) { + for (j = 0; j < pool->num_pipes_assigned_to_plane_for_mpcc_combine; j++) { + if (pool->pipes_assigned_to_plane[i][j] == pipe_idx && pool->pipe_used[i][j]) + return true; + } + } + + return false; +} + +static void free_pipe(struct pipe_ctx *pipe) +{ + memset(pipe, 0, sizeof(struct pipe_ctx)); +} + +static void free_unused_pipes_for_plane(struct dml2_context *ctx, struct dc_state *state, + const struct dc_plane_state *plane, const struct dc_plane_pipe_pool *pool, unsigned int stream_id, int plane_index) +{ + int i; + bool is_plane_duplicate = ctx->v20.scratch.plane_duplicate_exists; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + if (state->res_ctx.pipe_ctx[i].plane_state == plane && + state->res_ctx.pipe_ctx[i].stream->stream_id == stream_id && + (!is_plane_duplicate || + ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[state->res_ctx.pipe_ctx[i].pipe_idx] == plane_index) && + !is_pipe_used(pool, state->res_ctx.pipe_ctx[i].pipe_idx)) { + free_pipe(&state->res_ctx.pipe_ctx[i]); + } + } +} + +static void remove_pipes_from_blend_trees(struct dml2_context *ctx, struct dc_state *state, struct dc_plane_pipe_pool *pipe_pool, unsigned int odm_slice) +{ + struct pipe_ctx *pipe; + int i; + + for (i = 0; i < pipe_pool->num_pipes_assigned_to_plane_for_mpcc_combine; i++) { + pipe = &state->res_ctx.pipe_ctx[pipe_pool->pipes_assigned_to_plane[odm_slice][0]]; + if (pipe->top_pipe) + pipe->top_pipe->bottom_pipe = pipe->bottom_pipe; + + if (pipe->bottom_pipe) + pipe->bottom_pipe = pipe->top_pipe; + + pipe_pool->pipe_used[odm_slice][i] = true; + } +} + +static void map_pipes_for_stream(struct dml2_context *ctx, struct dc_state *state, const struct dc_stream_state *stream, + struct dc_pipe_mapping_scratch *scratch, const struct dc_state *existing_state) +{ + int odm_slice_index; + struct pipe_ctx *master_pipe = NULL; + + + master_pipe = assign_pipes_to_stream(ctx, state, stream, scratch->odm_info.odm_factor, &scratch->pipe_pool, existing_state); + sort_pipes_for_splitting(&scratch->pipe_pool); + + for (odm_slice_index = 0; odm_slice_index < scratch->odm_info.odm_factor; odm_slice_index++) { + remove_pipes_from_blend_trees(ctx, state, &scratch->pipe_pool, odm_slice_index); + + add_odm_slice_to_odm_tree(ctx, state, scratch, odm_slice_index); + + ctx->config.callbacks.acquire_secondary_pipe_for_mpc_odm(ctx->config.callbacks.dc, state, + master_pipe, &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][0]], true); + } +} + +static void map_pipes_for_plane(struct dml2_context *ctx, struct dc_state *state, const struct dc_stream_state *stream, const struct dc_plane_state *plane, + int plane_index, struct dc_pipe_mapping_scratch *scratch, const struct dc_state *existing_state) +{ + int odm_slice_index; + unsigned int plane_id; + struct pipe_ctx *master_pipe = NULL; + int i; + + if (!get_plane_id(ctx, state, plane, stream->stream_id, plane_index, &plane_id)) { + ASSERT(false); + return; + } + + master_pipe = assign_pipes_to_plane(ctx, state, stream, plane, scratch->odm_info.odm_factor, + scratch->mpc_info.mpc_factor, plane_index, &scratch->pipe_pool, existing_state); + sort_pipes_for_splitting(&scratch->pipe_pool); + + for (odm_slice_index = 0; odm_slice_index < scratch->odm_info.odm_factor; odm_slice_index++) { + // Now we have a list of all pipes to be used for this plane/stream, now setup the tree. + scratch->odm_info.next_higher_pipe_for_odm_slice[odm_slice_index] = add_plane_to_blend_tree(ctx, state, + plane, + &scratch->pipe_pool, + odm_slice_index, + scratch->odm_info.next_higher_pipe_for_odm_slice[odm_slice_index]); + + add_odm_slice_to_odm_tree(ctx, state, scratch, odm_slice_index); + + for (i = 0; i < scratch->pipe_pool.num_pipes_assigned_to_plane_for_mpcc_combine; i++) { + + ctx->config.callbacks.acquire_secondary_pipe_for_mpc_odm(ctx->config.callbacks.dc, state, + master_pipe, &state->res_ctx.pipe_ctx[scratch->pipe_pool.pipes_assigned_to_plane[odm_slice_index][i]], true); + } + } + + free_unused_pipes_for_plane(ctx, state, plane, &scratch->pipe_pool, stream->stream_id, plane_index); +} + +static unsigned int get_target_mpc_factor(struct dml2_context *ctx, + struct dc_state *state, + const struct dml_display_cfg_st *disp_cfg, + struct dml2_dml_to_dc_pipe_mapping *mapping, + const struct dc_stream_status *status, + const struct dc_stream_state *stream, + int plane_idx) +{ + unsigned int plane_id; + unsigned int cfg_idx; + unsigned int mpc_factor; + + if (ctx->architecture == dml2_architecture_20) { + get_plane_id(ctx, state, status->plane_states[plane_idx], + stream->stream_id, plane_idx, &plane_id); + cfg_idx = find_disp_cfg_idx_by_plane_id(mapping, plane_id); + mpc_factor = (unsigned int)disp_cfg->hw.DPPPerSurface[cfg_idx]; + } else if (ctx->architecture == dml2_architecture_21) { + if (ctx->config.svp_pstate.callbacks.get_stream_subvp_type(state, stream) == SUBVP_PHANTOM) { + struct dc_stream_state *main_stream; + struct dc_stream_status *main_stream_status; + + /* get stream id of main stream */ + main_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(state, stream); + if (!main_stream) { + ASSERT(false); + return 1; + } + + main_stream_status = ctx->config.callbacks.get_stream_status(state, main_stream); + if (!main_stream_status) { + ASSERT(false); + return 1; + } + + /* get plane id for associated main plane */ + get_plane_id(ctx, state, main_stream_status->plane_states[plane_idx], + main_stream->stream_id, plane_idx, &plane_id); + } else { + get_plane_id(ctx, state, status->plane_states[plane_idx], + stream->stream_id, plane_idx, &plane_id); + } + + cfg_idx = find_disp_cfg_idx_by_plane_id(mapping, plane_id); + mpc_factor = ctx->v21.mode_programming.programming->plane_programming[cfg_idx].num_dpps_required; + } else { + mpc_factor = 1; + ASSERT(false); + } + + /* For stereo timings, we need to pipe split */ + if (dml2_is_stereo_timing(stream)) + mpc_factor = 2; + + return mpc_factor; +} + +static unsigned int get_target_odm_factor( + const struct dml2_context *ctx, + struct dc_state *state, + const struct dml_display_cfg_st *disp_cfg, + struct dml2_dml_to_dc_pipe_mapping *mapping, + const struct dc_stream_state *stream) +{ + unsigned int cfg_idx; + + if (ctx->architecture == dml2_architecture_20) { + cfg_idx = find_disp_cfg_idx_by_stream_id( + mapping, stream->stream_id); + switch (disp_cfg->hw.ODMMode[cfg_idx]) { + case dml_odm_mode_bypass: + return 1; + case dml_odm_mode_combine_2to1: + return 2; + case dml_odm_mode_combine_4to1: + return 4; + default: + break; + } + } else if (ctx->architecture == dml2_architecture_21) { + if (ctx->config.svp_pstate.callbacks.get_stream_subvp_type(state, stream) == SUBVP_PHANTOM) { + struct dc_stream_state *main_stream; + + /* get stream id of main stream */ + main_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(state, stream); + if (!main_stream) + goto failed; + + /* get cfg idx for associated main stream */ + cfg_idx = find_disp_cfg_idx_by_stream_id( + mapping, main_stream->stream_id); + } else { + cfg_idx = find_disp_cfg_idx_by_stream_id( + mapping, stream->stream_id); + } + + return ctx->v21.mode_programming.programming->stream_programming[cfg_idx].num_odms_required; + } + +failed: + ASSERT(false); + return 1; +} + +static unsigned int get_source_odm_factor(const struct dml2_context *ctx, + struct dc_state *state, + const struct dc_stream_state *stream) +{ + struct pipe_ctx *otg_master = ctx->config.callbacks.get_otg_master_for_stream(&state->res_ctx, stream); + + if (!otg_master) + return 0; + + return ctx->config.callbacks.get_odm_slice_count(otg_master); +} + +static unsigned int get_source_mpc_factor(const struct dml2_context *ctx, + struct dc_state *state, + const struct dc_plane_state *plane) +{ + struct pipe_ctx *dpp_pipes[MAX_PIPES] = {0}; + int dpp_pipe_count = ctx->config.callbacks.get_dpp_pipes_for_plane(plane, + &state->res_ctx, dpp_pipes); + + ASSERT(dpp_pipe_count > 0); + return ctx->config.callbacks.get_mpc_slice_count(dpp_pipes[0]); +} + + +static void populate_mpc_factors_for_stream( + struct dml2_context *ctx, + const struct dml_display_cfg_st *disp_cfg, + struct dml2_dml_to_dc_pipe_mapping *mapping, + struct dc_state *state, + unsigned int stream_idx, + struct dml2_pipe_combine_factor odm_factor, + struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) +{ + const struct dc_stream_status *status = &state->stream_status[stream_idx]; + int i; + + for (i = 0; i < status->plane_count; i++) { + mpc_factors[i].source = get_source_mpc_factor(ctx, state, status->plane_states[i]); + mpc_factors[i].target = (odm_factor.target == 1) ? + get_target_mpc_factor(ctx, state, disp_cfg, mapping, status, state->streams[stream_idx], i) : 1; + } +} + +static void populate_odm_factors(const struct dml2_context *ctx, + const struct dml_display_cfg_st *disp_cfg, + struct dml2_dml_to_dc_pipe_mapping *mapping, + struct dc_state *state, + struct dml2_pipe_combine_factor odm_factors[MAX_PIPES]) +{ + int i; + + for (i = 0; i < state->stream_count; i++) { + odm_factors[i].source = get_source_odm_factor(ctx, state, state->streams[i]); + odm_factors[i].target = get_target_odm_factor( + ctx, state, disp_cfg, mapping, state->streams[i]); + } +} + +static bool unmap_dc_pipes_for_stream(struct dml2_context *ctx, + struct dc_state *state, + const struct dc_state *existing_state, + const struct dc_stream_state *stream, + const struct dc_stream_status *status, + struct dml2_pipe_combine_factor odm_factor, + struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) +{ + int plane_idx; + bool result = true; + + for (plane_idx = 0; plane_idx < status->plane_count; plane_idx++) + if (mpc_factors[plane_idx].target < mpc_factors[plane_idx].source) + result &= ctx->config.callbacks.update_pipes_for_plane_with_slice_count( + state, + existing_state, + ctx->config.callbacks.dc->res_pool, + status->plane_states[plane_idx], + mpc_factors[plane_idx].target); + if (odm_factor.target < odm_factor.source) + result &= ctx->config.callbacks.update_pipes_for_stream_with_slice_count( + state, + existing_state, + ctx->config.callbacks.dc->res_pool, + stream, + odm_factor.target); + return result; +} + +static bool map_dc_pipes_for_stream(struct dml2_context *ctx, + struct dc_state *state, + const struct dc_state *existing_state, + const struct dc_stream_state *stream, + const struct dc_stream_status *status, + struct dml2_pipe_combine_factor odm_factor, + struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES]) +{ + int plane_idx; + bool result = true; + + for (plane_idx = 0; plane_idx < status->plane_count; plane_idx++) + if (mpc_factors[plane_idx].target > mpc_factors[plane_idx].source) + result &= ctx->config.callbacks.update_pipes_for_plane_with_slice_count( + state, + existing_state, + ctx->config.callbacks.dc->res_pool, + status->plane_states[plane_idx], + mpc_factors[plane_idx].target); + if (odm_factor.target > odm_factor.source) + result &= ctx->config.callbacks.update_pipes_for_stream_with_slice_count( + state, + existing_state, + ctx->config.callbacks.dc->res_pool, + stream, + odm_factor.target); + return result; +} + +static bool map_dc_pipes_with_callbacks(struct dml2_context *ctx, + struct dc_state *state, + const struct dml_display_cfg_st *disp_cfg, + struct dml2_dml_to_dc_pipe_mapping *mapping, + const struct dc_state *existing_state) +{ + int i; + bool result = true; + + populate_odm_factors(ctx, disp_cfg, mapping, state, ctx->pipe_combine_scratch.odm_factors); + for (i = 0; i < state->stream_count; i++) + populate_mpc_factors_for_stream(ctx, disp_cfg, mapping, state, + i, ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); + for (i = 0; i < state->stream_count; i++) + result &= unmap_dc_pipes_for_stream(ctx, state, existing_state, state->streams[i], + &state->stream_status[i], ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); + for (i = 0; i < state->stream_count; i++) + result &= map_dc_pipes_for_stream(ctx, state, existing_state, state->streams[i], + &state->stream_status[i], ctx->pipe_combine_scratch.odm_factors[i], ctx->pipe_combine_scratch.mpc_factors[i]); + + return result; +} + +bool dml2_map_dc_pipes(struct dml2_context *ctx, struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, struct dml2_dml_to_dc_pipe_mapping *mapping, const struct dc_state *existing_state) +{ + int stream_index, plane_index, i; + + unsigned int stream_disp_cfg_index; + unsigned int plane_disp_cfg_index; + unsigned int disp_cfg_index_max; + + unsigned int plane_id; + unsigned int stream_id; + + const unsigned int *ODMMode, *DPPPerSurface; + unsigned int odm_mode_array[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}, dpp_per_surface_array[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; + struct dc_pipe_mapping_scratch scratch; + + if (ctx->config.map_dc_pipes_with_callbacks) + return map_dc_pipes_with_callbacks( + ctx, state, disp_cfg, mapping, existing_state); + + if (ctx->architecture == dml2_architecture_21) { + /* + * Extract ODM and DPP outputs from DML2.1 and map them in an array as required for pipe mapping in dml2_map_dc_pipes. + * As data cannot be directly extracted in const pointers, assign these arrays to const pointers before proceeding to + * maximize the reuse of existing code. Const pointers are required because dml2.0 dml_display_cfg_st is const. + * + */ + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + odm_mode_array[i] = ctx->v21.mode_programming.programming->stream_programming[i].num_odms_required; + dpp_per_surface_array[i] = ctx->v21.mode_programming.programming->plane_programming[i].num_dpps_required; + } + + ODMMode = (const unsigned int *)odm_mode_array; + DPPPerSurface = (const unsigned int *)dpp_per_surface_array; + disp_cfg_index_max = __DML2_WRAPPER_MAX_STREAMS_PLANES__; + } else { + ODMMode = (unsigned int *)disp_cfg->hw.ODMMode; + DPPPerSurface = disp_cfg->hw.DPPPerSurface; + disp_cfg_index_max = __DML_NUM_PLANES__; + } + + for (stream_index = 0; stream_index < state->stream_count; stream_index++) { + memset(&scratch, 0, sizeof(struct dc_pipe_mapping_scratch)); + + stream_id = state->streams[stream_index]->stream_id; + stream_disp_cfg_index = find_disp_cfg_idx_by_stream_id(mapping, stream_id); + if (stream_disp_cfg_index >= disp_cfg_index_max) + continue; + + if (ctx->architecture == dml2_architecture_20) { + if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_bypass) { + scratch.odm_info.odm_factor = 1; + } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_2to1) { + scratch.odm_info.odm_factor = 2; + } else if (ODMMode[stream_disp_cfg_index] == dml_odm_mode_combine_4to1) { + scratch.odm_info.odm_factor = 4; + } else { + ASSERT(false); + scratch.odm_info.odm_factor = 1; + } + } else if (ctx->architecture == dml2_architecture_21) { + /* After DML2.1 update, ODM interpretation needs to change and is no longer same as for DML2.0. + * This is not an issue with new resource management logic. This block ensure backcompat + * with legacy pipe management with updated DML. + * */ + if (ODMMode[stream_disp_cfg_index] == 1) { + scratch.odm_info.odm_factor = 1; + } else if (ODMMode[stream_disp_cfg_index] == 2) { + scratch.odm_info.odm_factor = 2; + } else if (ODMMode[stream_disp_cfg_index] == 4) { + scratch.odm_info.odm_factor = 4; + } else { + ASSERT(false); + scratch.odm_info.odm_factor = 1; + } + } + calculate_odm_slices(state->streams[stream_index], scratch.odm_info.odm_factor, scratch.odm_info.odm_slice_end_x); + + // If there are no planes, you still want to setup ODM... + if (state->stream_status[stream_index].plane_count == 0) { + map_pipes_for_stream(ctx, state, state->streams[stream_index], &scratch, existing_state); + } + + for (plane_index = 0; plane_index < state->stream_status[stream_index].plane_count; plane_index++) { + // Planes are ordered top to bottom. + if (get_plane_id(ctx, state, state->stream_status[stream_index].plane_states[plane_index], + stream_id, plane_index, &plane_id)) { + plane_disp_cfg_index = find_disp_cfg_idx_by_plane_id(mapping, plane_id); + + // Setup mpc_info for this plane + scratch.mpc_info.prev_odm_pipe = NULL; + if (scratch.odm_info.odm_factor == 1 && plane_disp_cfg_index < disp_cfg_index_max) { + // If ODM combine is not inuse, then the number of pipes + // per plane is determined by MPC combine factor + scratch.mpc_info.mpc_factor = DPPPerSurface[plane_disp_cfg_index]; + + //For stereo timings, we need to pipe split + if (dml2_is_stereo_timing(state->streams[stream_index])) + scratch.mpc_info.mpc_factor = 2; + } else { + // If ODM combine is enabled, then we use at most 1 pipe per + // odm slice per plane, i.e. MPC combine is never used + scratch.mpc_info.mpc_factor = 1; + } + + ASSERT(scratch.odm_info.odm_factor * scratch.mpc_info.mpc_factor > 0); + + // Clear the pool assignment scratch (which is per plane) + memset(&scratch.pipe_pool, 0, sizeof(struct dc_plane_pipe_pool)); + + map_pipes_for_plane(ctx, state, state->streams[stream_index], + state->stream_status[stream_index].plane_states[plane_index], plane_index, &scratch, existing_state); + } else { + // Plane ID cannot be generated, therefore no DML mapping can be performed. + ASSERT(false); + } + } + + } + + if (!validate_pipe_assignment(ctx, state, disp_cfg, mapping)) + ASSERT(false); + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state) { + if (!ctx->config.callbacks.build_scaling_params(pipe)) { + ASSERT(false); + } + } + + if (ctx->config.callbacks.build_test_pattern_params && + pipe->stream && + pipe->prev_odm_pipe == NULL && + pipe->top_pipe == NULL) + ctx->config.callbacks.build_test_pattern_params(&state->res_ctx, pipe); + } + + return true; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h new file mode 100644 index 000000000000..1538b708d8be --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h @@ -0,0 +1,52 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML2_DC_RESOURCE_MGMT_H__ +#define __DML2_DC_RESOURCE_MGMT_H__ + +#include "dml2_dc_types.h" + +struct dml2_context; +struct dml2_dml_to_dc_pipe_mapping; +struct dml_display_cfg_st; + +/* + * dml2_map_dc_pipes - Creates a pipe linkage in dc_state based on current display config. + * @ctx: Input dml2 context + * @state: Current dc_state to be updated. + * @disp_cfg: Current display config. + * @mapping: Pipe mapping logic structure to keep a track of pipes to be used. + * + * Based on ODM and DPPPersurface outputs calculated by the DML for the current display + * config, create a pipe linkage in dc_state which is then used by DC core. + * Make this function generic to be used by multiple DML versions. + * + * Return: True if pipe mapping and linking is successful, false otherwise. + */ + +bool dml2_map_dc_pipes(struct dml2_context *ctx, struct dc_state *state, const struct dml_display_cfg_st *disp_cfg, struct dml2_dml_to_dc_pipe_mapping *mapping, const struct dc_state *existing_state); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h new file mode 100644 index 000000000000..7ca7f2a743c2 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h @@ -0,0 +1,43 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + + +/* + * Wrapper header for externally defined types from DC. These types come from + * dc headers when building DML2 as part of DC, but are defined here when building + * DML2 as a standalone library (such as for unit testing). + */ + +#ifndef __DML2_DC_TYPES_H__ +#define __DML2_DC_TYPES_H__ + +#include "resource.h" +#include "core_types.h" +#include "dsc.h" +#include "clk_mgr.h" +#include "dc_state_priv.h" + +#endif //__DML2_DC_TYPES_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h new file mode 100644 index 000000000000..55b3e3ca54f7 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h @@ -0,0 +1,157 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML2_INTERNAL_TYPES_H__ +#define __DML2_INTERNAL_TYPES_H__ + +#include "dml2_dc_types.h" +#include "display_mode_core.h" +#include "dml2_wrapper.h" +#include "dml2_policy.h" + +#include "dml_top.h" +#include "dml21_wrapper.h" + +struct dml2_wrapper_optimize_configuration_params { + struct display_mode_lib_st *dml_core_ctx; + struct dml2_configuration_options *config; + struct ip_params_st *ip_params; + struct dml_display_cfg_st *cur_display_config; + struct dml_display_cfg_st *new_display_config; + const struct dml_mode_support_info_st *cur_mode_support_info; + struct dml_mode_eval_policy_st *cur_policy; + struct dml_mode_eval_policy_st *new_policy; +}; + +struct dml2_calculate_lowest_supported_state_for_temp_read_scratch { + struct dml_mode_support_info_st evaluation_info; + dml_float_t uclk_change_latencies[__DML_MAX_STATE_ARRAY_SIZE__]; + struct dml_display_cfg_st cur_display_config; + struct dml_display_cfg_st new_display_config; + struct dml_mode_eval_policy_st new_policy; + struct dml_mode_eval_policy_st cur_policy; +}; + +struct dml2_create_scratch { + struct dml2_policy_build_synthetic_soc_states_scratch build_synthetic_socbb_scratch; + struct soc_states_st in_states; +}; + +struct dml2_calculate_rq_and_dlg_params_scratch { + struct _vcs_dpi_dml_display_rq_regs_st rq_regs; + struct _vcs_dpi_dml_display_dlg_regs_st disp_dlg_regs; + struct _vcs_dpi_dml_display_ttu_regs_st disp_ttu_regs; +}; + +#define __DML2_WRAPPER_MAX_STREAMS_PLANES__ 6 + +struct dml2_dml_to_dc_pipe_mapping { + unsigned int disp_cfg_to_stream_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + bool disp_cfg_to_stream_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + unsigned int disp_cfg_to_plane_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + bool disp_cfg_to_plane_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + unsigned int dml_pipe_idx_to_stream_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + bool dml_pipe_idx_to_stream_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + unsigned int dml_pipe_idx_to_plane_id[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + bool dml_pipe_idx_to_plane_id_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + unsigned int dml_pipe_idx_to_plane_index[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; + bool dml_pipe_idx_to_plane_index_valid[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; +}; + +struct dml2_wrapper_scratch { + struct dml_display_cfg_st cur_display_config; + struct dml_display_cfg_st new_display_config; + struct dml_mode_eval_policy_st new_policy; + struct dml_mode_eval_policy_st cur_policy; + struct dml_mode_support_info_st mode_support_info; + struct dml_mode_support_ex_params_st mode_support_params; + + struct dummy_pstate_entry dummy_pstate_table[4]; + + struct dml2_create_scratch create_scratch; + struct dml2_calculate_lowest_supported_state_for_temp_read_scratch dml2_calculate_lowest_supported_state_for_temp_read_scratch; + struct dml2_calculate_rq_and_dlg_params_scratch calculate_rq_and_dlg_params_scratch; + + struct dml2_wrapper_optimize_configuration_params optimize_configuration_params; + struct dml2_policy_build_synthetic_soc_states_params build_synthetic_socbb_params; + + struct dml2_dml_to_dc_pipe_mapping dml_to_dc_pipe_mapping; + bool enable_flexible_pipe_mapping; + bool plane_duplicate_exists; + int hpo_stream_to_link_encoder_mapping[MAX_HPO_DP2_ENCODERS]; +}; + +struct dml2_helper_det_policy_scratch { + int dpps_per_surface[MAX_PLANES]; +}; + +enum dml2_architecture { + dml2_architecture_20, + dml2_architecture_21 +}; + +struct prepare_mcache_programming_locals { + struct dml2_build_mcache_programming_in_out build_mcache_programming_params; +}; + +struct dml21_wrapper_scratch { + struct prepare_mcache_programming_locals prepare_mcache_locals; + struct pipe_ctx temp_pipe; +}; + +struct dml2_pipe_combine_factor { + unsigned int source; + unsigned int target; +}; + +struct dml2_pipe_combine_scratch { + struct dml2_pipe_combine_factor odm_factors[MAX_PIPES]; + struct dml2_pipe_combine_factor mpc_factors[MAX_PIPES][MAX_PIPES]; +}; + +struct dml2_context { + enum dml2_architecture architecture; + struct dml2_configuration_options config; + struct dml2_helper_det_policy_scratch det_helper_scratch; + struct dml2_pipe_combine_scratch pipe_combine_scratch; + union { + struct { + struct display_mode_lib_st dml_core_ctx; + struct dml2_wrapper_scratch scratch; + struct dcn_watermarks g6_temp_read_watermark_set; + } v20; + struct { + struct dml21_wrapper_scratch scratch; + struct dml2_initialize_instance_in_out dml_init; + struct dml2_display_cfg display_config; + struct dml2_check_mode_supported_in_out mode_support; + struct dml2_build_mode_programming_in_out mode_programming; + struct dml2_dml_to_dc_pipe_mapping dml_to_dc_pipe_mapping; + } v21; + }; +}; + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c new file mode 100644 index 000000000000..66040c877d68 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c @@ -0,0 +1,911 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + + +#include "dml2_dc_types.h" +#include "dml2_internal_types.h" +#include "dml2_utils.h" +#include "dml2_mall_phantom.h" + +unsigned int dml2_helper_calculate_num_ways_for_subvp(struct dml2_context *ctx, struct dc_state *context) +{ + uint32_t num_ways = 0; + uint32_t bytes_per_pixel = 0; + uint32_t cache_lines_used = 0; + uint32_t lines_per_way = 0; + uint32_t total_cache_lines = 0; + uint32_t bytes_in_mall = 0; + uint32_t num_mblks = 0; + uint32_t cache_lines_per_plane = 0; + uint32_t i = 0; + uint32_t mblk_width = 0; + uint32_t mblk_height = 0; + uint32_t full_vp_width_blk_aligned = 0; + uint32_t mall_alloc_width_blk_aligned = 0; + uint32_t mall_alloc_height_blk_aligned = 0; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + + // Find the phantom pipes + if (pipe->stream && pipe->plane_state && !pipe->top_pipe && !pipe->prev_odm_pipe && + ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + bytes_per_pixel = pipe->plane_state->format >= SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616 ? 8 : 4; + mblk_width = ctx->config.mall_cfg.mblk_width_pixels; + mblk_height = bytes_per_pixel == 4 ? mblk_width = ctx->config.mall_cfg.mblk_height_4bpe_pixels : ctx->config.mall_cfg.mblk_height_8bpe_pixels; + + /* full_vp_width_blk_aligned = FLOOR(vp_x_start + full_vp_width + blk_width - 1, blk_width) - + * FLOOR(vp_x_start, blk_width) + */ + full_vp_width_blk_aligned = ((pipe->plane_res.scl_data.viewport.x + + pipe->plane_res.scl_data.viewport.width + mblk_width - 1) / mblk_width * mblk_width) + + (pipe->plane_res.scl_data.viewport.x / mblk_width * mblk_width); + + /* mall_alloc_width_blk_aligned_l/c = full_vp_width_blk_aligned_l/c */ + mall_alloc_width_blk_aligned = full_vp_width_blk_aligned; + + /* mall_alloc_height_blk_aligned_l/c = CEILING(sub_vp_height_l/c - 1, blk_height_l/c) + blk_height_l/c */ + mall_alloc_height_blk_aligned = (pipe->stream->timing.v_addressable - 1 + mblk_height - 1) / + mblk_height * mblk_height + mblk_height; + + /* full_mblk_width_ub_l/c = malldml2_mall_phantom.c_alloc_width_blk_aligned_l/c; + * full_mblk_height_ub_l/c = mall_alloc_height_blk_aligned_l/c; + * num_mblk_l/c = (full_mblk_width_ub_l/c / mblk_width_l/c) * (full_mblk_height_ub_l/c / mblk_height_l/c); + * (Should be divisible, but round up if not) + */ + num_mblks = ((mall_alloc_width_blk_aligned + mblk_width - 1) / mblk_width) * + ((mall_alloc_height_blk_aligned + mblk_height - 1) / mblk_height); + bytes_in_mall = num_mblks * ctx->config.mall_cfg.mblk_size_bytes; + // cache lines used is total bytes / cache_line size. Add +2 for worst case alignment + // (MALL is 64-byte aligned) + cache_lines_per_plane = bytes_in_mall / ctx->config.mall_cfg.cache_line_size_bytes + 2; + + // For DCC we must cache the meat surface, so double cache lines required + if (pipe->plane_state->dcc.enable) + cache_lines_per_plane *= 2; + cache_lines_used += cache_lines_per_plane; + } + } + + total_cache_lines = ctx->config.mall_cfg.max_cab_allocation_bytes / ctx->config.mall_cfg.cache_line_size_bytes; + lines_per_way = total_cache_lines / ctx->config.mall_cfg.cache_num_ways; + num_ways = cache_lines_used / lines_per_way; + if (cache_lines_used % lines_per_way > 0) + num_ways++; + + return num_ways; +} + +static void merge_pipes_for_subvp(struct dml2_context *ctx, struct dc_state *context) +{ + int i; + + /* merge pipes if necessary */ + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + + // For now merge all pipes for SubVP since pipe split case isn't supported yet + + /* if ODM merge we ignore mpc tree, mpo pipes will have their own flags */ + if (pipe->prev_odm_pipe) { + /*split off odm pipe*/ + pipe->prev_odm_pipe->next_odm_pipe = pipe->next_odm_pipe; + if (pipe->next_odm_pipe) + pipe->next_odm_pipe->prev_odm_pipe = pipe->prev_odm_pipe; + + pipe->bottom_pipe = NULL; + pipe->next_odm_pipe = NULL; + pipe->plane_state = NULL; + pipe->stream = NULL; + pipe->top_pipe = NULL; + pipe->prev_odm_pipe = NULL; + if (pipe->stream_res.dsc) + ctx->config.svp_pstate.callbacks.release_dsc(&context->res_ctx, ctx->config.svp_pstate.callbacks.dc->res_pool, &pipe->stream_res.dsc); + memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); + memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); + } else if (pipe->top_pipe && pipe->top_pipe->plane_state == pipe->plane_state) { + struct pipe_ctx *top_pipe = pipe->top_pipe; + struct pipe_ctx *bottom_pipe = pipe->bottom_pipe; + + top_pipe->bottom_pipe = bottom_pipe; + if (bottom_pipe) + bottom_pipe->top_pipe = top_pipe; + + pipe->top_pipe = NULL; + pipe->bottom_pipe = NULL; + pipe->plane_state = NULL; + pipe->stream = NULL; + memset(&pipe->plane_res, 0, sizeof(pipe->plane_res)); + memset(&pipe->stream_res, 0, sizeof(pipe->stream_res)); + } + } +} + +static bool all_pipes_have_stream_and_plane(struct dml2_context *ctx, const struct dc_state *context) +{ + int i; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + + if (!pipe->stream) + continue; + + if (!pipe->plane_state) + return false; + } + return true; +} + +static bool mpo_in_use(const struct dc_state *context) +{ + int i; + + for (i = 0; i < context->stream_count; i++) { + if (context->stream_status[i].plane_count > 1) + return true; + } + return false; +} + +/* + * dcn32_get_num_free_pipes: Calculate number of free pipes + * + * This function assumes that a "used" pipe is a pipe that has + * both a stream and a plane assigned to it. + * + * @dc: current dc state + * @context: new dc state + * + * Return: + * Number of free pipes available in the context + */ +static unsigned int get_num_free_pipes(struct dml2_context *ctx, struct dc_state *state) +{ + unsigned int i; + unsigned int free_pipes = 0; + unsigned int num_pipes = 0; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + if (pipe->stream && !pipe->top_pipe) { + while (pipe) { + num_pipes++; + pipe = pipe->bottom_pipe; + } + } + } + + free_pipes = ctx->config.dcn_pipe_count - num_pipes; + return free_pipes; +} + +/* + * dcn32_assign_subvp_pipe: Function to decide which pipe will use Sub-VP. + * + * We enter this function if we are Sub-VP capable (i.e. enough pipes available) + * and regular P-State switching (i.e. VACTIVE/VBLANK) is not supported, or if + * we are forcing SubVP P-State switching on the current config. + * + * The number of pipes used for the chosen surface must be less than or equal to the + * number of free pipes available. + * + * In general we choose surfaces with the longest frame time first (better for SubVP + VBLANK). + * For multi-display cases the ActiveDRAMClockChangeMargin doesn't provide enough info on its own + * for determining which should be the SubVP pipe (need a way to determine if a pipe / plane doesn't + * support MCLK switching naturally [i.e. ACTIVE or VBLANK]). + * + * @param dc: current dc state + * @param context: new dc state + * @param index: [out] dc pipe index for the pipe chosen to have phantom pipes assigned + * + * Return: + * True if a valid pipe assignment was found for Sub-VP. Otherwise false. + */ +static bool assign_subvp_pipe(struct dml2_context *ctx, struct dc_state *context, unsigned int *index) +{ + unsigned int i, pipe_idx; + unsigned int max_frame_time = 0; + bool valid_assignment_found = false; + unsigned int free_pipes = 2; //dcn32_get_num_free_pipes(dc, context); + bool current_assignment_freesync = false; + struct vba_vars_st *vba = &context->bw_ctx.dml.vba; + + for (i = 0, pipe_idx = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + unsigned int num_pipes = 0; + unsigned int refresh_rate = 0; + + if (!pipe->stream) + continue; + + // Round up + refresh_rate = (pipe->stream->timing.pix_clk_100hz * 100 + + pipe->stream->timing.v_total * pipe->stream->timing.h_total - 1) + / (double)(pipe->stream->timing.v_total * pipe->stream->timing.h_total); + /* SubVP pipe candidate requirements: + * - Refresh rate < 120hz + * - Not able to switch in vactive naturally (switching in active means the + * DET provides enough buffer to hide the P-State switch latency -- trying + * to combine this with SubVP can cause issues with the scheduling). + */ + if (pipe->plane_state && !pipe->top_pipe && + ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_NONE && refresh_rate < 120 && + vba->ActiveDRAMClockChangeLatencyMarginPerState[vba->VoltageLevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]] <= 0) { + while (pipe) { + num_pipes++; + pipe = pipe->bottom_pipe; + } + + pipe = &context->res_ctx.pipe_ctx[i]; + if (num_pipes <= free_pipes) { + struct dc_stream_state *stream = pipe->stream; + unsigned int frame_us = (stream->timing.v_total * stream->timing.h_total / + (double)(stream->timing.pix_clk_100hz * 100)) * 1000000; + if (frame_us > max_frame_time && !stream->ignore_msa_timing_param) { + *index = i; + max_frame_time = frame_us; + valid_assignment_found = true; + current_assignment_freesync = false; + /* For the 2-Freesync display case, still choose the one with the + * longest frame time + */ + } else if (stream->ignore_msa_timing_param && (!valid_assignment_found || + (current_assignment_freesync && frame_us > max_frame_time))) { + *index = i; + valid_assignment_found = true; + current_assignment_freesync = true; + } + } + } + pipe_idx++; + } + return valid_assignment_found; +} + +/* + * enough_pipes_for_subvp: Function to check if there are "enough" pipes for SubVP. + * + * This function returns true if there are enough free pipes + * to create the required phantom pipes for any given stream + * (that does not already have phantom pipe assigned). + * + * e.g. For a 2 stream config where the first stream uses one + * pipe and the second stream uses 2 pipes (i.e. pipe split), + * this function will return true because there is 1 remaining + * pipe which can be used as the phantom pipe for the non pipe + * split pipe. + * + * @dc: current dc state + * @context: new dc state + * + * Return: + * True if there are enough free pipes to assign phantom pipes to at least one + * stream that does not already have phantom pipes assigned. Otherwise false. + */ +static bool enough_pipes_for_subvp(struct dml2_context *ctx, struct dc_state *state) +{ + unsigned int i, split_cnt, free_pipes; + unsigned int min_pipe_split = ctx->config.dcn_pipe_count + 1; // init as max number of pipes + 1 + bool subvp_possible = false; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + // Find the minimum pipe split count for non SubVP pipes + if (pipe->stream && !pipe->top_pipe && + ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_NONE) { + split_cnt = 0; + while (pipe) { + split_cnt++; + pipe = pipe->bottom_pipe; + } + + if (split_cnt < min_pipe_split) + min_pipe_split = split_cnt; + } + } + + free_pipes = get_num_free_pipes(ctx, state); + + // SubVP only possible if at least one pipe is being used (i.e. free_pipes + // should not equal to the pipe_count) + if (free_pipes >= min_pipe_split && free_pipes < ctx->config.dcn_pipe_count) + subvp_possible = true; + + return subvp_possible; +} + +/* + * subvp_subvp_schedulable: Determine if SubVP + SubVP config is schedulable + * + * High level algorithm: + * 1. Find longest microschedule length (in us) between the two SubVP pipes + * 2. Check if the worst case overlap (VBLANK in middle of ACTIVE) for both + * pipes still allows for the maximum microschedule to fit in the active + * region for both pipes. + * + * @dc: current dc state + * @context: new dc state + * + * Return: + * bool - True if the SubVP + SubVP config is schedulable, false otherwise + */ +static bool subvp_subvp_schedulable(struct dml2_context *ctx, struct dc_state *context) +{ + struct pipe_ctx *subvp_pipes[2]; + struct dc_stream_state *phantom = NULL; + uint32_t microschedule_lines = 0; + uint32_t index = 0; + uint32_t i; + uint32_t max_microschedule_us = 0; + int32_t vactive1_us, vactive2_us, vblank1_us, vblank2_us; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + uint32_t time_us = 0; + + /* Loop to calculate the maximum microschedule time between the two SubVP pipes, + * and also to store the two main SubVP pipe pointers in subvp_pipes[2]. + */ + if (pipe->stream && pipe->plane_state && !pipe->top_pipe && + ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) { + phantom = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, pipe->stream); + microschedule_lines = (phantom->timing.v_total - phantom->timing.v_front_porch) + + phantom->timing.v_addressable; + + // Round up when calculating microschedule time (+ 1 at the end) + time_us = (microschedule_lines * phantom->timing.h_total) / + (double)(phantom->timing.pix_clk_100hz * 100) * 1000000 + + ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us + + ctx->config.svp_pstate.subvp_fw_processing_delay_us + 1; + if (time_us > max_microschedule_us) + max_microschedule_us = time_us; + + subvp_pipes[index] = pipe; + index++; + + // Maximum 2 SubVP pipes + if (index == 2) + break; + } + } + vactive1_us = ((subvp_pipes[0]->stream->timing.v_addressable * subvp_pipes[0]->stream->timing.h_total) / + (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; + vactive2_us = ((subvp_pipes[1]->stream->timing.v_addressable * subvp_pipes[1]->stream->timing.h_total) / + (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; + vblank1_us = (((subvp_pipes[0]->stream->timing.v_total - subvp_pipes[0]->stream->timing.v_addressable) * + subvp_pipes[0]->stream->timing.h_total) / + (double)(subvp_pipes[0]->stream->timing.pix_clk_100hz * 100)) * 1000000; + vblank2_us = (((subvp_pipes[1]->stream->timing.v_total - subvp_pipes[1]->stream->timing.v_addressable) * + subvp_pipes[1]->stream->timing.h_total) / + (double)(subvp_pipes[1]->stream->timing.pix_clk_100hz * 100)) * 1000000; + + if ((vactive1_us - vblank2_us) / 2 > max_microschedule_us && + (vactive2_us - vblank1_us) / 2 > max_microschedule_us) + return true; + + return false; +} + +/* + * dml2_svp_drr_schedulable: Determine if SubVP + DRR config is schedulable + * + * High level algorithm: + * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe + * 2. Determine the frame time for the DRR display when adding required margin for MCLK switching + * (the margin is equal to the MALL region + DRR margin (500us)) + * 3.If (SubVP Active - Prefetch > Stretched DRR frame + max(MALL region, Stretched DRR frame)) + * then report the configuration as supported + * + * @dc: current dc state + * @context: new dc state + * @drr_pipe: DRR pipe_ctx for the SubVP + DRR config + * + * Return: + * bool - True if the SubVP + DRR config is schedulable, false otherwise + */ +bool dml2_svp_drr_schedulable(struct dml2_context *ctx, struct dc_state *context, struct dc_crtc_timing *drr_timing) +{ + bool schedulable = false; + uint32_t i; + struct pipe_ctx *pipe = NULL; + struct dc_crtc_timing *main_timing = NULL; + struct dc_crtc_timing *phantom_timing = NULL; + struct dc_stream_state *phantom_stream; + int16_t prefetch_us = 0; + int16_t mall_region_us = 0; + int16_t drr_frame_us = 0; // nominal frame time + int16_t subvp_active_us = 0; + int16_t stretched_drr_us = 0; + int16_t drr_stretched_vblank_us = 0; + int16_t max_vblank_mallregion = 0; + + // Find SubVP pipe + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + pipe = &context->res_ctx.pipe_ctx[i]; + + // We check for master pipe, but it shouldn't matter since we only need + // the pipe for timing info (stream should be same for any pipe splits) + if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) + continue; + + // Find the SubVP pipe + if (ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) + break; + } + + phantom_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, pipe->stream); + main_timing = &pipe->stream->timing; + phantom_timing = &phantom_stream->timing; + prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / + (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + + ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us; + subvp_active_us = main_timing->v_addressable * main_timing->h_total / + (double)(main_timing->pix_clk_100hz * 100) * 1000000; + drr_frame_us = drr_timing->v_total * drr_timing->h_total / + (double)(drr_timing->pix_clk_100hz * 100) * 1000000; + // P-State allow width and FW delays already included phantom_timing->v_addressable + mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / + (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; + stretched_drr_us = drr_frame_us + mall_region_us + SUBVP_DRR_MARGIN_US; + drr_stretched_vblank_us = (drr_timing->v_total - drr_timing->v_addressable) * drr_timing->h_total / + (double)(drr_timing->pix_clk_100hz * 100) * 1000000 + (stretched_drr_us - drr_frame_us); + max_vblank_mallregion = drr_stretched_vblank_us > mall_region_us ? drr_stretched_vblank_us : mall_region_us; + + /* We consider SubVP + DRR schedulable if the stretched frame duration of the DRR display (i.e. the + * highest refresh rate + margin that can support UCLK P-State switch) passes the static analysis + * for VBLANK: (VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, + * and the max of (VBLANK blanking time, MALL region)). + */ + if (stretched_drr_us < (1 / (double)drr_timing->min_refresh_in_uhz) * 1000000 * 1000000 && + subvp_active_us - prefetch_us - stretched_drr_us - max_vblank_mallregion > 0) + schedulable = true; + + return schedulable; +} + + +/* + * subvp_vblank_schedulable: Determine if SubVP + VBLANK config is schedulable + * + * High level algorithm: + * 1. Get timing for SubVP pipe, phantom pipe, and VBLANK pipe + * 2. If (SubVP Active - Prefetch > Vblank Frame Time + max(MALL region, Vblank blanking time)) + * then report the configuration as supported + * 3. If the VBLANK display is DRR, then take the DRR static schedulability path + * + * @dc: current dc state + * @context: new dc state + * + * Return: + * bool - True if the SubVP + VBLANK/DRR config is schedulable, false otherwise + */ +static bool subvp_vblank_schedulable(struct dml2_context *ctx, struct dc_state *context) +{ + struct pipe_ctx *pipe = NULL; + struct pipe_ctx *subvp_pipe = NULL; + bool found = false; + bool schedulable = false; + uint32_t i = 0; + uint8_t vblank_index = 0; + uint16_t prefetch_us = 0; + uint16_t mall_region_us = 0; + uint16_t vblank_frame_us = 0; + uint16_t subvp_active_us = 0; + uint16_t vblank_blank_us = 0; + uint16_t max_vblank_mallregion = 0; + struct dc_crtc_timing *main_timing = NULL; + struct dc_crtc_timing *phantom_timing = NULL; + struct dc_crtc_timing *vblank_timing = NULL; + struct dc_stream_state *phantom_stream; + enum mall_stream_type pipe_mall_type; + + /* For SubVP + VBLANK/DRR cases, we assume there can only be + * a single VBLANK/DRR display. If DML outputs SubVP + VBLANK + * is supported, it is either a single VBLANK case or two VBLANK + * displays which are synchronized (in which case they have identical + * timings). + */ + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + pipe = &context->res_ctx.pipe_ctx[i]; + pipe_mall_type = ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe); + + // We check for master pipe, but it shouldn't matter since we only need + // the pipe for timing info (stream should be same for any pipe splits) + if (!pipe->stream || !pipe->plane_state || pipe->top_pipe || pipe->prev_odm_pipe) + continue; + + if (!found && pipe_mall_type == SUBVP_NONE) { + // Found pipe which is not SubVP or Phantom (i.e. the VBLANK pipe). + vblank_index = i; + found = true; + } + + if (!subvp_pipe && pipe_mall_type == SUBVP_MAIN) + subvp_pipe = pipe; + } + // Use ignore_msa_timing_param flag to identify as DRR + if (found && context->res_ctx.pipe_ctx[vblank_index].stream->ignore_msa_timing_param) { + // SUBVP + DRR case + schedulable = dml2_svp_drr_schedulable(ctx, context, &context->res_ctx.pipe_ctx[vblank_index].stream->timing); + } else if (found) { + phantom_stream = ctx->config.svp_pstate.callbacks.get_paired_subvp_stream(context, subvp_pipe->stream); + main_timing = &subvp_pipe->stream->timing; + phantom_timing = &phantom_stream->timing; + vblank_timing = &context->res_ctx.pipe_ctx[vblank_index].stream->timing; + // Prefetch time is equal to VACTIVE + BP + VSYNC of the phantom pipe + // Also include the prefetch end to mallstart delay time + prefetch_us = (phantom_timing->v_total - phantom_timing->v_front_porch) * phantom_timing->h_total / + (double)(phantom_timing->pix_clk_100hz * 100) * 1000000 + + ctx->config.svp_pstate.subvp_prefetch_end_to_mall_start_us; + // P-State allow width and FW delays already included phantom_timing->v_addressable + mall_region_us = phantom_timing->v_addressable * phantom_timing->h_total / + (double)(phantom_timing->pix_clk_100hz * 100) * 1000000; + vblank_frame_us = vblank_timing->v_total * vblank_timing->h_total / + (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; + vblank_blank_us = (vblank_timing->v_total - vblank_timing->v_addressable) * vblank_timing->h_total / + (double)(vblank_timing->pix_clk_100hz * 100) * 1000000; + subvp_active_us = main_timing->v_addressable * main_timing->h_total / + (double)(main_timing->pix_clk_100hz * 100) * 1000000; + max_vblank_mallregion = vblank_blank_us > mall_region_us ? vblank_blank_us : mall_region_us; + + // Schedulable if VACTIVE region of the SubVP pipe can fit the MALL prefetch, VBLANK frame time, + // and the max of (VBLANK blanking time, MALL region) + // TODO: Possibly add some margin (i.e. the below conditions should be [...] > X instead of [...] > 0) + if (subvp_active_us - prefetch_us - vblank_frame_us - max_vblank_mallregion > 0) + schedulable = true; + } + return schedulable; +} + +/* + * subvp_validate_static_schedulability: Check which SubVP case is calculated and handle + * static analysis based on the case. + * + * Three cases: + * 1. SubVP + SubVP + * 2. SubVP + VBLANK (DRR checked internally) + * 3. SubVP + VACTIVE (currently unsupported) + * + * @dc: current dc state + * @context: new dc state + * @vlevel: Voltage level calculated by DML + * + * Return: + * bool - True if statically schedulable, false otherwise + */ +bool dml2_svp_validate_static_schedulability(struct dml2_context *ctx, struct dc_state *context, enum dml_dram_clock_change_support pstate_change_type) +{ + bool schedulable = true; // true by default for single display case + struct vba_vars_st *vba = &context->bw_ctx.dml.vba; + uint32_t i, pipe_idx; + uint8_t subvp_count = 0; + uint8_t vactive_count = 0; + + for (i = 0, pipe_idx = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + enum mall_stream_type pipe_mall_type = ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe); + + if (!pipe->stream) + continue; + + if (pipe->plane_state && !pipe->top_pipe && + pipe_mall_type == SUBVP_MAIN) + subvp_count++; + + // Count how many planes that aren't SubVP/phantom are capable of VACTIVE + // switching (SubVP + VACTIVE unsupported). In situations where we force + // SubVP for a VACTIVE plane, we don't want to increment the vactive_count. + if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] > 0 && + pipe_mall_type == SUBVP_NONE) { + vactive_count++; + } + pipe_idx++; + } + + if (subvp_count == 2) { + // Static schedulability check for SubVP + SubVP case + schedulable = subvp_subvp_schedulable(ctx, context); + } else if (pstate_change_type == dml_dram_clock_change_vblank_w_mall_sub_vp) { + // Static schedulability check for SubVP + VBLANK case. Also handle the case where + // DML outputs SubVP + VBLANK + VACTIVE (DML will report as SubVP + VBLANK) + if (vactive_count > 0) + schedulable = false; + else + schedulable = subvp_vblank_schedulable(ctx, context); + } else if (pstate_change_type == dml_dram_clock_change_vactive_w_mall_sub_vp && + vactive_count > 0) { + // For single display SubVP cases, DML will output dm_dram_clock_change_vactive_w_mall_sub_vp by default. + // We tell the difference between SubVP vs. SubVP + VACTIVE by checking the vactive_count. + // SubVP + VACTIVE currently unsupported + schedulable = false; + } + return schedulable; +} + +static void set_phantom_stream_timing(struct dml2_context *ctx, struct dc_state *state, + struct pipe_ctx *ref_pipe, + struct dc_stream_state *phantom_stream, + unsigned int dc_pipe_idx, + unsigned int svp_height, + unsigned int svp_vstartup) +{ + unsigned int i; + double line_time, fp_and_sync_width_time; + struct pipe_ctx *pipe; + uint32_t phantom_vactive, phantom_bp, pstate_width_fw_delay_lines; + static const double cvt_rb_vblank_max = ((double) 460 / (1000 * 1000)); + + // Find DML pipe index (pipe_idx) using dc_pipe_idx + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + pipe = &state->res_ctx.pipe_ctx[i]; + + if (!pipe->stream) + continue; + + if (i == dc_pipe_idx) + break; + } + + // Calculate lines required for pstate allow width and FW processing delays + pstate_width_fw_delay_lines = ((double)(ctx->config.svp_pstate.subvp_fw_processing_delay_us + + ctx->config.svp_pstate.subvp_pstate_allow_width_us) / 1000000) * + (ref_pipe->stream->timing.pix_clk_100hz * 100) / + (double)ref_pipe->stream->timing.h_total; + + // DML calculation for MALL region doesn't take into account FW delay + // and required pstate allow width for multi-display cases + /* Add 16 lines margin to the MALL REGION because SUB_VP_START_LINE must be aligned + * to 2 swaths (i.e. 16 lines) + */ + phantom_vactive = svp_height + pstate_width_fw_delay_lines + ctx->config.svp_pstate.subvp_swath_height_margin_lines; + + phantom_stream->timing.v_front_porch = 1; + + line_time = phantom_stream->timing.h_total / ((double)phantom_stream->timing.pix_clk_100hz * 100); + fp_and_sync_width_time = (phantom_stream->timing.v_front_porch + phantom_stream->timing.v_sync_width) * line_time; + + if ((svp_vstartup * line_time) + fp_and_sync_width_time > cvt_rb_vblank_max) { + svp_vstartup = (cvt_rb_vblank_max - fp_and_sync_width_time) / line_time; + } + + // For backporch of phantom pipe, use vstartup of the main pipe + phantom_bp = svp_vstartup; + + phantom_stream->dst.y = 0; + phantom_stream->dst.height = phantom_vactive; + phantom_stream->src.y = 0; + phantom_stream->src.height = phantom_vactive; + + phantom_stream->timing.v_addressable = phantom_vactive; + + phantom_stream->timing.v_total = phantom_stream->timing.v_addressable + + phantom_stream->timing.v_front_porch + + phantom_stream->timing.v_sync_width + + phantom_bp; + phantom_stream->timing.flags.DSC = 0; // Don't need DSC for phantom timing +} + +static struct dc_stream_state *enable_phantom_stream(struct dml2_context *ctx, struct dc_state *state, unsigned int dc_pipe_idx, unsigned int svp_height, unsigned int vstartup) +{ + struct pipe_ctx *ref_pipe = &state->res_ctx.pipe_ctx[dc_pipe_idx]; + struct dc_stream_state *phantom_stream = ctx->config.svp_pstate.callbacks.create_phantom_stream( + ctx->config.svp_pstate.callbacks.dc, + state, + ref_pipe->stream); + + /* stream has limited viewport and small timing */ + memcpy(&phantom_stream->timing, &ref_pipe->stream->timing, sizeof(phantom_stream->timing)); + memcpy(&phantom_stream->src, &ref_pipe->stream->src, sizeof(phantom_stream->src)); + memcpy(&phantom_stream->dst, &ref_pipe->stream->dst, sizeof(phantom_stream->dst)); + set_phantom_stream_timing(ctx, state, ref_pipe, phantom_stream, dc_pipe_idx, svp_height, vstartup); + + ctx->config.svp_pstate.callbacks.add_phantom_stream(ctx->config.svp_pstate.callbacks.dc, + state, + phantom_stream, + ref_pipe->stream); + return phantom_stream; +} + +static void enable_phantom_plane(struct dml2_context *ctx, + struct dc_state *state, + struct dc_stream_state *phantom_stream, + unsigned int dc_pipe_idx) +{ + struct dc_plane_state *phantom_plane = NULL; + struct dc_plane_state *prev_phantom_plane = NULL; + struct pipe_ctx *curr_pipe = &state->res_ctx.pipe_ctx[dc_pipe_idx]; + + while (curr_pipe) { + if (curr_pipe->top_pipe && curr_pipe->top_pipe->plane_state == curr_pipe->plane_state) { + phantom_plane = prev_phantom_plane; + } else { + phantom_plane = ctx->config.svp_pstate.callbacks.create_phantom_plane( + ctx->config.svp_pstate.callbacks.dc, + state, + curr_pipe->plane_state); + if (!phantom_plane) + return; + } + + memcpy(&phantom_plane->address, &curr_pipe->plane_state->address, sizeof(phantom_plane->address)); + memcpy(&phantom_plane->scaling_quality, &curr_pipe->plane_state->scaling_quality, + sizeof(phantom_plane->scaling_quality)); + memcpy(&phantom_plane->src_rect, &curr_pipe->plane_state->src_rect, sizeof(phantom_plane->src_rect)); + memcpy(&phantom_plane->dst_rect, &curr_pipe->plane_state->dst_rect, sizeof(phantom_plane->dst_rect)); + memcpy(&phantom_plane->clip_rect, &curr_pipe->plane_state->clip_rect, sizeof(phantom_plane->clip_rect)); + memcpy(&phantom_plane->plane_size, &curr_pipe->plane_state->plane_size, + sizeof(phantom_plane->plane_size)); + memcpy(&phantom_plane->tiling_info, &curr_pipe->plane_state->tiling_info, + sizeof(phantom_plane->tiling_info)); + memcpy(&phantom_plane->dcc, &curr_pipe->plane_state->dcc, sizeof(phantom_plane->dcc)); + //phantom_plane->tiling_info.gfx10compatible.compat_level = curr_pipe->plane_state->tiling_info.gfx10compatible.compat_level; + phantom_plane->format = curr_pipe->plane_state->format; + phantom_plane->rotation = curr_pipe->plane_state->rotation; + phantom_plane->visible = curr_pipe->plane_state->visible; + + /* Shadow pipe has small viewport. */ + phantom_plane->clip_rect.y = 0; + phantom_plane->clip_rect.height = phantom_stream->timing.v_addressable; + + ctx->config.svp_pstate.callbacks.add_phantom_plane(ctx->config.svp_pstate.callbacks.dc, phantom_stream, phantom_plane, state); + + curr_pipe = curr_pipe->bottom_pipe; + prev_phantom_plane = phantom_plane; + } +} + +static void add_phantom_pipes_for_main_pipe(struct dml2_context *ctx, struct dc_state *state, unsigned int main_pipe_idx, unsigned int svp_height, unsigned int vstartup) +{ + struct dc_stream_state *phantom_stream = NULL; + unsigned int i; + + // The index of the DC pipe passed into this function is guarenteed to + // be a valid candidate for SubVP (i.e. has a plane, stream, doesn't + // already have phantom pipe assigned, etc.) by previous checks. + phantom_stream = enable_phantom_stream(ctx, state, main_pipe_idx, svp_height, vstartup); + enable_phantom_plane(ctx, state, phantom_stream, main_pipe_idx); + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + // Build scaling params for phantom pipes which were newly added. + // We determine which phantom pipes were added by comparing with + // the phantom stream. + if (pipe->plane_state && pipe->stream && pipe->stream == phantom_stream && + ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_PHANTOM) { + pipe->stream->use_dynamic_meta = false; + pipe->plane_state->flip_immediate = false; + if (!ctx->config.svp_pstate.callbacks.build_scaling_params(pipe)) { + // Log / remove phantom pipes since failed to build scaling params + } + } + } +} + +static bool remove_all_phantom_planes_for_stream(struct dml2_context *ctx, struct dc_stream_state *stream, struct dc_state *context) +{ + int i, old_plane_count; + struct dc_stream_status *stream_status = NULL; + struct dc_plane_state *del_planes[MAX_SURFACES] = { 0 }; + + for (i = 0; i < context->stream_count; i++) + if (context->streams[i] == stream) { + stream_status = &context->stream_status[i]; + break; + } + + if (stream_status == NULL) { + return false; + } + + old_plane_count = stream_status->plane_count; + + for (i = 0; i < old_plane_count; i++) + del_planes[i] = stream_status->plane_states[i]; + + for (i = 0; i < old_plane_count; i++) { + if (!ctx->config.svp_pstate.callbacks.remove_phantom_plane(ctx->config.svp_pstate.callbacks.dc, stream, del_planes[i], context)) + return false; + ctx->config.svp_pstate.callbacks.release_phantom_plane(ctx->config.svp_pstate.callbacks.dc, context, del_planes[i]); + } + + return true; +} + +bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state *state) +{ + int i; + bool removed_pipe = false; + struct dc_stream_state *phantom_stream = NULL; + + for (i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + // build scaling params for phantom pipes + if (pipe->plane_state && pipe->stream && ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(state, pipe) == SUBVP_PHANTOM) { + phantom_stream = pipe->stream; + + remove_all_phantom_planes_for_stream(ctx, phantom_stream, state); + ctx->config.svp_pstate.callbacks.remove_phantom_stream(ctx->config.svp_pstate.callbacks.dc, state, phantom_stream); + ctx->config.svp_pstate.callbacks.release_phantom_stream(ctx->config.svp_pstate.callbacks.dc, state, phantom_stream); + + removed_pipe = true; + } + + if (pipe->plane_state) { + pipe->plane_state->is_phantom = false; + } + } + return removed_pipe; +} + + +/* Conditions for setting up phantom pipes for SubVP: + * 1. Not force disable SubVP + * 2. Full update (i.e. DC_VALIDATE_MODE_AND_PROGRAMMING) + * 3. Enough pipes are available to support SubVP (TODO: Which pipes will use VACTIVE / VBLANK / SUBVP?) + * 4. Display configuration passes validation + * 5. (Config doesn't support MCLK in VACTIVE/VBLANK || dc->debug.force_subvp_mclk_switch) + */ +bool dml2_svp_add_phantom_pipe_to_dc_state(struct dml2_context *ctx, struct dc_state *state, struct dml_mode_support_info_st *mode_support_info) +{ + unsigned int dc_pipe_idx, dml_pipe_idx; + unsigned int svp_height, vstartup; + + if (ctx->config.svp_pstate.force_disable_subvp) + return false; + + if (!all_pipes_have_stream_and_plane(ctx, state)) + return false; + + if (mpo_in_use(state)) + return false; + + merge_pipes_for_subvp(ctx, state); + // to re-initialize viewport after the pipe merge + for (int i = 0; i < ctx->config.dcn_pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &state->res_ctx.pipe_ctx[i]; + + if (!pipe_ctx->plane_state || !pipe_ctx->stream) + continue; + + ctx->config.svp_pstate.callbacks.build_scaling_params(pipe_ctx); + } + + if (enough_pipes_for_subvp(ctx, state) && assign_subvp_pipe(ctx, state, &dc_pipe_idx)) { + dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(ctx, state->res_ctx.pipe_ctx[dc_pipe_idx].stream->stream_id); + svp_height = mode_support_info->SubViewportLinesNeededInMALL[dml_pipe_idx]; + vstartup = dml_get_vstartup_calculated(&ctx->v20.dml_core_ctx, dml_pipe_idx); + + add_phantom_pipes_for_main_pipe(ctx, state, dc_pipe_idx, svp_height, vstartup); + + return true; + } + + return false; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h new file mode 100644 index 000000000000..9d64851f54e7 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h @@ -0,0 +1,52 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML2_MALL_PHANTOM_H__ +#define __DML2_MALL_PHANTOM_H__ + +#include "dml2_dc_types.h" +#include "display_mode_core_structs.h" + +struct dml2_svp_helper_select_best_svp_candidate_params { + const struct dml_display_cfg_st *dml_config; + const struct dml_mode_support_info_st *mode_support_info; + const unsigned int blacklist; + unsigned int *candidate_index; +}; + +struct dml2_context; + +unsigned int dml2_helper_calculate_num_ways_for_subvp(struct dml2_context *ctx, struct dc_state *context); + +bool dml2_svp_add_phantom_pipe_to_dc_state(struct dml2_context *ctx, struct dc_state *state, struct dml_mode_support_info_st *mode_support_info); + +bool dml2_svp_remove_all_phantom_pipes(struct dml2_context *ctx, struct dc_state *state); + +bool dml2_svp_validate_static_schedulability(struct dml2_context *ctx, struct dc_state *context, enum dml_dram_clock_change_support pstate_change_type); + +bool dml2_svp_drr_schedulable(struct dml2_context *ctx, struct dc_state *context, struct dc_crtc_timing *drr_timing); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c new file mode 100644 index 000000000000..ef693f608d59 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c @@ -0,0 +1,311 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "dml2_policy.h" + +static void get_optimal_ntuple( + const struct soc_bounding_box_st *socbb, + struct soc_state_bounding_box_st *entry) +{ + if (entry->dcfclk_mhz > 0) { + float bw_on_sdp = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); + + entry->fabricclk_mhz = bw_on_sdp / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); + entry->dram_speed_mts = bw_on_sdp / (socbb->num_chans * + socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); + } else if (entry->fabricclk_mhz > 0) { + float bw_on_fabric = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); + + entry->dcfclk_mhz = bw_on_fabric / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); + entry->dram_speed_mts = bw_on_fabric / (socbb->num_chans * + socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); + } else if (entry->dram_speed_mts > 0) { + float bw_on_dram = (float)(entry->dram_speed_mts * socbb->num_chans * + socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); + + entry->fabricclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); + entry->dcfclk_mhz = bw_on_dram / (socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); + } +} + +static float calculate_net_bw_in_mbytes_sec(const struct soc_bounding_box_st *socbb, + struct soc_state_bounding_box_st *entry) +{ + float memory_bw_mbytes_sec = (float)(entry->dram_speed_mts * socbb->num_chans * + socbb->dram_channel_width_bytes * ((float)socbb->pct_ideal_dram_bw_after_urgent_pixel_only / 100)); + + float fabric_bw_mbytes_sec = (float)(entry->fabricclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_fabric_bw_after_urgent / 100)); + + float sdp_bw_mbytes_sec = (float)(entry->dcfclk_mhz * socbb->return_bus_width_bytes * ((float)socbb->pct_ideal_sdp_bw_after_urgent / 100)); + + float limiting_bw_mbytes_sec = memory_bw_mbytes_sec; + + if (fabric_bw_mbytes_sec < limiting_bw_mbytes_sec) + limiting_bw_mbytes_sec = fabric_bw_mbytes_sec; + + if (sdp_bw_mbytes_sec < limiting_bw_mbytes_sec) + limiting_bw_mbytes_sec = sdp_bw_mbytes_sec; + + return limiting_bw_mbytes_sec; +} + +static void insert_entry_into_table_sorted(const struct soc_bounding_box_st *socbb, + struct soc_states_st *table, + struct soc_state_bounding_box_st *entry) +{ + int index = 0; + int i = 0; + float net_bw_of_new_state = 0; + + get_optimal_ntuple(socbb, entry); + + if (table->num_states == 0) { + index = 0; + } else { + net_bw_of_new_state = calculate_net_bw_in_mbytes_sec(socbb, entry); + while (net_bw_of_new_state > calculate_net_bw_in_mbytes_sec(socbb, &table->state_array[index])) { + index++; + if (index >= (int) table->num_states) + break; + } + + for (i = table->num_states; i > index; i--) { + table->state_array[i] = table->state_array[i - 1]; + } + //ASSERT(index < MAX_CLK_TABLE_SIZE); + } + + table->state_array[index] = *entry; + table->state_array[index].dcfclk_mhz = (int)entry->dcfclk_mhz; + table->state_array[index].fabricclk_mhz = (int)entry->fabricclk_mhz; + table->state_array[index].dram_speed_mts = (int)entry->dram_speed_mts; + table->num_states++; +} + +static void remove_entry_from_table_at_index(struct soc_states_st *table, + unsigned int index) +{ + int i; + + if (table->num_states == 0) + return; + + for (i = index; i < (int) table->num_states - 1; i++) { + table->state_array[i] = table->state_array[i + 1]; + } + memset(&table->state_array[--table->num_states], 0, sizeof(struct soc_state_bounding_box_st)); +} + +int dml2_policy_build_synthetic_soc_states(struct dml2_policy_build_synthetic_soc_states_scratch *s, + struct dml2_policy_build_synthetic_soc_states_params *p) +{ + int i, j; + unsigned int min_fclk_mhz = p->in_states->state_array[0].fabricclk_mhz; + unsigned int min_dcfclk_mhz = p->in_states->state_array[0].dcfclk_mhz; + unsigned int min_socclk_mhz = p->in_states->state_array[0].socclk_mhz; + + int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, + max_phyclk_mhz = 0, max_dtbclk_mhz = 0, max_fclk_mhz = 0, + max_uclk_mhz = 0, max_socclk_mhz = 0; + + int num_uclk_dpms = 0, num_fclk_dpms = 0; + + for (i = 0; i < __DML_MAX_STATE_ARRAY_SIZE__; i++) { + if (p->in_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz) + max_dcfclk_mhz = (int) p->in_states->state_array[i].dcfclk_mhz; + if (p->in_states->state_array[i].fabricclk_mhz > max_fclk_mhz) + max_fclk_mhz = (int) p->in_states->state_array[i].fabricclk_mhz; + if (p->in_states->state_array[i].socclk_mhz > max_socclk_mhz) + max_socclk_mhz = (int) p->in_states->state_array[i].socclk_mhz; + if (p->in_states->state_array[i].dram_speed_mts > max_uclk_mhz) + max_uclk_mhz = (int) p->in_states->state_array[i].dram_speed_mts; + if (p->in_states->state_array[i].dispclk_mhz > max_dispclk_mhz) + max_dispclk_mhz = (int) p->in_states->state_array[i].dispclk_mhz; + if (p->in_states->state_array[i].dppclk_mhz > max_dppclk_mhz) + max_dppclk_mhz = (int) p->in_states->state_array[i].dppclk_mhz; + if (p->in_states->state_array[i].phyclk_mhz > max_phyclk_mhz) + max_phyclk_mhz = (int)p->in_states->state_array[i].phyclk_mhz; + if (p->in_states->state_array[i].dtbclk_mhz > max_dtbclk_mhz) + max_dtbclk_mhz = (int)p->in_states->state_array[i].dtbclk_mhz; + + if (p->in_states->state_array[i].fabricclk_mhz > 0) + num_fclk_dpms++; + if (p->in_states->state_array[i].dram_speed_mts > 0) + num_uclk_dpms++; + } + + if (!max_dcfclk_mhz || !max_dispclk_mhz || !max_dppclk_mhz || !max_phyclk_mhz || !max_dtbclk_mhz) + return -1; + + p->out_states->num_states = 0; + + s->entry = p->in_states->state_array[0]; + + s->entry.dispclk_mhz = max_dispclk_mhz; + s->entry.dppclk_mhz = max_dppclk_mhz; + s->entry.dtbclk_mhz = max_dtbclk_mhz; + s->entry.phyclk_mhz = max_phyclk_mhz; + + s->entry.dscclk_mhz = max_dispclk_mhz / 3; + s->entry.phyclk_mhz = max_phyclk_mhz; + s->entry.dtbclk_mhz = max_dtbclk_mhz; + + // Insert all the DCFCLK STAs first + for (i = 0; i < p->num_dcfclk_stas; i++) { + s->entry.dcfclk_mhz = p->dcfclk_stas_mhz[i]; + s->entry.fabricclk_mhz = 0; + s->entry.dram_speed_mts = 0; + if (i > 0) + s->entry.socclk_mhz = max_socclk_mhz; + + insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); + } + + // Insert the UCLK DPMS + for (i = 0; i < num_uclk_dpms; i++) { + s->entry.dcfclk_mhz = 0; + s->entry.fabricclk_mhz = 0; + s->entry.dram_speed_mts = p->in_states->state_array[i].dram_speed_mts; + if (i == 0) { + s->entry.socclk_mhz = min_socclk_mhz; + } else { + s->entry.socclk_mhz = max_socclk_mhz; + } + + insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); + } + + // Insert FCLK DPMs (if present) + if (num_fclk_dpms > 2) { + for (i = 0; i < num_fclk_dpms; i++) { + s->entry.dcfclk_mhz = 0; + s->entry.fabricclk_mhz = p->in_states->state_array[i].fabricclk_mhz; + s->entry.dram_speed_mts = 0; + + insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); + } + } + // Add max FCLK + else { + s->entry.dcfclk_mhz = 0; + s->entry.fabricclk_mhz = p->in_states->state_array[num_fclk_dpms - 1].fabricclk_mhz; + s->entry.dram_speed_mts = 0; + + insert_entry_into_table_sorted(p->in_bbox, p->out_states, &s->entry); + } + + // Remove states that require higher clocks than are supported + for (i = p->out_states->num_states - 1; i >= 0; i--) { + if (p->out_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz || + p->out_states->state_array[i].fabricclk_mhz > max_fclk_mhz || + p->out_states->state_array[i].dram_speed_mts > max_uclk_mhz) + remove_entry_from_table_at_index(p->out_states, i); + } + + // At this point, the table contains all "points of interest" based on + // DPMs from PMFW, and STAs. Table is sorted by BW, and all clock + // ratios (by derate, are exact). + + // Round up UCLK to DPMs + for (i = p->out_states->num_states - 1; i >= 0; i--) { + for (j = 0; j < num_uclk_dpms; j++) { + if (p->in_states->state_array[j].dram_speed_mts >= p->out_states->state_array[i].dram_speed_mts) { + p->out_states->state_array[i].dram_speed_mts = p->in_states->state_array[j].dram_speed_mts; + break; + } + } + } + + // If FCLK is coarse grained, round up to next DPMs + if (num_fclk_dpms > 2) { + for (i = p->out_states->num_states - 1; i >= 0; i--) { + for (j = 0; j < num_fclk_dpms; j++) { + if (p->in_states->state_array[j].fabricclk_mhz >= p->out_states->state_array[i].fabricclk_mhz) { + p->out_states->state_array[i].fabricclk_mhz = p->in_states->state_array[j].fabricclk_mhz; + break; + } + } + } + } + + // Clamp to min FCLK/DCFCLK + for (i = p->out_states->num_states - 1; i >= 0; i--) { + if (p->out_states->state_array[i].fabricclk_mhz < min_fclk_mhz) { + p->out_states->state_array[i].fabricclk_mhz = min_fclk_mhz; + } + if (p->out_states->state_array[i].dcfclk_mhz < min_dcfclk_mhz) { + p->out_states->state_array[i].dcfclk_mhz = min_dcfclk_mhz; + } + } + + // Remove duplicate states, note duplicate states are always neighbouring since table is sorted. + i = 0; + while (i < (int) p->out_states->num_states - 1) { + if (p->out_states->state_array[i].dcfclk_mhz == p->out_states->state_array[i + 1].dcfclk_mhz && + p->out_states->state_array[i].fabricclk_mhz == p->out_states->state_array[i + 1].fabricclk_mhz && + p->out_states->state_array[i].dram_speed_mts == p->out_states->state_array[i + 1].dram_speed_mts) + remove_entry_from_table_at_index(p->out_states, i); + else + i++; + } + + return 0; +} + +void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_mode_eval_policy_st *policy) +{ + for (int i = 0; i < __DML_NUM_PLANES__; i++) { + policy->MPCCombineUse[i] = dml_mpc_as_needed_for_voltage; // TOREVIEW: Is this still needed? When is MPCC useful for pstate given CRB? + policy->ODMUse[i] = dml_odm_use_policy_combine_as_needed; + policy->ImmediateFlipRequirement[i] = dml_immediate_flip_required; + policy->AllowForPStateChangeOrStutterInVBlank[i] = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; + } + + /* Change the default policy initializations as per spreadsheet. We might need to + * review and change them later on as per Jun's earlier comments. + */ + policy->UseUnboundedRequesting = dml_unbounded_requesting_enable; + policy->UseMinimumRequiredDCFCLK = false; + policy->DRAMClockChangeRequirementFinal = true; // TOREVIEW: What does this mean? + policy->FCLKChangeRequirementFinal = true; // TOREVIEW: What does this mean? + policy->USRRetrainingRequiredFinal = true; + policy->EnhancedPrefetchScheduleAccelerationFinal = true; // TOREVIEW: What does this mean? + policy->NomDETInKByteOverrideEnable = false; + policy->NomDETInKByteOverrideValue = 0; + policy->DCCProgrammingAssumesScanDirectionUnknownFinal = true; + policy->SynchronizeTimingsFinal = true; + policy->SynchronizeDRRDisplaysForUCLKPStateChangeFinal = true; + policy->AssumeModeSupportAtMaxPwrStateEvenDRAMClockChangeNotSupported = true; // TOREVIEW: What does this mean? + policy->AssumeModeSupportAtMaxPwrStateEvenFClockChangeNotSupported = true; // TOREVIEW: What does this mean? + if (project == dml_project_dcn35 || + project == dml_project_dcn36 || + project == dml_project_dcn351) { + policy->DCCProgrammingAssumesScanDirectionUnknownFinal = false; + policy->EnhancedPrefetchScheduleAccelerationFinal = 0; + policy->AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; /*new*/ + policy->UseOnlyMaxPrefetchModes = 1; + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h new file mode 100644 index 000000000000..e83e05248592 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h @@ -0,0 +1,47 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __DML2_POLICY_H__ +#define __DML2_POLICY_H__ + +#include "display_mode_core_structs.h" + +struct dml2_policy_build_synthetic_soc_states_params { + const struct soc_bounding_box_st *in_bbox; + struct soc_states_st *in_states; + struct soc_states_st *out_states; + int *dcfclk_stas_mhz; + int num_dcfclk_stas; +}; + +struct dml2_policy_build_synthetic_soc_states_scratch { + struct soc_state_bounding_box_st entry; +}; + +int dml2_policy_build_synthetic_soc_states(struct dml2_policy_build_synthetic_soc_states_scratch *s, + struct dml2_policy_build_synthetic_soc_states_params *p); + +void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_mode_eval_policy_st *policy); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c new file mode 100644 index 000000000000..d834cb595afa --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c @@ -0,0 +1,1528 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "display_mode_core.h" +#include "dml2_internal_types.h" +#include "dml2_translation_helper.h" + +#define NUM_DCFCLK_STAS 5 +#define NUM_DCFCLK_STAS_NEW 8 + +void dml2_init_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out) +{ + switch (dml2->v20.dml_core_ctx.project) { + case dml_project_dcn32: + case dml_project_dcn321: + default: + // Hardcoded values for DCN32x + out->vblank_nom_default_us = 600; + out->rob_buffer_size_kbytes = 128; + out->config_return_buffer_size_in_kbytes = 1280; + out->config_return_buffer_segment_size_in_kbytes = 64; + out->compressed_buffer_segment_size_in_kbytes = 64; + out->meta_fifo_size_in_kentries = 22; + out->zero_size_buffer_entries = 512; + out->dpte_buffer_size_in_pte_reqs_luma = 68; + out->dpte_buffer_size_in_pte_reqs_chroma = 36; + out->dcc_meta_buffer_size_bytes = 6272; + out->gpuvm_max_page_table_levels = 4; + out->hostvm_max_page_table_levels = 0; + out->pixel_chunk_size_kbytes = 8; + //out->alpha_pixel_chunk_size_kbytes; + out->min_pixel_chunk_size_bytes = 1024; + out->meta_chunk_size_kbytes = 2; + out->min_meta_chunk_size_bytes = 256; + out->writeback_chunk_size_kbytes = 8; + out->line_buffer_size_bits = 1171920; + out->max_line_buffer_lines = 32; + out->writeback_interface_buffer_size_kbytes = 90; + //Number of pipes after DCN Pipe harvesting + out->max_num_dpp = dml2->config.dcn_pipe_count; + out->max_num_otg = dml2->config.dcn_pipe_count; + out->max_num_wb = 1; + out->max_dchub_pscl_bw_pix_per_clk = 4; + out->max_pscl_lb_bw_pix_per_clk = 2; + out->max_lb_vscl_bw_pix_per_clk = 4; + out->max_vscl_hscl_bw_pix_per_clk = 4; + out->max_hscl_ratio = 6; + out->max_vscl_ratio = 6; + out->max_hscl_taps = 8; + out->max_vscl_taps = 8; + out->dispclk_ramp_margin_percent = 1; + out->dppclk_delay_subtotal = 47; + out->dppclk_delay_scl = 50; + out->dppclk_delay_scl_lb_only = 16; + out->dppclk_delay_cnvc_formatter = 28; + out->dppclk_delay_cnvc_cursor = 6; + out->cursor_buffer_size = 16; + out->cursor_chunk_size = 2; + out->dispclk_delay_subtotal = 125; + out->max_inter_dcn_tile_repeaters = 8; + out->writeback_max_hscl_ratio = 1; + out->writeback_max_vscl_ratio = 1; + out->writeback_min_hscl_ratio = 1; + out->writeback_min_vscl_ratio = 1; + out->writeback_max_hscl_taps = 1; + out->writeback_max_vscl_taps = 1; + out->writeback_line_buffer_buffer_size = 0; + out->num_dsc = 4; + out->maximum_dsc_bits_per_component = 12; + out->maximum_pixels_per_line_per_dsc_unit = 6016; + out->dsc422_native_support = true; + out->dcc_supported = true; + out->ptoi_supported = false; + + out->gpuvm_enable = false; + out->hostvm_enable = false; + out->cursor_64bpp_support = false; + out->dynamic_metadata_vm_enabled = false; + + out->max_num_hdmi_frl_outputs = 1; + out->max_num_dp2p0_outputs = 2; + out->max_num_dp2p0_streams = 4; + break; + + case dml_project_dcn35: + case dml_project_dcn351: + case dml_project_dcn36: + out->rob_buffer_size_kbytes = 64; + out->config_return_buffer_size_in_kbytes = 1792; + out->compressed_buffer_segment_size_in_kbytes = 64; + out->meta_fifo_size_in_kentries = 32; + out->zero_size_buffer_entries = 512; + out->pixel_chunk_size_kbytes = 8; + out->alpha_pixel_chunk_size_kbytes = 4; + out->min_pixel_chunk_size_bytes = 1024; + out->meta_chunk_size_kbytes = 2; + out->min_meta_chunk_size_bytes = 256; + out->writeback_chunk_size_kbytes = 8; + out->dpte_buffer_size_in_pte_reqs_luma = 68; + out->dpte_buffer_size_in_pte_reqs_chroma = 36; + out->dcc_meta_buffer_size_bytes = 6272; + out->gpuvm_enable = 1; + out->hostvm_enable = 1; + out->gpuvm_max_page_table_levels = 1; + out->hostvm_max_page_table_levels = 2; + out->num_dsc = 4; + out->maximum_dsc_bits_per_component = 12; + out->maximum_pixels_per_line_per_dsc_unit = 6016; + out->dsc422_native_support = 1; + out->line_buffer_size_bits = 986880; + out->dcc_supported = 1; + out->max_line_buffer_lines = 32; + out->writeback_interface_buffer_size_kbytes = 90; + out->max_num_dpp = 4; + out->max_num_otg = 4; + out->max_num_hdmi_frl_outputs = 1; + out->max_num_dp2p0_outputs = 2; + out->max_num_dp2p0_streams = 4; + out->max_num_wb = 1; + + out->max_dchub_pscl_bw_pix_per_clk = 4; + out->max_pscl_lb_bw_pix_per_clk = 2; + out->max_lb_vscl_bw_pix_per_clk = 4; + out->max_vscl_hscl_bw_pix_per_clk = 4; + out->max_hscl_ratio = 6; + out->max_vscl_ratio = 6; + out->max_hscl_taps = 8; + out->max_vscl_taps = 8; + out->dispclk_ramp_margin_percent = 1.11; + + out->dppclk_delay_subtotal = 47; + out->dppclk_delay_scl = 50; + out->dppclk_delay_scl_lb_only = 16; + out->dppclk_delay_cnvc_formatter = 28; + out->dppclk_delay_cnvc_cursor = 6; + out->dispclk_delay_subtotal = 125; + + out->dynamic_metadata_vm_enabled = false; + out->max_inter_dcn_tile_repeaters = 8; + out->cursor_buffer_size = 16; // kBytes + out->cursor_chunk_size = 2; // kBytes + + out->writeback_line_buffer_buffer_size = 0; + out->writeback_max_hscl_ratio = 1; + out->writeback_max_vscl_ratio = 1; + out->writeback_min_hscl_ratio = 1; + out->writeback_min_vscl_ratio = 1; + out->writeback_max_hscl_taps = 1; + out->writeback_max_vscl_taps = 1; + out->ptoi_supported = 0; + + out->vblank_nom_default_us = 668; /*not in dml, but in programming guide, hard coded in dml2_translate_ip_params*/ + out->config_return_buffer_segment_size_in_kbytes = 64; /*required, but not exist,, hard coded in dml2_translate_ip_params*/ + break; + + case dml_project_dcn401: + // Hardcoded values for DCN4m + out->vblank_nom_default_us = 668; //600; + out->rob_buffer_size_kbytes = 192; //128; + out->config_return_buffer_size_in_kbytes = 1344; //1280; + out->config_return_buffer_segment_size_in_kbytes = 64; + out->compressed_buffer_segment_size_in_kbytes = 64; + out->meta_fifo_size_in_kentries = 22; + out->dpte_buffer_size_in_pte_reqs_luma = 68; + out->dpte_buffer_size_in_pte_reqs_chroma = 36; + out->gpuvm_max_page_table_levels = 4; + out->pixel_chunk_size_kbytes = 8; + out->alpha_pixel_chunk_size_kbytes = 4; + out->min_pixel_chunk_size_bytes = 1024; + out->writeback_chunk_size_kbytes = 8; + out->line_buffer_size_bits = 1171920; + out->max_line_buffer_lines = 32; + out->writeback_interface_buffer_size_kbytes = 90; + //Number of pipes after DCN Pipe harvesting + out->max_num_dpp = dml2->config.dcn_pipe_count; + out->max_num_otg = dml2->config.dcn_pipe_count; + out->max_num_wb = 1; + out->max_dchub_pscl_bw_pix_per_clk = 4; + out->max_pscl_lb_bw_pix_per_clk = 2; + out->max_lb_vscl_bw_pix_per_clk = 4; + out->max_vscl_hscl_bw_pix_per_clk = 4; + out->max_hscl_ratio = 6; + out->max_vscl_ratio = 6; + out->max_hscl_taps = 8; + out->max_vscl_taps = 8; + out->dispclk_ramp_margin_percent = 1; + out->dppclk_delay_subtotal = 47; + out->dppclk_delay_scl = 50; + out->dppclk_delay_scl_lb_only = 16; + out->dppclk_delay_cnvc_formatter = 28; + out->dppclk_delay_cnvc_cursor = 6; + out->dispclk_delay_subtotal = 125; + out->cursor_buffer_size = 24; //16 + out->cursor_chunk_size = 2; + out->max_inter_dcn_tile_repeaters = 8; + out->writeback_max_hscl_ratio = 1; + out->writeback_max_vscl_ratio = 1; + out->writeback_min_hscl_ratio = 1; + out->writeback_min_vscl_ratio = 1; + out->writeback_max_hscl_taps = 1; + out->writeback_max_vscl_taps = 1; + out->writeback_line_buffer_buffer_size = 0; + out->num_dsc = 4; + out->maximum_dsc_bits_per_component = 12; + out->maximum_pixels_per_line_per_dsc_unit = 5760; + out->dsc422_native_support = true; + out->dcc_supported = true; + out->ptoi_supported = false; + + out->gpuvm_enable = false; + out->hostvm_enable = false; + out->cursor_64bpp_support = true; //false; + out->dynamic_metadata_vm_enabled = false; + + out->max_num_hdmi_frl_outputs = 1; + out->max_num_dp2p0_outputs = 4; //2; + out->max_num_dp2p0_streams = 4; + break; + } +} + +void dml2_init_socbb_params(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out) +{ + out->dprefclk_mhz = dml2->config.bbox_overrides.dprefclk_mhz; + out->xtalclk_mhz = dml2->config.bbox_overrides.xtalclk_mhz; + out->pcierefclk_mhz = 100; + out->refclk_mhz = dml2->config.bbox_overrides.dchub_refclk_mhz; + + out->max_outstanding_reqs = 512; + out->pct_ideal_sdp_bw_after_urgent = 100; + out->pct_ideal_fabric_bw_after_urgent = 67; + out->pct_ideal_dram_bw_after_urgent_pixel_only = 20; + out->pct_ideal_dram_bw_after_urgent_pixel_and_vm = 60; + out->pct_ideal_dram_bw_after_urgent_vm_only = 30; + out->pct_ideal_dram_bw_after_urgent_strobe = 67; + out->max_avg_sdp_bw_use_normal_percent = 80; + out->max_avg_fabric_bw_use_normal_percent = 60; + out->max_avg_dram_bw_use_normal_percent = 15; + out->max_avg_dram_bw_use_normal_strobe_percent = 50; + + out->urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096; + out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096; + out->urgent_out_of_order_return_per_channel_vm_only_bytes = 4096; + out->return_bus_width_bytes = 64; + out->dram_channel_width_bytes = 2; + out->fabric_datapath_to_dcn_data_return_bytes = 64; + out->hostvm_min_page_size_kbytes = 0; + out->gpuvm_min_page_size_kbytes = 256; + out->phy_downspread_percent = 0.38; + out->dcn_downspread_percent = 0.5; + out->dispclk_dppclk_vco_speed_mhz = dml2->config.bbox_overrides.disp_pll_vco_speed_mhz; + out->mall_allocated_for_dcn_mbytes = dml2->config.mall_cfg.max_cab_allocation_bytes / 1048576; // 64 or 32 MB; + + out->do_urgent_latency_adjustment = true; + + switch (dml2->v20.dml_core_ctx.project) { + + case dml_project_dcn32: + default: + out->num_chans = 24; + out->round_trip_ping_latency_dcfclk_cycles = 263; + out->smn_latency_us = 2; + break; + + case dml_project_dcn321: + out->num_chans = 8; + out->round_trip_ping_latency_dcfclk_cycles = 207; + out->smn_latency_us = 0; + break; + + case dml_project_dcn35: + case dml_project_dcn351: + case dml_project_dcn36: + out->num_chans = 4; + out->round_trip_ping_latency_dcfclk_cycles = 106; + out->smn_latency_us = 2; + out->dispclk_dppclk_vco_speed_mhz = 3600; + out->pct_ideal_dram_bw_after_urgent_pixel_only = 65.0; + break; + + + case dml_project_dcn401: + out->pct_ideal_fabric_bw_after_urgent = 76; //67; + out->max_avg_sdp_bw_use_normal_percent = 75; //80; + out->max_avg_fabric_bw_use_normal_percent = 57; //60; + + out->urgent_out_of_order_return_per_channel_pixel_only_bytes = 0; //4096; + out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 0; //4096; + out->urgent_out_of_order_return_per_channel_vm_only_bytes = 0; //4096; + + out->num_chans = 16; + out->round_trip_ping_latency_dcfclk_cycles = 1000; //263; + out->smn_latency_us = 0; //2 us + out->mall_allocated_for_dcn_mbytes = dml2->config.mall_cfg.max_cab_allocation_bytes / 1048576; // 64; + break; + } + /* ---Overrides if available--- */ + if (dml2->config.bbox_overrides.dram_num_chan) + out->num_chans = dml2->config.bbox_overrides.dram_num_chan; + + if (dml2->config.bbox_overrides.dram_chanel_width_bytes) + out->dram_channel_width_bytes = dml2->config.bbox_overrides.dram_chanel_width_bytes; +} + +void dml2_init_soc_states(struct dml2_context *dml2, const struct dc *in_dc, + const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out) +{ + struct dml2_policy_build_synthetic_soc_states_scratch *s = &dml2->v20.scratch.create_scratch.build_synthetic_socbb_scratch; + struct dml2_policy_build_synthetic_soc_states_params *p = &dml2->v20.scratch.build_synthetic_socbb_params; + unsigned int dcfclk_stas_mhz[NUM_DCFCLK_STAS] = {0}; + unsigned int dcfclk_stas_mhz_new[NUM_DCFCLK_STAS_NEW] = {0}; + unsigned int dml_project = dml2->v20.dml_core_ctx.project; + + unsigned int i = 0; + unsigned int transactions_per_mem_clock = 16; // project specific, depends on used Memory type + + if (dml_project == dml_project_dcn351) { + p->dcfclk_stas_mhz = dcfclk_stas_mhz_new; + p->num_dcfclk_stas = NUM_DCFCLK_STAS_NEW; + } else { + p->dcfclk_stas_mhz = dcfclk_stas_mhz; + p->num_dcfclk_stas = NUM_DCFCLK_STAS; + } + + p->in_bbox = in_bbox; + p->out_states = out; + p->in_states = &dml2->v20.scratch.create_scratch.in_states; + + + /* Initial hardcoded values */ + switch (dml2->v20.dml_core_ctx.project) { + + case dml_project_dcn32: + default: + p->in_states->num_states = 2; + transactions_per_mem_clock = 16; + p->in_states->state_array[0].socclk_mhz = 620.0; + p->in_states->state_array[0].dscclk_mhz = 716.667; + p->in_states->state_array[0].phyclk_mhz = 810; + p->in_states->state_array[0].phyclk_d18_mhz = 667; + p->in_states->state_array[0].phyclk_d32_mhz = 625; + p->in_states->state_array[0].dtbclk_mhz = 1564.0; + p->in_states->state_array[0].fabricclk_mhz = 450.0; + p->in_states->state_array[0].dcfclk_mhz = 300.0; + p->in_states->state_array[0].dispclk_mhz = 2150.0; + p->in_states->state_array[0].dppclk_mhz = 2150.0; + p->in_states->state_array[0].dram_speed_mts = 100 * transactions_per_mem_clock; + + p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; + p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; + p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; + p->in_states->state_array[0].writeback_latency_us = 12; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 3000; + p->in_states->state_array[0].sr_exit_z8_time_us = 0; + p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; + p->in_states->state_array[0].dram_clock_change_latency_us = 400; + p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; + p->in_states->state_array[0].sr_exit_time_us = 42.97; + p->in_states->state_array[0].sr_enter_plus_exit_time_us = 49.94; + p->in_states->state_array[0].fclk_change_latency_us = 20; + p->in_states->state_array[0].usr_retraining_latency_us = 2; + + p->in_states->state_array[1].socclk_mhz = 1200.0; + p->in_states->state_array[1].fabricclk_mhz = 2500.0; + p->in_states->state_array[1].dcfclk_mhz = 1564.0; + p->in_states->state_array[1].dram_speed_mts = 1125 * transactions_per_mem_clock; + break; + + case dml_project_dcn321: + p->in_states->num_states = 2; + transactions_per_mem_clock = 16; + p->in_states->state_array[0].socclk_mhz = 582.0; + p->in_states->state_array[0].dscclk_mhz = 573.333; + p->in_states->state_array[0].phyclk_mhz = 810; + p->in_states->state_array[0].phyclk_d18_mhz = 667; + p->in_states->state_array[0].phyclk_d32_mhz = 313; + p->in_states->state_array[0].dtbclk_mhz = 1564.0; + p->in_states->state_array[0].fabricclk_mhz = 450.0; + p->in_states->state_array[0].dcfclk_mhz = 300.0; + p->in_states->state_array[0].dispclk_mhz = 1720.0; + p->in_states->state_array[0].dppclk_mhz = 1720.0; + p->in_states->state_array[0].dram_speed_mts = 100 * transactions_per_mem_clock; + + p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; + p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; + p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; + p->in_states->state_array[0].writeback_latency_us = 12; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 3000; + p->in_states->state_array[0].sr_exit_z8_time_us = 0; + p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; + p->in_states->state_array[0].dram_clock_change_latency_us = 400; + p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; + p->in_states->state_array[0].sr_exit_time_us = 19.95; + p->in_states->state_array[0].sr_enter_plus_exit_time_us = 24.36; + p->in_states->state_array[0].fclk_change_latency_us = 7; + p->in_states->state_array[0].usr_retraining_latency_us = 0; + + p->in_states->state_array[1].socclk_mhz = 1200.0; + p->in_states->state_array[1].fabricclk_mhz = 2250.0; + p->in_states->state_array[1].dcfclk_mhz = 1434.0; + p->in_states->state_array[1].dram_speed_mts = 1000 * transactions_per_mem_clock; + break; + + + case dml_project_dcn401: + p->in_states->num_states = 2; + transactions_per_mem_clock = 16; + p->in_states->state_array[0].socclk_mhz = 300; //620.0; + p->in_states->state_array[0].dscclk_mhz = 666.667; //716.667; + p->in_states->state_array[0].phyclk_mhz = 810; + p->in_states->state_array[0].phyclk_d18_mhz = 667; + p->in_states->state_array[0].phyclk_d32_mhz = 625; + p->in_states->state_array[0].dtbclk_mhz = 2000; //1564.0; + p->in_states->state_array[0].fabricclk_mhz = 300; //450.0; + p->in_states->state_array[0].dcfclk_mhz = 200; //300.0; + p->in_states->state_array[0].dispclk_mhz = 2000; //2150.0; + p->in_states->state_array[0].dppclk_mhz = 2000; //2150.0; + p->in_states->state_array[0].dram_speed_mts = 97 * transactions_per_mem_clock; //100 * + + p->in_states->state_array[0].urgent_latency_pixel_data_only_us = 4; + p->in_states->state_array[0].urgent_latency_pixel_mixed_with_vm_data_us = 0; + p->in_states->state_array[0].urgent_latency_vm_data_only_us = 0; + p->in_states->state_array[0].writeback_latency_us = 12; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_component_us = 1; + p->in_states->state_array[0].urgent_latency_adjustment_fabric_clock_reference_mhz = 1000; //3000; + p->in_states->state_array[0].sr_exit_z8_time_us = 0; + p->in_states->state_array[0].sr_enter_plus_exit_z8_time_us = 0; + p->in_states->state_array[0].dram_clock_change_latency_us = 400; + p->in_states->state_array[0].use_ideal_dram_bw_strobe = true; + p->in_states->state_array[0].sr_exit_time_us = 15.70; //42.97; + p->in_states->state_array[0].sr_enter_plus_exit_time_us = 20.20; //49.94; + p->in_states->state_array[0].fclk_change_latency_us = 0; //20; + p->in_states->state_array[0].usr_retraining_latency_us = 0; //2; + + p->in_states->state_array[1].socclk_mhz = 1600; //1200.0; + p->in_states->state_array[1].fabricclk_mhz = 2500; //2500.0; + p->in_states->state_array[1].dcfclk_mhz = 1800; //1564.0; + p->in_states->state_array[1].dram_speed_mts = 1125 * transactions_per_mem_clock; + break; + } + + /* Override from passed values, if available */ + for (i = 0; i < p->in_states->num_states; i++) { + if (dml2->config.bbox_overrides.sr_exit_latency_us) { + p->in_states->state_array[i].sr_exit_time_us = + dml2->config.bbox_overrides.sr_exit_latency_us; + } + + if (dml2->config.bbox_overrides.sr_enter_plus_exit_latency_us) { + p->in_states->state_array[i].sr_enter_plus_exit_time_us = + dml2->config.bbox_overrides.sr_enter_plus_exit_latency_us; + } + + if (dml2->config.bbox_overrides.sr_exit_z8_time_us) { + p->in_states->state_array[i].sr_exit_z8_time_us = + dml2->config.bbox_overrides.sr_exit_z8_time_us; + } + + if (dml2->config.bbox_overrides.sr_enter_plus_exit_z8_time_us) { + p->in_states->state_array[i].sr_enter_plus_exit_z8_time_us = + dml2->config.bbox_overrides.sr_enter_plus_exit_z8_time_us; + } + + if (dml2->config.bbox_overrides.urgent_latency_us) { + p->in_states->state_array[i].urgent_latency_pixel_data_only_us = + dml2->config.bbox_overrides.urgent_latency_us; + } + + if (dml2->config.bbox_overrides.dram_clock_change_latency_us) { + p->in_states->state_array[i].dram_clock_change_latency_us = + dml2->config.bbox_overrides.dram_clock_change_latency_us; + } + + if (dml2->config.bbox_overrides.fclk_change_latency_us) { + p->in_states->state_array[i].fclk_change_latency_us = + dml2->config.bbox_overrides.fclk_change_latency_us; + } + } + + /* DCFCLK stas values are project specific */ + if ((dml2->v20.dml_core_ctx.project == dml_project_dcn32) || + (dml2->v20.dml_core_ctx.project == dml_project_dcn321)) { + p->dcfclk_stas_mhz[0] = p->in_states->state_array[0].dcfclk_mhz; + p->dcfclk_stas_mhz[1] = 615; + p->dcfclk_stas_mhz[2] = 906; + p->dcfclk_stas_mhz[3] = 1324; + p->dcfclk_stas_mhz[4] = p->in_states->state_array[1].dcfclk_mhz; + } else if (dml2->v20.dml_core_ctx.project != dml_project_dcn35 && + dml2->v20.dml_core_ctx.project != dml_project_dcn36 && + dml2->v20.dml_core_ctx.project != dml_project_dcn351) { + p->dcfclk_stas_mhz[0] = 300; + p->dcfclk_stas_mhz[1] = 615; + p->dcfclk_stas_mhz[2] = 906; + p->dcfclk_stas_mhz[3] = 1324; + p->dcfclk_stas_mhz[4] = 1500; + } + /* Copy clocks tables entries, if available */ + if (dml2->config.bbox_overrides.clks_table.num_states) { + p->in_states->num_states = dml2->config.bbox_overrides.clks_table.num_states; + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dcfclk_levels; i++) { + p->in_states->state_array[i].dcfclk_mhz = dml2->config.bbox_overrides.clks_table.clk_entries[i].dcfclk_mhz; + } + + p->dcfclk_stas_mhz[0] = dml2->config.bbox_overrides.clks_table.clk_entries[0].dcfclk_mhz; + if (i > 1) + p->dcfclk_stas_mhz[4] = dml2->config.bbox_overrides.clks_table.clk_entries[i-1].dcfclk_mhz; + + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_fclk_levels; i++) { + p->in_states->state_array[i].fabricclk_mhz = + dml2->config.bbox_overrides.clks_table.clk_entries[i].fclk_mhz; + } + + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_memclk_levels; i++) { + p->in_states->state_array[i].dram_speed_mts = + dml2->config.bbox_overrides.clks_table.clk_entries[i].memclk_mhz * transactions_per_mem_clock; + } + + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_socclk_levels; i++) { + p->in_states->state_array[i].socclk_mhz = + dml2->config.bbox_overrides.clks_table.clk_entries[i].socclk_mhz; + } + + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dtbclk_levels; i++) { + if (dml2->config.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz > 0) + p->in_states->state_array[i].dtbclk_mhz = + dml2->config.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz; + } + + for (i = 0; i < dml2->config.bbox_overrides.clks_table.num_entries_per_clk.num_dispclk_levels; i++) { + p->in_states->state_array[i].dispclk_mhz = + dml2->config.bbox_overrides.clks_table.clk_entries[i].dispclk_mhz; + p->in_states->state_array[i].dppclk_mhz = + dml2->config.bbox_overrides.clks_table.clk_entries[i].dppclk_mhz; + } + } + + if (dml2->v20.dml_core_ctx.project == dml_project_dcn35 || + dml2->v20.dml_core_ctx.project == dml_project_dcn36 || + dml2->v20.dml_core_ctx.project == dml_project_dcn351) { + int max_dcfclk_mhz = 0, max_dispclk_mhz = 0, max_dppclk_mhz = 0, max_phyclk_mhz = 0, + max_dtbclk_mhz = 0, max_fclk_mhz = 0, max_uclk_mhz = 0, max_socclk_mhz = 0; + + for (i = 0; i < p->in_states->num_states; i++) { + if (p->in_states->state_array[i].dcfclk_mhz > max_dcfclk_mhz) + max_dcfclk_mhz = (int)p->in_states->state_array[i].dcfclk_mhz; + if (p->in_states->state_array[i].fabricclk_mhz > max_fclk_mhz) + max_fclk_mhz = (int)p->in_states->state_array[i].fabricclk_mhz; + if (p->in_states->state_array[i].socclk_mhz > max_socclk_mhz) + max_socclk_mhz = (int)p->in_states->state_array[i].socclk_mhz; + if (p->in_states->state_array[i].dram_speed_mts > max_uclk_mhz) + max_uclk_mhz = (int)p->in_states->state_array[i].dram_speed_mts; + if (p->in_states->state_array[i].dispclk_mhz > max_dispclk_mhz) + max_dispclk_mhz = (int)p->in_states->state_array[i].dispclk_mhz; + if (p->in_states->state_array[i].dppclk_mhz > max_dppclk_mhz) + max_dppclk_mhz = (int)p->in_states->state_array[i].dppclk_mhz; + if (p->in_states->state_array[i].phyclk_mhz > max_phyclk_mhz) + max_phyclk_mhz = (int)p->in_states->state_array[i].phyclk_mhz; + if (p->in_states->state_array[i].dtbclk_mhz > max_dtbclk_mhz) + max_dtbclk_mhz = (int)p->in_states->state_array[i].dtbclk_mhz; + } + + for (i = 0; i < p->in_states->num_states; i++) { + /* Independent states - including base (unlisted) parameters from state 0. */ + p->out_states->state_array[i] = p->in_states->state_array[0]; + + p->out_states->state_array[i].dispclk_mhz = max_dispclk_mhz; + p->out_states->state_array[i].dppclk_mhz = max_dppclk_mhz; + p->out_states->state_array[i].dtbclk_mhz = max_dtbclk_mhz; + p->out_states->state_array[i].phyclk_mhz = max_phyclk_mhz; + + p->out_states->state_array[i].dscclk_mhz = max_dispclk_mhz / 3.0; + p->out_states->state_array[i].phyclk_mhz = max_phyclk_mhz; + p->out_states->state_array[i].dtbclk_mhz = max_dtbclk_mhz; + + /* Dependent states. */ + p->out_states->state_array[i].dram_speed_mts = p->in_states->state_array[i].dram_speed_mts; + p->out_states->state_array[i].fabricclk_mhz = p->in_states->state_array[i].fabricclk_mhz; + p->out_states->state_array[i].socclk_mhz = p->in_states->state_array[i].socclk_mhz; + p->out_states->state_array[i].dcfclk_mhz = p->in_states->state_array[i].dcfclk_mhz; + } + + p->out_states->num_states = p->in_states->num_states; + } else { + dml2_policy_build_synthetic_soc_states(s, p); + } +} + +void dml2_translate_ip_params(const struct dc *in, struct ip_params_st *out) +{ + const struct _vcs_dpi_ip_params_st *in_ip_params = &in->dml.ip; + /* Copy over the IP params tp dml2_ctx */ + out->compressed_buffer_segment_size_in_kbytes = in_ip_params->compressed_buffer_segment_size_in_kbytes; + out->config_return_buffer_size_in_kbytes = in_ip_params->config_return_buffer_size_in_kbytes; + out->cursor_buffer_size = in_ip_params->cursor_buffer_size; + out->cursor_chunk_size = in_ip_params->cursor_chunk_size; + out->dcc_meta_buffer_size_bytes = in_ip_params->dcc_meta_buffer_size_bytes; + out->dcc_supported = in_ip_params->dcc_supported; + out->dispclk_delay_subtotal = in_ip_params->dispclk_delay_subtotal; + out->dispclk_ramp_margin_percent = in_ip_params->dispclk_ramp_margin_percent; + out->dppclk_delay_cnvc_cursor = in_ip_params->dppclk_delay_cnvc_cursor; + out->dppclk_delay_cnvc_formatter = in_ip_params->dppclk_delay_cnvc_formatter; + out->dppclk_delay_scl = in_ip_params->dppclk_delay_scl; + out->dppclk_delay_scl_lb_only = in_ip_params->dppclk_delay_scl_lb_only; + out->dppclk_delay_subtotal = in_ip_params->dppclk_delay_subtotal; + out->dpte_buffer_size_in_pte_reqs_chroma = in_ip_params->dpte_buffer_size_in_pte_reqs_chroma; + out->dpte_buffer_size_in_pte_reqs_luma = in_ip_params->dpte_buffer_size_in_pte_reqs_luma; + out->dsc422_native_support = in_ip_params->dsc422_native_support; + out->dynamic_metadata_vm_enabled = in_ip_params->dynamic_metadata_vm_enabled; + out->gpuvm_enable = in_ip_params->gpuvm_enable; + out->gpuvm_max_page_table_levels = in_ip_params->gpuvm_max_page_table_levels; + out->hostvm_enable = in_ip_params->hostvm_enable; + out->hostvm_max_page_table_levels = in_ip_params->hostvm_max_page_table_levels; + out->line_buffer_size_bits = in_ip_params->line_buffer_size_bits; + out->maximum_dsc_bits_per_component = in_ip_params->maximum_dsc_bits_per_component; + out->maximum_pixels_per_line_per_dsc_unit = in_ip_params->maximum_pixels_per_line_per_dsc_unit; + out->max_dchub_pscl_bw_pix_per_clk = in_ip_params->max_dchub_pscl_bw_pix_per_clk; + out->max_hscl_ratio = in_ip_params->max_hscl_ratio; + out->max_hscl_taps = in_ip_params->max_hscl_taps; + out->max_inter_dcn_tile_repeaters = in_ip_params->max_inter_dcn_tile_repeaters; + out->max_lb_vscl_bw_pix_per_clk = in_ip_params->max_lb_vscl_bw_pix_per_clk; + out->max_line_buffer_lines = in_ip_params->max_line_buffer_lines; + out->max_num_dp2p0_outputs = in_ip_params->max_num_dp2p0_outputs; + out->max_num_dp2p0_streams = in_ip_params->max_num_dp2p0_streams; + out->max_num_dpp = in_ip_params->max_num_dpp; + out->max_num_hdmi_frl_outputs = in_ip_params->max_num_hdmi_frl_outputs; + out->max_num_otg = in_ip_params->max_num_otg; + out->max_num_wb = in_ip_params->max_num_wb; + out->max_pscl_lb_bw_pix_per_clk = in_ip_params->max_pscl_lb_bw_pix_per_clk; + out->max_vscl_hscl_bw_pix_per_clk = in_ip_params->max_vscl_hscl_bw_pix_per_clk; + out->max_vscl_ratio = in_ip_params->max_vscl_ratio; + out->max_vscl_taps = in_ip_params->max_vscl_taps; + out->meta_chunk_size_kbytes = in_ip_params->meta_chunk_size_kbytes; + out->meta_fifo_size_in_kentries = in_ip_params->meta_fifo_size_in_kentries; + out->min_meta_chunk_size_bytes = in_ip_params->min_meta_chunk_size_bytes; + out->min_pixel_chunk_size_bytes = in_ip_params->min_pixel_chunk_size_bytes; + out->num_dsc = in_ip_params->num_dsc; + out->pixel_chunk_size_kbytes = in_ip_params->pixel_chunk_size_kbytes; + out->ptoi_supported = in_ip_params->ptoi_supported; + out->rob_buffer_size_kbytes = in_ip_params->rob_buffer_size_kbytes; + out->writeback_chunk_size_kbytes = in_ip_params->writeback_chunk_size_kbytes; + out->writeback_interface_buffer_size_kbytes = in_ip_params->writeback_interface_buffer_size_kbytes; + out->writeback_line_buffer_buffer_size = in_ip_params->writeback_line_buffer_buffer_size; + out->writeback_max_hscl_ratio = in_ip_params->writeback_max_hscl_ratio; + out->writeback_max_hscl_taps = in_ip_params->writeback_max_hscl_taps; + out->writeback_max_vscl_ratio = in_ip_params->writeback_max_vscl_ratio; + out->writeback_max_vscl_taps = in_ip_params->writeback_max_vscl_taps; + out->writeback_min_hscl_ratio = in_ip_params->writeback_min_hscl_ratio; + out->writeback_min_vscl_ratio = in_ip_params->writeback_min_vscl_ratio; + out->zero_size_buffer_entries = in_ip_params->zero_size_buffer_entries; + + /* As per hardcoded reference / discussions */ + out->config_return_buffer_segment_size_in_kbytes = 64; + //out->vblank_nom_default_us = 600; + out->vblank_nom_default_us = in_ip_params->VBlankNomDefaultUS; +} + +void dml2_translate_socbb_params(const struct dc *in, struct soc_bounding_box_st *out) +{ + const struct _vcs_dpi_soc_bounding_box_st *in_soc_params = &in->dml.soc; + /* Copy over the SOCBB params to dml2_ctx */ + out->dispclk_dppclk_vco_speed_mhz = in_soc_params->dispclk_dppclk_vco_speed_mhz; + out->do_urgent_latency_adjustment = in_soc_params->do_urgent_latency_adjustment; + out->dram_channel_width_bytes = (dml_uint_t)in_soc_params->dram_channel_width_bytes; + out->fabric_datapath_to_dcn_data_return_bytes = (dml_uint_t)in_soc_params->fabric_datapath_to_dcn_data_return_bytes; + out->gpuvm_min_page_size_kbytes = in_soc_params->gpuvm_min_page_size_bytes / 1024; + out->hostvm_min_page_size_kbytes = in_soc_params->hostvm_min_page_size_bytes / 1024; + out->mall_allocated_for_dcn_mbytes = (dml_uint_t)in_soc_params->mall_allocated_for_dcn_mbytes; + out->max_avg_dram_bw_use_normal_percent = in_soc_params->max_avg_dram_bw_use_normal_percent; + out->max_avg_fabric_bw_use_normal_percent = in_soc_params->max_avg_fabric_bw_use_normal_percent; + out->max_avg_dram_bw_use_normal_strobe_percent = in_soc_params->max_avg_dram_bw_use_normal_strobe_percent; + out->max_avg_sdp_bw_use_normal_percent = in_soc_params->max_avg_sdp_bw_use_normal_percent; + out->max_outstanding_reqs = in_soc_params->max_request_size_bytes; + out->num_chans = in_soc_params->num_chans; + out->pct_ideal_dram_bw_after_urgent_strobe = in_soc_params->pct_ideal_dram_bw_after_urgent_strobe; + out->pct_ideal_dram_bw_after_urgent_vm_only = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_vm_only; + out->pct_ideal_fabric_bw_after_urgent = in_soc_params->pct_ideal_fabric_bw_after_urgent; + out->pct_ideal_sdp_bw_after_urgent = in_soc_params->pct_ideal_sdp_bw_after_urgent; + out->phy_downspread_percent = in_soc_params->downspread_percent; + out->refclk_mhz = 50; // As per hardcoded reference. + out->return_bus_width_bytes = in_soc_params->return_bus_width_bytes; + out->round_trip_ping_latency_dcfclk_cycles = in_soc_params->round_trip_ping_latency_dcfclk_cycles; + out->smn_latency_us = in_soc_params->smn_latency_us; + out->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = in_soc_params->urgent_out_of_order_return_per_channel_pixel_and_vm_bytes; + out->urgent_out_of_order_return_per_channel_pixel_only_bytes = in_soc_params->urgent_out_of_order_return_per_channel_pixel_only_bytes; + out->urgent_out_of_order_return_per_channel_vm_only_bytes = in_soc_params->urgent_out_of_order_return_per_channel_vm_only_bytes; + out->pct_ideal_dram_bw_after_urgent_pixel_and_vm = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm; + out->pct_ideal_dram_bw_after_urgent_pixel_only = in_soc_params->pct_ideal_dram_sdp_bw_after_urgent_pixel_only; + out->dcn_downspread_percent = in_soc_params->dcn_downspread_percent; +} + +void dml2_translate_soc_states(const struct dc *dc, struct soc_states_st *out, int num_states) +{ + unsigned int i = 0; + out->num_states = num_states; + + for (i = 0; i < out->num_states; i++) { + out->state_array[i].dcfclk_mhz = dc->dml.soc.clock_limits[i].dcfclk_mhz; + out->state_array[i].dispclk_mhz = dc->dml.soc.clock_limits[i].dispclk_mhz; + out->state_array[i].dppclk_mhz = dc->dml.soc.clock_limits[i].dppclk_mhz; + out->state_array[i].dram_speed_mts = dc->dml.soc.clock_limits[i].dram_speed_mts; + out->state_array[i].dtbclk_mhz = dc->dml.soc.clock_limits[i].dtbclk_mhz; + out->state_array[i].socclk_mhz = dc->dml.soc.clock_limits[i].socclk_mhz; + out->state_array[i].fabricclk_mhz = dc->dml.soc.clock_limits[i].fabricclk_mhz; + out->state_array[i].dscclk_mhz = dc->dml.soc.clock_limits[i].dscclk_mhz; + out->state_array[i].phyclk_d18_mhz = dc->dml.soc.clock_limits[i].phyclk_d18_mhz; + out->state_array[i].phyclk_d32_mhz = dc->dml.soc.clock_limits[i].phyclk_d32_mhz; + out->state_array[i].phyclk_mhz = dc->dml.soc.clock_limits[i].phyclk_mhz; + out->state_array[i].sr_enter_plus_exit_time_us = dc->dml.soc.sr_enter_plus_exit_time_us; + out->state_array[i].sr_exit_time_us = dc->dml.soc.sr_exit_time_us; + out->state_array[i].fclk_change_latency_us = dc->dml.soc.fclk_change_latency_us; + out->state_array[i].dram_clock_change_latency_us = dc->dml.soc.dram_clock_change_latency_us; + out->state_array[i].usr_retraining_latency_us = dc->dml.soc.usr_retraining_latency_us; + out->state_array[i].writeback_latency_us = dc->dml.soc.writeback_latency_us; + /* Driver initialized values for these are different than the spreadsheet. Use the + * spreadsheet ones for now. We need to decided which ones to use. + */ + out->state_array[i].sr_exit_z8_time_us = dc->dml.soc.sr_exit_z8_time_us; + out->state_array[i].sr_enter_plus_exit_z8_time_us = dc->dml.soc.sr_enter_plus_exit_z8_time_us; + //out->state_array[i].sr_exit_z8_time_us = 5.20; + //out->state_array[i].sr_enter_plus_exit_z8_time_us = 9.60; + out->state_array[i].use_ideal_dram_bw_strobe = true; + out->state_array[i].urgent_latency_pixel_data_only_us = dc->dml.soc.urgent_latency_pixel_data_only_us; + out->state_array[i].urgent_latency_pixel_mixed_with_vm_data_us = dc->dml.soc.urgent_latency_pixel_mixed_with_vm_data_us; + out->state_array[i].urgent_latency_vm_data_only_us = dc->dml.soc.urgent_latency_vm_data_only_us; + out->state_array[i].urgent_latency_adjustment_fabric_clock_component_us = dc->dml.soc.urgent_latency_adjustment_fabric_clock_component_us; + out->state_array[i].urgent_latency_adjustment_fabric_clock_reference_mhz = dc->dml.soc.urgent_latency_adjustment_fabric_clock_reference_mhz; + } +} + +static void populate_dml_timing_cfg_from_stream_state(struct dml_timing_cfg_st *out, unsigned int location, const struct dc_stream_state *in) +{ + dml_uint_t hblank_start, vblank_start; + + out->HActive[location] = in->timing.h_addressable + in->timing.h_border_left + in->timing.h_border_right; + out->VActive[location] = in->timing.v_addressable + in->timing.v_border_bottom + in->timing.v_border_top; + out->RefreshRate[location] = ((in->timing.pix_clk_100hz * 100) / in->timing.h_total) / in->timing.v_total; + out->VFrontPorch[location] = in->timing.v_front_porch; + out->PixelClock[location] = in->timing.pix_clk_100hz / 10000.00; + if (in->timing.timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING) + out->PixelClock[location] *= 2; + out->HTotal[location] = in->timing.h_total; + out->VTotal[location] = in->timing.v_total; + out->Interlace[location] = in->timing.flags.INTERLACE; + hblank_start = in->timing.h_total - in->timing.h_front_porch; + out->HBlankEnd[location] = hblank_start + - in->timing.h_addressable + - in->timing.h_border_left + - in->timing.h_border_right; + vblank_start = in->timing.v_total - in->timing.v_front_porch; + out->VBlankEnd[location] = vblank_start + - in->timing.v_addressable + - in->timing.v_border_top + - in->timing.v_border_bottom; + out->DRRDisplay[location] = false; +} + +static void populate_dml_output_cfg_from_stream_state(struct dml_output_cfg_st *out, unsigned int location, + const struct dc_stream_state *in, const struct pipe_ctx *pipe, struct dml2_context *dml2) +{ + unsigned int output_bpc; + + out->DSCEnable[location] = (enum dml_dsc_enable)in->timing.flags.DSC; + out->OutputLinkDPLanes[location] = 4; // As per code in dcn20_resource.c + out->DSCInputBitPerComponent[location] = 12; // As per code in dcn20_resource.c + out->DSCSlices[location] = in->timing.dsc_cfg.num_slices_h; + + switch (in->signal) { + case SIGNAL_TYPE_DISPLAY_PORT_MST: + case SIGNAL_TYPE_DISPLAY_PORT: + out->OutputEncoder[location] = dml_dp; + if (location < MAX_HPO_DP2_ENCODERS && dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[location] != -1) + out->OutputEncoder[dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[location]] = dml_dp2p0; + break; + case SIGNAL_TYPE_EDP: + out->OutputEncoder[location] = dml_edp; + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + out->OutputEncoder[location] = dml_hdmi; + break; + default: + out->OutputEncoder[location] = dml_dp; + } + + switch (in->timing.display_color_depth) { + case COLOR_DEPTH_666: + output_bpc = 6; + break; + case COLOR_DEPTH_888: + output_bpc = 8; + break; + case COLOR_DEPTH_101010: + output_bpc = 10; + break; + case COLOR_DEPTH_121212: + output_bpc = 12; + break; + case COLOR_DEPTH_141414: + output_bpc = 14; + break; + case COLOR_DEPTH_161616: + output_bpc = 16; + break; + case COLOR_DEPTH_999: + output_bpc = 9; + break; + case COLOR_DEPTH_111111: + output_bpc = 11; + break; + default: + output_bpc = 8; + break; + } + + switch (in->timing.pixel_encoding) { + case PIXEL_ENCODING_RGB: + case PIXEL_ENCODING_YCBCR444: + out->OutputFormat[location] = dml_444; + out->OutputBpp[location] = (dml_float_t)output_bpc * 3; + break; + case PIXEL_ENCODING_YCBCR420: + out->OutputFormat[location] = dml_420; + out->OutputBpp[location] = (output_bpc * 3.0) / 2; + break; + case PIXEL_ENCODING_YCBCR422: + if (in->timing.flags.DSC && !in->timing.dsc_cfg.ycbcr422_simple) + out->OutputFormat[location] = dml_n422; + else + out->OutputFormat[location] = dml_s422; + out->OutputBpp[location] = (dml_float_t)output_bpc * 2; + break; + default: + out->OutputFormat[location] = dml_444; + out->OutputBpp[location] = (dml_float_t)output_bpc * 3; + break; + } + + if (in->timing.flags.DSC) { + out->OutputBpp[location] = in->timing.dsc_cfg.bits_per_pixel / 16.0; + } + + // This has been false throughout DCN32x development. If needed we can change this later on. + out->OutputMultistreamEn[location] = false; + + switch (in->signal) { + case SIGNAL_TYPE_NONE: + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + case SIGNAL_TYPE_HDMI_TYPE_A: + case SIGNAL_TYPE_LVDS: + case SIGNAL_TYPE_RGB: + case SIGNAL_TYPE_DISPLAY_PORT: + case SIGNAL_TYPE_DISPLAY_PORT_MST: + case SIGNAL_TYPE_EDP: + case SIGNAL_TYPE_VIRTUAL: + default: + out->OutputLinkDPRate[location] = dml_dp_rate_na; + break; + } + + out->PixelClockBackEnd[location] = in->timing.pix_clk_100hz / 10000.00; + + out->AudioSampleLayout[location] = in->audio_info.modes->sample_size; + out->AudioSampleRate[location] = in->audio_info.modes->max_bit_rate; + + out->OutputDisabled[location] = true; +} + +static void populate_dummy_dml_surface_cfg(struct dml_surface_cfg_st *out, unsigned int location, const struct dc_stream_state *in) +{ + out->SurfaceWidthY[location] = in->timing.h_addressable; + out->SurfaceHeightY[location] = in->timing.v_addressable; + out->SurfaceWidthC[location] = in->timing.h_addressable; + out->SurfaceHeightC[location] = in->timing.v_addressable; + out->PitchY[location] = ((out->SurfaceWidthY[location] + 127) / 128) * 128; + out->PitchC[location] = 1; + out->DCCEnable[location] = false; + out->DCCMetaPitchY[location] = 0; + out->DCCMetaPitchC[location] = 0; + out->DCCRateLuma[location] = 1.0; + out->DCCRateChroma[location] = 1.0; + out->DCCFractionOfZeroSizeRequestsLuma[location] = 0; + out->DCCFractionOfZeroSizeRequestsChroma[location] = 0; + out->SurfaceTiling[location] = dml_sw_64kb_r_x; + out->SourcePixelFormat[location] = dml_444_32; +} + +static void populate_dml_surface_cfg_from_plane_state(enum dml_project_id dml2_project, struct dml_surface_cfg_st *out, unsigned int location, const struct dc_plane_state *in) +{ + out->PitchY[location] = in->plane_size.surface_pitch; + out->SurfaceHeightY[location] = in->plane_size.surface_size.height; + out->SurfaceWidthY[location] = in->plane_size.surface_size.width; + out->SurfaceHeightC[location] = in->plane_size.chroma_size.height; + out->SurfaceWidthC[location] = in->plane_size.chroma_size.width; + out->PitchC[location] = in->plane_size.chroma_pitch; + out->DCCEnable[location] = in->dcc.enable; + out->DCCMetaPitchY[location] = in->dcc.meta_pitch; + out->DCCMetaPitchC[location] = in->dcc.meta_pitch_c; + out->DCCRateLuma[location] = 1.0; + out->DCCRateChroma[location] = 1.0; + out->DCCFractionOfZeroSizeRequestsLuma[location] = in->dcc.independent_64b_blks; + out->DCCFractionOfZeroSizeRequestsChroma[location] = in->dcc.independent_64b_blks_c; + + switch (dml2_project) { + default: + out->SurfaceTiling[location] = (enum dml_swizzle_mode)in->tiling_info.gfx9.swizzle; + break; + case dml_project_dcn401: + // Temporary use gfx11 swizzle in dml, until proper dml for DCN4x is integrated/implemented + switch (in->tiling_info.gfx_addr3.swizzle) { + case DC_ADDR3_SW_4KB_2D: + case DC_ADDR3_SW_64KB_2D: + case DC_ADDR3_SW_256KB_2D: + default: + out->SurfaceTiling[location] = dml_sw_64kb_r_x; + break; + case DC_ADDR3_SW_LINEAR: + out->SurfaceTiling[location] = dml_sw_linear; + break; + } + } + + switch (in->format) { + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCbCr: + case SURFACE_PIXEL_FORMAT_VIDEO_420_YCrCb: + out->SourcePixelFormat[location] = dml_420_8; + break; + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCbCr: + case SURFACE_PIXEL_FORMAT_VIDEO_420_10bpc_YCrCb: + out->SourcePixelFormat[location] = dml_420_10; + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616: + case SURFACE_PIXEL_FORMAT_GRPH_ARGB16161616F: + case SURFACE_PIXEL_FORMAT_GRPH_ABGR16161616F: + out->SourcePixelFormat[location] = dml_444_64; + break; + case SURFACE_PIXEL_FORMAT_GRPH_ARGB1555: + case SURFACE_PIXEL_FORMAT_GRPH_RGB565: + out->SourcePixelFormat[location] = dml_444_16; + break; + case SURFACE_PIXEL_FORMAT_GRPH_PALETA_256_COLORS: + out->SourcePixelFormat[location] = dml_444_8; + break; + case SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA: + out->SourcePixelFormat[location] = dml_rgbe_alpha; + break; + default: + out->SourcePixelFormat[location] = dml_444_32; + break; + } +} + +static struct scaler_data *get_scaler_data_for_plane( + const struct dc_plane_state *in, + struct dc_state *context) +{ + int i; + struct pipe_ctx *temp_pipe = &context->res_ctx.temp_pipe; + + memset(temp_pipe, 0, sizeof(struct pipe_ctx)); + + for (i = 0; i < MAX_PIPES; i++) { + const struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state == in && !pipe->prev_odm_pipe) { + temp_pipe->stream = pipe->stream; + temp_pipe->plane_state = pipe->plane_state; + temp_pipe->plane_res.scl_data.taps = pipe->plane_res.scl_data.taps; + temp_pipe->stream_res = pipe->stream_res; + resource_build_scaling_params(temp_pipe); + break; + } + } + + ASSERT(i < MAX_PIPES); + return &temp_pipe->plane_res.scl_data; +} + +static void populate_dummy_dml_plane_cfg(struct dml_plane_cfg_st *out, unsigned int location, + const struct dc_stream_state *in, + const struct soc_bounding_box_st *soc) +{ + dml_uint_t width, height; + + if (in->timing.h_addressable > 3840) + width = 3840; + else + width = in->timing.h_addressable; // 4K max + + if (in->timing.v_addressable > 2160) + height = 2160; + else + height = in->timing.v_addressable; // 4K max + + out->CursorBPP[location] = dml_cur_32bit; + out->CursorWidth[location] = 256; + + out->GPUVMMinPageSizeKBytes[location] = soc->gpuvm_min_page_size_kbytes; + + out->ViewportWidth[location] = width; + out->ViewportHeight[location] = height; + out->ViewportStationary[location] = false; + out->ViewportWidthChroma[location] = 0; + out->ViewportHeightChroma[location] = 0; + out->ViewportXStart[location] = 0; + out->ViewportXStartC[location] = 0; + out->ViewportYStart[location] = 0; + out->ViewportYStartC[location] = 0; + + out->ScalerEnabled[location] = false; + out->HRatio[location] = 1.0; + out->VRatio[location] = 1.0; + out->HRatioChroma[location] = 0; + out->VRatioChroma[location] = 0; + out->HTaps[location] = 1; + out->VTaps[location] = 1; + out->HTapsChroma[location] = 0; + out->VTapsChroma[location] = 0; + out->SourceScan[location] = dml_rotation_0; + out->ScalerRecoutWidth[location] = width; + + out->LBBitPerPixel[location] = 57; + + out->DynamicMetadataEnable[location] = false; + + out->NumberOfCursors[location] = 1; + out->UseMALLForStaticScreen[location] = dml_use_mall_static_screen_disable; + out->UseMALLForPStateChange[location] = dml_use_mall_pstate_change_disable; + + out->DETSizeOverride[location] = 256; + + out->ScalerEnabled[location] = false; +} + +static void populate_dml_plane_cfg_from_plane_state(struct dml_plane_cfg_st *out, unsigned int location, + const struct dc_plane_state *in, struct dc_state *context, + const struct soc_bounding_box_st *soc) +{ + struct scaler_data *scaler_data = get_scaler_data_for_plane(in, context); + + out->CursorBPP[location] = dml_cur_32bit; + out->CursorWidth[location] = 256; + + out->GPUVMMinPageSizeKBytes[location] = soc->gpuvm_min_page_size_kbytes; + + out->ViewportWidth[location] = scaler_data->viewport.width; + out->ViewportHeight[location] = scaler_data->viewport.height; + out->ViewportWidthChroma[location] = scaler_data->viewport_c.width; + out->ViewportHeightChroma[location] = scaler_data->viewport_c.height; + out->ViewportXStart[location] = scaler_data->viewport.x; + out->ViewportYStart[location] = scaler_data->viewport.y; + out->ViewportXStartC[location] = scaler_data->viewport_c.x; + out->ViewportYStartC[location] = scaler_data->viewport_c.y; + out->ViewportStationary[location] = false; + + out->ScalerEnabled[location] = scaler_data->ratios.horz.value != dc_fixpt_one.value || + scaler_data->ratios.horz_c.value != dc_fixpt_one.value || + scaler_data->ratios.vert.value != dc_fixpt_one.value || + scaler_data->ratios.vert_c.value != dc_fixpt_one.value; + + /* Current driver code base uses LBBitPerPixel as 57. There is a discrepancy + * from the HW/DML teams about this value. Initialize LBBitPerPixel with the + * value current used in Navi3x . + */ + + out->LBBitPerPixel[location] = 57; + + if (out->ScalerEnabled[location] == false) { + out->HRatio[location] = 1; + out->HRatioChroma[location] = 1; + out->VRatio[location] = 1; + out->VRatioChroma[location] = 1; + } else { + /* Follow the original dml_wrapper.c code direction to fix scaling issues */ + out->HRatio[location] = (dml_float_t)scaler_data->ratios.horz.value / (1ULL << 32); + out->HRatioChroma[location] = (dml_float_t)scaler_data->ratios.horz_c.value / (1ULL << 32); + out->VRatio[location] = (dml_float_t)scaler_data->ratios.vert.value / (1ULL << 32); + out->VRatioChroma[location] = (dml_float_t)scaler_data->ratios.vert_c.value / (1ULL << 32); + } + + if (!scaler_data->taps.h_taps) { + out->HTaps[location] = 1; + out->HTapsChroma[location] = 1; + } else { + out->HTaps[location] = scaler_data->taps.h_taps; + out->HTapsChroma[location] = scaler_data->taps.h_taps_c; + } + if (!scaler_data->taps.v_taps) { + out->VTaps[location] = 1; + out->VTapsChroma[location] = 1; + } else { + out->VTaps[location] = scaler_data->taps.v_taps; + out->VTapsChroma[location] = scaler_data->taps.v_taps_c; + } + + out->SourceScan[location] = (enum dml_rotation_angle)in->rotation; + out->ScalerRecoutWidth[location] = in->dst_rect.width; + + out->DynamicMetadataEnable[location] = false; + out->DynamicMetadataLinesBeforeActiveRequired[location] = 0; + out->DynamicMetadataTransmittedBytes[location] = 0; + + out->NumberOfCursors[location] = 1; +} + +static unsigned int map_stream_to_dml_display_cfg(const struct dml2_context *dml2, + const struct dc_stream_state *stream, const struct dml_display_cfg_st *dml_dispcfg) +{ + int i = 0; + int location = -1; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] && dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i] == stream->stream_id) { + location = i; + break; + } + } + + return location; +} + +static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *context, const struct dc_plane_state *plane, + unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) +{ + int i, j; + bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; + + if (!plane_id) + return false; + + for (i = 0; i < context->stream_count; i++) { + if (context->streams[i]->stream_id == stream_id) { + for (j = 0; j < context->stream_status[i].plane_count; j++) { + if (context->stream_status[i].plane_states[j] == plane && + (!is_plane_duplicate || (j == plane_index))) { + *plane_id = (i << 16) | j; + return true; + } + } + } + } + + return false; +} + +static unsigned int map_plane_to_dml_display_cfg(const struct dml2_context *dml2, const struct dc_plane_state *plane, + const struct dc_state *context, const struct dml_display_cfg_st *dml_dispcfg, unsigned int stream_id, int plane_index) +{ + unsigned int plane_id; + int i = 0; + int location = -1; + + if (!get_plane_id(context->bw_ctx.dml2, context, plane, stream_id, plane_index, &plane_id)) { + ASSERT(false); + return -1; + } + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] && dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i] == plane_id) { + location = i; + break; + } + } + + return location; +} + +static void dml2_populate_pipe_to_plane_index_mapping(struct dml2_context *dml2, struct dc_state *state) +{ + unsigned int i; + unsigned int pipe_index = 0; + unsigned int plane_index = 0; + struct dml2_dml_to_dc_pipe_mapping *dml_to_dc_pipe_mapping = &dml2->v20.scratch.dml_to_dc_pipe_mapping; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[i] = false; + dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index[i] = 0; + } + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + struct pipe_ctx *pipe = &state->res_ctx.pipe_ctx[i]; + + if (!pipe || !pipe->stream || !pipe->plane_state) + continue; + + while (pipe) { + pipe_index = pipe->pipe_idx; + + if (pipe->stream && dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[pipe_index] == false) { + dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index[pipe_index] = plane_index; + plane_index++; + dml_to_dc_pipe_mapping->dml_pipe_idx_to_plane_index_valid[pipe_index] = true; + } + + pipe = pipe->bottom_pipe; + } + + plane_index = 0; + } +} + +static void populate_dml_writeback_cfg_from_stream_state(struct dml_writeback_cfg_st *out, + unsigned int location, const struct dc_stream_state *in) +{ + if (in->num_wb_info > 0) { + for (int i = 0; i < __DML_NUM_DMB__; i++) { + const struct dc_writeback_info *wb_info = &in->writeback_info[i]; + /*current dml support 1 dwb per stream, limitation*/ + if (wb_info->wb_enabled) { + out->WritebackEnable[location] = wb_info->wb_enabled; + out->ActiveWritebacksPerSurface[location] = wb_info->dwb_params.cnv_params.src_width; + out->WritebackDestinationWidth[location] = wb_info->dwb_params.dest_width; + out->WritebackDestinationHeight[location] = wb_info->dwb_params.dest_height; + + out->WritebackSourceWidth[location] = wb_info->dwb_params.cnv_params.crop_en ? + wb_info->dwb_params.cnv_params.crop_width : + wb_info->dwb_params.cnv_params.src_width; + + out->WritebackSourceHeight[location] = wb_info->dwb_params.cnv_params.crop_en ? + wb_info->dwb_params.cnv_params.crop_height : + wb_info->dwb_params.cnv_params.src_height; + /*current design does not have chroma scaling, need to follow up*/ + out->WritebackHTaps[location] = wb_info->dwb_params.scaler_taps.h_taps > 0 ? + wb_info->dwb_params.scaler_taps.h_taps : 1; + out->WritebackVTaps[location] = wb_info->dwb_params.scaler_taps.v_taps > 0 ? + wb_info->dwb_params.scaler_taps.v_taps : 1; + + out->WritebackHRatio[location] = wb_info->dwb_params.cnv_params.crop_en ? + (double)wb_info->dwb_params.cnv_params.crop_width / + (double)wb_info->dwb_params.dest_width : + (double)wb_info->dwb_params.cnv_params.src_width / + (double)wb_info->dwb_params.dest_width; + out->WritebackVRatio[location] = wb_info->dwb_params.cnv_params.crop_en ? + (double)wb_info->dwb_params.cnv_params.crop_height / + (double)wb_info->dwb_params.dest_height : + (double)wb_info->dwb_params.cnv_params.src_height / + (double)wb_info->dwb_params.dest_height; + } + } + } +} + +static void dml2_map_hpo_stream_encoder_to_hpo_link_encoder_index(struct dml2_context *dml2, struct dc_state *context) +{ + int i; + struct pipe_ctx *current_pipe_context; + + /* Scratch gets reset to zero in dml, but link encoder instance can be zero, so reset to -1 */ + for (i = 0; i < MAX_HPO_DP2_ENCODERS; i++) { + dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[i] = -1; + } + + /* If an HPO stream encoder is allocated to a pipe, get the instance of it's allocated HPO Link encoder */ + for (i = 0; i < MAX_PIPES; i++) { + current_pipe_context = &context->res_ctx.pipe_ctx[i]; + if (current_pipe_context->stream && + current_pipe_context->stream_res.hpo_dp_stream_enc && + current_pipe_context->link_res.hpo_dp_link_enc && + dc_is_dp_signal(current_pipe_context->stream->signal)) { + dml2->v20.scratch.hpo_stream_to_link_encoder_mapping[current_pipe_context->stream_res.hpo_dp_stream_enc->inst] = + current_pipe_context->link_res.hpo_dp_link_enc->inst; + } + } +} + +void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_state *context, struct dml_display_cfg_st *dml_dispcfg) +{ + int i = 0, j = 0, k = 0; + int disp_cfg_stream_location, disp_cfg_plane_location; + enum mall_stream_type stream_mall_type; + struct pipe_ctx *current_pipe_context; + + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[i] = false; + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[i] = false; + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] = false; + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] = false; + } + + //Generally these are set by referencing our latest BB/IP params in dcn32_resource.c file + dml_dispcfg->plane.GPUVMEnable = dml2->v20.dml_core_ctx.ip.gpuvm_enable; + dml_dispcfg->plane.GPUVMMaxPageTableLevels = dml2->v20.dml_core_ctx.ip.gpuvm_max_page_table_levels; + dml_dispcfg->plane.HostVMEnable = dml2->v20.dml_core_ctx.ip.hostvm_enable; + dml_dispcfg->plane.HostVMMaxPageTableLevels = dml2->v20.dml_core_ctx.ip.hostvm_max_page_table_levels; + if (dml2->v20.dml_core_ctx.ip.hostvm_enable) + dml2->v20.dml_core_ctx.policy.AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter; + + dml2_populate_pipe_to_plane_index_mapping(dml2, context); + dml2_map_hpo_stream_encoder_to_hpo_link_encoder_index(dml2, context); + + for (i = 0; i < context->stream_count; i++) { + current_pipe_context = NULL; + for (k = 0; k < MAX_PIPES; k++) { + /* find one pipe allocated to this stream for the purpose of getting + info about the link later */ + if (context->streams[i] == context->res_ctx.pipe_ctx[k].stream) { + current_pipe_context = &context->res_ctx.pipe_ctx[k]; + break; + } + } + disp_cfg_stream_location = map_stream_to_dml_display_cfg(dml2, context->streams[i], dml_dispcfg); + stream_mall_type = dc_state_get_stream_subvp_type(context, context->streams[i]); + + if (disp_cfg_stream_location < 0) + disp_cfg_stream_location = dml_dispcfg->num_timings++; + + ASSERT(disp_cfg_stream_location >= 0 && disp_cfg_stream_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); + + populate_dml_timing_cfg_from_stream_state(&dml_dispcfg->timing, disp_cfg_stream_location, context->streams[i]); + populate_dml_output_cfg_from_stream_state(&dml_dispcfg->output, disp_cfg_stream_location, context->streams[i], current_pipe_context, dml2); + /*Call site for populate_dml_writeback_cfg_from_stream_state*/ + populate_dml_writeback_cfg_from_stream_state(&dml_dispcfg->writeback, + disp_cfg_stream_location, context->streams[i]); + + switch (context->streams[i]->debug.force_odm_combine_segments) { + case 2: + dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_stream_location] = dml_odm_use_policy_combine_2to1; + break; + case 4: + dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_stream_location] = dml_odm_use_policy_combine_4to1; + break; + default: + break; + } + + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_stream_location] = context->streams[i]->stream_id; + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_stream_location] = true; + + if (context->stream_status[i].plane_count == 0) { + disp_cfg_plane_location = dml_dispcfg->num_surfaces++; + + populate_dummy_dml_surface_cfg(&dml_dispcfg->surface, disp_cfg_plane_location, context->streams[i]); + populate_dummy_dml_plane_cfg(&dml_dispcfg->plane, disp_cfg_plane_location, + context->streams[i], &dml2->v20.dml_core_ctx.soc); + + dml_dispcfg->plane.BlendingAndTiming[disp_cfg_plane_location] = disp_cfg_stream_location; + + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; + } else { + for (j = 0; j < context->stream_status[i].plane_count; j++) { + disp_cfg_plane_location = map_plane_to_dml_display_cfg(dml2, + context->stream_status[i].plane_states[j], context, dml_dispcfg, context->streams[i]->stream_id, j); + + if (disp_cfg_plane_location < 0) + disp_cfg_plane_location = dml_dispcfg->num_surfaces++; + + ASSERT(disp_cfg_plane_location >= 0 && disp_cfg_plane_location < __DML2_WRAPPER_MAX_STREAMS_PLANES__); + + populate_dml_surface_cfg_from_plane_state(dml2->v20.dml_core_ctx.project, &dml_dispcfg->surface, disp_cfg_plane_location, context->stream_status[i].plane_states[j]); + populate_dml_plane_cfg_from_plane_state( + &dml_dispcfg->plane, disp_cfg_plane_location, + context->stream_status[i].plane_states[j], context, + &dml2->v20.dml_core_ctx.soc); + + if (stream_mall_type == SUBVP_MAIN) { + dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_sub_viewport; + dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_optimize; + } else if (stream_mall_type == SUBVP_PHANTOM) { + dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_phantom_pipe; + dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_disable; + dml2->v20.dml_core_ctx.policy.ImmediateFlipRequirement[disp_cfg_plane_location] = dml_immediate_flip_not_required; + } else { + dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_disable; + dml_dispcfg->plane.UseMALLForStaticScreen[disp_cfg_plane_location] = dml_use_mall_static_screen_optimize; + } + + dml_dispcfg->plane.BlendingAndTiming[disp_cfg_plane_location] = disp_cfg_stream_location; + + if (get_plane_id(dml2, context, context->stream_status[i].plane_states[j], context->streams[i]->stream_id, j, + &dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; + + if (j >= 1) { + populate_dml_timing_cfg_from_stream_state(&dml_dispcfg->timing, disp_cfg_plane_location, context->streams[i]); + populate_dml_output_cfg_from_stream_state(&dml_dispcfg->output, disp_cfg_plane_location, context->streams[i], current_pipe_context, dml2); + switch (context->streams[i]->debug.force_odm_combine_segments) { + case 2: + dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_plane_location] = dml_odm_use_policy_combine_2to1; + break; + case 4: + dml2->v20.dml_core_ctx.policy.ODMUse[disp_cfg_plane_location] = dml_odm_use_policy_combine_4to1; + break; + default: + break; + } + + if (stream_mall_type == SUBVP_MAIN) + dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_sub_viewport; + else if (stream_mall_type == SUBVP_PHANTOM) + dml_dispcfg->plane.UseMALLForPStateChange[disp_cfg_plane_location] = dml_use_mall_pstate_change_phantom_pipe; + + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[disp_cfg_plane_location] = context->streams[i]->stream_id; + dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id_valid[disp_cfg_plane_location] = true; + + dml_dispcfg->num_timings++; + } + } + } + } +} + +void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, + struct _vcs_dpi_dml_display_dlg_regs_st *disp_dlg_regs, + struct _vcs_dpi_dml_display_ttu_regs_st *disp_ttu_regs, + struct pipe_ctx *out) +{ + memset(&out->rq_regs, 0, sizeof(out->rq_regs)); + out->rq_regs.rq_regs_l.chunk_size = rq_regs->rq_regs_l.chunk_size; + out->rq_regs.rq_regs_l.min_chunk_size = rq_regs->rq_regs_l.min_chunk_size; + out->rq_regs.rq_regs_l.meta_chunk_size = rq_regs->rq_regs_l.meta_chunk_size; + out->rq_regs.rq_regs_l.min_meta_chunk_size = rq_regs->rq_regs_l.min_meta_chunk_size; + out->rq_regs.rq_regs_l.dpte_group_size = rq_regs->rq_regs_l.dpte_group_size; + out->rq_regs.rq_regs_l.mpte_group_size = rq_regs->rq_regs_l.mpte_group_size; + out->rq_regs.rq_regs_l.swath_height = rq_regs->rq_regs_l.swath_height; + out->rq_regs.rq_regs_l.pte_row_height_linear = rq_regs->rq_regs_l.pte_row_height_linear; + + out->rq_regs.rq_regs_c.chunk_size = rq_regs->rq_regs_c.chunk_size; + out->rq_regs.rq_regs_c.min_chunk_size = rq_regs->rq_regs_c.min_chunk_size; + out->rq_regs.rq_regs_c.meta_chunk_size = rq_regs->rq_regs_c.meta_chunk_size; + out->rq_regs.rq_regs_c.min_meta_chunk_size = rq_regs->rq_regs_c.min_meta_chunk_size; + out->rq_regs.rq_regs_c.dpte_group_size = rq_regs->rq_regs_c.dpte_group_size; + out->rq_regs.rq_regs_c.mpte_group_size = rq_regs->rq_regs_c.mpte_group_size; + out->rq_regs.rq_regs_c.swath_height = rq_regs->rq_regs_c.swath_height; + out->rq_regs.rq_regs_c.pte_row_height_linear = rq_regs->rq_regs_c.pte_row_height_linear; + + out->rq_regs.drq_expansion_mode = rq_regs->drq_expansion_mode; + out->rq_regs.prq_expansion_mode = rq_regs->prq_expansion_mode; + out->rq_regs.mrq_expansion_mode = rq_regs->mrq_expansion_mode; + out->rq_regs.crq_expansion_mode = rq_regs->crq_expansion_mode; + out->rq_regs.plane1_base_address = rq_regs->plane1_base_address; + + memset(&out->dlg_regs, 0, sizeof(out->dlg_regs)); + out->dlg_regs.refcyc_h_blank_end = disp_dlg_regs->refcyc_h_blank_end; + out->dlg_regs.dlg_vblank_end = disp_dlg_regs->dlg_vblank_end; + out->dlg_regs.min_dst_y_next_start = disp_dlg_regs->min_dst_y_next_start; + out->dlg_regs.refcyc_per_htotal = disp_dlg_regs->refcyc_per_htotal; + out->dlg_regs.refcyc_x_after_scaler = disp_dlg_regs->refcyc_x_after_scaler; + out->dlg_regs.dst_y_after_scaler = disp_dlg_regs->dst_y_after_scaler; + out->dlg_regs.dst_y_prefetch = disp_dlg_regs->dst_y_prefetch; + out->dlg_regs.dst_y_per_vm_vblank = disp_dlg_regs->dst_y_per_vm_vblank; + out->dlg_regs.dst_y_per_row_vblank = disp_dlg_regs->dst_y_per_row_vblank; + out->dlg_regs.dst_y_per_vm_flip = disp_dlg_regs->dst_y_per_vm_flip; + out->dlg_regs.dst_y_per_row_flip = disp_dlg_regs->dst_y_per_row_flip; + out->dlg_regs.ref_freq_to_pix_freq = disp_dlg_regs->ref_freq_to_pix_freq; + out->dlg_regs.vratio_prefetch = disp_dlg_regs->vratio_prefetch; + out->dlg_regs.vratio_prefetch_c = disp_dlg_regs->vratio_prefetch_c; + out->dlg_regs.refcyc_per_pte_group_vblank_l = disp_dlg_regs->refcyc_per_pte_group_vblank_l; + out->dlg_regs.refcyc_per_pte_group_vblank_c = disp_dlg_regs->refcyc_per_pte_group_vblank_c; + out->dlg_regs.refcyc_per_meta_chunk_vblank_l = disp_dlg_regs->refcyc_per_meta_chunk_vblank_l; + out->dlg_regs.refcyc_per_meta_chunk_vblank_c = disp_dlg_regs->refcyc_per_meta_chunk_vblank_c; + out->dlg_regs.refcyc_per_pte_group_flip_l = disp_dlg_regs->refcyc_per_pte_group_flip_l; + out->dlg_regs.refcyc_per_pte_group_flip_c = disp_dlg_regs->refcyc_per_pte_group_flip_c; + out->dlg_regs.refcyc_per_meta_chunk_flip_l = disp_dlg_regs->refcyc_per_meta_chunk_flip_l; + out->dlg_regs.refcyc_per_meta_chunk_flip_c = disp_dlg_regs->refcyc_per_meta_chunk_flip_c; + out->dlg_regs.dst_y_per_pte_row_nom_l = disp_dlg_regs->dst_y_per_pte_row_nom_l; + out->dlg_regs.dst_y_per_pte_row_nom_c = disp_dlg_regs->dst_y_per_pte_row_nom_c; + out->dlg_regs.refcyc_per_pte_group_nom_l = disp_dlg_regs->refcyc_per_pte_group_nom_l; + out->dlg_regs.refcyc_per_pte_group_nom_c = disp_dlg_regs->refcyc_per_pte_group_nom_c; + out->dlg_regs.dst_y_per_meta_row_nom_l = disp_dlg_regs->dst_y_per_meta_row_nom_l; + out->dlg_regs.dst_y_per_meta_row_nom_c = disp_dlg_regs->dst_y_per_meta_row_nom_c; + out->dlg_regs.refcyc_per_meta_chunk_nom_l = disp_dlg_regs->refcyc_per_meta_chunk_nom_l; + out->dlg_regs.refcyc_per_meta_chunk_nom_c = disp_dlg_regs->refcyc_per_meta_chunk_nom_c; + out->dlg_regs.refcyc_per_line_delivery_pre_l = disp_dlg_regs->refcyc_per_line_delivery_pre_l; + out->dlg_regs.refcyc_per_line_delivery_pre_c = disp_dlg_regs->refcyc_per_line_delivery_pre_c; + out->dlg_regs.refcyc_per_line_delivery_l = disp_dlg_regs->refcyc_per_line_delivery_l; + out->dlg_regs.refcyc_per_line_delivery_c = disp_dlg_regs->refcyc_per_line_delivery_c; + out->dlg_regs.refcyc_per_vm_group_vblank = disp_dlg_regs->refcyc_per_vm_group_vblank; + out->dlg_regs.refcyc_per_vm_group_flip = disp_dlg_regs->refcyc_per_vm_group_flip; + out->dlg_regs.refcyc_per_vm_req_vblank = disp_dlg_regs->refcyc_per_vm_req_vblank; + out->dlg_regs.refcyc_per_vm_req_flip = disp_dlg_regs->refcyc_per_vm_req_flip; + out->dlg_regs.dst_y_offset_cur0 = disp_dlg_regs->dst_y_offset_cur0; + out->dlg_regs.chunk_hdl_adjust_cur0 = disp_dlg_regs->chunk_hdl_adjust_cur0; + out->dlg_regs.dst_y_offset_cur1 = disp_dlg_regs->dst_y_offset_cur1; + out->dlg_regs.chunk_hdl_adjust_cur1 = disp_dlg_regs->chunk_hdl_adjust_cur1; + out->dlg_regs.vready_after_vcount0 = disp_dlg_regs->vready_after_vcount0; + out->dlg_regs.dst_y_delta_drq_limit = disp_dlg_regs->dst_y_delta_drq_limit; + out->dlg_regs.refcyc_per_vm_dmdata = disp_dlg_regs->refcyc_per_vm_dmdata; + out->dlg_regs.dmdata_dl_delta = disp_dlg_regs->dmdata_dl_delta; + + memset(&out->ttu_regs, 0, sizeof(out->ttu_regs)); + out->ttu_regs.qos_level_low_wm = disp_ttu_regs->qos_level_low_wm; + out->ttu_regs.qos_level_high_wm = disp_ttu_regs->qos_level_high_wm; + out->ttu_regs.min_ttu_vblank = disp_ttu_regs->min_ttu_vblank; + out->ttu_regs.qos_level_flip = disp_ttu_regs->qos_level_flip; + out->ttu_regs.refcyc_per_req_delivery_l = disp_ttu_regs->refcyc_per_req_delivery_l; + out->ttu_regs.refcyc_per_req_delivery_c = disp_ttu_regs->refcyc_per_req_delivery_c; + out->ttu_regs.refcyc_per_req_delivery_cur0 = disp_ttu_regs->refcyc_per_req_delivery_cur0; + out->ttu_regs.refcyc_per_req_delivery_cur1 = disp_ttu_regs->refcyc_per_req_delivery_cur1; + out->ttu_regs.refcyc_per_req_delivery_pre_l = disp_ttu_regs->refcyc_per_req_delivery_pre_l; + out->ttu_regs.refcyc_per_req_delivery_pre_c = disp_ttu_regs->refcyc_per_req_delivery_pre_c; + out->ttu_regs.refcyc_per_req_delivery_pre_cur0 = disp_ttu_regs->refcyc_per_req_delivery_pre_cur0; + out->ttu_regs.refcyc_per_req_delivery_pre_cur1 = disp_ttu_regs->refcyc_per_req_delivery_pre_cur1; + out->ttu_regs.qos_level_fixed_l = disp_ttu_regs->qos_level_fixed_l; + out->ttu_regs.qos_level_fixed_c = disp_ttu_regs->qos_level_fixed_c; + out->ttu_regs.qos_level_fixed_cur0 = disp_ttu_regs->qos_level_fixed_cur0; + out->ttu_regs.qos_level_fixed_cur1 = disp_ttu_regs->qos_level_fixed_cur1; + out->ttu_regs.qos_ramp_disable_l = disp_ttu_regs->qos_ramp_disable_l; + out->ttu_regs.qos_ramp_disable_c = disp_ttu_regs->qos_ramp_disable_c; + out->ttu_regs.qos_ramp_disable_cur0 = disp_ttu_regs->qos_ramp_disable_cur0; + out->ttu_regs.qos_ramp_disable_cur1 = disp_ttu_regs->qos_ramp_disable_cur1; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h new file mode 100644 index 000000000000..d764773938f4 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h @@ -0,0 +1,41 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML2_TRANSLATION_HELPER_H__ +#define __DML2_TRANSLATION_HELPER_H__ + +void dml2_init_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out); +void dml2_init_socbb_params(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out); +void dml2_init_soc_states(struct dml2_context *dml2, const struct dc *in_dc, + const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out); +void dml2_translate_ip_params(const struct dc *in_dc, struct ip_params_st *out); +void dml2_translate_socbb_params(const struct dc *in_dc, struct soc_bounding_box_st *out); +void dml2_translate_soc_states(const struct dc *in_dc, struct soc_states_st *out, int num_states); +void map_dc_state_into_dml_display_cfg(struct dml2_context *dml2, struct dc_state *context, struct dml_display_cfg_st *dml_dispcfg); +void dml2_update_pipe_ctx_dchub_regs(struct _vcs_dpi_dml_display_rq_regs_st *rq_regs, struct _vcs_dpi_dml_display_dlg_regs_st *disp_dlg_regs, struct _vcs_dpi_dml_display_ttu_regs_st *disp_ttu_regs, struct pipe_ctx *out); +bool is_dp2p0_output_encoder(const struct pipe_ctx *pipe); + +#endif //__DML2_TRANSLATION_HELPER_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c new file mode 100644 index 000000000000..9a33158b63bf --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c @@ -0,0 +1,560 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +//#include "dml2_utils.h" +#include "display_mode_core.h" +#include "dml_display_rq_dlg_calc.h" +#include "dml2_internal_types.h" +#include "dml2_translation_helper.h" +#include "dml2_utils.h" + +void dml2_util_copy_dml_timing(struct dml_timing_cfg_st *dml_timing_array, unsigned int dst_index, unsigned int src_index) +{ + dml_timing_array->HTotal[dst_index] = dml_timing_array->HTotal[src_index]; + dml_timing_array->VTotal[dst_index] = dml_timing_array->VTotal[src_index]; + dml_timing_array->HBlankEnd[dst_index] = dml_timing_array->HBlankEnd[src_index]; + dml_timing_array->VBlankEnd[dst_index] = dml_timing_array->VBlankEnd[src_index]; + dml_timing_array->RefreshRate[dst_index] = dml_timing_array->RefreshRate[src_index]; + dml_timing_array->VFrontPorch[dst_index] = dml_timing_array->VFrontPorch[src_index]; + dml_timing_array->PixelClock[dst_index] = dml_timing_array->PixelClock[src_index]; + dml_timing_array->HActive[dst_index] = dml_timing_array->HActive[src_index]; + dml_timing_array->VActive[dst_index] = dml_timing_array->VActive[src_index]; + dml_timing_array->Interlace[dst_index] = dml_timing_array->Interlace[src_index]; + dml_timing_array->DRRDisplay[dst_index] = dml_timing_array->DRRDisplay[src_index]; + dml_timing_array->VBlankNom[dst_index] = dml_timing_array->VBlankNom[src_index]; +} + +void dml2_util_copy_dml_plane(struct dml_plane_cfg_st *dml_plane_array, unsigned int dst_index, unsigned int src_index) +{ + dml_plane_array->GPUVMMinPageSizeKBytes[dst_index] = dml_plane_array->GPUVMMinPageSizeKBytes[src_index]; + dml_plane_array->ForceOneRowForFrame[dst_index] = dml_plane_array->ForceOneRowForFrame[src_index]; + dml_plane_array->PTEBufferModeOverrideEn[dst_index] = dml_plane_array->PTEBufferModeOverrideEn[src_index]; + dml_plane_array->PTEBufferMode[dst_index] = dml_plane_array->PTEBufferMode[src_index]; + dml_plane_array->ViewportWidth[dst_index] = dml_plane_array->ViewportWidth[src_index]; + dml_plane_array->ViewportHeight[dst_index] = dml_plane_array->ViewportHeight[src_index]; + dml_plane_array->ViewportWidthChroma[dst_index] = dml_plane_array->ViewportWidthChroma[src_index]; + dml_plane_array->ViewportHeightChroma[dst_index] = dml_plane_array->ViewportHeightChroma[src_index]; + dml_plane_array->ViewportXStart[dst_index] = dml_plane_array->ViewportXStart[src_index]; + dml_plane_array->ViewportXStartC[dst_index] = dml_plane_array->ViewportXStartC[src_index]; + dml_plane_array->ViewportYStart[dst_index] = dml_plane_array->ViewportYStart[src_index]; + dml_plane_array->ViewportYStartC[dst_index] = dml_plane_array->ViewportYStartC[src_index]; + dml_plane_array->ViewportStationary[dst_index] = dml_plane_array->ViewportStationary[src_index]; + + dml_plane_array->ScalerEnabled[dst_index] = dml_plane_array->ScalerEnabled[src_index]; + dml_plane_array->HRatio[dst_index] = dml_plane_array->HRatio[src_index]; + dml_plane_array->VRatio[dst_index] = dml_plane_array->VRatio[src_index]; + dml_plane_array->HRatioChroma[dst_index] = dml_plane_array->HRatioChroma[src_index]; + dml_plane_array->VRatioChroma[dst_index] = dml_plane_array->VRatioChroma[src_index]; + dml_plane_array->HTaps[dst_index] = dml_plane_array->HTaps[src_index]; + dml_plane_array->VTaps[dst_index] = dml_plane_array->VTaps[src_index]; + dml_plane_array->HTapsChroma[dst_index] = dml_plane_array->HTapsChroma[src_index]; + dml_plane_array->VTapsChroma[dst_index] = dml_plane_array->VTapsChroma[src_index]; + dml_plane_array->LBBitPerPixel[dst_index] = dml_plane_array->LBBitPerPixel[src_index]; + + dml_plane_array->SourceScan[dst_index] = dml_plane_array->SourceScan[src_index]; + dml_plane_array->ScalerRecoutWidth[dst_index] = dml_plane_array->ScalerRecoutWidth[src_index]; + + dml_plane_array->DynamicMetadataEnable[dst_index] = dml_plane_array->DynamicMetadataEnable[src_index]; + dml_plane_array->DynamicMetadataLinesBeforeActiveRequired[dst_index] = dml_plane_array->DynamicMetadataLinesBeforeActiveRequired[src_index]; + dml_plane_array->DynamicMetadataTransmittedBytes[dst_index] = dml_plane_array->DynamicMetadataTransmittedBytes[src_index]; + dml_plane_array->DETSizeOverride[dst_index] = dml_plane_array->DETSizeOverride[src_index]; + + dml_plane_array->NumberOfCursors[dst_index] = dml_plane_array->NumberOfCursors[src_index]; + dml_plane_array->CursorWidth[dst_index] = dml_plane_array->CursorWidth[src_index]; + dml_plane_array->CursorBPP[dst_index] = dml_plane_array->CursorBPP[src_index]; + + dml_plane_array->UseMALLForStaticScreen[dst_index] = dml_plane_array->UseMALLForStaticScreen[src_index]; + dml_plane_array->UseMALLForPStateChange[dst_index] = dml_plane_array->UseMALLForPStateChange[src_index]; + + dml_plane_array->BlendingAndTiming[dst_index] = dml_plane_array->BlendingAndTiming[src_index]; +} + +void dml2_util_copy_dml_surface(struct dml_surface_cfg_st *dml_surface_array, unsigned int dst_index, unsigned int src_index) +{ + dml_surface_array->SurfaceTiling[dst_index] = dml_surface_array->SurfaceTiling[src_index]; + dml_surface_array->SourcePixelFormat[dst_index] = dml_surface_array->SourcePixelFormat[src_index]; + dml_surface_array->PitchY[dst_index] = dml_surface_array->PitchY[src_index]; + dml_surface_array->SurfaceWidthY[dst_index] = dml_surface_array->SurfaceWidthY[src_index]; + dml_surface_array->SurfaceHeightY[dst_index] = dml_surface_array->SurfaceHeightY[src_index]; + dml_surface_array->PitchC[dst_index] = dml_surface_array->PitchC[src_index]; + dml_surface_array->SurfaceWidthC[dst_index] = dml_surface_array->SurfaceWidthC[src_index]; + dml_surface_array->SurfaceHeightC[dst_index] = dml_surface_array->SurfaceHeightC[src_index]; + + dml_surface_array->DCCEnable[dst_index] = dml_surface_array->DCCEnable[src_index]; + dml_surface_array->DCCMetaPitchY[dst_index] = dml_surface_array->DCCMetaPitchY[src_index]; + dml_surface_array->DCCMetaPitchC[dst_index] = dml_surface_array->DCCMetaPitchC[src_index]; + + dml_surface_array->DCCRateLuma[dst_index] = dml_surface_array->DCCRateLuma[src_index]; + dml_surface_array->DCCRateChroma[dst_index] = dml_surface_array->DCCRateChroma[src_index]; + dml_surface_array->DCCFractionOfZeroSizeRequestsLuma[dst_index] = dml_surface_array->DCCFractionOfZeroSizeRequestsLuma[src_index]; + dml_surface_array->DCCFractionOfZeroSizeRequestsChroma[dst_index] = dml_surface_array->DCCFractionOfZeroSizeRequestsChroma[src_index]; +} + +void dml2_util_copy_dml_output(struct dml_output_cfg_st *dml_output_array, unsigned int dst_index, unsigned int src_index) +{ + dml_output_array->DSCInputBitPerComponent[dst_index] = dml_output_array->DSCInputBitPerComponent[src_index]; + dml_output_array->OutputFormat[dst_index] = dml_output_array->OutputFormat[src_index]; + dml_output_array->OutputEncoder[dst_index] = dml_output_array->OutputEncoder[src_index]; + dml_output_array->OutputMultistreamId[dst_index] = dml_output_array->OutputMultistreamId[src_index]; + dml_output_array->OutputMultistreamEn[dst_index] = dml_output_array->OutputMultistreamEn[src_index]; + dml_output_array->OutputBpp[dst_index] = dml_output_array->OutputBpp[src_index]; + dml_output_array->PixelClockBackEnd[dst_index] = dml_output_array->PixelClockBackEnd[src_index]; + dml_output_array->DSCEnable[dst_index] = dml_output_array->DSCEnable[src_index]; + dml_output_array->OutputLinkDPLanes[dst_index] = dml_output_array->OutputLinkDPLanes[src_index]; + dml_output_array->OutputLinkDPRate[dst_index] = dml_output_array->OutputLinkDPRate[src_index]; + dml_output_array->ForcedOutputLinkBPP[dst_index] = dml_output_array->ForcedOutputLinkBPP[src_index]; + dml_output_array->AudioSampleRate[dst_index] = dml_output_array->AudioSampleRate[src_index]; + dml_output_array->AudioSampleLayout[dst_index] = dml_output_array->AudioSampleLayout[src_index]; +} + +unsigned int dml2_util_get_maximum_odm_combine_for_output(bool force_odm_4to1, enum dml_output_encoder_class encoder, bool dsc_enabled) +{ + switch (encoder) { + case dml_dp: + case dml_edp: + return 2; + case dml_dp2p0: + if (dsc_enabled || force_odm_4to1) + return 4; + else + return 2; + case dml_hdmi: + return 1; + case dml_hdmifrl: + if (force_odm_4to1) + return 4; + else + return 2; + default: + return 1; + } +} + +bool is_dp2p0_output_encoder(const struct pipe_ctx *pipe_ctx) +{ + if (pipe_ctx == NULL || pipe_ctx->stream == NULL) + return false; + + /* If this assert is hit then we have a link encoder dynamic management issue */ + ASSERT(pipe_ctx->stream_res.hpo_dp_stream_enc ? pipe_ctx->link_res.hpo_dp_link_enc != NULL : true); + + return (pipe_ctx->stream_res.hpo_dp_stream_enc && + pipe_ctx->link_res.hpo_dp_link_enc && + dc_is_dp_signal(pipe_ctx->stream->signal)); +} + +bool is_dtbclk_required(const struct dc *dc, struct dc_state *context) +{ + int i; + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + if (!context->res_ctx.pipe_ctx[i].stream) + continue; + if (is_dp2p0_output_encoder(&context->res_ctx.pipe_ctx[i])) + return true; + } + return false; +} + +void dml2_copy_clocks_to_dc_state(struct dml2_dcn_clocks *out_clks, struct dc_state *context) +{ + context->bw_ctx.bw.dcn.clk.dispclk_khz = out_clks->dispclk_khz; + context->bw_ctx.bw.dcn.clk.dcfclk_khz = out_clks->dcfclk_khz; + context->bw_ctx.bw.dcn.clk.dramclk_khz = out_clks->uclk_mts / 16; + context->bw_ctx.bw.dcn.clk.fclk_khz = out_clks->fclk_khz; + context->bw_ctx.bw.dcn.clk.phyclk_khz = out_clks->phyclk_khz; + context->bw_ctx.bw.dcn.clk.socclk_khz = out_clks->socclk_khz; + context->bw_ctx.bw.dcn.clk.ref_dtbclk_khz = out_clks->ref_dtbclk_khz; + context->bw_ctx.bw.dcn.clk.p_state_change_support = out_clks->p_state_supported; +} + +int dml2_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id) +{ + int i; + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[i] && ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[i] == stream_id) + return i; + } + + return -1; +} + +static int find_dml_pipe_idx_by_plane_id(struct dml2_context *ctx, unsigned int plane_id) +{ + int i; + for (i = 0; i < __DML2_WRAPPER_MAX_STREAMS_PLANES__; i++) { + if (ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[i] && ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[i] == plane_id) + return i; + } + + return -1; +} + +static bool get_plane_id(struct dml2_context *dml2, const struct dc_state *state, const struct dc_plane_state *plane, + unsigned int stream_id, unsigned int plane_index, unsigned int *plane_id) +{ + unsigned int i, j; + bool is_plane_duplicate = dml2->v20.scratch.plane_duplicate_exists; + + if (!plane_id) + return false; + + for (i = 0; i < state->stream_count; i++) { + if (state->streams[i]->stream_id == stream_id) { + for (j = 0; j < state->stream_status[i].plane_count; j++) { + if (state->stream_status[i].plane_states[j] == plane && + (!is_plane_duplicate || (j == plane_index))) { + *plane_id = (i << 16) | j; + return true; + } + } + } + } + + return false; +} + +static void populate_pipe_ctx_dlg_params_from_dml(struct pipe_ctx *pipe_ctx, struct display_mode_lib_st *mode_lib, dml_uint_t pipe_idx) +{ + unsigned int hactive, vactive, hblank_start, vblank_start, hblank_end, vblank_end; + struct dc_crtc_timing *timing = &pipe_ctx->stream->timing; + + hactive = timing->h_addressable + timing->h_border_left + timing->h_border_right; + vactive = timing->v_addressable + timing->v_border_bottom + timing->v_border_top; + hblank_start = pipe_ctx->stream->timing.h_total - pipe_ctx->stream->timing.h_front_porch; + vblank_start = pipe_ctx->stream->timing.v_total - pipe_ctx->stream->timing.v_front_porch; + + hblank_end = hblank_start - timing->h_addressable - timing->h_border_left - timing->h_border_right; + vblank_end = vblank_start - timing->v_addressable - timing->v_border_top - timing->v_border_bottom; + + pipe_ctx->pipe_dlg_param.vstartup_start = dml_get_vstartup_calculated(mode_lib, pipe_idx); + pipe_ctx->pipe_dlg_param.vupdate_offset = dml_get_vupdate_offset(mode_lib, pipe_idx); + pipe_ctx->pipe_dlg_param.vupdate_width = dml_get_vupdate_width(mode_lib, pipe_idx); + pipe_ctx->pipe_dlg_param.vready_offset = dml_get_vready_offset(mode_lib, pipe_idx); + + pipe_ctx->pipe_dlg_param.otg_inst = pipe_ctx->stream_res.tg->inst; + + pipe_ctx->pipe_dlg_param.hactive = hactive; + pipe_ctx->pipe_dlg_param.vactive = vactive; + pipe_ctx->pipe_dlg_param.htotal = pipe_ctx->stream->timing.h_total; + pipe_ctx->pipe_dlg_param.vtotal = pipe_ctx->stream->timing.v_total; + pipe_ctx->pipe_dlg_param.hblank_end = hblank_end; + pipe_ctx->pipe_dlg_param.vblank_end = vblank_end; + pipe_ctx->pipe_dlg_param.hblank_start = hblank_start; + pipe_ctx->pipe_dlg_param.vblank_start = vblank_start; + pipe_ctx->pipe_dlg_param.vfront_porch = pipe_ctx->stream->timing.v_front_porch; + pipe_ctx->pipe_dlg_param.pixel_rate_mhz = pipe_ctx->stream->timing.pix_clk_100hz / 10000.00; + pipe_ctx->pipe_dlg_param.refresh_rate = ((timing->pix_clk_100hz * 100) / timing->h_total) / timing->v_total; + pipe_ctx->pipe_dlg_param.vtotal_max = pipe_ctx->stream->adjust.v_total_max; + pipe_ctx->pipe_dlg_param.vtotal_min = pipe_ctx->stream->adjust.v_total_min; + pipe_ctx->pipe_dlg_param.recout_height = pipe_ctx->plane_res.scl_data.recout.height; + pipe_ctx->pipe_dlg_param.recout_width = pipe_ctx->plane_res.scl_data.recout.width; + pipe_ctx->pipe_dlg_param.full_recout_height = pipe_ctx->plane_res.scl_data.recout.height; + pipe_ctx->pipe_dlg_param.full_recout_width = pipe_ctx->plane_res.scl_data.recout.width; +} + +void dml2_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, struct dml2_context *in_ctx, unsigned int pipe_cnt) +{ + unsigned int dc_pipe_ctx_index, dml_pipe_idx, plane_id; + enum mall_stream_type pipe_mall_type; + struct dml2_calculate_rq_and_dlg_params_scratch *s = &in_ctx->v20.scratch.calculate_rq_and_dlg_params_scratch; + + context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = (unsigned int)in_ctx->v20.dml_core_ctx.mp.DCFCLKDeepSleep * 1000; + context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; + + if (in_ctx->v20.dml_core_ctx.ms.support.FCLKChangeSupport[0] == dml_fclock_change_unsupported) + context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = false; + else + context->bw_ctx.bw.dcn.clk.fclk_p_state_change_support = true; + + if (context->bw_ctx.bw.dcn.clk.dispclk_khz < dc->debug.min_disp_clk_khz) + context->bw_ctx.bw.dcn.clk.dispclk_khz = dc->debug.min_disp_clk_khz; + + context->bw_ctx.bw.dcn.compbuf_size_kb = in_ctx->v20.dml_core_ctx.ip.config_return_buffer_size_in_kbytes; + + for (dc_pipe_ctx_index = 0; dc_pipe_ctx_index < pipe_cnt; dc_pipe_ctx_index++) { + if (!context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream) + continue; + /* The DML2 and the DC logic of determining pipe indices are different from each other so + * there is a need to know which DML pipe index maps to which DC pipe. The code below + * finds a dml_pipe_index from the plane id if a plane is valid. If a plane is not valid then + * it finds a dml_pipe_index from the stream id. */ + if (get_plane_id(in_ctx, context, context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state, + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id, + in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[context->res_ctx.pipe_ctx[dc_pipe_ctx_index].pipe_idx], &plane_id)) { + dml_pipe_idx = find_dml_pipe_idx_by_plane_id(in_ctx, plane_id); + } else { + dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(in_ctx, context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id); + } + + if (dml_pipe_idx == 0xFFFFFFFF) + continue; + ASSERT(in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[dml_pipe_idx]); + ASSERT(in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[dml_pipe_idx] == context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream->stream_id); + + /* Use the dml_pipe_index here for the getters to fetch the correct values and dc_pipe_index in the pipe_ctx to populate them + * at the right locations. + */ + populate_pipe_ctx_dlg_params_from_dml(&context->res_ctx.pipe_ctx[dc_pipe_ctx_index], &context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); + + pipe_mall_type = dc_state_get_pipe_subvp_type(context, &context->res_ctx.pipe_ctx[dc_pipe_ctx_index]); + if (pipe_mall_type == SUBVP_PHANTOM) { + // Phantom pipe requires that DET_SIZE = 0 and no unbounded requests + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb = 0; + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].unbounded_req = false; + } else { + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb = dml_get_det_buffer_size_kbytes(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); + // Unbounded requesting should not ever be used when more than 1 pipe is enabled. + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].unbounded_req = in_ctx->v20.dml_core_ctx.ms.UnboundedRequestEnabledThisState; + } + + context->bw_ctx.bw.dcn.compbuf_size_kb -= context->res_ctx.pipe_ctx[dc_pipe_ctx_index].det_buffer_size_kb; + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz = dml_get_dppclk_calculated(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx) * 1000; + if (context->bw_ctx.bw.dcn.clk.dppclk_khz < context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz) + context->bw_ctx.bw.dcn.clk.dppclk_khz = context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_res.bw.dppclk_khz; + + dml_rq_dlg_get_rq_reg(&s->rq_regs, &in_ctx->v20.dml_core_ctx, dml_pipe_idx); + dml_rq_dlg_get_dlg_reg(&s->disp_dlg_regs, &s->disp_ttu_regs, &in_ctx->v20.dml_core_ctx, dml_pipe_idx); + dml2_update_pipe_ctx_dchub_regs(&s->rq_regs, &s->disp_dlg_regs, &s->disp_ttu_regs, &out_new_hw_state->pipe_ctx[dc_pipe_ctx_index]); + + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes = dml_get_surface_size_for_mall(&context->bw_ctx.dml2->v20.dml_core_ctx, dml_pipe_idx); + + /* Reuse MALL Allocation Sizes logic from dcn32_fpu.c */ + /* Count from active, top pipes per plane only. Only add mall_ss_size_bytes for each unique plane. */ + if (context->res_ctx.pipe_ctx[dc_pipe_ctx_index].stream && context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state && + (context->res_ctx.pipe_ctx[dc_pipe_ctx_index].top_pipe == NULL || + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].plane_state != context->res_ctx.pipe_ctx[dc_pipe_ctx_index].top_pipe->plane_state) && + context->res_ctx.pipe_ctx[dc_pipe_ctx_index].prev_odm_pipe == NULL) { + /* SS: all active surfaces stored in MALL */ + if (pipe_mall_type != SUBVP_PHANTOM) { + context->bw_ctx.bw.dcn.mall_ss_size_bytes += context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes; + } else { + /* SUBVP: phantom surfaces only stored in MALL */ + context->bw_ctx.bw.dcn.mall_subvp_size_bytes += context->res_ctx.pipe_ctx[dc_pipe_ctx_index].surface_size_in_mall_bytes; + } + } + } + + context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; + context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; + + context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz = in_ctx->v20.dml_core_ctx.states.state_array[in_ctx->v20.scratch.mode_support_params.out_lowest_state_idx].dppclk_mhz + * 1000; + context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz = in_ctx->v20.dml_core_ctx.states.state_array[in_ctx->v20.scratch.mode_support_params.out_lowest_state_idx].dispclk_mhz + * 1000; + + if (dc->config.forced_clocks || dc->debug.max_disp_clk) { + context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.max_supported_dispclk_khz; + context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.max_supported_dppclk_khz ; + } +} + +void dml2_extract_watermark_set(struct dcn_watermarks *watermark, struct display_mode_lib_st *dml_core_ctx) +{ + watermark->urgent_ns = dml_get_wm_urgent(dml_core_ctx) * 1000; + watermark->cstate_pstate.cstate_enter_plus_exit_ns = dml_get_wm_stutter_enter_exit(dml_core_ctx) * 1000; + watermark->cstate_pstate.cstate_exit_ns = dml_get_wm_stutter_exit(dml_core_ctx) * 1000; + watermark->cstate_pstate.pstate_change_ns = dml_get_wm_dram_clock_change(dml_core_ctx) * 1000; + watermark->pte_meta_urgent_ns = dml_get_wm_memory_trip(dml_core_ctx) * 1000; + watermark->frac_urg_bw_nom = dml_get_fraction_of_urgent_bandwidth(dml_core_ctx) * 1000; + watermark->frac_urg_bw_flip = dml_get_fraction_of_urgent_bandwidth_imm_flip(dml_core_ctx) * 1000; + watermark->urgent_latency_ns = dml_get_urgent_latency(dml_core_ctx) * 1000; + watermark->cstate_pstate.fclk_pstate_change_ns = dml_get_wm_fclk_change(dml_core_ctx) * 1000; + watermark->usr_retraining_ns = dml_get_wm_usr_retraining(dml_core_ctx) * 1000; + watermark->cstate_pstate.cstate_enter_plus_exit_z8_ns = dml_get_wm_z8_stutter_enter_exit(dml_core_ctx) * 1000; + watermark->cstate_pstate.cstate_exit_z8_ns = dml_get_wm_z8_stutter(dml_core_ctx) * 1000; +} + +unsigned int dml2_calc_max_scaled_time( + unsigned int time_per_pixel, + enum mmhubbub_wbif_mode mode, + unsigned int urgent_watermark) +{ + unsigned int time_per_byte = 0; + unsigned int total_free_entry = 0xb40; + unsigned int buf_lh_capability; + unsigned int max_scaled_time; + + if (mode == PACKED_444) /* packed mode 32 bpp */ + time_per_byte = time_per_pixel/4; + else if (mode == PACKED_444_FP16) /* packed mode 64 bpp */ + time_per_byte = time_per_pixel/8; + + if (time_per_byte == 0) + time_per_byte = 1; + + buf_lh_capability = (total_free_entry*time_per_byte*32) >> 6; /* time_per_byte is in u6.6*/ + max_scaled_time = buf_lh_capability - urgent_watermark; + return max_scaled_time; +} + +void dml2_extract_writeback_wm(struct dc_state *context, struct display_mode_lib_st *dml_core_ctx) +{ + int i, j = 0; + struct mcif_arb_params *wb_arb_params = NULL; + struct dcn_bw_writeback *bw_writeback = NULL; + enum mmhubbub_wbif_mode wbif_mode = PACKED_444_FP16; /*for now*/ + + if (context->stream_count != 0) { + for (i = 0; i < context->stream_count; i++) { + if (context->streams[i]->num_wb_info != 0) + j++; + } + } + if (j == 0) /*no dwb */ + return; + for (i = 0; i < __DML_NUM_DMB__; i++) { + bw_writeback = &context->bw_ctx.bw.dcn.bw_writeback; + wb_arb_params = &context->bw_ctx.bw.dcn.bw_writeback.mcif_wb_arb[i]; + + for (j = 0 ; j < 4; j++) { + /*current dml only has one set of watermark, need to follow up*/ + bw_writeback->mcif_wb_arb[i].cli_watermark[j] = + dml_get_wm_writeback_urgent(dml_core_ctx) * 1000; + bw_writeback->mcif_wb_arb[i].pstate_watermark[j] = + dml_get_wm_writeback_dram_clock_change(dml_core_ctx) * 1000; + } + if (context->res_ctx.pipe_ctx[i].stream->phy_pix_clk != 0) { + /* time_per_pixel should be in u6.6 format */ + bw_writeback->mcif_wb_arb[i].time_per_pixel = + (1000000 << 6) / context->res_ctx.pipe_ctx[i].stream->phy_pix_clk; + } + bw_writeback->mcif_wb_arb[i].slice_lines = 32; + bw_writeback->mcif_wb_arb[i].arbitration_slice = 2; + bw_writeback->mcif_wb_arb[i].max_scaled_time = + dml2_calc_max_scaled_time(wb_arb_params->time_per_pixel, + wbif_mode, wb_arb_params->cli_watermark[0]); + /*not required any more*/ + bw_writeback->mcif_wb_arb[i].dram_speed_change_duration = + dml_get_wm_writeback_dram_clock_change(dml_core_ctx) * 1000; + + } +} +void dml2_initialize_det_scratch(struct dml2_context *in_ctx) +{ + int i; + + for (i = 0; i < MAX_PLANES; i++) { + in_ctx->det_helper_scratch.dpps_per_surface[i] = 1; + } +} + +static unsigned int find_planes_per_stream_and_stream_count(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg, int *num_of_planes_per_stream) +{ + unsigned int plane_index, stream_index = 0, num_of_streams; + + for (plane_index = 0; plane_index < dml_dispcfg->num_surfaces; plane_index++) { + /* Number of planes per stream */ + num_of_planes_per_stream[stream_index] += 1; + + if (plane_index + 1 < dml_dispcfg->num_surfaces && dml_dispcfg->plane.BlendingAndTiming[plane_index] != dml_dispcfg->plane.BlendingAndTiming[plane_index + 1]) + stream_index++; + } + + num_of_streams = stream_index + 1; + + return num_of_streams; +} + +void dml2_apply_det_buffer_allocation_policy(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg) +{ + unsigned int num_of_streams = 0, plane_index = 0, max_det_size, stream_index = 0; + int num_of_planes_per_stream[__DML_NUM_PLANES__] = { 0 }; + + max_det_size = in_ctx->config.det_segment_size * in_ctx->config.max_segments_per_hubp; + + num_of_streams = find_planes_per_stream_and_stream_count(in_ctx, dml_dispcfg, num_of_planes_per_stream); + + for (plane_index = 0; plane_index < dml_dispcfg->num_surfaces; plane_index++) { + + if (in_ctx->config.override_det_buffer_size_kbytes) + dml_dispcfg->plane.DETSizeOverride[plane_index] = max_det_size / in_ctx->config.dcn_pipe_count; + else { + dml_dispcfg->plane.DETSizeOverride[plane_index] = ((max_det_size / num_of_streams) / num_of_planes_per_stream[stream_index] / in_ctx->det_helper_scratch.dpps_per_surface[plane_index]); + + /* If the override size is not divisible by det_segment_size then round off to nearest number divisible by det_segment_size as + * this is a requirement. + */ + if (dml_dispcfg->plane.DETSizeOverride[plane_index] % in_ctx->config.det_segment_size != 0) { + dml_dispcfg->plane.DETSizeOverride[plane_index] = dml_dispcfg->plane.DETSizeOverride[plane_index] & ~0x3F; + } + + if (plane_index + 1 < dml_dispcfg->num_surfaces && dml_dispcfg->plane.BlendingAndTiming[plane_index] != dml_dispcfg->plane.BlendingAndTiming[plane_index + 1]) + stream_index++; + } + } +} + +bool dml2_verify_det_buffer_configuration(struct dml2_context *in_ctx, struct dc_state *display_state, struct dml2_helper_det_policy_scratch *det_scratch) +{ + unsigned int i = 0, dml_pipe_idx = 0, plane_id = 0; + unsigned int max_det_size, total_det_allocated = 0; + bool need_recalculation = false; + + max_det_size = in_ctx->config.det_segment_size * in_ctx->config.max_segments_per_hubp; + + for (i = 0; i < MAX_PIPES; i++) { + if (!display_state->res_ctx.pipe_ctx[i].stream) + continue; + if (get_plane_id(in_ctx, display_state, display_state->res_ctx.pipe_ctx[i].plane_state, + display_state->res_ctx.pipe_ctx[i].stream->stream_id, + in_ctx->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_index[display_state->res_ctx.pipe_ctx[i].pipe_idx], &plane_id)) + dml_pipe_idx = find_dml_pipe_idx_by_plane_id(in_ctx, plane_id); + else + dml_pipe_idx = dml2_helper_find_dml_pipe_idx_by_stream_id(in_ctx, display_state->res_ctx.pipe_ctx[i].stream->stream_id); + + if (dml_pipe_idx == 0xFFFFFFFF) + continue; + total_det_allocated += dml_get_det_buffer_size_kbytes(&in_ctx->v20.dml_core_ctx, dml_pipe_idx); + if (total_det_allocated > max_det_size) { + need_recalculation = true; + } + } + + /* Store the DPPPerSurface for correctly determining the number of planes in the next call. */ + for (i = 0; i < MAX_PLANES; i++) { + det_scratch->dpps_per_surface[i] = in_ctx->v20.scratch.cur_display_config.hw.DPPPerSurface[i]; + } + + return need_recalculation; +} + +bool dml2_is_stereo_timing(const struct dc_stream_state *stream) +{ + bool is_stereo = false; + + if ((stream->view_format == + VIEW_3D_FORMAT_SIDE_BY_SIDE || + stream->view_format == + VIEW_3D_FORMAT_TOP_AND_BOTTOM) && + (stream->timing.timing_3d_format == + TIMING_3D_FORMAT_TOP_AND_BOTTOM || + stream->timing.timing_3d_format == + TIMING_3D_FORMAT_SIDE_BY_SIDE)) + is_stereo = true; + + return is_stereo; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h new file mode 100644 index 000000000000..04fcfe637119 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h @@ -0,0 +1,149 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef _DML2_UTILS_H_ +#define _DML2_UTILS_H_ + +#include "os_types.h" +#include "dml2_dc_types.h" + +struct dc; +struct dml_timing_cfg_st; +struct dml2_dcn_clocks; +struct dc_state; + +void dml2_util_copy_dml_timing(struct dml_timing_cfg_st *dml_timing_array, unsigned int dst_index, unsigned int src_index); +void dml2_util_copy_dml_plane(struct dml_plane_cfg_st *dml_plane_array, unsigned int dst_index, unsigned int src_index); +void dml2_util_copy_dml_surface(struct dml_surface_cfg_st *dml_surface_array, unsigned int dst_index, unsigned int src_index); +void dml2_util_copy_dml_output(struct dml_output_cfg_st *dml_output_array, unsigned int dst_index, unsigned int src_index); +unsigned int dml2_util_get_maximum_odm_combine_for_output(bool force_odm_4to1, enum dml_output_encoder_class encoder, bool dsc_enabled); +void dml2_copy_clocks_to_dc_state(struct dml2_dcn_clocks *out_clks, struct dc_state *context); +void dml2_extract_watermark_set(struct dcn_watermarks *watermark, struct display_mode_lib_st *dml_core_ctx); +void dml2_extract_writeback_wm(struct dc_state *context, struct display_mode_lib_st *dml_core_ctx); +int dml2_helper_find_dml_pipe_idx_by_stream_id(struct dml2_context *ctx, unsigned int stream_id); +bool is_dtbclk_required(const struct dc *dc, struct dc_state *context); +bool dml2_is_stereo_timing(const struct dc_stream_state *stream); +unsigned int dml2_calc_max_scaled_time( + unsigned int time_per_pixel, + enum mmhubbub_wbif_mode mode, + unsigned int urgent_watermark); + +/* + * dml2_dc_construct_pipes - This function will determine if we need additional pipes based + * on the DML calculated outputs for MPC, ODM and allocate them as necessary. This function + * could be called after in dml_validate_build_resource after dml_mode_pragramming like : + * { + * ... + * map_hw_resources(&s->cur_display_config, &s->mode_support_info); + * result = dml_mode_programming(&in_ctx->dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); + * dml2_dc_construct_pipes(in_display_state, s->mode_support_info, out_hw_context); + * ... + * } + * + * @context: To obtain res_ctx and read other information like stream ID etc. + * @dml_mode_support_st : To get the ODM, MPC outputs as determined by the DML. + * @out_hw_context : Handle to the new hardware context. + * + * + * Return: None. + */ +void dml2_dc_construct_pipes(struct dc_state *context, struct dml_mode_support_info_st *dml_mode_support_st, + struct resource_context *out_hw_context); + +/* + * dml2_predict_pipe_split - This function is the dml2 version of predict split pipe. It predicts a + * if pipe split is required or not and returns the output as a bool. + * @context : dc_state. + * @pipe : old_index is the index of the pipe as derived from pipe_idx. + * @index : index of the pipe + * + * + * Return: Returns the result in boolean. + */ +bool dml2_predict_pipe_split(struct dc_state *context, display_pipe_params_st pipe, int index); + +/* + * dml2_build_mapped_resource - This function is the dml2 version of build_mapped_resource. + * In case of ODM, we need to build pipe hardware params again as done in dcn20_build_mapped_resource. + * @dc : struct dc + * @context : struct dc_state. + * @stream : stream whoose corresponding pipe params need to be modified. + * + * + * Return: Returns DC_OK if successful. + */ +enum dc_status dml2_build_mapped_resource(const struct dc *dc, struct dc_state *context, struct dc_stream_state *stream); + +/* + * dml2_extract_rq_regs - This function will extract information needed for struct _vcs_dpi_display_rq_regs_st + * and populate it. + * @context: To obtain and populate the res_ctx->pipe_ctx->rq_regs with DML outputs. + * @support : This structure has the DML intermediate outputs required to populate rq_regs. + * + * + * Return: None. + */ + + /* + * dml2_calculate_rq_and_dlg_params - This function will call into DML2 functions needed + * for populating rq, ttu and dlg param structures and populate it. + * @dc : struct dc + * @context : dc_state provides a handle to selectively populate pipe_ctx + * @out_new_hw_state: To obtain and populate the rq, dlg and ttu regs in + * out_new_hw_state->pipe_ctx with DML outputs. + * @in_ctx : This structure has the pointer to display_mode_lib_st. + * @pipe_cnt : DML functions to obtain RQ, TTu and DLG params need a pipe_index. + * This helps provide pipe_index in the pipe_cnt loop. + * + * + * Return: None. + */ +void dml2_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_state *context, struct resource_context *out_new_hw_state, struct dml2_context *in_ctx, unsigned int pipe_cnt); + +/* + * dml2_apply_det_buffer_allocation_policy - This function will determine the DET Buffer size + * and return the number of streams. + * @dml2 : Handle for dml2 context + * @dml_dispcfg : dml_dispcfg is the DML2 struct representing the current display config + * Return : None. + */ +void dml2_apply_det_buffer_allocation_policy(struct dml2_context *in_ctx, struct dml_display_cfg_st *dml_dispcfg); + +/* + * dml2_verify_det_buffer_configuration - This function will verify if the allocated DET buffer exceeds + * the total available DET size available and outputs a boolean to indicate if recalulation is needed. + * @dml2 : Handle for dml2 context + * @dml_dispcfg : dml_dispcfg is the DML2 struct representing the current display config + * @struct dml2_helper_det_policy_scratch : Pointer to DET helper scratch + * Return : returns true if recalculation is required, false otherwise. + */ +bool dml2_verify_det_buffer_configuration(struct dml2_context *in_ctx, struct dc_state *display_state, struct dml2_helper_det_policy_scratch *det_scratch); + +/* + * dml2_initialize_det_scratch - This function will initialize the DET scratch space as per requirements. + * @dml2 : Handle for dml2 context + * Return : None + */ +void dml2_initialize_det_scratch(struct dml2_context *in_ctx); +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c new file mode 100644 index 000000000000..9deb03a18ccc --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c @@ -0,0 +1,704 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include "display_mode_core.h" +#include "dml2_internal_types.h" +#include "dml2_utils.h" +#include "dml2_policy.h" +#include "dml2_translation_helper.h" +#include "dml2_mall_phantom.h" +#include "dml2_dc_resource_mgmt.h" +#include "dml21_wrapper.h" + +static void initialize_dml2_ip_params(struct dml2_context *dml2, const struct dc *in_dc, struct ip_params_st *out) +{ + if (dml2->config.use_native_soc_bb_construction) + dml2_init_ip_params(dml2, in_dc, out); + else + dml2_translate_ip_params(in_dc, out); +} + +static void initialize_dml2_soc_bbox(struct dml2_context *dml2, const struct dc *in_dc, struct soc_bounding_box_st *out) +{ + if (dml2->config.use_native_soc_bb_construction) + dml2_init_socbb_params(dml2, in_dc, out); + else + dml2_translate_socbb_params(in_dc, out); +} + +static void initialize_dml2_soc_states(struct dml2_context *dml2, + const struct dc *in_dc, const struct soc_bounding_box_st *in_bbox, struct soc_states_st *out) +{ + if (dml2->config.use_native_soc_bb_construction) + dml2_init_soc_states(dml2, in_dc, in_bbox, out); + else + dml2_translate_soc_states(in_dc, out, in_dc->dml.soc.num_states); +} + +static void map_hw_resources(struct dml2_context *dml2, + struct dml_display_cfg_st *in_out_display_cfg, struct dml_mode_support_info_st *mode_support_info) +{ + unsigned int num_pipes = 0; + int i, j; + + for (i = 0; i < __DML_NUM_PLANES__; i++) { + in_out_display_cfg->hw.ODMMode[i] = mode_support_info->ODMMode[i]; + in_out_display_cfg->hw.DPPPerSurface[i] = mode_support_info->DPPPerSurface[i]; + in_out_display_cfg->hw.DSCEnabled[i] = mode_support_info->DSCEnabled[i]; + in_out_display_cfg->hw.NumberOfDSCSlices[i] = mode_support_info->NumberOfDSCSlices[i]; + in_out_display_cfg->hw.DLGRefClkFreqMHz = 24; + if (dml2->v20.dml_core_ctx.project != dml_project_dcn35 && + dml2->v20.dml_core_ctx.project != dml_project_dcn36 && + dml2->v20.dml_core_ctx.project != dml_project_dcn351) { + /*dGPU default as 50Mhz*/ + in_out_display_cfg->hw.DLGRefClkFreqMHz = 50; + } + for (j = 0; j < mode_support_info->DPPPerSurface[i]; j++) { + if (i >= __DML2_WRAPPER_MAX_STREAMS_PLANES__) { + dml_print("DML::%s: Index out of bounds: i=%d, __DML2_WRAPPER_MAX_STREAMS_PLANES__=%d\n", + __func__, i, __DML2_WRAPPER_MAX_STREAMS_PLANES__); + break; + } + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id[num_pipes] = dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_stream_id[i]; + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_stream_id_valid[num_pipes] = true; + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id[num_pipes] = dml2->v20.scratch.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[i]; + dml2->v20.scratch.dml_to_dc_pipe_mapping.dml_pipe_idx_to_plane_id_valid[num_pipes] = true; + num_pipes++; + } + } +} + +static unsigned int pack_and_call_dml_mode_support_ex(struct dml2_context *dml2, + const struct dml_display_cfg_st *display_cfg, + struct dml_mode_support_info_st *evaluation_info, + enum dc_validate_mode validate_mode) +{ + struct dml2_wrapper_scratch *s = &dml2->v20.scratch; + + s->mode_support_params.mode_lib = &dml2->v20.dml_core_ctx; + s->mode_support_params.in_display_cfg = display_cfg; + if (validate_mode == DC_VALIDATE_MODE_ONLY) + s->mode_support_params.in_start_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; + else + s->mode_support_params.in_start_state_idx = 0; + s->mode_support_params.out_evaluation_info = evaluation_info; + + memset(evaluation_info, 0, sizeof(struct dml_mode_support_info_st)); + s->mode_support_params.out_lowest_state_idx = 0; + + return dml_mode_support_ex(&s->mode_support_params); +} + +static bool optimize_configuration(struct dml2_context *dml2, struct dml2_wrapper_optimize_configuration_params *p) +{ + int unused_dpps = p->ip_params->max_num_dpp; + int i; + int odms_needed; + int largest_blend_and_timing = 0; + bool optimization_done = false; + + for (i = 0; i < (int) p->cur_display_config->num_timings; i++) { + if (p->cur_display_config->plane.BlendingAndTiming[i] > largest_blend_and_timing) + largest_blend_and_timing = p->cur_display_config->plane.BlendingAndTiming[i]; + } + + if (p->new_policy != p->cur_policy) + *p->new_policy = *p->cur_policy; + + if (p->new_display_config != p->cur_display_config) + *p->new_display_config = *p->cur_display_config; + + + // Optimize Clocks + if (!optimization_done) { + if (largest_blend_and_timing == 0 && p->cur_policy->ODMUse[0] == dml_odm_use_policy_combine_as_needed && dml2->config.minimize_dispclk_using_odm) { + odms_needed = dml2_util_get_maximum_odm_combine_for_output(dml2->config.optimize_odm_4to1, + p->cur_display_config->output.OutputEncoder[0], p->cur_mode_support_info->DSCEnabled[0]) - 1; + + if (odms_needed <= unused_dpps) { + if (odms_needed == 1) { + p->new_policy->ODMUse[0] = dml_odm_use_policy_combine_2to1; + optimization_done = true; + } else if (odms_needed == 3) { + p->new_policy->ODMUse[0] = dml_odm_use_policy_combine_4to1; + optimization_done = true; + } else + optimization_done = false; + } + } + } + + return optimization_done; +} + +static int calculate_lowest_supported_state_for_temp_read(struct dml2_context *dml2, struct dc_state *display_state, + enum dc_validate_mode validate_mode) +{ + struct dml2_calculate_lowest_supported_state_for_temp_read_scratch *s = &dml2->v20.scratch.dml2_calculate_lowest_supported_state_for_temp_read_scratch; + struct dml2_wrapper_scratch *s_global = &dml2->v20.scratch; + + unsigned int dml_result = 0; + int result = -1, i, j; + + build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); + + /* Zero out before each call before proceeding */ + memset(s, 0, sizeof(struct dml2_calculate_lowest_supported_state_for_temp_read_scratch)); + memset(&s_global->mode_support_params, 0, sizeof(struct dml_mode_support_ex_params_st)); + memset(&s_global->dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); + + for (i = 0; i < dml2->config.dcn_pipe_count; i++) { + /* Calling resource_build_scaling_params will populate the pipe params + * with the necessary information needed for correct DML calculations + * This is also done in DML1 driver code path and hence display_state + * cannot be const. + */ + struct pipe_ctx *pipe = &display_state->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state) { + if (!dml2->config.callbacks.build_scaling_params(pipe)) { + ASSERT(false); + return false; + } + } + } + + map_dc_state_into_dml_display_cfg(dml2, display_state, &s->cur_display_config); + + for (i = 0; i < dml2->v20.dml_core_ctx.states.num_states; i++) { + s->uclk_change_latencies[i] = dml2->v20.dml_core_ctx.states.state_array[i].dram_clock_change_latency_us; + } + + for (i = 0; i < 4; i++) { + for (j = 0; j < dml2->v20.dml_core_ctx.states.num_states; j++) { + dml2->v20.dml_core_ctx.states.state_array[j].dram_clock_change_latency_us = s_global->dummy_pstate_table[i].dummy_pstate_latency_us; + } + + dml_result = pack_and_call_dml_mode_support_ex(dml2, &s->cur_display_config, &s->evaluation_info, + validate_mode); + + if (dml_result && s->evaluation_info.DRAMClockChangeSupport[0] == dml_dram_clock_change_vactive) { + map_hw_resources(dml2, &s->cur_display_config, &s->evaluation_info); + dml_result = dml_mode_programming(&dml2->v20.dml_core_ctx, s_global->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); + + ASSERT(dml_result); + + dml2_extract_watermark_set(&dml2->v20.g6_temp_read_watermark_set, &dml2->v20.dml_core_ctx); + dml2->v20.g6_temp_read_watermark_set.cstate_pstate.fclk_pstate_change_ns = dml2->v20.g6_temp_read_watermark_set.cstate_pstate.pstate_change_ns; + + result = s_global->mode_support_params.out_lowest_state_idx; + + while (dml2->v20.dml_core_ctx.states.state_array[result].dram_speed_mts < s_global->dummy_pstate_table[i].dram_speed_mts) + result++; + + break; + } + } + + for (i = 0; i < dml2->v20.dml_core_ctx.states.num_states; i++) { + dml2->v20.dml_core_ctx.states.state_array[i].dram_clock_change_latency_us = s->uclk_change_latencies[i]; + } + + return result; +} + +static void copy_dummy_pstate_table(struct dummy_pstate_entry *dest, struct dummy_pstate_entry *src, unsigned int num_entries) +{ + for (int i = 0; i < num_entries; i++) { + dest[i] = src[i]; + } +} + +static bool are_timings_requiring_odm_doing_blending(const struct dml_display_cfg_st *display_cfg, + const struct dml_mode_support_info_st *evaluation_info) +{ + unsigned int planes_per_timing[__DML_NUM_PLANES__] = {0}; + int i; + + for (i = 0; i < display_cfg->num_surfaces; i++) + planes_per_timing[display_cfg->plane.BlendingAndTiming[i]]++; + + for (i = 0; i < __DML_NUM_PLANES__; i++) { + if (planes_per_timing[i] > 1 && evaluation_info->ODMMode[i] != dml_odm_mode_bypass) + return true; + } + + return false; +} + +static bool does_configuration_meet_sw_policies(struct dml2_context *ctx, const struct dml_display_cfg_st *display_cfg, + const struct dml_mode_support_info_st *evaluation_info) +{ + bool pass = true; + + if (!ctx->config.enable_windowed_mpo_odm) { + if (are_timings_requiring_odm_doing_blending(display_cfg, evaluation_info)) + pass = false; + } + + return pass; +} + +static bool dml_mode_support_wrapper(struct dml2_context *dml2, + struct dc_state *display_state, + enum dc_validate_mode validate_mode) +{ + struct dml2_wrapper_scratch *s = &dml2->v20.scratch; + unsigned int result = 0, i; + unsigned int optimized_result = true; + + build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); + + /* Zero out before each call before proceeding */ + memset(&s->cur_display_config, 0, sizeof(struct dml_display_cfg_st)); + memset(&s->mode_support_params, 0, sizeof(struct dml_mode_support_ex_params_st)); + memset(&s->dml_to_dc_pipe_mapping, 0, sizeof(struct dml2_dml_to_dc_pipe_mapping)); + memset(&s->optimize_configuration_params, 0, sizeof(struct dml2_wrapper_optimize_configuration_params)); + + for (i = 0; i < dml2->config.dcn_pipe_count; i++) { + /* Calling resource_build_scaling_params will populate the pipe params + * with the necessary information needed for correct DML calculations + * This is also done in DML1 driver code path and hence display_state + * cannot be const. + */ + struct pipe_ctx *pipe = &display_state->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state) { + if (!dml2->config.callbacks.build_scaling_params(pipe)) { + ASSERT(false); + return false; + } + } + } + + map_dc_state_into_dml_display_cfg(dml2, display_state, &s->cur_display_config); + if (!dml2->config.skip_hw_state_mapping) + dml2_apply_det_buffer_allocation_policy(dml2, &s->cur_display_config); + + result = pack_and_call_dml_mode_support_ex(dml2, + &s->cur_display_config, + &s->mode_support_info, + validate_mode); + + if (result) + result = does_configuration_meet_sw_policies(dml2, &s->cur_display_config, &s->mode_support_info); + + // Try to optimize + if (result) { + s->cur_policy = dml2->v20.dml_core_ctx.policy; + s->optimize_configuration_params.dml_core_ctx = &dml2->v20.dml_core_ctx; + s->optimize_configuration_params.config = &dml2->config; + s->optimize_configuration_params.ip_params = &dml2->v20.dml_core_ctx.ip; + s->optimize_configuration_params.cur_display_config = &s->cur_display_config; + s->optimize_configuration_params.cur_mode_support_info = &s->mode_support_info; + s->optimize_configuration_params.cur_policy = &s->cur_policy; + s->optimize_configuration_params.new_display_config = &s->new_display_config; + s->optimize_configuration_params.new_policy = &s->new_policy; + + while (optimized_result && optimize_configuration(dml2, &s->optimize_configuration_params)) { + dml2->v20.dml_core_ctx.policy = s->new_policy; + optimized_result = pack_and_call_dml_mode_support_ex(dml2, + &s->new_display_config, + &s->mode_support_info, + validate_mode); + + if (optimized_result) + optimized_result = does_configuration_meet_sw_policies(dml2, &s->new_display_config, &s->mode_support_info); + + // If the new optimized state is supposed, then set current = new + if (optimized_result) { + s->cur_display_config = s->new_display_config; + s->cur_policy = s->new_policy; + } else { + // Else, restore policy to current + dml2->v20.dml_core_ctx.policy = s->cur_policy; + } + } + + // Optimize ended with a failed config, so we need to restore DML state to last passing + if (!optimized_result) { + result = pack_and_call_dml_mode_support_ex(dml2, + &s->cur_display_config, + &s->mode_support_info, + validate_mode); + } + } + + if (result) + map_hw_resources(dml2, &s->cur_display_config, &s->mode_support_info); + + return result; +} + +static bool call_dml_mode_support_and_programming(struct dc_state *context, enum dc_validate_mode validate_mode) +{ + unsigned int result = 0; + unsigned int min_state = 0; + int min_state_for_g6_temp_read = 0; + + + if (!context) + return false; + + struct dml2_context *dml2 = context->bw_ctx.dml2; + struct dml2_wrapper_scratch *s = &dml2->v20.scratch; + + if (!context->streams[0]->sink->link->dc->caps.is_apu) { + min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context, + validate_mode); + + ASSERT(min_state_for_g6_temp_read >= 0); + } + + result = dml_mode_support_wrapper(dml2, context, validate_mode); + + /* Upon trying to sett certain frequencies in FRL, min_state_for_g6_temp_read is reported as -1. This leads to an invalid value of min_state causing crashes later on. + * Use the default logic for min_state only when min_state_for_g6_temp_read is a valid value. In other cases, use the value calculated by the DML directly. + */ + if (!context->streams[0]->sink->link->dc->caps.is_apu) { + if (min_state_for_g6_temp_read >= 0) + min_state = min_state_for_g6_temp_read > s->mode_support_params.out_lowest_state_idx ? min_state_for_g6_temp_read : s->mode_support_params.out_lowest_state_idx; + else + min_state = s->mode_support_params.out_lowest_state_idx; + } + + if (result) { + if (!context->streams[0]->sink->link->dc->caps.is_apu) { + result = dml_mode_programming(&dml2->v20.dml_core_ctx, min_state, &s->cur_display_config, true); + } else { + result = dml_mode_programming(&dml2->v20.dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true); + } + } + return result; +} + +static bool dml2_validate_and_build_resource(const struct dc *in_dc, struct dc_state *context, + enum dc_validate_mode validate_mode) +{ + struct dml2_context *dml2 = context->bw_ctx.dml2; + struct dml2_wrapper_scratch *s = &dml2->v20.scratch; + struct dml2_dcn_clocks out_clks; + unsigned int result = 0; + bool need_recalculation = false; + uint32_t cstate_enter_plus_exit_z8_ns; + + if (context->stream_count == 0) { + unsigned int lowest_state_idx = 0; + + out_clks.p_state_supported = true; + out_clks.dispclk_khz = 0; /* No requirement, and lowest index will generally be maximum dispclk. */ + out_clks.dcfclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dcfclk_mhz * 1000; + out_clks.fclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].fabricclk_mhz * 1000; + out_clks.uclk_mts = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dram_speed_mts; + out_clks.phyclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].phyclk_mhz * 1000; + out_clks.socclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].socclk_mhz * 1000; + out_clks.ref_dtbclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dtbclk_mhz * 1000; + context->bw_ctx.bw.dcn.clk.dtbclk_en = false; + dml2_copy_clocks_to_dc_state(&out_clks, context); + return true; + } + + /* Zero out before each call before proceeding */ + memset(&dml2->v20.scratch, 0, sizeof(struct dml2_wrapper_scratch)); + memset(&dml2->v20.dml_core_ctx.policy, 0, sizeof(struct dml_mode_eval_policy_st)); + memset(&dml2->v20.dml_core_ctx.ms, 0, sizeof(struct mode_support_st)); + memset(&dml2->v20.dml_core_ctx.mp, 0, sizeof(struct mode_program_st)); + + /* Initialize DET scratch */ + dml2_initialize_det_scratch(dml2); + + copy_dummy_pstate_table(s->dummy_pstate_table, in_dc->clk_mgr->bw_params->dummy_pstate_table, 4); + + result = call_dml_mode_support_and_programming(context, validate_mode); + /* Call map dc pipes to map the pipes based on the DML output. For correctly determining if recalculation + * is required or not, the resource context needs to correctly reflect the number of active pipes. We would + * only know the correct number if active pipes after dml2_map_dc_pipes is called. + */ + if (result && !dml2->config.skip_hw_state_mapping) + dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); + + /* Verify and update DET Buffer configuration if needed. dml2_verify_det_buffer_configuration will check if DET Buffer + * size needs to be updated. If yes it will update the DETOverride variable and set need_recalculation flag to true. + * Based on that flag, run mode support again. Verification needs to be run after dml_mode_programming because the getters + * return correct det buffer values only after dml_mode_programming is called. + */ + if (result && !dml2->config.skip_hw_state_mapping) { + need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); + if (need_recalculation) { + /* Engage the DML again if recalculation is required. */ + call_dml_mode_support_and_programming(context, validate_mode); + if (!dml2->config.skip_hw_state_mapping) { + dml2_map_dc_pipes(dml2, context, &s->cur_display_config, &s->dml_to_dc_pipe_mapping, in_dc->current_state); + } + need_recalculation = dml2_verify_det_buffer_configuration(dml2, context, &dml2->det_helper_scratch); + ASSERT(need_recalculation == false); + } + } + + if (result) { + unsigned int lowest_state_idx = s->mode_support_params.out_lowest_state_idx; + out_clks.dispclk_khz = (unsigned int)dml2->v20.dml_core_ctx.mp.Dispclk_calculated * 1000; + out_clks.p_state_supported = s->mode_support_info.DRAMClockChangeSupport[0] != dml_dram_clock_change_unsupported; + if (in_dc->config.use_default_clock_table && + (lowest_state_idx < dml2->v20.dml_core_ctx.states.num_states - 1)) { + lowest_state_idx = dml2->v20.dml_core_ctx.states.num_states - 1; + out_clks.dispclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dispclk_mhz * 1000; + } + + out_clks.dcfclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dcfclk_mhz * 1000; + out_clks.fclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].fabricclk_mhz * 1000; + out_clks.uclk_mts = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dram_speed_mts; + out_clks.phyclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].phyclk_mhz * 1000; + out_clks.socclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].socclk_mhz * 1000; + out_clks.ref_dtbclk_khz = (unsigned int)dml2->v20.dml_core_ctx.states.state_array[lowest_state_idx].dtbclk_mhz * 1000; + context->bw_ctx.bw.dcn.clk.dtbclk_en = is_dtbclk_required(in_dc, context); + + if (!dml2->config.skip_hw_state_mapping) { + /* Call dml2_calculate_rq_and_dlg_params */ + dml2_calculate_rq_and_dlg_params(in_dc, context, &context->res_ctx, dml2, in_dc->res_pool->pipe_count); + } + + dml2_copy_clocks_to_dc_state(&out_clks, context); + dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.a, &dml2->v20.dml_core_ctx); + dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.b, &dml2->v20.dml_core_ctx); + if (context->streams[0]->sink->link->dc->caps.is_apu) + dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.dml_core_ctx); + else + memcpy(&context->bw_ctx.bw.dcn.watermarks.c, &dml2->v20.g6_temp_read_watermark_set, sizeof(context->bw_ctx.bw.dcn.watermarks.c)); + dml2_extract_watermark_set(&context->bw_ctx.bw.dcn.watermarks.d, &dml2->v20.dml_core_ctx); + dml2_extract_writeback_wm(context, &dml2->v20.dml_core_ctx); + //copy for deciding zstate use + context->bw_ctx.dml.vba.StutterPeriod = context->bw_ctx.dml2->v20.dml_core_ctx.mp.StutterPeriod; + + cstate_enter_plus_exit_z8_ns = context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_z8_ns; + + if (context->bw_ctx.dml.vba.StutterPeriod < in_dc->debug.minimum_z8_residency_time && + cstate_enter_plus_exit_z8_ns < in_dc->debug.minimum_z8_residency_time * 1000) + cstate_enter_plus_exit_z8_ns = in_dc->debug.minimum_z8_residency_time * 1000; + + context->bw_ctx.bw.dcn.watermarks.a.cstate_pstate.cstate_enter_plus_exit_z8_ns = cstate_enter_plus_exit_z8_ns; + } + + return result; +} + +static bool dml2_validate_only(struct dc_state *context, enum dc_validate_mode validate_mode) +{ + struct dml2_context *dml2; + unsigned int result = 0; + + if (!context || context->stream_count == 0) + return true; + + dml2 = context->bw_ctx.dml2; + + /* Zero out before each call before proceeding */ + memset(&dml2->v20.scratch, 0, sizeof(struct dml2_wrapper_scratch)); + memset(&dml2->v20.dml_core_ctx.policy, 0, sizeof(struct dml_mode_eval_policy_st)); + memset(&dml2->v20.dml_core_ctx.ms, 0, sizeof(struct mode_support_st)); + memset(&dml2->v20.dml_core_ctx.mp, 0, sizeof(struct mode_program_st)); + + build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy); + + map_dc_state_into_dml_display_cfg(dml2, context, &dml2->v20.scratch.cur_display_config); + if (!dml2->config.skip_hw_state_mapping) + dml2_apply_det_buffer_allocation_policy(dml2, &dml2->v20.scratch.cur_display_config); + + result = pack_and_call_dml_mode_support_ex(dml2, + &dml2->v20.scratch.cur_display_config, + &dml2->v20.scratch.mode_support_info, + validate_mode); + + if (result) + result = does_configuration_meet_sw_policies(dml2, &dml2->v20.scratch.cur_display_config, &dml2->v20.scratch.mode_support_info); + + return result == 1; +} + +static void dml2_apply_debug_options(const struct dc *dc, struct dml2_context *dml2) +{ + if (dc->debug.override_odm_optimization) { + dml2->config.minimize_dispclk_using_odm = dc->debug.minimize_dispclk_using_odm; + } +} + +bool dml2_validate(const struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2, + enum dc_validate_mode validate_mode) +{ + bool out = false; + + if (!dml2) + return false; + dml2_apply_debug_options(in_dc, dml2); + + /* DML2.1 validation path */ + if (dml2->architecture == dml2_architecture_21) { + out = dml21_validate(in_dc, context, dml2, validate_mode); + return out; + } + + DC_FP_START(); + + /* Use dml_validate_only for DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX path */ + if (validate_mode != DC_VALIDATE_MODE_AND_PROGRAMMING) + out = dml2_validate_only(context, validate_mode); + else + out = dml2_validate_and_build_resource(in_dc, context, validate_mode); + + DC_FP_END(); + + return out; +} + +static inline struct dml2_context *dml2_allocate_memory(void) +{ + return (struct dml2_context *) vzalloc(sizeof(struct dml2_context)); +} + +static void dml2_init(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) +{ + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); + return; + } + + // Store config options + (*dml2)->config = *config; + + switch (in_dc->ctx->dce_version) { + case DCN_VERSION_3_5: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn35; + break; + case DCN_VERSION_3_51: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn351; + break; + case DCN_VERSION_3_6: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn36; + break; + case DCN_VERSION_3_2: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn32; + break; + case DCN_VERSION_3_21: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn321; + break; + case DCN_VERSION_4_01: + (*dml2)->v20.dml_core_ctx.project = dml_project_dcn401; + break; + default: + (*dml2)->v20.dml_core_ctx.project = dml_project_default; + break; + } + + DC_FP_START(); + + initialize_dml2_ip_params(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.ip); + + initialize_dml2_soc_bbox(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.soc); + + initialize_dml2_soc_states(*dml2, in_dc, &(*dml2)->v20.dml_core_ctx.soc, &(*dml2)->v20.dml_core_ctx.states); + + DC_FP_END(); +} + +bool dml2_create(const struct dc *in_dc, const struct dml2_configuration_options *config, struct dml2_context **dml2) +{ + // TODO : Temporarily add DCN_VERSION_3_2 for N-1 validation. Remove DCN_VERSION_3_2 after N-1 validation phase is complete. + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) + return dml21_create(in_dc, dml2, config); + + // Allocate Mode Lib Ctx + *dml2 = dml2_allocate_memory(); + + if (!(*dml2)) + return false; + + dml2_init(in_dc, config, dml2); + + return true; +} + +void dml2_destroy(struct dml2_context *dml2) +{ + if (!dml2) + return; + + if (dml2->architecture == dml2_architecture_21) + dml21_destroy(dml2); + vfree(dml2); +} + +void dml2_extract_dram_and_fclk_change_support(struct dml2_context *dml2, + unsigned int *fclk_change_support, unsigned int *dram_clk_change_support) +{ + *fclk_change_support = (unsigned int) dml2->v20.dml_core_ctx.ms.support.FCLKChangeSupport[0]; + *dram_clk_change_support = (unsigned int) dml2->v20.dml_core_ctx.ms.support.DRAMClockChangeSupport[0]; +} + +void dml2_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2) +{ + if (dml2->architecture == dml2_architecture_21) + dml21_prepare_mcache_programming(in_dc, context, dml2); +} + +void dml2_copy(struct dml2_context *dst_dml2, + struct dml2_context *src_dml2) +{ + if (src_dml2->architecture == dml2_architecture_21) { + dml21_copy(dst_dml2, src_dml2); + return; + } + /* copy Mode Lib Ctx */ + memcpy(dst_dml2, src_dml2, sizeof(struct dml2_context)); +} + +bool dml2_create_copy(struct dml2_context **dst_dml2, + struct dml2_context *src_dml2) +{ + if (src_dml2->architecture == dml2_architecture_21) + return dml21_create_copy(dst_dml2, src_dml2); + /* Allocate Mode Lib Ctx */ + *dst_dml2 = dml2_allocate_memory(); + + if (!(*dst_dml2)) + return false; + + /* copy Mode Lib Ctx */ + dml2_copy(*dst_dml2, src_dml2); + + return true; +} + +void dml2_reinit(const struct dc *in_dc, + const struct dml2_configuration_options *config, + struct dml2_context **dml2) +{ + if ((in_dc->debug.using_dml21) && (in_dc->ctx->dce_version >= DCN_VERSION_4_01)) { + dml21_reinit(in_dc, *dml2, config); + return; + } + + dml2_init(in_dc, config, dml2); +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h new file mode 100644 index 000000000000..c384e141cebc --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h @@ -0,0 +1,309 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef _DML2_WRAPPER_H_ +#define _DML2_WRAPPER_H_ + +#include "os_types.h" + +#define DML2_MAX_NUM_DPM_LVL 30 + +struct dml2_context; +struct display_mode_lib_st; +struct dc; +struct pipe_ctx; +struct dc_plane_state; +struct dc_sink; +struct dc_stream_state; +struct resource_context; +struct display_stream_compressor; +struct dc_mcache_params; + +// Configuration of the MALL on the SoC +struct dml2_soc_mall_info { + // Cache line size of 0 means MALL is not enabled/present + unsigned int cache_line_size_bytes; + unsigned int cache_num_ways; + unsigned int max_cab_allocation_bytes; + + unsigned int mblk_width_pixels; + unsigned int mblk_size_bytes; + unsigned int mblk_height_4bpe_pixels; + unsigned int mblk_height_8bpe_pixels; +}; + +// Output of DML2 for clock requirements +struct dml2_dcn_clocks { + unsigned int dispclk_khz; + unsigned int dcfclk_khz; + unsigned int fclk_khz; + unsigned int uclk_mts; + unsigned int phyclk_khz; + unsigned int socclk_khz; + unsigned int ref_dtbclk_khz; + bool p_state_supported; + unsigned int cab_num_ways_required; + unsigned int dcfclk_khz_ds; +}; + +struct dml2_dc_callbacks { + struct dc *dc; + bool (*build_scaling_params)(struct pipe_ctx *pipe_ctx); + void (*build_test_pattern_params)(struct resource_context *res_ctx, struct pipe_ctx *otg_master); + bool (*can_support_mclk_switch_using_fw_based_vblank_stretch)(struct dc *dc, struct dc_state *context); + bool (*acquire_secondary_pipe_for_mpc_odm)(const struct dc *dc, struct dc_state *state, struct pipe_ctx *pri_pipe, struct pipe_ctx *sec_pipe, bool odm); + bool (*update_pipes_for_stream_with_slice_count)( + struct dc_state *new_ctx, + const struct dc_state *cur_ctx, + const struct resource_pool *pool, + const struct dc_stream_state *stream, + int new_slice_count); + bool (*update_pipes_for_plane_with_slice_count)( + struct dc_state *new_ctx, + const struct dc_state *cur_ctx, + const struct resource_pool *pool, + const struct dc_plane_state *plane, + int slice_count); + int (*get_odm_slice_index)(const struct pipe_ctx *opp_head); + int (*get_odm_slice_count)(const struct pipe_ctx *opp_head); + int (*get_mpc_slice_index)(const struct pipe_ctx *dpp_pipe); + int (*get_mpc_slice_count)(const struct pipe_ctx *dpp_pipe); + struct pipe_ctx *(*get_opp_head)(const struct pipe_ctx *pipe_ctx); + struct pipe_ctx *(*get_otg_master_for_stream)( + struct resource_context *res_ctx, + const struct dc_stream_state *stream); + int (*get_opp_heads_for_otg_master)(const struct pipe_ctx *otg_master, + struct resource_context *res_ctx, + struct pipe_ctx *opp_heads[MAX_PIPES]); + int (*get_dpp_pipes_for_plane)(const struct dc_plane_state *plane, + struct resource_context *res_ctx, + struct pipe_ctx *dpp_pipes[MAX_PIPES]); + struct dc_stream_status *(*get_stream_status)( + struct dc_state *state, + const struct dc_stream_state *stream); + struct dc_stream_state *(*get_stream_from_id)(const struct dc_state *state, unsigned int id); + unsigned int (*get_max_flickerless_instant_vtotal_increase)( + struct dc_stream_state *stream, + bool is_gaming); + bool (*allocate_mcache)(struct dc_state *context, const struct dc_mcache_params *mcache_params); +}; + +struct dml2_dc_svp_callbacks { + struct dc *dc; + bool (*build_scaling_params)(struct pipe_ctx *pipe_ctx); + struct dc_stream_state* (*create_phantom_stream)(const struct dc *dc, + struct dc_state *state, + struct dc_stream_state *main_stream); + struct dc_plane_state* (*create_phantom_plane)(const struct dc *dc, + struct dc_state *state, + struct dc_plane_state *main_plane); + enum dc_status (*add_phantom_stream)(const struct dc *dc, + struct dc_state *state, + struct dc_stream_state *phantom_stream, + struct dc_stream_state *main_stream); + bool (*add_phantom_plane)(const struct dc *dc, struct dc_stream_state *stream, struct dc_plane_state *plane_state, struct dc_state *context); + bool (*remove_phantom_plane)(const struct dc *dc, + struct dc_stream_state *stream, + struct dc_plane_state *plane_state, + struct dc_state *context); + enum dc_status (*remove_phantom_stream)(const struct dc *dc, + struct dc_state *state, + struct dc_stream_state *stream); + void (*release_phantom_plane)(const struct dc *dc, + struct dc_state *state, + struct dc_plane_state *plane); + void (*release_phantom_stream)(const struct dc *dc, + struct dc_state *state, + struct dc_stream_state *stream); + void (*release_dsc)(struct resource_context *res_ctx, const struct resource_pool *pool, struct display_stream_compressor **dsc); + enum mall_stream_type (*get_pipe_subvp_type)(const struct dc_state *state, const struct pipe_ctx *pipe_ctx); + enum mall_stream_type (*get_stream_subvp_type)(const struct dc_state *state, const struct dc_stream_state *stream); + struct dc_stream_state *(*get_paired_subvp_stream)(const struct dc_state *state, const struct dc_stream_state *stream); + bool (*remove_phantom_streams_and_planes)( + const struct dc *dc, + struct dc_state *state); + void (*release_phantom_streams_and_planes)( + const struct dc *dc, + struct dc_state *state); + unsigned int (*calculate_mall_ways_from_bytes)( + const struct dc *dc, + unsigned int total_size_in_mall_bytes); +}; + +struct dml2_clks_table_entry { + unsigned int dcfclk_mhz; + unsigned int fclk_mhz; + unsigned int memclk_mhz; + unsigned int socclk_mhz; + unsigned int dtbclk_mhz; + unsigned int dispclk_mhz; + unsigned int dppclk_mhz; + unsigned int dram_speed_mts; /*which is based on wck_ratio*/ +}; + +struct dml2_clks_num_entries { + unsigned int num_dcfclk_levels; + unsigned int num_fclk_levels; + unsigned int num_memclk_levels; + unsigned int num_socclk_levels; + unsigned int num_dtbclk_levels; + unsigned int num_dispclk_levels; + unsigned int num_dppclk_levels; +}; + +struct dml2_clks_limit_table { + struct dml2_clks_table_entry clk_entries[DML2_MAX_NUM_DPM_LVL]; + struct dml2_clks_num_entries num_entries_per_clk; + unsigned int num_states; +}; + +// Various overrides, per ASIC or per SKU specific, or for debugging purpose when/if available +struct dml2_soc_bbox_overrides { + double xtalclk_mhz; + double dchub_refclk_mhz; + double dprefclk_mhz; + double disp_pll_vco_speed_mhz; + double urgent_latency_us; + double sr_exit_latency_us; + double sr_enter_plus_exit_latency_us; + double sr_exit_z8_time_us; + double sr_enter_plus_exit_z8_time_us; + double dram_clock_change_latency_us; + double fclk_change_latency_us; + unsigned int dram_num_chan; + unsigned int dram_chanel_width_bytes; + struct dml2_clks_limit_table clks_table; +}; + +enum dml2_force_pstate_methods { + dml2_force_pstate_method_auto = 0, + dml2_force_pstate_method_vactive, + dml2_force_pstate_method_vblank, + dml2_force_pstate_method_drr, + dml2_force_pstate_method_subvp, +}; + +struct dml2_configuration_options { + int dcn_pipe_count; + bool use_native_pstate_optimization; + bool enable_windowed_mpo_odm; + bool use_native_soc_bb_construction; + bool skip_hw_state_mapping; + bool optimize_odm_4to1; + bool minimize_dispclk_using_odm; + bool override_det_buffer_size_kbytes; + struct dml2_dc_callbacks callbacks; + struct { + bool force_disable_subvp; + bool force_enable_subvp; + unsigned int subvp_fw_processing_delay_us; + unsigned int subvp_pstate_allow_width_us; + unsigned int subvp_prefetch_end_to_mall_start_us; + unsigned int subvp_swath_height_margin_lines; + struct dml2_dc_svp_callbacks callbacks; + } svp_pstate; + struct dml2_soc_mall_info mall_cfg; + struct dml2_soc_bbox_overrides bbox_overrides; + unsigned int max_segments_per_hubp; + unsigned int det_segment_size; + /* Only for debugging purposes when initializing SOCBB params via tool for DML21. */ + struct socbb_ip_params_external *external_socbb_ip_params; + struct { + bool force_pstate_method_enable; + enum dml2_force_pstate_methods force_pstate_method_values[MAX_PIPES]; + } pmo; + bool map_dc_pipes_with_callbacks; + + bool use_clock_dc_limits; + bool gpuvm_enable; + bool force_tdlut_enable; + void *bb_from_dmub; +}; + +/* + * dml2_create - Creates dml2_context. + * @in_dc: dc. + * @config: dml2 configuration options. + * @dml2: Created dml2 context. + * + * Create and destroy of DML2 is done as part of dc_state creation + * and dc_state_free. DML2 IP, SOC and STATES are initialized at + * creation time. + * + * Return: True if dml2 is successfully created, false otherwise. + */ +bool dml2_create(const struct dc *in_dc, + const struct dml2_configuration_options *config, + struct dml2_context **dml2); + +void dml2_destroy(struct dml2_context *dml2); +void dml2_copy(struct dml2_context *dst_dml2, + struct dml2_context *src_dml2); +bool dml2_create_copy(struct dml2_context **dst_dml2, + struct dml2_context *src_dml2); +void dml2_reinit(const struct dc *in_dc, + const struct dml2_configuration_options *config, + struct dml2_context **dml2); + +/* + * dml2_validate - Determines if a display configuration is supported or not. + * @in_dc: dc. + * @context: dc_state to be validated. + * @validate_mode: DC_VALIDATE_MODE_ONLY and DC_VALIDATE_MODE_AND_STATE_INDEX will not populate context.res_ctx. + * + * DML1.0 compatible interface for validation. + * + * Based on fast_validate option internally would call: + * + * -dml2_validate_and_build_resource - for non fast_validate option + * Calculates if dc_state can be supported on the SOC, and attempts to + * optimize the power management feature supports versus minimum clocks. + * If supported, also builds out_new_hw_state to represent the hw programming + * for the new dc state. + * + * -dml2_validate_only - for fast_validate option + * Calculates if dc_state can be supported on the SOC (i.e. at maximum + * clocks) with all mandatory power features enabled. + + * Context: Two threads may not invoke this function concurrently unless they reference + * separate dc_states for validation. + * Return: True if mode is supported, false otherwise. + */ +bool dml2_validate(const struct dc *in_dc, + struct dc_state *context, + struct dml2_context *dml2, + enum dc_validate_mode validate_mode); + +/* + * dml2_extract_dram_and_fclk_change_support - Extracts the FCLK and UCLK change support info. + * @dml2: input dml2 context pointer. + * @fclk_change_support: output pointer holding the fclk change support info (vactive, vblank, unsupported). + * @dram_clk_change_support: output pointer holding the uclk change support info (vactive, vblank, unsupported). + */ +void dml2_extract_dram_and_fclk_change_support(struct dml2_context *dml2, + unsigned int *fclk_change_support, unsigned int *dram_clk_change_support); +void dml2_prepare_mcache_programming(struct dc *in_dc, struct dc_state *context, struct dml2_context *dml2); +#endif //_DML2_WRAPPER_H_ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h new file mode 100644 index 000000000000..17f0972b1af7 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h @@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML_ASSERT_H__ +#define __DML_ASSERT_H__ + +#include "os_types.h" + +#endif //__DML_ASSERT_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h new file mode 100644 index 000000000000..d459f93cf40b --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h @@ -0,0 +1,34 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +/* This header intentinally does not include an #ifdef guard as it only contains includes for other headers*/ + +/* + * Standard Types + */ +#include "os_types.h" +#include "cmntypes.h" + diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c new file mode 100644 index 000000000000..00d22e542469 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c @@ -0,0 +1,573 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "dml_display_rq_dlg_calc.h" +#include "display_mode_core.h" +#include "display_mode_util.h" + +static dml_bool_t is_dual_plane(enum dml_source_format_class source_format) +{ + dml_bool_t ret_val = 0; + + if ((source_format == dml_420_12) || (source_format == dml_420_8) || (source_format == dml_420_10) || (source_format == dml_rgbe_alpha)) + ret_val = 1; + + return ret_val; +} + +void dml_rq_dlg_get_rq_reg(dml_display_rq_regs_st *rq_regs, + struct display_mode_lib_st *mode_lib, + const dml_uint_t pipe_idx) +{ + dml_uint_t plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); + enum dml_source_format_class source_format = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[plane_idx]; + enum dml_swizzle_mode sw_mode = mode_lib->ms.cache_display_cfg.surface.SurfaceTiling[plane_idx]; + dml_bool_t dual_plane = is_dual_plane((enum dml_source_format_class)(source_format)); + + uint32 pixel_chunk_bytes = 0; + uint32 min_pixel_chunk_bytes = 0; + uint32 meta_chunk_bytes = 0; + uint32 min_meta_chunk_bytes = 0; + uint32 dpte_group_bytes = 0; + uint32 mpte_group_bytes = 0; + + uint32 p1_pixel_chunk_bytes = 0; + uint32 p1_min_pixel_chunk_bytes = 0; + uint32 p1_meta_chunk_bytes = 0; + uint32 p1_min_meta_chunk_bytes = 0; + uint32 p1_dpte_group_bytes = 0; + uint32 p1_mpte_group_bytes = 0; + + dml_uint_t detile_buf_size_in_bytes; + dml_uint_t detile_buf_plane1_addr = 0; + + dml_float_t stored_swath_l_bytes; + dml_float_t stored_swath_c_bytes; + dml_bool_t is_phantom_pipe; + + dml_uint_t pte_row_height_linear; + + dml_print("DML_DLG::%s: Calculation for pipe[%d] start\n", __func__, pipe_idx); + + memset(rq_regs, 0, sizeof(*rq_regs)); + + pixel_chunk_bytes = (dml_uint_t)(dml_get_pixel_chunk_size_in_kbyte(mode_lib) * 1024); + min_pixel_chunk_bytes = (dml_uint_t)(dml_get_min_pixel_chunk_size_in_byte(mode_lib)); + + if (pixel_chunk_bytes == 64 * 1024) + min_pixel_chunk_bytes = 0; + + meta_chunk_bytes = (dml_uint_t)(dml_get_meta_chunk_size_in_kbyte(mode_lib) * 1024); + min_meta_chunk_bytes = (dml_uint_t)(dml_get_min_meta_chunk_size_in_byte(mode_lib)); + + dpte_group_bytes = (dml_uint_t)(dml_get_dpte_group_size_in_bytes(mode_lib, pipe_idx)); + mpte_group_bytes = (dml_uint_t)(dml_get_vm_group_size_in_bytes(mode_lib, pipe_idx)); + + p1_pixel_chunk_bytes = pixel_chunk_bytes; + p1_min_pixel_chunk_bytes = min_pixel_chunk_bytes; + p1_meta_chunk_bytes = meta_chunk_bytes; + p1_min_meta_chunk_bytes = min_meta_chunk_bytes; + p1_dpte_group_bytes = dpte_group_bytes; + p1_mpte_group_bytes = mpte_group_bytes; + + if (source_format == dml_rgbe_alpha) + p1_pixel_chunk_bytes = (dml_uint_t)(dml_get_alpha_pixel_chunk_size_in_kbyte(mode_lib) * 1024); + + rq_regs->rq_regs_l.chunk_size = (dml_uint_t)(dml_log2((dml_float_t) pixel_chunk_bytes) - 10); + rq_regs->rq_regs_c.chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_pixel_chunk_bytes) - 10); + + if (min_pixel_chunk_bytes == 0) + rq_regs->rq_regs_l.min_chunk_size = 0; + else + rq_regs->rq_regs_l.min_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) min_pixel_chunk_bytes) - 8 + 1); + + if (p1_min_pixel_chunk_bytes == 0) + rq_regs->rq_regs_c.min_chunk_size = 0; + else + rq_regs->rq_regs_c.min_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_min_pixel_chunk_bytes) - 8 + 1); + + rq_regs->rq_regs_l.meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) meta_chunk_bytes) - 10); + rq_regs->rq_regs_c.meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_meta_chunk_bytes) - 10); + + if (min_meta_chunk_bytes == 0) + rq_regs->rq_regs_l.min_meta_chunk_size = 0; + else + rq_regs->rq_regs_l.min_meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) min_meta_chunk_bytes) - 6 + 1); + + if (min_meta_chunk_bytes == 0) + rq_regs->rq_regs_c.min_meta_chunk_size = 0; + else + rq_regs->rq_regs_c.min_meta_chunk_size = (dml_uint_t)(dml_log2((dml_float_t) p1_min_meta_chunk_bytes) - 6 + 1); + + rq_regs->rq_regs_l.dpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) dpte_group_bytes) - 6); + rq_regs->rq_regs_l.mpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) mpte_group_bytes) - 6); + rq_regs->rq_regs_c.dpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) p1_dpte_group_bytes) - 6); + rq_regs->rq_regs_c.mpte_group_size = (dml_uint_t)(dml_log2((dml_float_t) p1_mpte_group_bytes) - 6); + + detile_buf_size_in_bytes = (dml_uint_t)(dml_get_det_buffer_size_kbytes(mode_lib, pipe_idx) * 1024); + + pte_row_height_linear = (dml_uint_t)(dml_get_dpte_row_height_linear_l(mode_lib, pipe_idx)); + + if (sw_mode == dml_sw_linear) + ASSERT(pte_row_height_linear >= 8); + + rq_regs->rq_regs_l.pte_row_height_linear = (dml_uint_t)(dml_floor(dml_log2((dml_float_t) pte_row_height_linear), 1) - 3); + + if (dual_plane) { + dml_uint_t p1_pte_row_height_linear = (dml_uint_t)(dml_get_dpte_row_height_linear_c(mode_lib, pipe_idx)); + if (sw_mode == dml_sw_linear) + ASSERT(p1_pte_row_height_linear >= 8); + + rq_regs->rq_regs_c.pte_row_height_linear = (dml_uint_t)(dml_floor(dml_log2((dml_float_t) p1_pte_row_height_linear), 1) - 3); + } + + rq_regs->rq_regs_l.swath_height = (dml_uint_t)(dml_log2((dml_float_t) dml_get_swath_height_l(mode_lib, pipe_idx))); + rq_regs->rq_regs_c.swath_height = (dml_uint_t)(dml_log2((dml_float_t) dml_get_swath_height_c(mode_lib, pipe_idx))); + + if (pixel_chunk_bytes >= 32 * 1024 || (dual_plane && p1_pixel_chunk_bytes >= 32 * 1024)) { //32kb + rq_regs->drq_expansion_mode = 0; + } else { + rq_regs->drq_expansion_mode = 2; + } + rq_regs->prq_expansion_mode = 1; + rq_regs->mrq_expansion_mode = 1; + rq_regs->crq_expansion_mode = 1; + + stored_swath_l_bytes = dml_get_det_stored_buffer_size_l_bytes(mode_lib, pipe_idx); + stored_swath_c_bytes = dml_get_det_stored_buffer_size_c_bytes(mode_lib, pipe_idx); + is_phantom_pipe = dml_get_is_phantom_pipe(mode_lib, pipe_idx); + + // Note: detile_buf_plane1_addr is in unit of 1KB + if (dual_plane) { + if (is_phantom_pipe) { + detile_buf_plane1_addr = (dml_uint_t)((1024.0*1024.0) / 2.0 / 1024.0); // half to chroma + } else { + if (stored_swath_l_bytes / stored_swath_c_bytes <= 1.5) { + detile_buf_plane1_addr = (dml_uint_t)(detile_buf_size_in_bytes / 2.0 / 1024.0); // half to chroma +#ifdef __DML_VBA_DEBUG__ + dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d (1/2 to chroma)\n", __func__, detile_buf_plane1_addr); +#endif + } else { + detile_buf_plane1_addr = (dml_uint_t)(dml_round_to_multiple((dml_uint_t)((2.0 * detile_buf_size_in_bytes) / 3.0), 1024, 0) / 1024.0); // 2/3 to luma +#ifdef __DML_VBA_DEBUG__ + dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d (1/3 chroma)\n", __func__, detile_buf_plane1_addr); +#endif + } + } + } + rq_regs->plane1_base_address = detile_buf_plane1_addr; + +#ifdef __DML_VBA_DEBUG__ + dml_print("DML_DLG: %s: is_phantom_pipe = %d\n", __func__, is_phantom_pipe); + dml_print("DML_DLG: %s: stored_swath_l_bytes = %f\n", __func__, stored_swath_l_bytes); + dml_print("DML_DLG: %s: stored_swath_c_bytes = %f\n", __func__, stored_swath_c_bytes); + dml_print("DML_DLG: %s: detile_buf_size_in_bytes = %d\n", __func__, detile_buf_size_in_bytes); + dml_print("DML_DLG: %s: detile_buf_plane1_addr = %d\n", __func__, detile_buf_plane1_addr); + dml_print("DML_DLG: %s: plane1_base_address = %d\n", __func__, rq_regs->plane1_base_address); +#endif + dml_print_rq_regs_st(rq_regs); + dml_print("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); +} + +// Note: currently taken in as is. +// Nice to decouple code from hw register implement and extract code that are repeated for luma and chroma. + + +void dml_rq_dlg_get_dlg_reg(dml_display_dlg_regs_st *disp_dlg_regs, + dml_display_ttu_regs_st *disp_ttu_regs, + struct display_mode_lib_st *mode_lib, + const dml_uint_t pipe_idx) +{ + dml_uint_t plane_idx = dml_get_plane_idx(mode_lib, pipe_idx); + enum dml_source_format_class source_format = mode_lib->ms.cache_display_cfg.surface.SourcePixelFormat[plane_idx]; + struct dml_timing_cfg_st *timing = &mode_lib->ms.cache_display_cfg.timing; + struct dml_plane_cfg_st *plane = &mode_lib->ms.cache_display_cfg.plane; + struct dml_hw_resource_st *hw = &mode_lib->ms.cache_display_cfg.hw; + dml_bool_t dual_plane = is_dual_plane(source_format); + dml_uint_t num_cursors = plane->NumberOfCursors[plane_idx]; + enum dml_odm_mode odm_mode = hw->ODMMode[plane_idx]; + + dml_uint_t htotal = timing->HTotal[plane_idx]; + dml_uint_t hactive = timing->HActive[plane_idx]; + dml_uint_t hblank_end = timing->HBlankEnd[plane_idx]; + dml_uint_t vblank_end = timing->VBlankEnd[plane_idx]; + dml_bool_t interlaced = timing->Interlace[plane_idx]; + dml_float_t pclk_freq_in_mhz = (dml_float_t) timing->PixelClock[plane_idx]; + dml_float_t refclk_freq_in_mhz = (hw->DLGRefClkFreqMHz > 0) ? (dml_float_t) hw->DLGRefClkFreqMHz : mode_lib->soc.refclk_mhz; + dml_float_t ref_freq_to_pix_freq = refclk_freq_in_mhz / pclk_freq_in_mhz; + + dml_uint_t vready_after_vcount0; + + dml_uint_t dst_x_after_scaler; + dml_uint_t dst_y_after_scaler; + + dml_float_t dst_y_prefetch; + dml_float_t dst_y_per_vm_vblank; + dml_float_t dst_y_per_row_vblank; + dml_float_t dst_y_per_vm_flip; + dml_float_t dst_y_per_row_flip; + + dml_float_t max_dst_y_per_vm_vblank = 32.0; //U5.2 + dml_float_t max_dst_y_per_row_vblank = 16.0; //U4.2 + + dml_float_t vratio_pre_l; + dml_float_t vratio_pre_c; + + dml_float_t refcyc_per_line_delivery_pre_l; + dml_float_t refcyc_per_line_delivery_l; + dml_float_t refcyc_per_line_delivery_pre_c = 0.; + dml_float_t refcyc_per_line_delivery_c = 0.; + dml_float_t refcyc_per_req_delivery_pre_l; + dml_float_t refcyc_per_req_delivery_l; + dml_float_t refcyc_per_req_delivery_pre_c = 0.; + dml_float_t refcyc_per_req_delivery_c = 0.; + dml_float_t refcyc_per_req_delivery_pre_cur0 = 0.; + dml_float_t refcyc_per_req_delivery_cur0 = 0.; + + dml_float_t dst_y_per_pte_row_nom_l; + dml_float_t dst_y_per_pte_row_nom_c; + dml_float_t dst_y_per_meta_row_nom_l; + dml_float_t dst_y_per_meta_row_nom_c; + dml_float_t refcyc_per_pte_group_nom_l; + dml_float_t refcyc_per_pte_group_nom_c; + dml_float_t refcyc_per_pte_group_vblank_l; + dml_float_t refcyc_per_pte_group_vblank_c; + dml_float_t refcyc_per_pte_group_flip_l; + dml_float_t refcyc_per_pte_group_flip_c; + dml_float_t refcyc_per_meta_chunk_nom_l; + dml_float_t refcyc_per_meta_chunk_nom_c; + dml_float_t refcyc_per_meta_chunk_vblank_l; + dml_float_t refcyc_per_meta_chunk_vblank_c; + dml_float_t refcyc_per_meta_chunk_flip_l; + dml_float_t refcyc_per_meta_chunk_flip_c; + + dml_float_t temp; + dml_float_t min_ttu_vblank; + dml_uint_t min_dst_y_next_start; + + dml_print("DML_DLG::%s: Calculation for pipe_idx=%d\n", __func__, pipe_idx); + dml_print("DML_DLG::%s: plane_idx = %d\n", __func__, plane_idx); + dml_print("DML_DLG: %s: htotal = %d\n", __func__, htotal); + dml_print("DML_DLG: %s: refclk_freq_in_mhz = %3.2f\n", __func__, refclk_freq_in_mhz); + dml_print("DML_DLG: %s: hw->DLGRefClkFreqMHz = %3.2f\n", __func__, hw->DLGRefClkFreqMHz); + dml_print("DML_DLG: %s: soc.refclk_mhz = %3.2f\n", __func__, mode_lib->soc.refclk_mhz); + dml_print("DML_DLG: %s: pclk_freq_in_mhz = %3.2f\n", __func__, pclk_freq_in_mhz); + dml_print("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, ref_freq_to_pix_freq); + dml_print("DML_DLG: %s: interlaced = %d\n", __func__, interlaced); + + memset(disp_dlg_regs, 0, sizeof(*disp_dlg_regs)); + memset(disp_ttu_regs, 0, sizeof(*disp_ttu_regs)); + + ASSERT(refclk_freq_in_mhz != 0); + ASSERT(pclk_freq_in_mhz != 0); + ASSERT(ref_freq_to_pix_freq < 4.0); + + // Need to figure out which side of odm combine we're in + // Assume the pipe instance under the same plane is in order + + if (odm_mode == dml_odm_mode_bypass) { + disp_dlg_regs->refcyc_h_blank_end = (dml_uint_t)((dml_float_t) hblank_end * ref_freq_to_pix_freq); + } else if (odm_mode == dml_odm_mode_combine_2to1 || odm_mode == dml_odm_mode_combine_4to1) { + // find out how many pipe are in this plane + dml_uint_t num_active_pipes = dml_get_num_active_pipes(&mode_lib->ms.cache_display_cfg); + dml_uint_t first_pipe_idx_in_plane = __DML_NUM_PLANES__; + dml_uint_t pipe_idx_in_combine = 0; // pipe index within the plane + dml_uint_t odm_combine_factor = (odm_mode == dml_odm_mode_combine_2to1 ? 2 : 4); + + for (dml_uint_t i = 0; i < num_active_pipes; i++) { + if (dml_get_plane_idx(mode_lib, i) == plane_idx) { + if (i < first_pipe_idx_in_plane) { + first_pipe_idx_in_plane = i; + } + } + } + pipe_idx_in_combine = pipe_idx - first_pipe_idx_in_plane; // DML assumes the pipes in the same plane will have continuous indexing (i.e. plane 0 use pipe 0, 1, and plane 1 uses pipe 2, 3, etc.) + + disp_dlg_regs->refcyc_h_blank_end = (dml_uint_t)(((dml_float_t) hblank_end + (dml_float_t) pipe_idx_in_combine * (dml_float_t) hactive / (dml_float_t) odm_combine_factor) * ref_freq_to_pix_freq); + dml_print("DML_DLG: %s: pipe_idx = %d\n", __func__, pipe_idx); + dml_print("DML_DLG: %s: first_pipe_idx_in_plane = %d\n", __func__, first_pipe_idx_in_plane); + dml_print("DML_DLG: %s: pipe_idx_in_combine = %d\n", __func__, pipe_idx_in_combine); + dml_print("DML_DLG: %s: odm_combine_factor = %d\n", __func__, odm_combine_factor); + } + dml_print("DML_DLG: %s: refcyc_h_blank_end = %d\n", __func__, disp_dlg_regs->refcyc_h_blank_end); + + ASSERT(disp_dlg_regs->refcyc_h_blank_end < (dml_uint_t)dml_pow(2, 13)); + + disp_dlg_regs->ref_freq_to_pix_freq = (dml_uint_t)(ref_freq_to_pix_freq * dml_pow(2, 19)); + temp = dml_pow(2, 8); + disp_dlg_regs->refcyc_per_htotal = (dml_uint_t)(ref_freq_to_pix_freq * (dml_float_t)htotal * temp); + disp_dlg_regs->dlg_vblank_end = interlaced ? (vblank_end / 2) : vblank_end; // 15 bits + + min_ttu_vblank = dml_get_min_ttu_vblank_in_us(mode_lib, pipe_idx); + min_dst_y_next_start = (dml_uint_t)(dml_get_min_dst_y_next_start(mode_lib, pipe_idx)); + + dml_print("DML_DLG: %s: min_ttu_vblank (us) = %3.2f\n", __func__, min_ttu_vblank); + dml_print("DML_DLG: %s: min_dst_y_next_start = %d\n", __func__, min_dst_y_next_start); + dml_print("DML_DLG: %s: ref_freq_to_pix_freq = %3.2f\n", __func__, ref_freq_to_pix_freq); + + vready_after_vcount0 = (dml_uint_t)(dml_get_vready_at_or_after_vsync(mode_lib, pipe_idx)); + disp_dlg_regs->vready_after_vcount0 = vready_after_vcount0; + + dml_print("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, disp_dlg_regs->vready_after_vcount0); + + dst_x_after_scaler = (dml_uint_t)(dml_get_dst_x_after_scaler(mode_lib, pipe_idx)); + dst_y_after_scaler = (dml_uint_t)(dml_get_dst_y_after_scaler(mode_lib, pipe_idx)); + + dml_print("DML_DLG: %s: dst_x_after_scaler = %d\n", __func__, dst_x_after_scaler); + dml_print("DML_DLG: %s: dst_y_after_scaler = %d\n", __func__, dst_y_after_scaler); + + dst_y_prefetch = dml_get_dst_y_prefetch(mode_lib, pipe_idx); + dst_y_per_vm_vblank = dml_get_dst_y_per_vm_vblank(mode_lib, pipe_idx); + dst_y_per_row_vblank = dml_get_dst_y_per_row_vblank(mode_lib, pipe_idx); + dst_y_per_vm_flip = dml_get_dst_y_per_vm_flip(mode_lib, pipe_idx); + dst_y_per_row_flip = dml_get_dst_y_per_row_flip(mode_lib, pipe_idx); + + // magic! + if (htotal <= 75) { + max_dst_y_per_vm_vblank = 100.0; + max_dst_y_per_row_vblank = 100.0; + } + + dml_print("DML_DLG: %s: dst_y_prefetch (after rnd) = %3.2f\n", __func__, dst_y_prefetch); + dml_print("DML_DLG: %s: dst_y_per_vm_flip = %3.2f\n", __func__, dst_y_per_vm_flip); + dml_print("DML_DLG: %s: dst_y_per_row_flip = %3.2f\n", __func__, dst_y_per_row_flip); + dml_print("DML_DLG: %s: dst_y_per_vm_vblank = %3.2f\n", __func__, dst_y_per_vm_vblank); + dml_print("DML_DLG: %s: dst_y_per_row_vblank = %3.2f\n", __func__, dst_y_per_row_vblank); + + ASSERT(dst_y_per_vm_vblank < max_dst_y_per_vm_vblank); + ASSERT(dst_y_per_row_vblank < max_dst_y_per_row_vblank); + ASSERT(dst_y_prefetch > (dst_y_per_vm_vblank + dst_y_per_row_vblank)); + + vratio_pre_l = dml_get_vratio_prefetch_l(mode_lib, pipe_idx); + vratio_pre_c = dml_get_vratio_prefetch_c(mode_lib, pipe_idx); + + dml_print("DML_DLG: %s: vratio_pre_l = %3.2f\n", __func__, vratio_pre_l); + dml_print("DML_DLG: %s: vratio_pre_c = %3.2f\n", __func__, vratio_pre_c); + + // Active + refcyc_per_line_delivery_pre_l = dml_get_refcyc_per_line_delivery_pre_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_line_delivery_l = dml_get_refcyc_per_line_delivery_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_l = %3.2f\n", __func__, refcyc_per_line_delivery_pre_l); + dml_print("DML_DLG: %s: refcyc_per_line_delivery_l = %3.2f\n", __func__, refcyc_per_line_delivery_l); + + if (dual_plane) { + refcyc_per_line_delivery_pre_c = dml_get_refcyc_per_line_delivery_pre_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_line_delivery_c = dml_get_refcyc_per_line_delivery_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + dml_print("DML_DLG: %s: refcyc_per_line_delivery_pre_c = %3.2f\n", __func__, refcyc_per_line_delivery_pre_c); + dml_print("DML_DLG: %s: refcyc_per_line_delivery_c = %3.2f\n", __func__, refcyc_per_line_delivery_c); + } + + disp_dlg_regs->refcyc_per_vm_dmdata = (dml_uint_t)(dml_get_refcyc_per_vm_dmdata_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); + disp_dlg_regs->dmdata_dl_delta = (dml_uint_t)(dml_get_dmdata_dl_delta_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); + + refcyc_per_req_delivery_pre_l = dml_get_refcyc_per_req_delivery_pre_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_req_delivery_l = dml_get_refcyc_per_req_delivery_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_l = %3.2f\n", __func__, refcyc_per_req_delivery_pre_l); + dml_print("DML_DLG: %s: refcyc_per_req_delivery_l = %3.2f\n", __func__, refcyc_per_req_delivery_l); + + if (dual_plane) { + refcyc_per_req_delivery_pre_c = dml_get_refcyc_per_req_delivery_pre_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_req_delivery_c = dml_get_refcyc_per_req_delivery_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_c = %3.2f\n", __func__, refcyc_per_req_delivery_pre_c); + dml_print("DML_DLG: %s: refcyc_per_req_delivery_c = %3.2f\n", __func__, refcyc_per_req_delivery_c); + } + + // TTU - Cursor + ASSERT(num_cursors <= 1); + if (num_cursors > 0) { + refcyc_per_req_delivery_pre_cur0 = dml_get_refcyc_per_cursor_req_delivery_pre_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_req_delivery_cur0 = dml_get_refcyc_per_cursor_req_delivery_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + dml_print("DML_DLG: %s: refcyc_per_req_delivery_pre_cur0 = %3.2f\n", __func__, refcyc_per_req_delivery_pre_cur0); + dml_print("DML_DLG: %s: refcyc_per_req_delivery_cur0 = %3.2f\n", __func__, refcyc_per_req_delivery_cur0); + } + + // Assign to register structures + disp_dlg_regs->min_dst_y_next_start = (dml_uint_t)((dml_float_t) min_dst_y_next_start * dml_pow(2, 2)); + ASSERT(disp_dlg_regs->min_dst_y_next_start < (dml_uint_t)dml_pow(2, 18)); + + disp_dlg_regs->dst_y_after_scaler = dst_y_after_scaler; // in terms of line + disp_dlg_regs->refcyc_x_after_scaler = (dml_uint_t)((dml_float_t) dst_x_after_scaler * ref_freq_to_pix_freq); // in terms of refclk + disp_dlg_regs->dst_y_prefetch = (dml_uint_t)(dst_y_prefetch * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_vm_vblank = (dml_uint_t)(dst_y_per_vm_vblank * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_row_vblank = (dml_uint_t)(dst_y_per_row_vblank * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_vm_flip = (dml_uint_t)(dst_y_per_vm_flip * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_row_flip = (dml_uint_t)(dst_y_per_row_flip * dml_pow(2, 2)); + + disp_dlg_regs->vratio_prefetch = (dml_uint_t)(vratio_pre_l * dml_pow(2, 19)); + disp_dlg_regs->vratio_prefetch_c = (dml_uint_t)(vratio_pre_c * dml_pow(2, 19)); + + dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_vblank); + dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_vblank = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_vblank); + dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_vm_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_vm_flip); + dml_print("DML_DLG: %s: disp_dlg_regs->dst_y_per_row_flip = 0x%x\n", __func__, disp_dlg_regs->dst_y_per_row_flip); + + disp_dlg_regs->refcyc_per_vm_group_vblank = (dml_uint_t)(dml_get_refcyc_per_vm_group_vblank_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); + disp_dlg_regs->refcyc_per_vm_group_flip = (dml_uint_t)(dml_get_refcyc_per_vm_group_flip_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz); + disp_dlg_regs->refcyc_per_vm_req_vblank = (dml_uint_t)(dml_get_refcyc_per_vm_req_vblank_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10)); + disp_dlg_regs->refcyc_per_vm_req_flip = (dml_uint_t)(dml_get_refcyc_per_vm_req_flip_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz * dml_pow(2, 10)); + + dst_y_per_pte_row_nom_l = dml_get_dst_y_per_pte_row_nom_l(mode_lib, pipe_idx); + dst_y_per_pte_row_nom_c = dml_get_dst_y_per_pte_row_nom_c(mode_lib, pipe_idx); + dst_y_per_meta_row_nom_l = dml_get_dst_y_per_meta_row_nom_l(mode_lib, pipe_idx); + dst_y_per_meta_row_nom_c = dml_get_dst_y_per_meta_row_nom_c(mode_lib, pipe_idx); + + refcyc_per_pte_group_nom_l = dml_get_refcyc_per_pte_group_nom_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_pte_group_nom_c = dml_get_refcyc_per_pte_group_nom_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_pte_group_vblank_l = dml_get_refcyc_per_pte_group_vblank_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_pte_group_vblank_c = dml_get_refcyc_per_pte_group_vblank_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_pte_group_flip_l = dml_get_refcyc_per_pte_group_flip_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_pte_group_flip_c = dml_get_refcyc_per_pte_group_flip_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + refcyc_per_meta_chunk_nom_l = dml_get_refcyc_per_meta_chunk_nom_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_meta_chunk_nom_c = dml_get_refcyc_per_meta_chunk_nom_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_meta_chunk_vblank_l = dml_get_refcyc_per_meta_chunk_vblank_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_meta_chunk_vblank_c = dml_get_refcyc_per_meta_chunk_vblank_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_meta_chunk_flip_l = dml_get_refcyc_per_meta_chunk_flip_l_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + refcyc_per_meta_chunk_flip_c = dml_get_refcyc_per_meta_chunk_flip_c_in_us(mode_lib, pipe_idx) * refclk_freq_in_mhz; + + disp_dlg_regs->dst_y_per_pte_row_nom_l = (dml_uint_t)(dst_y_per_pte_row_nom_l * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_pte_row_nom_c = (dml_uint_t)(dst_y_per_pte_row_nom_c * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_meta_row_nom_l = (dml_uint_t)(dst_y_per_meta_row_nom_l * dml_pow(2, 2)); + disp_dlg_regs->dst_y_per_meta_row_nom_c = (dml_uint_t)(dst_y_per_meta_row_nom_c * dml_pow(2, 2)); + disp_dlg_regs->refcyc_per_pte_group_nom_l = (dml_uint_t)(refcyc_per_pte_group_nom_l); + disp_dlg_regs->refcyc_per_pte_group_nom_c = (dml_uint_t)(refcyc_per_pte_group_nom_c); + disp_dlg_regs->refcyc_per_pte_group_vblank_l = (dml_uint_t)(refcyc_per_pte_group_vblank_l); + disp_dlg_regs->refcyc_per_pte_group_vblank_c = (dml_uint_t)(refcyc_per_pte_group_vblank_c); + disp_dlg_regs->refcyc_per_pte_group_flip_l = (dml_uint_t)(refcyc_per_pte_group_flip_l); + disp_dlg_regs->refcyc_per_pte_group_flip_c = (dml_uint_t)(refcyc_per_pte_group_flip_c); + disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (dml_uint_t)(refcyc_per_meta_chunk_nom_l); + disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (dml_uint_t)(refcyc_per_meta_chunk_nom_c); + disp_dlg_regs->refcyc_per_meta_chunk_vblank_l = (dml_uint_t)(refcyc_per_meta_chunk_vblank_l); + disp_dlg_regs->refcyc_per_meta_chunk_vblank_c = (dml_uint_t)(refcyc_per_meta_chunk_vblank_c); + disp_dlg_regs->refcyc_per_meta_chunk_flip_l = (dml_uint_t)(refcyc_per_meta_chunk_flip_l); + disp_dlg_regs->refcyc_per_meta_chunk_flip_c = (dml_uint_t)(refcyc_per_meta_chunk_flip_c); + disp_dlg_regs->refcyc_per_line_delivery_pre_l = (dml_uint_t)dml_floor(refcyc_per_line_delivery_pre_l, 1); + disp_dlg_regs->refcyc_per_line_delivery_l = (dml_uint_t)dml_floor(refcyc_per_line_delivery_l, 1); + disp_dlg_regs->refcyc_per_line_delivery_pre_c = (dml_uint_t)dml_floor(refcyc_per_line_delivery_pre_c, 1); + disp_dlg_regs->refcyc_per_line_delivery_c = (dml_uint_t)dml_floor(refcyc_per_line_delivery_c, 1); + + disp_dlg_regs->chunk_hdl_adjust_cur0 = 3; + disp_dlg_regs->dst_y_offset_cur0 = 0; + disp_dlg_regs->chunk_hdl_adjust_cur1 = 3; + disp_dlg_regs->dst_y_offset_cur1 = 0; + + disp_dlg_regs->dst_y_delta_drq_limit = 0x7fff; // off + + disp_ttu_regs->refcyc_per_req_delivery_pre_l = (dml_uint_t)(refcyc_per_req_delivery_pre_l * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_l = (dml_uint_t)(refcyc_per_req_delivery_l * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_pre_c = (dml_uint_t)(refcyc_per_req_delivery_pre_c * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_c = (dml_uint_t)(refcyc_per_req_delivery_c * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_pre_cur0 = (dml_uint_t)(refcyc_per_req_delivery_pre_cur0 * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_cur0 = (dml_uint_t)(refcyc_per_req_delivery_cur0 * dml_pow(2, 10)); + disp_ttu_regs->refcyc_per_req_delivery_pre_cur1 = 0; + disp_ttu_regs->refcyc_per_req_delivery_cur1 = 0; + disp_ttu_regs->qos_level_low_wm = 0; + + disp_ttu_regs->qos_level_high_wm = (dml_uint_t)(4.0 * (dml_float_t)htotal * ref_freq_to_pix_freq); + + disp_ttu_regs->qos_level_flip = 14; + disp_ttu_regs->qos_level_fixed_l = 8; + disp_ttu_regs->qos_level_fixed_c = 8; + disp_ttu_regs->qos_level_fixed_cur0 = 8; + disp_ttu_regs->qos_ramp_disable_l = 0; + disp_ttu_regs->qos_ramp_disable_c = 0; + disp_ttu_regs->qos_ramp_disable_cur0 = 0; + disp_ttu_regs->min_ttu_vblank = (dml_uint_t)(min_ttu_vblank * refclk_freq_in_mhz); + + // CHECK for HW registers' range, assert or clamp + ASSERT(refcyc_per_req_delivery_pre_l < dml_pow(2, 13)); + ASSERT(refcyc_per_req_delivery_l < dml_pow(2, 13)); + ASSERT(refcyc_per_req_delivery_pre_c < dml_pow(2, 13)); + ASSERT(refcyc_per_req_delivery_c < dml_pow(2, 13)); + if (disp_dlg_regs->refcyc_per_vm_group_vblank >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_group_vblank = (dml_uint_t)(dml_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_group_flip >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_group_flip = (dml_uint_t)(dml_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_req_vblank >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_req_vblank = (dml_uint_t)(dml_pow(2, 23) - 1); + + if (disp_dlg_regs->refcyc_per_vm_req_flip >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_vm_req_flip = (dml_uint_t)(dml_pow(2, 23) - 1); + + + ASSERT(disp_dlg_regs->dst_y_after_scaler < (dml_uint_t)8); + ASSERT(disp_dlg_regs->refcyc_x_after_scaler < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->dst_y_per_pte_row_nom_l < (dml_uint_t)dml_pow(2, 17)); + if (dual_plane) { + if (disp_dlg_regs->dst_y_per_pte_row_nom_c >= (dml_uint_t)dml_pow(2, 17)) { // FIXME what so special about chroma, can we just assert? + dml_print("DML_DLG: %s: Warning dst_y_per_pte_row_nom_c %u > register max U15.2 %u\n", __func__, disp_dlg_regs->dst_y_per_pte_row_nom_c, (dml_uint_t)dml_pow(2, 17) - 1); + } + } + ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_l < (dml_uint_t)dml_pow(2, 17)); + ASSERT(disp_dlg_regs->dst_y_per_meta_row_nom_c < (dml_uint_t)dml_pow(2, 17)); + + if (disp_dlg_regs->refcyc_per_pte_group_nom_l >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_pte_group_nom_l = (dml_uint_t)(dml_pow(2, 23) - 1); + if (dual_plane) { + if (disp_dlg_regs->refcyc_per_pte_group_nom_c >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_pte_group_nom_c = (dml_uint_t)(dml_pow(2, 23) - 1); + } + ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_l < (dml_uint_t)dml_pow(2, 13)); + if (dual_plane) { + ASSERT(disp_dlg_regs->refcyc_per_pte_group_vblank_c < (dml_uint_t)dml_pow(2, 13)); + } + + if (disp_dlg_regs->refcyc_per_meta_chunk_nom_l >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_meta_chunk_nom_l = (dml_uint_t)(dml_pow(2, 23) - 1); + if (dual_plane) { + if (disp_dlg_regs->refcyc_per_meta_chunk_nom_c >= (dml_uint_t)dml_pow(2, 23)) + disp_dlg_regs->refcyc_per_meta_chunk_nom_c = (dml_uint_t)(dml_pow(2, 23) - 1); + } + ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_l < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->refcyc_per_meta_chunk_vblank_c < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_l < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->refcyc_per_line_delivery_l < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->refcyc_per_line_delivery_pre_c < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_dlg_regs->refcyc_per_line_delivery_c < (dml_uint_t)dml_pow(2, 13)); + ASSERT(disp_ttu_regs->qos_level_low_wm < (dml_uint_t) dml_pow(2, 14)); + ASSERT(disp_ttu_regs->qos_level_high_wm < (dml_uint_t) dml_pow(2, 14)); + ASSERT(disp_ttu_regs->min_ttu_vblank < (dml_uint_t) dml_pow(2, 24)); + + dml_print_ttu_regs_st(disp_ttu_regs); + dml_print_dlg_regs_st(disp_dlg_regs); + dml_print("DML_DLG::%s: Calculation for pipe[%d] done\n", __func__, pipe_idx); +} + +void dml_rq_dlg_get_arb_params(struct display_mode_lib_st *mode_lib, dml_display_arb_params_st *arb_param) +{ + memset(arb_param, 0, sizeof(*arb_param)); + arb_param->max_req_outstanding = 256; + arb_param->min_req_outstanding = 256; // turn off the sat level feature if this set to max + arb_param->sat_level_us = 60; + arb_param->hvm_max_qos_commit_threshold = 0xf; + arb_param->hvm_min_req_outstand_commit_threshold = 0xa; + arb_param->compbuf_reserved_space_kbytes = 2 * 8; // assume max data chunk size of 8K +} diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h new file mode 100644 index 000000000000..bf491cf0582d --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h @@ -0,0 +1,63 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __DML_DISPLAY_RQ_DLG_CALC_H__ +#define __DML_DISPLAY_RQ_DLG_CALC_H__ + +#include "display_mode_core_structs.h" +#include "display_mode_lib_defines.h" + +struct display_mode_lib_st; + +// Function: dml_rq_dlg_get_rq_reg +// Main entry point for test to get the register values out of this DML class. +// This function calls and fucntions to calculate +// and then populate the rq_regs struct +// Input: +// Assume mode_program is already called +// Output: +// rq_regs - struct that holds all the RQ registers field value. +// See also: + +void dml_rq_dlg_get_rq_reg(dml_display_rq_regs_st *rq_regs, + struct display_mode_lib_st *mode_lib, + const dml_uint_t pipe_idx); + +// Function: dml_rq_dlg_get_dlg_reg +// Calculate and return DLG and TTU register struct given the system setting +// Output: +// dlg_regs - output DLG register struct +// ttu_regs - output DLG TTU register struct +// Input: +// Assume mode_program is already called +// pipe_idx - index that identifies the e2e_pipe_param that corresponding to this dlg +void dml_rq_dlg_get_dlg_reg(dml_display_dlg_regs_st *dlg_regs, + dml_display_ttu_regs_st *ttu_regs, + struct display_mode_lib_st *mode_lib, + const dml_uint_t pipe_idx); + +// Function: dml_rq_dlg_get_arb_params +void dml_rq_dlg_get_arb_params(struct display_mode_lib_st *mode_lib, dml_display_arb_params_st *arb_param); + +#endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h new file mode 100644 index 000000000000..7fadbe6d7af4 --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h @@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __DML_LOGGING_H__ +#define __DML_LOGGING_H__ + +#define dml_print(...) ((void)0) + +#endif //__DML_LOGGING_H__ diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h index fdabbeec8ffa..4570b8016de5 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h @@ -31,7 +31,7 @@ #include "dcn30/dcn30_hubp.h" #include "dcn31/dcn31_hubp.h" #include "dcn32/dcn32_hubp.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" #define HUBP_3DLUT_FL_REG_LIST_DCN401(inst)\ SRI_ARR_US(_3DLUT_FL_CONFIG, HUBP, inst),\ diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index d11893f8c916..3fe590436e74 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -58,8 +58,8 @@ #include "transform.h" #include "dpp.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" -#include "dml2/dml21/inc/dml_top_types.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_types.h" struct resource_pool; struct dc_state; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h index b0c13b506c11..c62c4c1ed7c3 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h @@ -41,8 +41,8 @@ #include "mem_input.h" #include "cursor_reg_cache.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" -#include "dml2/dml21/inc/dml_top_types.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_types.h" #define OPP_ID_INVALID 0xf #define MAX_TTU 0xffffff diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h index 42fbc70f7056..d468bc85566a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h @@ -29,7 +29,7 @@ #include "include/grph_object_id.h" #include "dml/display_mode_structs.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" struct dchub_init_data; struct cstate_pstate_watermarks_st { diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c index 3965a7f1b64b..4c2e8aa96d1f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c @@ -92,7 +92,7 @@ #include "dc_state_priv.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define DC_LOGGER_INIT(logger) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c index fff57f23f4f7..3a83a63d2116 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c @@ -33,7 +33,7 @@ #include "resource.h" #include "include/irq_service_interface.h" #include "dcn35_resource.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dcn20/dcn20_resource.h" #include "dcn30/dcn30_resource.h" diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index 0abd163b425e..d00cc4030be3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -83,7 +83,7 @@ #include "vm_helper.h" #include "dcn20/dcn20_vmid.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "link_enc_cfg.h" #define DC_LOGGER_INIT(logger) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c index ca125ee6c2fb..f4b28617b2de 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c @@ -11,7 +11,7 @@ #include "resource.h" #include "include/irq_service_interface.h" #include "dcn36_resource.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dcn20/dcn20_resource.h" #include "dcn30/dcn30_resource.h" diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 1d18807e4749..8e2d3781752c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -73,7 +73,7 @@ #include "dc_state_priv.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define DC_LOGGER_INIT(logger) diff --git a/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h b/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h index 21d842857601..88c11b6be004 100644 --- a/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h +++ b/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h @@ -9,7 +9,7 @@ #include "dc.h" #include "clk_mgr.h" #include "soc_and_ip_translator.h" -#include "dml2/dml21/inc/dml_top_soc_parameter_types.h" +#include "dml2_0/dml21/inc/dml_top_soc_parameter_types.h" void dcn401_construct_soc_and_ip_translator(struct soc_and_ip_translator *soc_and_ip_translator); -- cgit v1.2.3 From dbf58a9dad4d80286c5c17e199f595eb0dd3be5a Mon Sep 17 00:00:00 2001 From: Karen Chen Date: Wed, 15 Oct 2025 11:13:07 -0400 Subject: drm/amd/display: Add more DC HW state info to underflow logging [Why] Debugging underflow issues frequently requires knowing the HW state at the time of underflow. To enable capturing this HW state information, interface functions are needed for the various DC HW blocks. [How] This change adds the interface functions to read HW state for the following DC HW blocks: - HUBBUB - HUBP - DPP - MPC - OPP - DSC - OPTC - DCCG Reviewed-by: George Shen Signed-off-by: Karen Chen Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc.h | 30 +++-- .../gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h | 64 ++++++++- .../gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c | 123 +++++++++++++++++ .../gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h | 2 + .../drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c | 3 +- .../drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h | 3 +- .../gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c | 1 + .../gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h | 5 +- .../drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c | 1 + .../gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h | 4 +- .../gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c | 16 +++ .../gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h | 2 + .../gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c | 1 + .../gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c | 1 + .../gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c | 1 + .../gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c | 8 ++ .../gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h | 1 + .../gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c | 1 + .../gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c | 1 + drivers/gpu/drm/amd/display/dc/dsc/dsc.h | 5 + .../drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c | 33 +---- .../drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h | 6 +- .../drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c | 3 +- .../drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c | 3 +- .../drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c | 3 +- .../amd/display/dc/hubbub/dcn401/dcn401_hubbub.c | 3 +- .../gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h | 136 ++++++++++++++++++- .../gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h | 8 +- .../gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c | 147 +++++++++++++++++---- .../gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h | 2 + .../gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c | 4 +- .../gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c | 4 +- .../gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c | 4 +- .../drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c | 4 +- .../drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c | 76 ++++++----- drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h | 121 ++++++++++++++++- drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h | 12 +- drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h | 16 ++- drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h | 1 + drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h | 27 ++++ drivers/gpu/drm/amd/display/dc/inc/hw/opp.h | 13 ++ .../drm/amd/display/dc/inc/hw/timing_generator.h | 130 ++++++++++++++++++ .../gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c | 16 +++ .../gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h | 5 + .../gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c | 1 + .../gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c | 1 + .../gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c | 14 +- .../gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h | 6 +- .../gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c | 13 ++ .../gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h | 6 +- .../gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c | 13 ++ .../gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h | 4 +- .../gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h | 38 +++++- .../gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c | 131 ++++++++++++++++++ .../gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h | 2 + .../drm/amd/display/dc/optc/dcn314/dcn314_optc.c | 1 + .../gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c | 1 + .../gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c | 1 + .../drm/amd/display/dc/optc/dcn401/dcn401_optc.c | 1 + 59 files changed, 1133 insertions(+), 150 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 182a35f6b439..93b39710db53 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -54,6 +54,14 @@ struct abm_save_restore; struct aux_payload; struct set_config_cmd_payload; struct dmub_notification; +struct dcn_hubbub_reg_state; +struct dcn_hubp_reg_state; +struct dcn_dpp_reg_state; +struct dcn_mpc_reg_state; +struct dcn_opp_reg_state; +struct dcn_dsc_reg_state; +struct dcn_optc_reg_state; +struct dcn_dccg_reg_state; #define DC_VER "3.2.355" @@ -1831,20 +1839,14 @@ struct dc_surface_update { }; struct dc_underflow_debug_data { - uint32_t otg_inst; - uint32_t otg_underflow; - uint32_t h_position; - uint32_t v_position; - uint32_t otg_frame_count; - struct dc_underflow_per_hubp_debug_data { - uint32_t hubp_underflow; - uint32_t hubp_in_blank; - uint32_t hubp_readline; - uint32_t det_config_error; - } hubps[MAX_PIPES]; - uint32_t curr_det_sizes[MAX_PIPES]; - uint32_t target_det_sizes[MAX_PIPES]; - uint32_t compbuf_config_error; + struct dcn_hubbub_reg_state *hubbub_reg_state; + struct dcn_hubp_reg_state *hubp_reg_state[MAX_PIPES]; + struct dcn_dpp_reg_state *dpp_reg_state[MAX_PIPES]; + struct dcn_mpc_reg_state *mpc_reg_state[MAX_PIPES]; + struct dcn_opp_reg_state *opp_reg_state[MAX_PIPES]; + struct dcn_dsc_reg_state *dsc_reg_state[MAX_PIPES]; + struct dcn_optc_reg_state *optc_reg_state[MAX_PIPES]; + struct dcn_dccg_reg_state *dccg_reg_state[MAX_PIPES]; }; /* diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h index a9b88f5e0c04..8bdffd9ff31b 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h @@ -425,7 +425,69 @@ struct dccg_mask { uint32_t SYMCLKD_CLOCK_ENABLE; \ uint32_t SYMCLKE_CLOCK_ENABLE; \ uint32_t DP_DTO_MODULO[MAX_PIPES]; \ - uint32_t DP_DTO_PHASE[MAX_PIPES] + uint32_t DP_DTO_PHASE[MAX_PIPES]; \ + uint32_t DC_MEM_GLOBAL_PWR_REQ_CNTL; \ + uint32_t DCCG_AUDIO_DTO0_MODULE; \ + uint32_t DCCG_AUDIO_DTO0_PHASE; \ + uint32_t DCCG_AUDIO_DTO1_MODULE; \ + uint32_t DCCG_AUDIO_DTO1_PHASE; \ + uint32_t DCCG_CAC_STATUS; \ + uint32_t DCCG_CAC_STATUS2; \ + uint32_t DCCG_DISP_CNTL_REG; \ + uint32_t DCCG_DS_CNTL; \ + uint32_t DCCG_DS_DTO_INCR; \ + uint32_t DCCG_DS_DTO_MODULO; \ + uint32_t DCCG_DS_HW_CAL_INTERVAL; \ + uint32_t DCCG_GTC_CNTL; \ + uint32_t DCCG_GTC_CURRENT; \ + uint32_t DCCG_GTC_DTO_INCR; \ + uint32_t DCCG_GTC_DTO_MODULO; \ + uint32_t DCCG_PERFMON_CNTL; \ + uint32_t DCCG_PERFMON_CNTL2; \ + uint32_t DCCG_SOFT_RESET; \ + uint32_t DCCG_TEST_CLK_SEL; \ + uint32_t DCCG_VSYNC_CNT_CTRL; \ + uint32_t DCCG_VSYNC_CNT_INT_CTRL; \ + uint32_t DCCG_VSYNC_OTG0_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG1_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG2_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG3_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG4_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG5_LATCH_VALUE; \ + uint32_t DISPCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DP_DTO_DBUF_EN; \ + uint32_t DPIACLK_540M_DTO_MODULO; \ + uint32_t DPIACLK_540M_DTO_PHASE; \ + uint32_t DPIACLK_810M_DTO_MODULO; \ + uint32_t DPIACLK_810M_DTO_PHASE; \ + uint32_t DPIACLK_DTO_CNTL; \ + uint32_t DPIASYMCLK_CNTL; \ + uint32_t DPPCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DPREFCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DPREFCLK_CNTL; \ + uint32_t DTBCLK_DTO_DBUF_EN; \ + uint32_t FORCE_SYMCLK_DISABLE; \ + uint32_t HDMICHARCLK0_CLOCK_CNTL; \ + uint32_t MICROSECOND_TIME_BASE_DIV; \ + uint32_t MILLISECOND_TIME_BASE_DIV; \ + uint32_t OTG0_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG0_PIXEL_RATE_CNTL; \ + uint32_t OTG1_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG1_PIXEL_RATE_CNTL; \ + uint32_t OTG2_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG2_PIXEL_RATE_CNTL; \ + uint32_t OTG3_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG3_PIXEL_RATE_CNTL; \ + uint32_t PHYPLLA_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLB_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLC_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLD_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLE_PIXCLK_RESYNC_CNTL; \ + uint32_t REFCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SOCCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SYMCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SYMCLK_PSP_CNTL + struct dccg_registers { DCCG_REG_VARIABLE_LIST; }; diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c index 8664f0c4c9b7..97df04b7e39d 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c @@ -709,6 +709,128 @@ void dccg31_otg_drop_pixel(struct dccg *dccg, OTG_DROP_PIXEL[otg_inst], 1); } +void dccg31_read_reg_state(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state) +{ + struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg); + + dccg_reg_state->dc_mem_global_pwr_req_cntl = REG_READ(DC_MEM_GLOBAL_PWR_REQ_CNTL); + dccg_reg_state->dccg_audio_dtbclk_dto_modulo = REG_READ(DCCG_AUDIO_DTBCLK_DTO_MODULO); + dccg_reg_state->dccg_audio_dtbclk_dto_phase = REG_READ(DCCG_AUDIO_DTBCLK_DTO_PHASE); + dccg_reg_state->dccg_audio_dto_source = REG_READ(DCCG_AUDIO_DTO_SOURCE); + dccg_reg_state->dccg_audio_dto0_module = REG_READ(DCCG_AUDIO_DTO0_MODULE); + dccg_reg_state->dccg_audio_dto0_phase = REG_READ(DCCG_AUDIO_DTO0_PHASE); + dccg_reg_state->dccg_audio_dto1_module = REG_READ(DCCG_AUDIO_DTO1_MODULE); + dccg_reg_state->dccg_audio_dto1_phase = REG_READ(DCCG_AUDIO_DTO1_PHASE); + dccg_reg_state->dccg_cac_status = REG_READ(DCCG_CAC_STATUS); + dccg_reg_state->dccg_cac_status2 = REG_READ(DCCG_CAC_STATUS2); + dccg_reg_state->dccg_disp_cntl_reg = REG_READ(DCCG_DISP_CNTL_REG); + dccg_reg_state->dccg_ds_cntl = REG_READ(DCCG_DS_CNTL); + dccg_reg_state->dccg_ds_dto_incr = REG_READ(DCCG_DS_DTO_INCR); + dccg_reg_state->dccg_ds_dto_modulo = REG_READ(DCCG_DS_DTO_MODULO); + dccg_reg_state->dccg_ds_hw_cal_interval = REG_READ(DCCG_DS_HW_CAL_INTERVAL); + dccg_reg_state->dccg_gate_disable_cntl = REG_READ(DCCG_GATE_DISABLE_CNTL); + dccg_reg_state->dccg_gate_disable_cntl2 = REG_READ(DCCG_GATE_DISABLE_CNTL2); + dccg_reg_state->dccg_gate_disable_cntl3 = REG_READ(DCCG_GATE_DISABLE_CNTL3); + dccg_reg_state->dccg_gate_disable_cntl4 = REG_READ(DCCG_GATE_DISABLE_CNTL4); + dccg_reg_state->dccg_gate_disable_cntl5 = REG_READ(DCCG_GATE_DISABLE_CNTL5); + dccg_reg_state->dccg_gate_disable_cntl6 = REG_READ(DCCG_GATE_DISABLE_CNTL6); + dccg_reg_state->dccg_global_fgcg_rep_cntl = REG_READ(DCCG_GLOBAL_FGCG_REP_CNTL); + dccg_reg_state->dccg_gtc_cntl = REG_READ(DCCG_GTC_CNTL); + dccg_reg_state->dccg_gtc_current = REG_READ(DCCG_GTC_CURRENT); + dccg_reg_state->dccg_gtc_dto_incr = REG_READ(DCCG_GTC_DTO_INCR); + dccg_reg_state->dccg_gtc_dto_modulo = REG_READ(DCCG_GTC_DTO_MODULO); + dccg_reg_state->dccg_perfmon_cntl = REG_READ(DCCG_PERFMON_CNTL); + dccg_reg_state->dccg_perfmon_cntl2 = REG_READ(DCCG_PERFMON_CNTL2); + dccg_reg_state->dccg_soft_reset = REG_READ(DCCG_SOFT_RESET); + dccg_reg_state->dccg_test_clk_sel = REG_READ(DCCG_TEST_CLK_SEL); + dccg_reg_state->dccg_vsync_cnt_ctrl = REG_READ(DCCG_VSYNC_CNT_CTRL); + dccg_reg_state->dccg_vsync_cnt_int_ctrl = REG_READ(DCCG_VSYNC_CNT_INT_CTRL); + dccg_reg_state->dccg_vsync_otg0_latch_value = REG_READ(DCCG_VSYNC_OTG0_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg1_latch_value = REG_READ(DCCG_VSYNC_OTG1_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg2_latch_value = REG_READ(DCCG_VSYNC_OTG2_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg3_latch_value = REG_READ(DCCG_VSYNC_OTG3_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg4_latch_value = REG_READ(DCCG_VSYNC_OTG4_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg5_latch_value = REG_READ(DCCG_VSYNC_OTG5_LATCH_VALUE); + dccg_reg_state->dispclk_cgtt_blk_ctrl_reg = REG_READ(DISPCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dispclk_freq_change_cntl = REG_READ(DISPCLK_FREQ_CHANGE_CNTL); + dccg_reg_state->dp_dto_dbuf_en = REG_READ(DP_DTO_DBUF_EN); + dccg_reg_state->dp_dto0_modulo = REG_READ(DP_DTO_MODULO[0]); + dccg_reg_state->dp_dto0_phase = REG_READ(DP_DTO_PHASE[0]); + dccg_reg_state->dp_dto1_modulo = REG_READ(DP_DTO_MODULO[1]); + dccg_reg_state->dp_dto1_phase = REG_READ(DP_DTO_PHASE[1]); + dccg_reg_state->dp_dto2_modulo = REG_READ(DP_DTO_MODULO[2]); + dccg_reg_state->dp_dto2_phase = REG_READ(DP_DTO_PHASE[2]); + dccg_reg_state->dp_dto3_modulo = REG_READ(DP_DTO_MODULO[3]); + dccg_reg_state->dp_dto3_phase = REG_READ(DP_DTO_PHASE[3]); + dccg_reg_state->dpiaclk_540m_dto_modulo = REG_READ(DPIACLK_540M_DTO_MODULO); + dccg_reg_state->dpiaclk_540m_dto_phase = REG_READ(DPIACLK_540M_DTO_PHASE); + dccg_reg_state->dpiaclk_810m_dto_modulo = REG_READ(DPIACLK_810M_DTO_MODULO); + dccg_reg_state->dpiaclk_810m_dto_phase = REG_READ(DPIACLK_810M_DTO_PHASE); + dccg_reg_state->dpiaclk_dto_cntl = REG_READ(DPIACLK_DTO_CNTL); + dccg_reg_state->dpiasymclk_cntl = REG_READ(DPIASYMCLK_CNTL); + dccg_reg_state->dppclk_cgtt_blk_ctrl_reg = REG_READ(DPPCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dppclk_ctrl = REG_READ(DPPCLK_CTRL); + dccg_reg_state->dppclk_dto_ctrl = REG_READ(DPPCLK_DTO_CTRL); + dccg_reg_state->dppclk0_dto_param = REG_READ(DPPCLK_DTO_PARAM[0]); + dccg_reg_state->dppclk1_dto_param = REG_READ(DPPCLK_DTO_PARAM[1]); + dccg_reg_state->dppclk2_dto_param = REG_READ(DPPCLK_DTO_PARAM[2]); + dccg_reg_state->dppclk3_dto_param = REG_READ(DPPCLK_DTO_PARAM[3]); + dccg_reg_state->dprefclk_cgtt_blk_ctrl_reg = REG_READ(DPREFCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dprefclk_cntl = REG_READ(DPREFCLK_CNTL); + dccg_reg_state->dpstreamclk_cntl = REG_READ(DPSTREAMCLK_CNTL); + dccg_reg_state->dscclk_dto_ctrl = REG_READ(DSCCLK_DTO_CTRL); + dccg_reg_state->dscclk0_dto_param = REG_READ(DSCCLK0_DTO_PARAM); + dccg_reg_state->dscclk1_dto_param = REG_READ(DSCCLK1_DTO_PARAM); + dccg_reg_state->dscclk2_dto_param = REG_READ(DSCCLK2_DTO_PARAM); + dccg_reg_state->dscclk3_dto_param = REG_READ(DSCCLK3_DTO_PARAM); + dccg_reg_state->dtbclk_dto_dbuf_en = REG_READ(DTBCLK_DTO_DBUF_EN); + dccg_reg_state->dtbclk_dto0_modulo = REG_READ(DTBCLK_DTO_MODULO[0]); + dccg_reg_state->dtbclk_dto0_phase = REG_READ(DTBCLK_DTO_PHASE[0]); + dccg_reg_state->dtbclk_dto1_modulo = REG_READ(DTBCLK_DTO_MODULO[1]); + dccg_reg_state->dtbclk_dto1_phase = REG_READ(DTBCLK_DTO_PHASE[1]); + dccg_reg_state->dtbclk_dto2_modulo = REG_READ(DTBCLK_DTO_MODULO[2]); + dccg_reg_state->dtbclk_dto2_phase = REG_READ(DTBCLK_DTO_PHASE[2]); + dccg_reg_state->dtbclk_dto3_modulo = REG_READ(DTBCLK_DTO_MODULO[3]); + dccg_reg_state->dtbclk_dto3_phase = REG_READ(DTBCLK_DTO_PHASE[3]); + dccg_reg_state->dtbclk_p_cntl = REG_READ(DTBCLK_P_CNTL); + dccg_reg_state->force_symclk_disable = REG_READ(FORCE_SYMCLK_DISABLE); + dccg_reg_state->hdmicharclk0_clock_cntl = REG_READ(HDMICHARCLK0_CLOCK_CNTL); + dccg_reg_state->hdmistreamclk_cntl = REG_READ(HDMISTREAMCLK_CNTL); + dccg_reg_state->hdmistreamclk0_dto_param = REG_READ(HDMISTREAMCLK0_DTO_PARAM); + dccg_reg_state->microsecond_time_base_div = REG_READ(MICROSECOND_TIME_BASE_DIV); + dccg_reg_state->millisecond_time_base_div = REG_READ(MILLISECOND_TIME_BASE_DIV); + dccg_reg_state->otg_pixel_rate_div = REG_READ(OTG_PIXEL_RATE_DIV); + dccg_reg_state->otg0_phypll_pixel_rate_cntl = REG_READ(OTG0_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg0_pixel_rate_cntl = REG_READ(OTG0_PIXEL_RATE_CNTL); + dccg_reg_state->otg1_phypll_pixel_rate_cntl = REG_READ(OTG1_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg1_pixel_rate_cntl = REG_READ(OTG1_PIXEL_RATE_CNTL); + dccg_reg_state->otg2_phypll_pixel_rate_cntl = REG_READ(OTG2_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg2_pixel_rate_cntl = REG_READ(OTG2_PIXEL_RATE_CNTL); + dccg_reg_state->otg3_phypll_pixel_rate_cntl = REG_READ(OTG3_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg3_pixel_rate_cntl = REG_READ(OTG3_PIXEL_RATE_CNTL); + dccg_reg_state->phyasymclk_clock_cntl = REG_READ(PHYASYMCLK_CLOCK_CNTL); + dccg_reg_state->phybsymclk_clock_cntl = REG_READ(PHYBSYMCLK_CLOCK_CNTL); + dccg_reg_state->phycsymclk_clock_cntl = REG_READ(PHYCSYMCLK_CLOCK_CNTL); + dccg_reg_state->phydsymclk_clock_cntl = REG_READ(PHYDSYMCLK_CLOCK_CNTL); + dccg_reg_state->phyesymclk_clock_cntl = REG_READ(PHYESYMCLK_CLOCK_CNTL); + dccg_reg_state->phyplla_pixclk_resync_cntl = REG_READ(PHYPLLA_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phypllb_pixclk_resync_cntl = REG_READ(PHYPLLB_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phypllc_pixclk_resync_cntl = REG_READ(PHYPLLC_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phyplld_pixclk_resync_cntl = REG_READ(PHYPLLD_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phyplle_pixclk_resync_cntl = REG_READ(PHYPLLE_PIXCLK_RESYNC_CNTL); + dccg_reg_state->refclk_cgtt_blk_ctrl_reg = REG_READ(REFCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->socclk_cgtt_blk_ctrl_reg = REG_READ(SOCCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->symclk_cgtt_blk_ctrl_reg = REG_READ(SYMCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->symclk_psp_cntl = REG_READ(SYMCLK_PSP_CNTL); + dccg_reg_state->symclk32_le_cntl = REG_READ(SYMCLK32_LE_CNTL); + dccg_reg_state->symclk32_se_cntl = REG_READ(SYMCLK32_SE_CNTL); + dccg_reg_state->symclka_clock_enable = REG_READ(SYMCLKA_CLOCK_ENABLE); + dccg_reg_state->symclkb_clock_enable = REG_READ(SYMCLKB_CLOCK_ENABLE); + dccg_reg_state->symclkc_clock_enable = REG_READ(SYMCLKC_CLOCK_ENABLE); + dccg_reg_state->symclkd_clock_enable = REG_READ(SYMCLKD_CLOCK_ENABLE); + dccg_reg_state->symclke_clock_enable = REG_READ(SYMCLKE_CLOCK_ENABLE); +} + static const struct dccg_funcs dccg31_funcs = { .update_dpp_dto = dccg31_update_dpp_dto, .get_dccg_ref_freq = dccg31_get_dccg_ref_freq, @@ -727,6 +849,7 @@ static const struct dccg_funcs dccg31_funcs = { .set_dispclk_change_mode = dccg31_set_dispclk_change_mode, .disable_dsc = dccg31_disable_dscclk, .enable_dsc = dccg31_enable_dscclk, + .dccg_read_reg_state = dccg31_read_reg_state, }; struct dccg *dccg31_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h index cd261051dc2c..bf659920d4cc 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h @@ -236,4 +236,6 @@ void dccg31_disable_dscclk(struct dccg *dccg, int inst); void dccg31_enable_dscclk(struct dccg *dccg, int inst); +void dccg31_read_reg_state(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state); + #endif //__DCN31_DCCG_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c index 8f6edd8e9beb..ef3db6beba25 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c @@ -377,7 +377,8 @@ static const struct dccg_funcs dccg314_funcs = { .get_pixel_rate_div = dccg314_get_pixel_rate_div, .trigger_dio_fifo_resync = dccg314_trigger_dio_fifo_resync, .set_valid_pixel_rate = dccg314_set_valid_pixel_rate, - .set_dtbclk_p_src = dccg314_set_dtbclk_p_src + .set_dtbclk_p_src = dccg314_set_dtbclk_p_src, + .dccg_read_reg_state = dccg31_read_reg_state }; struct dccg *dccg314_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h index 60ea1d248deb..a609635f35db 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h @@ -74,8 +74,7 @@ SR(DCCG_GATE_DISABLE_CNTL3),\ SR(HDMISTREAMCLK0_DTO_PARAM),\ SR(OTG_PIXEL_RATE_DIV),\ - SR(DTBCLK_P_CNTL),\ - SR(DCCG_AUDIO_DTO_SOURCE) + SR(DTBCLK_P_CNTL) #define DCCG_MASK_SH_LIST_DCN314_COMMON(mask_sh) \ DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c index 0aa41759f79d..bd2f528137b2 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c @@ -2453,6 +2453,7 @@ static const struct dccg_funcs dccg35_funcs = { .disable_symclk_se = dccg35_disable_symclk_se, .set_dtbclk_p_src = dccg35_set_dtbclk_p_src, .dccg_root_gate_disable_control = dccg35_root_gate_disable_control, + .dccg_read_reg_state = dccg31_read_reg_state, }; struct dccg *dccg35_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h index ad7a5850d31b..7b9c36456cd9 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h @@ -41,8 +41,9 @@ SR(SYMCLKA_CLOCK_ENABLE),\ SR(SYMCLKB_CLOCK_ENABLE),\ SR(SYMCLKC_CLOCK_ENABLE),\ - SR(SYMCLKD_CLOCK_ENABLE),\ - SR(SYMCLKE_CLOCK_ENABLE) + SR(SYMCLKD_CLOCK_ENABLE), \ + SR(SYMCLKE_CLOCK_ENABLE),\ + SR(SYMCLK_PSP_CNTL) #define DCCG_MASK_SH_LIST_DCN35(mask_sh) \ DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c index 0b8ed9b94d3c..663a18ee5162 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c @@ -886,6 +886,7 @@ static const struct dccg_funcs dccg401_funcs = { .enable_symclk_se = dccg401_enable_symclk_se, .disable_symclk_se = dccg401_disable_symclk_se, .set_dtbclk_p_src = dccg401_set_dtbclk_p_src, + .dccg_read_reg_state = dccg31_read_reg_state }; struct dccg *dccg401_create( diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h index f466182963f7..b12f34345a58 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h @@ -1348,7 +1348,8 @@ struct dcn_dpp_mask { uint32_t CURSOR0_COLOR1; \ uint32_t DPP_CONTROL; \ uint32_t CM_HDR_MULT_COEF; \ - uint32_t CURSOR0_FP_SCALE_BIAS; + uint32_t CURSOR0_FP_SCALE_BIAS; \ + uint32_t OBUF_CONTROL; struct dcn_dpp_registers { DPP_COMMON_REG_VARIABLE_LIST @@ -1450,7 +1451,6 @@ void dpp1_set_degamma( void dpp1_set_degamma_pwl(struct dpp *dpp_base, const struct pwl_params *params); - void dpp_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s); diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c index fe52df01caf7..ef4a16117181 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c @@ -84,6 +84,22 @@ void dpp30_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s) } } +void dpp30_read_reg_state(struct dpp *dpp_base, struct dcn_dpp_reg_state *dpp_reg_state) +{ + struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base); + + dpp_reg_state->recout_start = REG_READ(RECOUT_START); + dpp_reg_state->recout_size = REG_READ(RECOUT_SIZE); + dpp_reg_state->scl_horz_filter_scale_ratio = REG_READ(SCL_HORZ_FILTER_SCALE_RATIO); + dpp_reg_state->scl_vert_filter_scale_ratio = REG_READ(SCL_VERT_FILTER_SCALE_RATIO); + dpp_reg_state->scl_mode = REG_READ(SCL_MODE); + dpp_reg_state->cm_control = REG_READ(CM_CONTROL); + dpp_reg_state->dpp_control = REG_READ(DPP_CONTROL); + dpp_reg_state->dscl_control = REG_READ(DSCL_CONTROL); + dpp_reg_state->obuf_control = REG_READ(OBUF_CONTROL); + dpp_reg_state->mpc_size = REG_READ(MPC_SIZE); +} + /*program post scaler scs block in dpp CM*/ void dpp3_program_post_csc( struct dpp *dpp_base, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h index f236824126e9..d4a70b4379ea 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h @@ -594,6 +594,8 @@ void dpp3_program_CM_dealpha( void dpp30_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s); +void dpp30_read_reg_state(struct dpp *dpp_base, struct dcn_dpp_reg_state *dpp_reg_state); + bool dpp3_get_optimal_number_of_taps( struct dpp *dpp, struct scaler_data *scl_data, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c index fa67e54bf94e..8a5aa5e86850 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c @@ -134,6 +134,7 @@ static struct dpp_funcs dcn32_dpp_funcs = { .dpp_dppclk_control = dpp1_dppclk_control, .dpp_set_hdr_multiplier = dpp3_set_hdr_multiplier, .dpp_get_gamut_remap = dpp3_cm_get_gamut_remap, + .dpp_read_reg_state = dpp30_read_reg_state, }; diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c index f7a373a3d70a..977d83bf7741 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c @@ -95,6 +95,7 @@ void dpp35_program_bias_and_scale_fcnv( static struct dpp_funcs dcn35_dpp_funcs = { .dpp_program_gamcor_lut = dpp3_program_gamcor_lut, .dpp_read_state = dpp30_read_state, + .dpp_read_reg_state = dpp30_read_reg_state, .dpp_reset = dpp_reset, .dpp_set_scaler = dpp1_dscl_set_scaler_manual_scale, .dpp_get_optimal_number_of_taps = dpp3_get_optimal_number_of_taps, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c index 36187f890d5d..96c2c853de42 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c @@ -248,6 +248,7 @@ static struct dpp_funcs dcn401_dpp_funcs = { .set_optional_cursor_attributes = dpp401_set_optional_cursor_attributes, .dpp_dppclk_control = dpp1_dppclk_control, .dpp_set_hdr_multiplier = dpp3_set_hdr_multiplier, + .dpp_read_reg_state = dpp30_read_reg_state, .set_cursor_matrix = dpp401_set_cursor_matrix, }; diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c index ee8c11e085cd..242f1e6f0d8f 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c @@ -35,6 +35,7 @@ static void dsc_write_to_registers(struct display_stream_compressor *dsc, const static const struct dsc_funcs dcn20_dsc_funcs = { .dsc_get_enc_caps = dsc2_get_enc_caps, .dsc_read_state = dsc2_read_state, + .dsc_read_reg_state = dsc2_read_reg_state, .dsc_validate_stream = dsc2_validate_stream, .dsc_set_config = dsc2_set_config, .dsc_get_packed_pps = dsc2_get_packed_pps, @@ -155,6 +156,13 @@ void dsc2_read_state(struct display_stream_compressor *dsc, struct dcn_dsc_state DSCRM_DSC_OPP_PIPE_SOURCE, &s->dsc_opp_source); } +void dsc2_read_reg_state(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state) +{ + struct dcn20_dsc *dsc20 = TO_DCN20_DSC(dsc); + + dccg_reg_state->dsc_top_control = REG_READ(DSC_TOP_CONTROL); + dccg_reg_state->dscc_interrupt_control_status = REG_READ(DSCC_INTERRUPT_CONTROL_STATUS); +} bool dsc2_validate_stream(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg) { diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h index a9c04fc95bd1..2337c3a97235 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h @@ -606,6 +606,7 @@ bool dsc2_get_packed_pps(struct display_stream_compressor *dsc, uint8_t *dsc_packed_pps); void dsc2_read_state(struct display_stream_compressor *dsc, struct dcn_dsc_state *s); +void dsc2_read_reg_state(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state); bool dsc2_validate_stream(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg); void dsc2_set_config(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg, struct dsc_optc_config *dsc_optc_cfg); diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c index 6f4f5a3c4861..f9c6377ac66c 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c @@ -32,6 +32,7 @@ static void dsc35_enable(struct display_stream_compressor *dsc, int opp_pipe); static const struct dsc_funcs dcn35_dsc_funcs = { .dsc_get_enc_caps = dsc2_get_enc_caps, .dsc_read_state = dsc2_read_state, + .dsc_read_reg_state = dsc2_read_reg_state, .dsc_validate_stream = dsc2_validate_stream, .dsc_set_config = dsc2_set_config, .dsc_get_packed_pps = dsc2_get_packed_pps, diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c index 7bd92ae8b13e..c1bdbb38c690 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c @@ -26,6 +26,7 @@ static const struct dsc_funcs dcn401_dsc_funcs = { .dsc_disconnect = dsc401_disconnect, .dsc_wait_disconnect_pending_clear = dsc401_wait_disconnect_pending_clear, .dsc_get_single_enc_caps = dsc401_get_single_enc_caps, + .dsc_read_reg_state = dsc2_read_reg_state }; /* Macro definitios for REG_SET macros*/ diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dsc.h b/drivers/gpu/drm/amd/display/dc/dsc/dsc.h index b433e16842bf..81c83d5fe042 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dsc/dsc.h @@ -66,6 +66,10 @@ struct dcn_dsc_state { uint32_t dsc_opp_source; }; +struct dcn_dsc_reg_state { + uint32_t dsc_top_control; + uint32_t dscc_interrupt_control_status; +}; /* DSC encoder capabilities * They differ from the DPCD DSC caps because they are based on AMD DSC encoder caps. @@ -100,6 +104,7 @@ struct dsc_enc_caps { struct dsc_funcs { void (*dsc_get_enc_caps)(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz); void (*dsc_read_state)(struct display_stream_compressor *dsc, struct dcn_dsc_state *s); + void (*dsc_read_reg_state)(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state); bool (*dsc_validate_stream)(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg); void (*dsc_set_config)(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg, struct dsc_optc_config *dsc_optc_cfg); diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c index e7e5f6d4778e..181a93dc46e6 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c @@ -440,33 +440,15 @@ void hubbub3_init_watermarks(struct hubbub *hubbub) REG_WRITE(DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_D, reg); } -void hubbub3_get_det_sizes(struct hubbub *hubbub, uint32_t *curr_det_sizes, uint32_t *target_det_sizes) +void hubbub3_read_reg_state(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state) { struct dcn20_hubbub *hubbub1 = TO_DCN20_HUBBUB(hubbub); - REG_GET_2(DCHUBBUB_DET0_CTRL, DET0_SIZE_CURRENT, &curr_det_sizes[0], - DET0_SIZE, &target_det_sizes[0]); - - REG_GET_2(DCHUBBUB_DET1_CTRL, DET1_SIZE_CURRENT, &curr_det_sizes[1], - DET1_SIZE, &target_det_sizes[1]); - - REG_GET_2(DCHUBBUB_DET2_CTRL, DET2_SIZE_CURRENT, &curr_det_sizes[2], - DET2_SIZE, &target_det_sizes[2]); - - REG_GET_2(DCHUBBUB_DET3_CTRL, DET3_SIZE_CURRENT, &curr_det_sizes[3], - DET3_SIZE, &target_det_sizes[3]); - -} - -uint32_t hubbub3_compbuf_config_error(struct hubbub *hubbub) -{ - struct dcn20_hubbub *hubbub1 = TO_DCN20_HUBBUB(hubbub); - uint32_t compbuf_config_error = 0; - - REG_GET(DCHUBBUB_COMPBUF_CTRL, CONFIG_ERROR, - &compbuf_config_error); - - return compbuf_config_error; + hubbub_reg_state->det0_ctrl = REG_READ(DCHUBBUB_DET0_CTRL); + hubbub_reg_state->det1_ctrl = REG_READ(DCHUBBUB_DET1_CTRL); + hubbub_reg_state->det2_ctrl = REG_READ(DCHUBBUB_DET2_CTRL); + hubbub_reg_state->det3_ctrl = REG_READ(DCHUBBUB_DET3_CTRL); + hubbub_reg_state->compbuf_ctrl = REG_READ(DCHUBBUB_COMPBUF_CTRL); } static const struct hubbub_funcs hubbub30_funcs = { @@ -486,8 +468,7 @@ static const struct hubbub_funcs hubbub30_funcs = { .force_pstate_change_control = hubbub3_force_pstate_change_control, .init_watermarks = hubbub3_init_watermarks, .hubbub_read_state = hubbub2_read_state, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub3_construct(struct dcn20_hubbub *hubbub3, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h index 49a469969d36..9e14de3ccaee 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h @@ -133,10 +133,6 @@ void hubbub3_force_pstate_change_control(struct hubbub *hubbub, void hubbub3_init_watermarks(struct hubbub *hubbub); -void hubbub3_get_det_sizes(struct hubbub *hubbub, - uint32_t *curr_det_sizes, - uint32_t *target_det_sizes); - -uint32_t hubbub3_compbuf_config_error(struct hubbub *hubbub); +void hubbub3_read_reg_state(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state); #endif diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c index cdb20251a154..d1aaa58b7db3 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c @@ -1071,8 +1071,7 @@ static const struct hubbub_funcs hubbub31_funcs = { .program_compbuf_size = dcn31_program_compbuf_size, .init_crb = dcn31_init_crb, .hubbub_read_state = hubbub2_read_state, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub31_construct(struct dcn20_hubbub *hubbub31, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c index 4d4ca6d77bbd..237331b35378 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c @@ -1037,8 +1037,7 @@ static const struct hubbub_funcs hubbub32_funcs = { .force_usr_retraining_allow = hubbub32_force_usr_retraining_allow, .set_request_limit = hubbub32_set_request_limit, .get_mall_en = hubbub32_get_mall_en, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub32_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c index a443722a8632..1b7746a6549a 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c @@ -589,8 +589,7 @@ static const struct hubbub_funcs hubbub35_funcs = { .hubbub_read_state = hubbub2_read_state, .force_usr_retraining_allow = hubbub32_force_usr_retraining_allow, .dchubbub_init = hubbub35_init, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub35_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c index a36273a52880..d11afd1ce72a 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c @@ -1247,8 +1247,7 @@ static const struct hubbub_funcs hubbub4_01_funcs = { .program_compbuf_segments = dcn401_program_compbuf_segments, .wait_for_det_update = dcn401_wait_for_det_update, .program_arbiter = dcn401_program_arbiter, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub401_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h index cf2eb9793008..f2571076fc50 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h @@ -105,7 +105,9 @@ SRI(DCN_CUR0_TTU_CNTL0, HUBPREQ, id),\ SRI(DCN_CUR0_TTU_CNTL1, HUBPREQ, id),\ SRI(HUBP_CLK_CNTL, HUBP, id),\ - SRI(HUBPRET_READ_LINE_VALUE, HUBPRET, id) + SRI(HUBPRET_READ_LINE_VALUE, HUBPRET, id),\ + SRI(HUBP_MEASURE_WIN_CTRL_DCFCLK, HUBP, id),\ + SRI(HUBP_MEASURE_WIN_CTRL_DPPCLK, HUBP, id) /* Register address initialization macro for ASICs with VM */ #define HUBP_REG_LIST_DCN_VM(id)\ @@ -251,7 +253,19 @@ uint32_t CURSOR_HOT_SPOT; \ uint32_t CURSOR_DST_OFFSET; \ uint32_t HUBP_CLK_CNTL; \ - uint32_t HUBPRET_READ_LINE_VALUE + uint32_t HUBPRET_READ_LINE_VALUE; \ + uint32_t HUBP_MEASURE_WIN_CTRL_DCFCLK; \ + uint32_t HUBP_MEASURE_WIN_CTRL_DPPCLK; \ + uint32_t HUBPRET_INTERRUPT; \ + uint32_t HUBPRET_MEM_PWR_CTRL; \ + uint32_t HUBPRET_MEM_PWR_STATUS; \ + uint32_t HUBPRET_READ_LINE_CTRL0; \ + uint32_t HUBPRET_READ_LINE_CTRL1; \ + uint32_t HUBPRET_READ_LINE0; \ + uint32_t HUBPRET_READ_LINE1; \ + uint32_t HUBPREQ_MEM_PWR_CTRL; \ + uint32_t HUBPREQ_MEM_PWR_STATUS + #define HUBP_SF(reg_name, field_name, post_fix)\ .field_name = reg_name ## __ ## field_name ## post_fix @@ -688,6 +702,123 @@ struct dcn_fl_regs_st { uint32_t lut_fl_mode; uint32_t lut_fl_format; }; +struct dcn_hubp_reg_state { + uint32_t hubp_cntl; + uint32_t mall_config; + uint32_t mall_sub_vp; + uint32_t hubp_req_size_config; + uint32_t hubp_req_size_config_c; + uint32_t vmpg_config; + uint32_t addr_config; + uint32_t pri_viewport_dimension; + uint32_t pri_viewport_dimension_c; + uint32_t pri_viewport_start; + uint32_t pri_viewport_start_c; + uint32_t sec_viewport_dimension; + uint32_t sec_viewport_dimension_c; + uint32_t sec_viewport_start; + uint32_t sec_viewport_start_c; + uint32_t surface_config; + uint32_t tiling_config; + uint32_t clk_cntl; + uint32_t mall_status; + uint32_t measure_win_ctrl_dcfclk; + uint32_t measure_win_ctrl_dppclk; + + uint32_t blank_offset_0; + uint32_t blank_offset_1; + uint32_t cursor_settings; + uint32_t dcn_cur0_ttu_cntl0; + uint32_t dcn_cur0_ttu_cntl1; + uint32_t dcn_cur1_ttu_cntl0; + uint32_t dcn_cur1_ttu_cntl1; + uint32_t dcn_dmdat_vm_cntl; + uint32_t dcn_expansion_mode; + uint32_t dcn_global_ttu_cntl; + uint32_t dcn_surf0_ttu_cntl0; + uint32_t dcn_surf0_ttu_cntl1; + uint32_t dcn_surf1_ttu_cntl0; + uint32_t dcn_surf1_ttu_cntl1; + uint32_t dcn_ttu_qos_wm; + uint32_t dcn_vm_mx_l1_tlb_cntl; + uint32_t dcn_vm_system_aperture_high_addr; + uint32_t dcn_vm_system_aperture_low_addr; + uint32_t dcsurf_flip_control; + uint32_t dcsurf_flip_control2; + uint32_t dcsurf_primary_meta_surface_address; + uint32_t dcsurf_primary_meta_surface_address_c; + uint32_t dcsurf_primary_meta_surface_address_high; + uint32_t dcsurf_primary_meta_surface_address_high_c; + uint32_t dcsurf_primary_surface_address; + uint32_t dcsurf_primary_surface_address_c; + uint32_t dcsurf_primary_surface_address_high; + uint32_t dcsurf_primary_surface_address_high_c; + uint32_t dcsurf_secondary_meta_surface_address; + uint32_t dcsurf_secondary_meta_surface_address_c; + uint32_t dcsurf_secondary_meta_surface_address_high; + uint32_t dcsurf_secondary_meta_surface_address_high_c; + uint32_t dcsurf_secondary_surface_address; + uint32_t dcsurf_secondary_surface_address_c; + uint32_t dcsurf_secondary_surface_address_high; + uint32_t dcsurf_secondary_surface_address_high_c; + uint32_t dcsurf_surface_control; + uint32_t dcsurf_surface_earliest_inuse; + uint32_t dcsurf_surface_earliest_inuse_c; + uint32_t dcsurf_surface_earliest_inuse_high; + uint32_t dcsurf_surface_earliest_inuse_high_c; + uint32_t dcsurf_surface_flip_interrupt; + uint32_t dcsurf_surface_inuse; + uint32_t dcsurf_surface_inuse_c; + uint32_t dcsurf_surface_inuse_high; + uint32_t dcsurf_surface_inuse_high_c; + uint32_t dcsurf_surface_pitch; + uint32_t dcsurf_surface_pitch_c; + uint32_t dst_after_scaler; + uint32_t dst_dimensions; + uint32_t dst_y_delta_drq_limit; + uint32_t flip_parameters_0; + uint32_t flip_parameters_1; + uint32_t flip_parameters_2; + uint32_t flip_parameters_3; + uint32_t flip_parameters_4; + uint32_t flip_parameters_5; + uint32_t flip_parameters_6; + uint32_t hubpreq_mem_pwr_ctrl; + uint32_t hubpreq_mem_pwr_status; + uint32_t nom_parameters_0; + uint32_t nom_parameters_1; + uint32_t nom_parameters_2; + uint32_t nom_parameters_3; + uint32_t nom_parameters_4; + uint32_t nom_parameters_5; + uint32_t nom_parameters_6; + uint32_t nom_parameters_7; + uint32_t per_line_delivery; + uint32_t per_line_delivery_pre; + uint32_t prefetch_settings; + uint32_t prefetch_settings_c; + uint32_t ref_freq_to_pix_freq; + uint32_t uclk_pstate_force; + uint32_t vblank_parameters_0; + uint32_t vblank_parameters_1; + uint32_t vblank_parameters_2; + uint32_t vblank_parameters_3; + uint32_t vblank_parameters_4; + uint32_t vblank_parameters_5; + uint32_t vblank_parameters_6; + uint32_t vmid_settings_0; + + uint32_t hubpret_control; + uint32_t hubpret_interrupt; + uint32_t hubpret_mem_pwr_ctrl; + uint32_t hubpret_mem_pwr_status; + uint32_t hubpret_read_line_ctrl0; + uint32_t hubpret_read_line_ctrl1; + uint32_t hubpret_read_line_status; + uint32_t hubpret_read_line_value; + uint32_t hubpret_read_line0; + uint32_t hubpret_read_line1; +}; struct dcn_hubp_state { struct _vcs_dpi_display_dlg_regs_st dlg_attr; @@ -718,7 +849,6 @@ struct dcn_hubp_state { uint32_t hubp_cntl; uint32_t flip_control; }; - struct dcn10_hubp { struct hubp base; struct dcn_hubp_state state; diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h index f325db555102..7062e6653062 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h @@ -145,7 +145,8 @@ uint32_t FLIP_PARAMETERS_2;\ uint32_t DCN_CUR1_TTU_CNTL0;\ uint32_t DCN_CUR1_TTU_CNTL1;\ - uint32_t VMID_SETTINGS_0 + uint32_t VMID_SETTINGS_0;\ + uint32_t DST_Y_DELTA_DRQ_LIMIT /*shared with dcn3.x*/ #define DCN21_HUBP_REG_COMMON_VARIABLE_LIST \ @@ -176,7 +177,10 @@ uint32_t HUBP_3DLUT_CONTROL;\ uint32_t HUBP_3DLUT_DLG_PARAM;\ uint32_t DCSURF_VIEWPORT_MCACHE_SPLIT_COORDINATE;\ - uint32_t DCHUBP_MCACHEID_CONFIG + uint32_t DCHUBP_MCACHEID_CONFIG;\ + uint32_t DCHUBP_MALL_SUB_VP;\ + uint32_t DCHUBP_ADDR_CONFIG;\ + uint32_t HUBP_MALL_STATUS #define DCN2_HUBP_REG_FIELD_VARIABLE_LIST(type) \ DCN_HUBP_REG_FIELD_BASE_LIST(type); \ diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c index 556214b2227d..0cc6f4558989 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c @@ -476,6 +476,126 @@ void hubp3_read_state(struct hubp *hubp) } +void hubp3_read_reg_state(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + reg_state->hubp_cntl = REG_READ(DCHUBP_CNTL); + reg_state->mall_config = REG_READ(DCHUBP_MALL_CONFIG); + reg_state->mall_sub_vp = REG_READ(DCHUBP_MALL_SUB_VP); + reg_state->hubp_req_size_config = REG_READ(DCHUBP_REQ_SIZE_CONFIG); + reg_state->hubp_req_size_config_c = REG_READ(DCHUBP_REQ_SIZE_CONFIG_C); + reg_state->vmpg_config = REG_READ(DCHUBP_VMPG_CONFIG); + reg_state->addr_config = REG_READ(DCSURF_ADDR_CONFIG); + reg_state->pri_viewport_dimension = REG_READ(DCSURF_PRI_VIEWPORT_DIMENSION); + reg_state->pri_viewport_dimension_c = REG_READ(DCSURF_PRI_VIEWPORT_DIMENSION_C); + reg_state->pri_viewport_start = REG_READ(DCSURF_PRI_VIEWPORT_START); + reg_state->pri_viewport_start_c = REG_READ(DCSURF_PRI_VIEWPORT_START_C); + reg_state->sec_viewport_dimension = REG_READ(DCSURF_SEC_VIEWPORT_DIMENSION); + reg_state->sec_viewport_dimension_c = REG_READ(DCSURF_SEC_VIEWPORT_DIMENSION_C); + reg_state->sec_viewport_start = REG_READ(DCSURF_SEC_VIEWPORT_START); + reg_state->sec_viewport_start_c = REG_READ(DCSURF_SEC_VIEWPORT_START_C); + reg_state->surface_config = REG_READ(DCSURF_SURFACE_CONFIG); + reg_state->tiling_config = REG_READ(DCSURF_TILING_CONFIG); + reg_state->clk_cntl = REG_READ(HUBP_CLK_CNTL); + reg_state->mall_status = REG_READ(HUBP_MALL_STATUS); + reg_state->measure_win_ctrl_dcfclk = REG_READ(HUBP_MEASURE_WIN_CTRL_DCFCLK); + reg_state->measure_win_ctrl_dppclk = REG_READ(HUBP_MEASURE_WIN_CTRL_DPPCLK); + + reg_state->blank_offset_0 = REG_READ(BLANK_OFFSET_0); + reg_state->blank_offset_1 = REG_READ(BLANK_OFFSET_1); + reg_state->cursor_settings = REG_READ(CURSOR_SETTINGS); + reg_state->dcn_cur0_ttu_cntl0 = REG_READ(DCN_CUR0_TTU_CNTL0); + reg_state->dcn_cur0_ttu_cntl1 = REG_READ(DCN_CUR0_TTU_CNTL1); + reg_state->dcn_cur1_ttu_cntl0 = REG_READ(DCN_CUR1_TTU_CNTL0); + reg_state->dcn_cur1_ttu_cntl1 = REG_READ(DCN_CUR1_TTU_CNTL1); + reg_state->dcn_dmdat_vm_cntl = REG_READ(DCN_DMDATA_VM_CNTL); + reg_state->dcn_expansion_mode = REG_READ(DCN_EXPANSION_MODE); + reg_state->dcn_global_ttu_cntl = REG_READ(DCN_GLOBAL_TTU_CNTL); + reg_state->dcn_surf0_ttu_cntl0 = REG_READ(DCN_SURF0_TTU_CNTL0); + reg_state->dcn_surf0_ttu_cntl1 = REG_READ(DCN_SURF0_TTU_CNTL1); + reg_state->dcn_surf1_ttu_cntl0 = REG_READ(DCN_SURF1_TTU_CNTL0); + reg_state->dcn_surf1_ttu_cntl1 = REG_READ(DCN_SURF1_TTU_CNTL1); + reg_state->dcn_ttu_qos_wm = REG_READ(DCN_TTU_QOS_WM); + reg_state->dcn_vm_mx_l1_tlb_cntl = REG_READ(DCN_VM_MX_L1_TLB_CNTL); + reg_state->dcn_vm_system_aperture_high_addr = REG_READ(DCN_VM_SYSTEM_APERTURE_HIGH_ADDR); + reg_state->dcn_vm_system_aperture_low_addr = REG_READ(DCN_VM_SYSTEM_APERTURE_LOW_ADDR); + reg_state->dcsurf_flip_control = REG_READ(DCSURF_FLIP_CONTROL); + reg_state->dcsurf_flip_control2 = REG_READ(DCSURF_FLIP_CONTROL2); + reg_state->dcsurf_primary_meta_surface_address = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS); + reg_state->dcsurf_primary_meta_surface_address_c = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_C); + reg_state->dcsurf_primary_meta_surface_address_high = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_primary_meta_surface_address_high_c = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_primary_surface_address = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS); + reg_state->dcsurf_primary_surface_address_c = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_C); + reg_state->dcsurf_primary_surface_address_high = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_primary_surface_address_high_c = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_secondary_meta_surface_address = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS); + reg_state->dcsurf_secondary_meta_surface_address_c = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_C); + reg_state->dcsurf_secondary_meta_surface_address_high = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_secondary_meta_surface_address_high_c = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_secondary_surface_address = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS); + reg_state->dcsurf_secondary_surface_address_c = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_C); + reg_state->dcsurf_secondary_surface_address_high = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_secondary_surface_address_high_c = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_surface_control = REG_READ(DCSURF_SURFACE_CONTROL); + reg_state->dcsurf_surface_earliest_inuse = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE); + reg_state->dcsurf_surface_earliest_inuse_c = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_C); + reg_state->dcsurf_surface_earliest_inuse_high = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_HIGH); + reg_state->dcsurf_surface_earliest_inuse_high_c = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_HIGH_C); + reg_state->dcsurf_surface_flip_interrupt = REG_READ(DCSURF_SURFACE_FLIP_INTERRUPT); + reg_state->dcsurf_surface_inuse = REG_READ(DCSURF_SURFACE_INUSE); + reg_state->dcsurf_surface_inuse_c = REG_READ(DCSURF_SURFACE_INUSE_C); + reg_state->dcsurf_surface_inuse_high = REG_READ(DCSURF_SURFACE_INUSE_HIGH); + reg_state->dcsurf_surface_inuse_high_c = REG_READ(DCSURF_SURFACE_INUSE_HIGH_C); + reg_state->dcsurf_surface_pitch = REG_READ(DCSURF_SURFACE_PITCH); + reg_state->dcsurf_surface_pitch_c = REG_READ(DCSURF_SURFACE_PITCH_C); + reg_state->dst_after_scaler = REG_READ(DST_AFTER_SCALER); + reg_state->dst_dimensions = REG_READ(DST_DIMENSIONS); + reg_state->dst_y_delta_drq_limit = REG_READ(DST_Y_DELTA_DRQ_LIMIT); + reg_state->flip_parameters_0 = REG_READ(FLIP_PARAMETERS_0); + reg_state->flip_parameters_1 = REG_READ(FLIP_PARAMETERS_1); + reg_state->flip_parameters_2 = REG_READ(FLIP_PARAMETERS_2); + reg_state->flip_parameters_3 = REG_READ(FLIP_PARAMETERS_3); + reg_state->flip_parameters_4 = REG_READ(FLIP_PARAMETERS_4); + reg_state->flip_parameters_5 = REG_READ(FLIP_PARAMETERS_5); + reg_state->flip_parameters_6 = REG_READ(FLIP_PARAMETERS_6); + reg_state->hubpreq_mem_pwr_ctrl = REG_READ(HUBPREQ_MEM_PWR_CTRL); + reg_state->hubpreq_mem_pwr_status = REG_READ(HUBPREQ_MEM_PWR_STATUS); + reg_state->nom_parameters_0 = REG_READ(NOM_PARAMETERS_0); + reg_state->nom_parameters_1 = REG_READ(NOM_PARAMETERS_1); + reg_state->nom_parameters_2 = REG_READ(NOM_PARAMETERS_2); + reg_state->nom_parameters_3 = REG_READ(NOM_PARAMETERS_3); + reg_state->nom_parameters_4 = REG_READ(NOM_PARAMETERS_4); + reg_state->nom_parameters_5 = REG_READ(NOM_PARAMETERS_5); + reg_state->nom_parameters_6 = REG_READ(NOM_PARAMETERS_6); + reg_state->nom_parameters_7 = REG_READ(NOM_PARAMETERS_7); + reg_state->per_line_delivery = REG_READ(PER_LINE_DELIVERY); + reg_state->per_line_delivery_pre = REG_READ(PER_LINE_DELIVERY_PRE); + reg_state->prefetch_settings = REG_READ(PREFETCH_SETTINGS); + reg_state->prefetch_settings_c = REG_READ(PREFETCH_SETTINGS_C); + reg_state->ref_freq_to_pix_freq = REG_READ(REF_FREQ_TO_PIX_FREQ); + reg_state->uclk_pstate_force = REG_READ(UCLK_PSTATE_FORCE); + reg_state->vblank_parameters_0 = REG_READ(VBLANK_PARAMETERS_0); + reg_state->vblank_parameters_1 = REG_READ(VBLANK_PARAMETERS_1); + reg_state->vblank_parameters_2 = REG_READ(VBLANK_PARAMETERS_2); + reg_state->vblank_parameters_3 = REG_READ(VBLANK_PARAMETERS_3); + reg_state->vblank_parameters_4 = REG_READ(VBLANK_PARAMETERS_4); + reg_state->vblank_parameters_5 = REG_READ(VBLANK_PARAMETERS_5); + reg_state->vblank_parameters_6 = REG_READ(VBLANK_PARAMETERS_6); + reg_state->vmid_settings_0 = REG_READ(VMID_SETTINGS_0); + reg_state->hubpret_control = REG_READ(HUBPRET_CONTROL); + reg_state->hubpret_interrupt = REG_READ(HUBPRET_INTERRUPT); + reg_state->hubpret_mem_pwr_ctrl = REG_READ(HUBPRET_MEM_PWR_CTRL); + reg_state->hubpret_mem_pwr_status = REG_READ(HUBPRET_MEM_PWR_STATUS); + reg_state->hubpret_read_line_ctrl0 = REG_READ(HUBPRET_READ_LINE_CTRL0); + reg_state->hubpret_read_line_ctrl1 = REG_READ(HUBPRET_READ_LINE_CTRL1); + reg_state->hubpret_read_line_status = REG_READ(HUBPRET_READ_LINE_STATUS); + reg_state->hubpret_read_line_value = REG_READ(HUBPRET_READ_LINE_VALUE); + reg_state->hubpret_read_line0 = REG_READ(HUBPRET_READ_LINE0); + reg_state->hubpret_read_line1 = REG_READ(HUBPRET_READ_LINE1); +} + void hubp3_setup( struct hubp *hubp, struct _vcs_dpi_display_dlg_regs_st *dlg_attr, @@ -505,30 +625,6 @@ void hubp3_init(struct hubp *hubp) hubp_reset(hubp); } -uint32_t hubp3_get_current_read_line(struct hubp *hubp) -{ - uint32_t read_line = 0; - struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - - REG_GET(HUBPRET_READ_LINE_VALUE, - PIPE_READ_LINE, - &read_line); - - return read_line; -} - -unsigned int hubp3_get_underflow_status(struct hubp *hubp) -{ - uint32_t hubp_underflow = 0; - struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - - REG_GET(DCHUBP_CNTL, - HUBP_UNDERFLOW_STATUS, - &hubp_underflow); - - return hubp_underflow; -} - static struct hubp_funcs dcn30_hubp_funcs = { .hubp_enable_tripleBuffer = hubp2_enable_triplebuffer, .hubp_is_triplebuffer_enabled = hubp2_is_triplebuffer_enabled, @@ -558,8 +654,7 @@ static struct hubp_funcs dcn30_hubp_funcs = { .hubp_soft_reset = hubp1_soft_reset, .hubp_set_flip_int = hubp1_set_flip_int, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp3_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h index 842f4eb72cc8..c767e9f4f9b3 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h @@ -296,6 +296,8 @@ void hubp3_dmdata_set_attributes( void hubp3_read_state(struct hubp *hubp); +void hubp3_read_reg_state(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state); + void hubp3_init(struct hubp *hubp); void hubp3_clear_tiling(struct hubp *hubp); diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c index 47101847c2b7..189045f85039 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c @@ -110,9 +110,7 @@ static struct hubp_funcs dcn31_hubp_funcs = { .hubp_in_blank = hubp1_in_blank, .program_extended_blank = hubp31_program_extended_blank, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state, }; bool hubp31_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c index 41c8f78efdc3..a781085b046b 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c @@ -222,9 +222,7 @@ static struct hubp_funcs dcn32_hubp_funcs = { .hubp_update_mall_sel = hubp32_update_mall_sel, .hubp_prepare_subvp_buffering = hubp32_prepare_subvp_buffering, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp32_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c index b140808f21af..79c583e258c7 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c @@ -209,6 +209,7 @@ static struct hubp_funcs dcn35_hubp_funcs = { .dmdata_load = hubp2_dmdata_load, .dmdata_status_done = hubp2_dmdata_status_done, .hubp_read_state = hubp3_read_state, + .hubp_read_reg_state = hubp3_read_reg_state, .hubp_clear_underflow = hubp2_clear_underflow, .hubp_set_flip_control_surface_gsl = hubp2_set_flip_control_surface_gsl, .hubp_init = hubp35_init, @@ -218,9 +219,6 @@ static struct hubp_funcs dcn35_hubp_funcs = { .hubp_in_blank = hubp1_in_blank, .program_extended_blank = hubp31_program_extended_blank_value, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, }; bool hubp35_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c index 0fcbc6a35be6..3a2e0848173e 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c @@ -1071,9 +1071,7 @@ static struct hubp_funcs dcn401_hubp_funcs = { .hubp_get_3dlut_fl_done = hubp401_get_3dlut_fl_done, .hubp_clear_tiling = hubp401_clear_tiling, .hubp_program_3dlut_fl_config = hubp401_program_3dlut_fl_config, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp401_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index e47ed5571dfd..81bcadf5e57e 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -53,7 +53,8 @@ #include "link_service.h" #include "dc_state_priv.h" - +#define TO_DCN_DCCG(dccg)\ + container_of(dccg, struct dcn_dccg, base) #define DC_LOGGER_INIT(logger) @@ -1235,44 +1236,47 @@ void dcn30_get_underflow_debug_data(const struct dc *dc, { struct hubbub *hubbub = dc->res_pool->hubbub; - if (tg) { - uint32_t v_blank_start = 0, v_blank_end = 0; - - out_data->otg_inst = tg->inst; - - tg->funcs->get_scanoutpos(tg, - &v_blank_start, - &v_blank_end, - &out_data->h_position, - &out_data->v_position); - - out_data->otg_frame_count = tg->funcs->get_frame_count(tg); - - out_data->otg_underflow = tg->funcs->is_optc_underflow_occurred(tg); + if (hubbub) { + if (hubbub->funcs->hubbub_read_reg_state) { + hubbub->funcs->hubbub_read_reg_state(hubbub, out_data->hubbub_reg_state); + } } for (int i = 0; i < MAX_PIPES; i++) { struct hubp *hubp = dc->res_pool->hubps[i]; - - if (hubp) { - if (hubp->funcs->hubp_get_underflow_status) - out_data->hubps[i].hubp_underflow = hubp->funcs->hubp_get_underflow_status(hubp); - - if (hubp->funcs->hubp_in_blank) - out_data->hubps[i].hubp_in_blank = hubp->funcs->hubp_in_blank(hubp); - - if (hubp->funcs->hubp_get_current_read_line) - out_data->hubps[i].hubp_readline = hubp->funcs->hubp_get_current_read_line(hubp); - - if (hubp->funcs->hubp_get_det_config_error) - out_data->hubps[i].det_config_error = hubp->funcs->hubp_get_det_config_error(hubp); - } + struct dpp *dpp = dc->res_pool->dpps[i]; + struct output_pixel_processor *opp = dc->res_pool->opps[i]; + struct display_stream_compressor *dsc = dc->res_pool->dscs[i]; + struct mpc *mpc = dc->res_pool->mpc; + struct timing_generator *optc = dc->res_pool->timing_generators[i]; + struct dccg *dccg = dc->res_pool->dccg; + + if (hubp) + if (hubp->funcs->hubp_read_reg_state) + hubp->funcs->hubp_read_reg_state(hubp, out_data->hubp_reg_state[i]); + + if (dpp) + if (dpp->funcs->dpp_read_reg_state) + dpp->funcs->dpp_read_reg_state(dpp, out_data->dpp_reg_state[i]); + + if (opp) + if (opp->funcs->opp_read_reg_state) + opp->funcs->opp_read_reg_state(opp, out_data->opp_reg_state[i]); + + if (dsc) + if (dsc->funcs->dsc_read_reg_state) + dsc->funcs->dsc_read_reg_state(dsc, out_data->dsc_reg_state[i]); + + if (mpc) + if (mpc->funcs->mpc_read_reg_state) + mpc->funcs->mpc_read_reg_state(mpc, i, out_data->mpc_reg_state[i]); + + if (optc) + if (optc->funcs->optc_read_reg_state) + optc->funcs->optc_read_reg_state(optc, out_data->optc_reg_state[i]); + + if (dccg) + if (dccg->funcs->dccg_read_reg_state) + dccg->funcs->dccg_read_reg_state(dccg, out_data->dccg_reg_state[i]); } - - if (hubbub->funcs->get_det_sizes) - hubbub->funcs->get_det_sizes(hubbub, out_data->curr_det_sizes, out_data->target_det_sizes); - - if (hubbub->funcs->compbuf_config_error) - out_data->compbuf_config_error = hubbub->funcs->compbuf_config_error(hubbub); - } diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h index 61c4d2a7db1c..500a601e99b5 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h @@ -71,6 +71,125 @@ enum pixel_rate_div { PIXEL_RATE_DIV_NA = 0xF }; +struct dcn_dccg_reg_state { + uint32_t dc_mem_global_pwr_req_cntl; + uint32_t dccg_audio_dtbclk_dto_modulo; + uint32_t dccg_audio_dtbclk_dto_phase; + uint32_t dccg_audio_dto_source; + uint32_t dccg_audio_dto0_module; + uint32_t dccg_audio_dto0_phase; + uint32_t dccg_audio_dto1_module; + uint32_t dccg_audio_dto1_phase; + uint32_t dccg_cac_status; + uint32_t dccg_cac_status2; + uint32_t dccg_disp_cntl_reg; + uint32_t dccg_ds_cntl; + uint32_t dccg_ds_dto_incr; + uint32_t dccg_ds_dto_modulo; + uint32_t dccg_ds_hw_cal_interval; + uint32_t dccg_gate_disable_cntl; + uint32_t dccg_gate_disable_cntl2; + uint32_t dccg_gate_disable_cntl3; + uint32_t dccg_gate_disable_cntl4; + uint32_t dccg_gate_disable_cntl5; + uint32_t dccg_gate_disable_cntl6; + uint32_t dccg_global_fgcg_rep_cntl; + uint32_t dccg_gtc_cntl; + uint32_t dccg_gtc_current; + uint32_t dccg_gtc_dto_incr; + uint32_t dccg_gtc_dto_modulo; + uint32_t dccg_perfmon_cntl; + uint32_t dccg_perfmon_cntl2; + uint32_t dccg_soft_reset; + uint32_t dccg_test_clk_sel; + uint32_t dccg_vsync_cnt_ctrl; + uint32_t dccg_vsync_cnt_int_ctrl; + uint32_t dccg_vsync_otg0_latch_value; + uint32_t dccg_vsync_otg1_latch_value; + uint32_t dccg_vsync_otg2_latch_value; + uint32_t dccg_vsync_otg3_latch_value; + uint32_t dccg_vsync_otg4_latch_value; + uint32_t dccg_vsync_otg5_latch_value; + uint32_t dispclk_cgtt_blk_ctrl_reg; + uint32_t dispclk_freq_change_cntl; + uint32_t dp_dto_dbuf_en; + uint32_t dp_dto0_modulo; + uint32_t dp_dto0_phase; + uint32_t dp_dto1_modulo; + uint32_t dp_dto1_phase; + uint32_t dp_dto2_modulo; + uint32_t dp_dto2_phase; + uint32_t dp_dto3_modulo; + uint32_t dp_dto3_phase; + uint32_t dpiaclk_540m_dto_modulo; + uint32_t dpiaclk_540m_dto_phase; + uint32_t dpiaclk_810m_dto_modulo; + uint32_t dpiaclk_810m_dto_phase; + uint32_t dpiaclk_dto_cntl; + uint32_t dpiasymclk_cntl; + uint32_t dppclk_cgtt_blk_ctrl_reg; + uint32_t dppclk_ctrl; + uint32_t dppclk_dto_ctrl; + uint32_t dppclk0_dto_param; + uint32_t dppclk1_dto_param; + uint32_t dppclk2_dto_param; + uint32_t dppclk3_dto_param; + uint32_t dprefclk_cgtt_blk_ctrl_reg; + uint32_t dprefclk_cntl; + uint32_t dpstreamclk_cntl; + uint32_t dscclk_dto_ctrl; + uint32_t dscclk0_dto_param; + uint32_t dscclk1_dto_param; + uint32_t dscclk2_dto_param; + uint32_t dscclk3_dto_param; + uint32_t dtbclk_dto_dbuf_en; + uint32_t dtbclk_dto0_modulo; + uint32_t dtbclk_dto0_phase; + uint32_t dtbclk_dto1_modulo; + uint32_t dtbclk_dto1_phase; + uint32_t dtbclk_dto2_modulo; + uint32_t dtbclk_dto2_phase; + uint32_t dtbclk_dto3_modulo; + uint32_t dtbclk_dto3_phase; + uint32_t dtbclk_p_cntl; + uint32_t force_symclk_disable; + uint32_t hdmicharclk0_clock_cntl; + uint32_t hdmistreamclk_cntl; + uint32_t hdmistreamclk0_dto_param; + uint32_t microsecond_time_base_div; + uint32_t millisecond_time_base_div; + uint32_t otg_pixel_rate_div; + uint32_t otg0_phypll_pixel_rate_cntl; + uint32_t otg0_pixel_rate_cntl; + uint32_t otg1_phypll_pixel_rate_cntl; + uint32_t otg1_pixel_rate_cntl; + uint32_t otg2_phypll_pixel_rate_cntl; + uint32_t otg2_pixel_rate_cntl; + uint32_t otg3_phypll_pixel_rate_cntl; + uint32_t otg3_pixel_rate_cntl; + uint32_t phyasymclk_clock_cntl; + uint32_t phybsymclk_clock_cntl; + uint32_t phycsymclk_clock_cntl; + uint32_t phydsymclk_clock_cntl; + uint32_t phyesymclk_clock_cntl; + uint32_t phyplla_pixclk_resync_cntl; + uint32_t phypllb_pixclk_resync_cntl; + uint32_t phypllc_pixclk_resync_cntl; + uint32_t phyplld_pixclk_resync_cntl; + uint32_t phyplle_pixclk_resync_cntl; + uint32_t refclk_cgtt_blk_ctrl_reg; + uint32_t socclk_cgtt_blk_ctrl_reg; + uint32_t symclk_cgtt_blk_ctrl_reg; + uint32_t symclk_psp_cntl; + uint32_t symclk32_le_cntl; + uint32_t symclk32_se_cntl; + uint32_t symclka_clock_enable; + uint32_t symclkb_clock_enable; + uint32_t symclkc_clock_enable; + uint32_t symclkd_clock_enable; + uint32_t symclke_clock_enable; +}; + struct dccg { struct dc_context *ctx; const struct dccg_funcs *funcs; @@ -81,7 +200,6 @@ struct dccg { //int audio_dtbclk_khz;/* TODO needs to be removed */ //int ref_dtbclk_khz;/* TODO needs to be removed */ }; - struct dtbclk_dto_params { const struct dc_crtc_timing *timing; int otg_inst; @@ -214,6 +332,7 @@ struct dccg_funcs { void (*set_dto_dscclk)(struct dccg *dccg, uint32_t dsc_inst, uint32_t num_slices_h); void (*set_ref_dscclk)(struct dccg *dccg, uint32_t dsc_inst); void (*dccg_root_gate_disable_control)(struct dccg *dccg, uint32_t pipe_idx, uint32_t disable_clock_gating); + void (*dccg_read_reg_state)(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state); }; #endif //__DAL_DCCG_H__ diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h index 2ce47c403840..dafc8490efb5 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h @@ -137,6 +137,14 @@ struct dcn_hubbub_state { uint32_t dram_state_cntl; }; +struct dcn_hubbub_reg_state { + uint32_t det0_ctrl; + uint32_t det1_ctrl; + uint32_t det2_ctrl; + uint32_t det3_ctrl; + uint32_t compbuf_ctrl; +}; + struct hubbub_system_latencies { uint32_t max_latency_ns; uint32_t avg_latency_ns; @@ -216,6 +224,8 @@ struct hubbub_funcs { void (*init_watermarks)(struct hubbub *hubbub); + void (*hubbub_read_reg_state)(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state); + /** * @program_det_size: * @@ -242,8 +252,6 @@ struct hubbub_funcs { void (*program_compbuf_segments)(struct hubbub *hubbub, unsigned compbuf_size_seg, bool safe_to_increase); void (*wait_for_det_update)(struct hubbub *hubbub, int hubp_inst); bool (*program_arbiter)(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs, bool safe_to_lower); - void (*get_det_sizes)(struct hubbub *hubbub, uint32_t *curr_det_sizes, uint32_t *target_det_sizes); - uint32_t (*compbuf_config_error)(struct hubbub *hubbub); struct hubbub_perfmon_funcs { void (*reset)(struct hubbub *hubbub); void (*start_measuring_max_memory_latency_ns)( diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h index 09c224691618..d88b57d4f512 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h @@ -65,7 +65,6 @@ union defer_reg_writes { } bits; uint32_t raw; }; - struct dpp { const struct dpp_funcs *funcs; struct dc_context *ctx; @@ -203,6 +202,19 @@ struct dcn_dpp_state { uint32_t gamcor_mode; }; +struct dcn_dpp_reg_state { + uint32_t recout_start; + uint32_t recout_size; + uint32_t scl_horz_filter_scale_ratio; + uint32_t scl_vert_filter_scale_ratio; + uint32_t scl_mode; + uint32_t cm_control; + uint32_t dpp_control; + uint32_t dscl_control; + uint32_t obuf_control; + uint32_t mpc_size; +}; + struct CM_bias_params { uint32_t cm_bias_cr_r; uint32_t cm_bias_y_g; @@ -226,6 +238,8 @@ struct dpp_funcs { void (*dpp_read_state)(struct dpp *dpp, struct dcn_dpp_state *s); + void (*dpp_read_reg_state)(struct dpp *dpp, struct dcn_dpp_reg_state *dpp_reg_state); + void (*dpp_reset)(struct dpp *dpp); void (*dpp_set_scaler)(struct dpp *dpp, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h index c62c4c1ed7c3..a79019365af8 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h @@ -238,6 +238,7 @@ struct hubp_funcs { void (*hubp_clk_cntl)(struct hubp *hubp, bool enable); void (*hubp_vtg_sel)(struct hubp *hubp, uint32_t otg_inst); void (*hubp_read_state)(struct hubp *hubp); + void (*hubp_read_reg_state)(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state); void (*hubp_clear_underflow)(struct hubp *hubp); void (*hubp_disable_control)(struct hubp *hubp, bool disable_hubp); unsigned int (*hubp_get_underflow_status)(struct hubp *hubp); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h index 22960ee03dee..a8d1abe20f62 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h @@ -350,6 +350,15 @@ struct mpcc_state { struct mpc_rmcm_regs rmcm_regs; }; +struct dcn_mpc_reg_state { + uint32_t mpcc_bot_sel; + uint32_t mpcc_control; + uint32_t mpcc_status; + uint32_t mpcc_top_sel; + uint32_t mpcc_opp_id; + uint32_t mpcc_ogam_control; +}; + /** * struct mpc_funcs - funcs */ @@ -373,6 +382,24 @@ struct mpc_funcs { struct mpc *mpc, int mpcc_inst, struct mpcc_state *s); + /** + * @mpc_read_reg_state: + * + * Read MPC register state for debugging underflow purposes. + * + * Parameters: + * + * - [in] mpc - MPC context + * - [out] reg_state - MPC register state structure + * + * Return: + * + * void + */ + void (*mpc_read_reg_state)( + struct mpc *mpc, + int mpcc_inst, + struct dcn_mpc_reg_state *mpc_reg_state); /** * @insert_plane: diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h index 747679cb4944..e1428a83ecbc 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h @@ -297,6 +297,16 @@ struct oppbuf_params { uint32_t num_segment_padded_pixels; }; +struct dcn_opp_reg_state { + uint32_t dpg_control; + uint32_t fmt_control; + uint32_t oppbuf_control; + uint32_t opp_pipe_control; + uint32_t opp_pipe_crc_control; + uint32_t opp_abm_control; + uint32_t dscrm_dsc_forward_config; +}; + struct opp_funcs { @@ -368,6 +378,9 @@ struct opp_funcs { struct output_pixel_processor *opp, enum dc_pixel_encoding pixel_encoding, bool is_primary); + + void (*opp_read_reg_state)( + struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index f2de2cf23859..da7bf59c4b9d 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -175,6 +175,135 @@ struct dcn_otg_state { uint32_t otg_double_buffer_control; }; +struct dcn_optc_reg_state { + uint32_t optc_bytes_per_pixel; + uint32_t optc_data_format_control; + uint32_t optc_data_source_select; + uint32_t optc_input_clock_control; + uint32_t optc_input_global_control; + uint32_t optc_input_spare_register; + uint32_t optc_memory_config; + uint32_t optc_rsmu_underflow; + uint32_t optc_underflow_threshold; + uint32_t optc_width_control; + + uint32_t otg_3d_structure_control; + uint32_t otg_clock_control; + uint32_t otg_control; + uint32_t otg_count_control; + uint32_t otg_count_reset; + uint32_t otg_crc_cntl; + uint32_t otg_crc_sig_blue_control_mask; + uint32_t otg_crc_sig_red_green_mask; + uint32_t otg_crc0_data_b; + uint32_t otg_crc0_data_rg; + uint32_t otg_crc0_windowa_x_control; + uint32_t otg_crc0_windowa_x_control_readback; + uint32_t otg_crc0_windowa_y_control; + uint32_t otg_crc0_windowa_y_control_readback; + uint32_t otg_crc0_windowb_x_control; + uint32_t otg_crc0_windowb_x_control_readback; + uint32_t otg_crc0_windowb_y_control; + uint32_t otg_crc0_windowb_y_control_readback; + uint32_t otg_crc1_data_b; + uint32_t otg_crc1_data_rg; + uint32_t otg_crc1_windowa_x_control; + uint32_t otg_crc1_windowa_x_control_readback; + uint32_t otg_crc1_windowa_y_control; + uint32_t otg_crc1_windowa_y_control_readback; + uint32_t otg_crc1_windowb_x_control; + uint32_t otg_crc1_windowb_x_control_readback; + uint32_t otg_crc1_windowb_y_control; + uint32_t otg_crc1_windowb_y_control_readback; + uint32_t otg_crc2_data_b; + uint32_t otg_crc2_data_rg; + uint32_t otg_crc3_data_b; + uint32_t otg_crc3_data_rg; + uint32_t otg_dlpc_control; + uint32_t otg_double_buffer_control; + uint32_t otg_drr_control2; + uint32_t otg_drr_control; + uint32_t otg_drr_timing_int_status; + uint32_t otg_drr_trigger_window; + uint32_t otg_drr_v_total_change; + uint32_t otg_drr_v_total_reach_range; + uint32_t otg_dsc_start_position; + uint32_t otg_force_count_now_cntl; + uint32_t otg_global_control0; + uint32_t otg_global_control1; + uint32_t otg_global_control2; + uint32_t otg_global_control3; + uint32_t otg_global_control4; + uint32_t otg_global_sync_status; + uint32_t otg_gsl_control; + uint32_t otg_gsl_vsync_gap; + uint32_t otg_gsl_window_x; + uint32_t otg_gsl_window_y; + uint32_t otg_h_blank_start_end; + uint32_t otg_h_sync_a; + uint32_t otg_h_sync_a_cntl; + uint32_t otg_h_timing_cntl; + uint32_t otg_h_total; + uint32_t otg_interlace_control; + uint32_t otg_interlace_status; + uint32_t otg_interrupt_control; + uint32_t otg_long_vblank_status; + uint32_t otg_m_const_dto0; + uint32_t otg_m_const_dto1; + uint32_t otg_manual_force_vsync_next_line; + uint32_t otg_master_en; + uint32_t otg_master_update_lock; + uint32_t otg_master_update_mode; + uint32_t otg_nom_vert_position; + uint32_t otg_pipe_update_status; + uint32_t otg_pixel_data_readback0; + uint32_t otg_pixel_data_readback1; + uint32_t otg_request_control; + uint32_t otg_snapshot_control; + uint32_t otg_snapshot_frame; + uint32_t otg_snapshot_position; + uint32_t otg_snapshot_status; + uint32_t otg_spare_register; + uint32_t otg_static_screen_control; + uint32_t otg_status; + uint32_t otg_status_frame_count; + uint32_t otg_status_hv_count; + uint32_t otg_status_position; + uint32_t otg_status_vf_count; + uint32_t otg_stereo_control; + uint32_t otg_stereo_force_next_eye; + uint32_t otg_stereo_status; + uint32_t otg_trig_manual_control; + uint32_t otg_triga_cntl; + uint32_t otg_triga_manual_trig; + uint32_t otg_trigb_cntl; + uint32_t otg_trigb_manual_trig; + uint32_t otg_update_lock; + uint32_t otg_v_blank_start_end; + uint32_t otg_v_count_stop_control; + uint32_t otg_v_count_stop_control2; + uint32_t otg_v_sync_a; + uint32_t otg_v_sync_a_cntl; + uint32_t otg_v_total; + uint32_t otg_v_total_control; + uint32_t otg_v_total_int_status; + uint32_t otg_v_total_max; + uint32_t otg_v_total_mid; + uint32_t otg_v_total_min; + uint32_t otg_vert_sync_control; + uint32_t otg_vertical_interrupt0_control; + uint32_t otg_vertical_interrupt0_position; + uint32_t otg_vertical_interrupt1_control; + uint32_t otg_vertical_interrupt1_position; + uint32_t otg_vertical_interrupt2_control; + uint32_t otg_vertical_interrupt2_position; + uint32_t otg_vready_param; + uint32_t otg_vstartup_param; + uint32_t otg_vsync_nom_int_status; + uint32_t otg_vupdate_keepout; + uint32_t otg_vupdate_param; +}; + /** * struct timing_generator - Entry point to Output Timing Generator feature. */ @@ -381,6 +510,7 @@ struct timing_generator_funcs { void (*set_vupdate_keepout)(struct timing_generator *tg, bool enable); bool (*wait_update_lock_status)(struct timing_generator *tg, bool locked); void (*read_otg_state)(struct timing_generator *tg, struct dcn_otg_state *s); + void (*optc_read_reg_state)(struct timing_generator *tg, struct dcn_optc_reg_state *optc_reg_state); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c index 85298b8a1b5e..6bfd2c1294e5 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c @@ -1514,6 +1514,21 @@ static void mpc3_read_mpcc_state( MPCC_OGAM_SELECT_CURRENT, &s->rgam_lut); } +void mpc3_read_reg_state( + struct mpc *mpc, + int mpcc_inst, struct dcn_mpc_reg_state *mpc_reg_state) +{ + struct dcn30_mpc *mpc30 = TO_DCN30_MPC(mpc); + + mpc_reg_state->mpcc_bot_sel = REG_READ(MPCC_BOT_SEL[mpcc_inst]); + mpc_reg_state->mpcc_control = REG_READ(MPCC_CONTROL[mpcc_inst]); + mpc_reg_state->mpcc_ogam_control = REG_READ(MPCC_OGAM_CONTROL[mpcc_inst]); + mpc_reg_state->mpcc_opp_id = REG_READ(MPCC_OPP_ID[mpcc_inst]); + mpc_reg_state->mpcc_status = REG_READ(MPCC_STATUS[mpcc_inst]); + mpc_reg_state->mpcc_top_sel = REG_READ(MPCC_TOP_SEL[mpcc_inst]); + +} + static const struct mpc_funcs dcn30_mpc_funcs = { .read_mpcc_state = mpc3_read_mpcc_state, .insert_plane = mpc1_insert_plane, @@ -1544,6 +1559,7 @@ static const struct mpc_funcs dcn30_mpc_funcs = { .release_rmu = mpcc3_release_rmu, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, .set_mpc_mem_lp_mode = mpc3_set_mpc_mem_lp_mode, }; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h index 103f29900a2c..e2f147d17178 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h @@ -1096,6 +1096,11 @@ void mpc3_power_on_ogam_lut( struct mpc *mpc, int mpcc_id, bool power_on); +void mpc3_read_reg_state( + struct mpc *mpc, + int mpcc_inst, + struct dcn_mpc_reg_state *mpc_reg_state); + void mpc3_init_mpcc(struct mpcc *mpcc, int mpcc_inst); enum dc_lut_mode mpc3_get_ogam_current( diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c index 6f0e017a8ae2..83bbbf34bcac 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c @@ -1020,6 +1020,7 @@ static const struct mpc_funcs dcn32_mpc_funcs = { .release_rmu = NULL, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, }; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c index e1a0308dee57..eeac13fdd6f5 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c @@ -598,6 +598,7 @@ static const struct mpc_funcs dcn401_mpc_funcs = { .release_rmu = NULL, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, .set_movable_cm_location = mpc401_set_movable_cm_location, .update_3dlut_fast_load_select = mpc401_update_3dlut_fast_load_select, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c index 71e9288d60ed..45d418636d0c 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c @@ -372,6 +372,17 @@ void opp1_pipe_clock_control(struct output_pixel_processor *opp, bool enable) REG_UPDATE(OPP_PIPE_CONTROL, OPP_PIPE_CLOCK_EN, regval); } + +void opp1_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn10_opp *oppn10 = TO_DCN10_OPP(opp); + + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); +} + /*****************************************/ /* Constructor, Destructor */ /*****************************************/ @@ -392,7 +403,8 @@ static const struct opp_funcs dcn10_opp_funcs = { .opp_program_dpg_dimensions = NULL, .dpg_is_blanked = NULL, .dpg_is_pending = NULL, - .opp_destroy = opp1_destroy + .opp_destroy = opp1_destroy, + .opp_read_reg_state = opp1_read_reg_state }; void dcn10_opp_construct(struct dcn10_opp *oppn10, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h index c87de68a509e..38d0d530a9b7 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h @@ -63,7 +63,8 @@ uint32_t OPPBUF_CONTROL1; \ uint32_t OPPBUF_3D_PARAMETERS_0; \ uint32_t OPPBUF_3D_PARAMETERS_1; \ - uint32_t OPP_PIPE_CONTROL + uint32_t OPP_PIPE_CONTROL; \ + uint32_t OPP_PIPE_CRC_CONTROL #define OPP_MASK_SH_LIST_DCN(mask_sh) \ OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, mask_sh), \ @@ -153,7 +154,6 @@ struct dcn10_opp { const struct dcn10_opp_registers *regs; const struct dcn10_opp_shift *opp_shift; const struct dcn10_opp_mask *opp_mask; - bool is_write_to_ram_a_safe; }; @@ -188,4 +188,6 @@ void opp1_pipe_clock_control(struct output_pixel_processor *opp, bool enable); void opp1_destroy(struct output_pixel_processor **opp); +void opp1_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c index f5fe0cac7cb0..ce826a5be4c7 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c @@ -377,6 +377,18 @@ uint32_t opp2_get_left_edge_extra_pixel_count(struct output_pixel_processor *opp return 0; } +void opp2_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn20_opp *oppn20 = TO_DCN20_OPP(opp); + + opp_reg_state->dpg_control = REG_READ(DPG_CONTROL); + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); + opp_reg_state->dscrm_dsc_forward_config = REG_READ(DSCRM_DSC_FORWARD_CONFIG); +} + /*****************************************/ /* Constructor, Destructor */ /*****************************************/ @@ -395,6 +407,7 @@ static struct opp_funcs dcn20_opp_funcs = { .opp_destroy = opp1_destroy, .opp_program_left_edge_extra_pixel = opp2_program_left_edge_extra_pixel, .opp_get_left_edge_extra_pixel_count = opp2_get_left_edge_extra_pixel_count, + .opp_read_reg_state = opp2_read_reg_state }; void dcn20_opp_construct(struct dcn20_opp *oppn20, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h index 34936e6c49f3..fb0c047c1788 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h @@ -59,7 +59,8 @@ uint32_t DPG_COLOUR_G_Y; \ uint32_t DPG_COLOUR_R_CR; \ uint32_t DPG_RAMP_CONTROL; \ - uint32_t DPG_STATUS + uint32_t DPG_STATUS; \ + uint32_t DSCRM_DSC_FORWARD_CONFIG #define OPP_DPG_MASK_SH_LIST(mask_sh) \ OPP_SF(DPG0_DPG_CONTROL, DPG_EN, mask_sh), \ @@ -171,4 +172,7 @@ void opp2_program_left_edge_extra_pixel ( uint32_t opp2_get_left_edge_extra_pixel_count(struct output_pixel_processor *opp, enum dc_pixel_encoding pixel_encoding, bool is_primary); + +void opp2_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c index 3542b51c9aac..e11c4e16402f 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c @@ -51,3 +51,16 @@ void dcn35_opp_set_fgcg(struct dcn20_opp *oppn20, bool enable) { REG_UPDATE(OPP_TOP_CLK_CONTROL, OPP_FGCG_REP_DIS, !enable); } + +void dcn35_opp_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn20_opp *oppn20 = TO_DCN20_OPP(opp); + + opp_reg_state->dpg_control = REG_READ(DPG_CONTROL); + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_abm_control = REG_READ(OPP_ABM_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); + opp_reg_state->dscrm_dsc_forward_config = REG_READ(DSCRM_DSC_FORWARD_CONFIG); +} diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h index a9a413527801..c6cace90e8f2 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h @@ -31,7 +31,8 @@ #define OPP_REG_VARIABLE_LIST_DCN3_5 \ OPP_REG_VARIABLE_LIST_DCN2_0; \ - uint32_t OPP_TOP_CLK_CONTROL + uint32_t OPP_TOP_CLK_CONTROL; \ + uint32_t OPP_ABM_CONTROL #define OPP_MASK_SH_LIST_DCN35(mask_sh) \ OPP_MASK_SH_LIST_DCN20(mask_sh), \ @@ -64,4 +65,5 @@ void dcn35_opp_construct(struct dcn20_opp *oppn20, void dcn35_opp_set_fgcg(struct dcn20_opp *oppn20, bool enable); +void dcn35_opp_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); #endif diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h index 8b2a8455eb56..803bcc25601c 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h @@ -209,7 +209,43 @@ uint32_t OPTC_WIDTH_CONTROL2; \ uint32_t OTG_PSTATE_REGISTER; \ uint32_t OTG_PIPE_UPDATE_STATUS; \ - uint32_t INTERRUPT_DEST + uint32_t INTERRUPT_DEST; \ + uint32_t OPTC_INPUT_SPARE_REGISTER; \ + uint32_t OPTC_RSMU_UNDERFLOW; \ + uint32_t OPTC_UNDERFLOW_THRESHOLD; \ + uint32_t OTG_COUNT_CONTROL; \ + uint32_t OTG_COUNT_RESET; \ + uint32_t OTG_CRC_SIG_BLUE_CONTROL_MASK; \ + uint32_t OTG_CRC_SIG_RED_GREEN_MASK; \ + uint32_t OTG_DLPC_CONTROL; \ + uint32_t OTG_DRR_CONTROL2; \ + uint32_t OTG_DRR_TIMING_INT_STATUS; \ + uint32_t OTG_GLOBAL_CONTROL3; \ + uint32_t OTG_GLOBAL_SYNC_STATUS; \ + uint32_t OTG_GSL_VSYNC_GAP; \ + uint32_t OTG_INTERLACE_STATUS; \ + uint32_t OTG_INTERRUPT_CONTROL; \ + uint32_t OTG_LONG_VBLANK_STATUS; \ + uint32_t OTG_MANUAL_FORCE_VSYNC_NEXT_LINE; \ + uint32_t OTG_MASTER_EN; \ + uint32_t OTG_PIXEL_DATA_READBACK0; \ + uint32_t OTG_PIXEL_DATA_READBACK1; \ + uint32_t OTG_REQUEST_CONTROL; \ + uint32_t OTG_SNAPSHOT_CONTROL; \ + uint32_t OTG_SNAPSHOT_FRAME; \ + uint32_t OTG_SNAPSHOT_POSITION; \ + uint32_t OTG_SNAPSHOT_STATUS; \ + uint32_t OTG_SPARE_REGISTER; \ + uint32_t OTG_STATUS_HV_COUNT; \ + uint32_t OTG_STATUS_VF_COUNT; \ + uint32_t OTG_STEREO_FORCE_NEXT_EYE; \ + uint32_t OTG_TRIG_MANUAL_CONTROL; \ + uint32_t OTG_TRIGB_CNTL; \ + uint32_t OTG_TRIGB_MANUAL_TRIG; \ + uint32_t OTG_UPDATE_LOCK; \ + uint32_t OTG_V_TOTAL_INT_STATUS; \ + uint32_t OTG_VSYNC_NOM_INT_STATUS + struct dcn_optc_registers { OPTC_REG_VARIABLE_LIST_DCN; diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c index 4f1830ba619f..c6417538090f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c @@ -315,6 +315,136 @@ void optc31_read_otg_state(struct timing_generator *optc, s->otg_double_buffer_control = REG_READ(OTG_DOUBLE_BUFFER_CONTROL); } +void optc31_read_reg_state(struct timing_generator *optc, struct dcn_optc_reg_state *optc_reg_state) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + + optc_reg_state->optc_bytes_per_pixel = REG_READ(OPTC_BYTES_PER_PIXEL); + optc_reg_state->optc_data_format_control = REG_READ(OPTC_DATA_FORMAT_CONTROL); + optc_reg_state->optc_data_source_select = REG_READ(OPTC_DATA_SOURCE_SELECT); + optc_reg_state->optc_input_clock_control = REG_READ(OPTC_INPUT_CLOCK_CONTROL); + optc_reg_state->optc_input_global_control = REG_READ(OPTC_INPUT_GLOBAL_CONTROL); + optc_reg_state->optc_input_spare_register = REG_READ(OPTC_INPUT_SPARE_REGISTER); + optc_reg_state->optc_memory_config = REG_READ(OPTC_MEMORY_CONFIG); + optc_reg_state->optc_rsmu_underflow = REG_READ(OPTC_RSMU_UNDERFLOW); + optc_reg_state->optc_underflow_threshold = REG_READ(OPTC_UNDERFLOW_THRESHOLD); + optc_reg_state->optc_width_control = REG_READ(OPTC_WIDTH_CONTROL); + optc_reg_state->otg_3d_structure_control = REG_READ(OTG_3D_STRUCTURE_CONTROL); + optc_reg_state->otg_clock_control = REG_READ(OTG_CLOCK_CONTROL); + optc_reg_state->otg_control = REG_READ(OTG_CONTROL); + optc_reg_state->otg_count_control = REG_READ(OTG_COUNT_CONTROL); + optc_reg_state->otg_count_reset = REG_READ(OTG_COUNT_RESET); + optc_reg_state->otg_crc_cntl = REG_READ(OTG_CRC_CNTL); + optc_reg_state->otg_crc_sig_blue_control_mask = REG_READ(OTG_CRC_SIG_BLUE_CONTROL_MASK); + optc_reg_state->otg_crc_sig_red_green_mask = REG_READ(OTG_CRC_SIG_RED_GREEN_MASK); + optc_reg_state->otg_crc0_data_b = REG_READ(OTG_CRC0_DATA_B); + optc_reg_state->otg_crc0_data_rg = REG_READ(OTG_CRC0_DATA_RG); + optc_reg_state->otg_crc0_windowa_x_control = REG_READ(OTG_CRC0_WINDOWA_X_CONTROL); + optc_reg_state->otg_crc0_windowa_x_control_readback = REG_READ(OTG_CRC0_WINDOWA_X_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowa_y_control = REG_READ(OTG_CRC0_WINDOWA_Y_CONTROL); + optc_reg_state->otg_crc0_windowa_y_control_readback = REG_READ(OTG_CRC0_WINDOWA_Y_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowb_x_control = REG_READ(OTG_CRC0_WINDOWB_X_CONTROL); + optc_reg_state->otg_crc0_windowb_x_control_readback = REG_READ(OTG_CRC0_WINDOWB_X_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowb_y_control = REG_READ(OTG_CRC0_WINDOWB_Y_CONTROL); + optc_reg_state->otg_crc0_windowb_y_control_readback = REG_READ(OTG_CRC0_WINDOWB_Y_CONTROL_READBACK); + optc_reg_state->otg_crc1_data_b = REG_READ(OTG_CRC1_DATA_B); + optc_reg_state->otg_crc1_data_rg = REG_READ(OTG_CRC1_DATA_RG); + optc_reg_state->otg_crc1_windowa_x_control = REG_READ(OTG_CRC1_WINDOWA_X_CONTROL); + optc_reg_state->otg_crc1_windowa_x_control_readback = REG_READ(OTG_CRC1_WINDOWA_X_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowa_y_control = REG_READ(OTG_CRC1_WINDOWA_Y_CONTROL); + optc_reg_state->otg_crc1_windowa_y_control_readback = REG_READ(OTG_CRC1_WINDOWA_Y_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowb_x_control = REG_READ(OTG_CRC1_WINDOWB_X_CONTROL); + optc_reg_state->otg_crc1_windowb_x_control_readback = REG_READ(OTG_CRC1_WINDOWB_X_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowb_y_control = REG_READ(OTG_CRC1_WINDOWB_Y_CONTROL); + optc_reg_state->otg_crc1_windowb_y_control_readback = REG_READ(OTG_CRC1_WINDOWB_Y_CONTROL_READBACK); + optc_reg_state->otg_crc2_data_b = REG_READ(OTG_CRC2_DATA_B); + optc_reg_state->otg_crc2_data_rg = REG_READ(OTG_CRC2_DATA_RG); + optc_reg_state->otg_crc3_data_b = REG_READ(OTG_CRC3_DATA_B); + optc_reg_state->otg_crc3_data_rg = REG_READ(OTG_CRC3_DATA_RG); + optc_reg_state->otg_dlpc_control = REG_READ(OTG_DLPC_CONTROL); + optc_reg_state->otg_double_buffer_control = REG_READ(OTG_DOUBLE_BUFFER_CONTROL); + optc_reg_state->otg_drr_control2 = REG_READ(OTG_DRR_CONTROL2); + optc_reg_state->otg_drr_control = REG_READ(OTG_DRR_CONTROL); + optc_reg_state->otg_drr_timing_int_status = REG_READ(OTG_DRR_TIMING_INT_STATUS); + optc_reg_state->otg_drr_trigger_window = REG_READ(OTG_DRR_TRIGGER_WINDOW); + optc_reg_state->otg_drr_v_total_change = REG_READ(OTG_DRR_V_TOTAL_CHANGE); + optc_reg_state->otg_dsc_start_position = REG_READ(OTG_DSC_START_POSITION); + optc_reg_state->otg_force_count_now_cntl = REG_READ(OTG_FORCE_COUNT_NOW_CNTL); + optc_reg_state->otg_global_control0 = REG_READ(OTG_GLOBAL_CONTROL0); + optc_reg_state->otg_global_control1 = REG_READ(OTG_GLOBAL_CONTROL1); + optc_reg_state->otg_global_control2 = REG_READ(OTG_GLOBAL_CONTROL2); + optc_reg_state->otg_global_control3 = REG_READ(OTG_GLOBAL_CONTROL3); + optc_reg_state->otg_global_control4 = REG_READ(OTG_GLOBAL_CONTROL4); + optc_reg_state->otg_global_sync_status = REG_READ(OTG_GLOBAL_SYNC_STATUS); + optc_reg_state->otg_gsl_control = REG_READ(OTG_GSL_CONTROL); + optc_reg_state->otg_gsl_vsync_gap = REG_READ(OTG_GSL_VSYNC_GAP); + optc_reg_state->otg_gsl_window_x = REG_READ(OTG_GSL_WINDOW_X); + optc_reg_state->otg_gsl_window_y = REG_READ(OTG_GSL_WINDOW_Y); + optc_reg_state->otg_h_blank_start_end = REG_READ(OTG_H_BLANK_START_END); + optc_reg_state->otg_h_sync_a = REG_READ(OTG_H_SYNC_A); + optc_reg_state->otg_h_sync_a_cntl = REG_READ(OTG_H_SYNC_A_CNTL); + optc_reg_state->otg_h_timing_cntl = REG_READ(OTG_H_TIMING_CNTL); + optc_reg_state->otg_h_total = REG_READ(OTG_H_TOTAL); + optc_reg_state->otg_interlace_control = REG_READ(OTG_INTERLACE_CONTROL); + optc_reg_state->otg_interlace_status = REG_READ(OTG_INTERLACE_STATUS); + optc_reg_state->otg_interrupt_control = REG_READ(OTG_INTERRUPT_CONTROL); + optc_reg_state->otg_long_vblank_status = REG_READ(OTG_LONG_VBLANK_STATUS); + optc_reg_state->otg_m_const_dto0 = REG_READ(OTG_M_CONST_DTO0); + optc_reg_state->otg_m_const_dto1 = REG_READ(OTG_M_CONST_DTO1); + optc_reg_state->otg_manual_force_vsync_next_line = REG_READ(OTG_MANUAL_FORCE_VSYNC_NEXT_LINE); + optc_reg_state->otg_master_en = REG_READ(OTG_MASTER_EN); + optc_reg_state->otg_master_update_lock = REG_READ(OTG_MASTER_UPDATE_LOCK); + optc_reg_state->otg_master_update_mode = REG_READ(OTG_MASTER_UPDATE_MODE); + optc_reg_state->otg_nom_vert_position = REG_READ(OTG_NOM_VERT_POSITION); + optc_reg_state->otg_pipe_update_status = REG_READ(OTG_PIPE_UPDATE_STATUS); + optc_reg_state->otg_pixel_data_readback0 = REG_READ(OTG_PIXEL_DATA_READBACK0); + optc_reg_state->otg_pixel_data_readback1 = REG_READ(OTG_PIXEL_DATA_READBACK1); + optc_reg_state->otg_request_control = REG_READ(OTG_REQUEST_CONTROL); + optc_reg_state->otg_snapshot_control = REG_READ(OTG_SNAPSHOT_CONTROL); + optc_reg_state->otg_snapshot_frame = REG_READ(OTG_SNAPSHOT_FRAME); + optc_reg_state->otg_snapshot_position = REG_READ(OTG_SNAPSHOT_POSITION); + optc_reg_state->otg_snapshot_status = REG_READ(OTG_SNAPSHOT_STATUS); + optc_reg_state->otg_spare_register = REG_READ(OTG_SPARE_REGISTER); + optc_reg_state->otg_static_screen_control = REG_READ(OTG_STATIC_SCREEN_CONTROL); + optc_reg_state->otg_status = REG_READ(OTG_STATUS); + optc_reg_state->otg_status_frame_count = REG_READ(OTG_STATUS_FRAME_COUNT); + optc_reg_state->otg_status_hv_count = REG_READ(OTG_STATUS_HV_COUNT); + optc_reg_state->otg_status_position = REG_READ(OTG_STATUS_POSITION); + optc_reg_state->otg_status_vf_count = REG_READ(OTG_STATUS_VF_COUNT); + optc_reg_state->otg_stereo_control = REG_READ(OTG_STEREO_CONTROL); + optc_reg_state->otg_stereo_force_next_eye = REG_READ(OTG_STEREO_FORCE_NEXT_EYE); + optc_reg_state->otg_stereo_status = REG_READ(OTG_STEREO_STATUS); + optc_reg_state->otg_trig_manual_control = REG_READ(OTG_TRIG_MANUAL_CONTROL); + optc_reg_state->otg_triga_cntl = REG_READ(OTG_TRIGA_CNTL); + optc_reg_state->otg_triga_manual_trig = REG_READ(OTG_TRIGA_MANUAL_TRIG); + optc_reg_state->otg_trigb_cntl = REG_READ(OTG_TRIGB_CNTL); + optc_reg_state->otg_trigb_manual_trig = REG_READ(OTG_TRIGB_MANUAL_TRIG); + optc_reg_state->otg_update_lock = REG_READ(OTG_UPDATE_LOCK); + optc_reg_state->otg_v_blank_start_end = REG_READ(OTG_V_BLANK_START_END); + optc_reg_state->otg_v_count_stop_control = REG_READ(OTG_V_COUNT_STOP_CONTROL); + optc_reg_state->otg_v_count_stop_control2 = REG_READ(OTG_V_COUNT_STOP_CONTROL2); + optc_reg_state->otg_v_sync_a = REG_READ(OTG_V_SYNC_A); + optc_reg_state->otg_v_sync_a_cntl = REG_READ(OTG_V_SYNC_A_CNTL); + optc_reg_state->otg_v_total = REG_READ(OTG_V_TOTAL); + optc_reg_state->otg_v_total_control = REG_READ(OTG_V_TOTAL_CONTROL); + optc_reg_state->otg_v_total_int_status = REG_READ(OTG_V_TOTAL_INT_STATUS); + optc_reg_state->otg_v_total_max = REG_READ(OTG_V_TOTAL_MAX); + optc_reg_state->otg_v_total_mid = REG_READ(OTG_V_TOTAL_MID); + optc_reg_state->otg_v_total_min = REG_READ(OTG_V_TOTAL_MIN); + optc_reg_state->otg_vert_sync_control = REG_READ(OTG_VERT_SYNC_CONTROL); + optc_reg_state->otg_vertical_interrupt0_control = REG_READ(OTG_VERTICAL_INTERRUPT0_CONTROL); + optc_reg_state->otg_vertical_interrupt0_position = REG_READ(OTG_VERTICAL_INTERRUPT0_POSITION); + optc_reg_state->otg_vertical_interrupt1_control = REG_READ(OTG_VERTICAL_INTERRUPT1_CONTROL); + optc_reg_state->otg_vertical_interrupt1_position = REG_READ(OTG_VERTICAL_INTERRUPT1_POSITION); + optc_reg_state->otg_vertical_interrupt2_control = REG_READ(OTG_VERTICAL_INTERRUPT2_CONTROL); + optc_reg_state->otg_vertical_interrupt2_position = REG_READ(OTG_VERTICAL_INTERRUPT2_POSITION); + optc_reg_state->otg_vready_param = REG_READ(OTG_VREADY_PARAM); + optc_reg_state->otg_vstartup_param = REG_READ(OTG_VSTARTUP_PARAM); + optc_reg_state->otg_vsync_nom_int_status = REG_READ(OTG_VSYNC_NOM_INT_STATUS); + optc_reg_state->otg_vupdate_keepout = REG_READ(OTG_VUPDATE_KEEPOUT); + optc_reg_state->otg_vupdate_param = REG_READ(OTG_VUPDATE_PARAM); +} + static const struct timing_generator_funcs dcn31_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, @@ -377,6 +507,7 @@ static const struct timing_generator_funcs dcn31_tg_funcs = { .init_odm = optc3_init_odm, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn31_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h index 0f72c274f40b..98f7d2e299c5 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h @@ -274,4 +274,6 @@ void optc3_init_odm(struct timing_generator *optc); void optc31_read_otg_state(struct timing_generator *optc, struct dcn_otg_state *s); +void optc31_read_reg_state(struct timing_generator *optc, struct dcn_optc_reg_state *optc_reg_state); + #endif /* __DC_OPTC_DCN31_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c index 4a2caca37255..43ff957288b2 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c @@ -256,6 +256,7 @@ static const struct timing_generator_funcs dcn314_tg_funcs = { .set_h_timing_div_manual_mode = optc314_set_h_timing_div_manual_mode, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn314_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c index b2b226bcd871..3dcb0d0c931c 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c @@ -365,6 +365,7 @@ static const struct timing_generator_funcs dcn32_tg_funcs = { .get_otg_double_buffer_pending = optc3_get_otg_update_pending, .get_pipe_update_pending = optc3_get_pipe_update_pending, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn32_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c index 52d5ea98c86b..f699e95059f3 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c @@ -511,6 +511,7 @@ static const struct timing_generator_funcs dcn35_tg_funcs = { .set_long_vtotal = optc35_set_long_vtotal, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn35_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index 5af13706e601..a8e978d1fae8 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -533,6 +533,7 @@ static const struct timing_generator_funcs dcn401_tg_funcs = { .set_vupdate_keepout = optc401_set_vupdate_keepout, .wait_update_lock_status = optc401_wait_update_lock_status, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn401_timing_generator_init(struct optc *optc1) -- cgit v1.2.3 From 11cc0bdc5bf7ebd8a14a137ee63e1a4bfd94f5d1 Mon Sep 17 00:00:00 2001 From: Meenakshikumar Somasundaram Date: Thu, 9 Oct 2025 09:13:49 -0400 Subject: drm/amd/display: update link encoder assignment [Why] Map a link encoder instance matching stream encoder instance if possible. [How] Get the stream encoder instance and assign the same link encoder instance if available. Reviewed-by: PeiChen Huang Signed-off-by: Meenakshikumar Somasundaram Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 43 +++++++++++++++++------ 1 file changed, 33 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 8d0cdbf14e17..89fe72820f6a 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -2690,17 +2690,40 @@ static inline int find_fixed_dio_link_enc(const struct dc_link *link) } static inline int find_free_dio_link_enc(const struct resource_context *res_ctx, - const struct dc_link *link, const struct resource_pool *pool) + const struct dc_link *link, const struct resource_pool *pool, struct dc_stream_state *stream) { - int i; + int i, j = -1; + int stream_enc_inst = -1; int enc_count = pool->dig_link_enc_count; - /* for dpia, check preferred encoder first and then the next one */ - for (i = 0; i < enc_count; i++) - if (res_ctx->dio_link_enc_ref_cnts[(link->dpia_preferred_eng_id + i) % enc_count] == 0) - break; + /* Find stream encoder instance for the stream */ + if (stream) { + for (i = 0; i < pool->pipe_count; i++) { + if ((res_ctx->pipe_ctx[i].stream == stream) && + (res_ctx->pipe_ctx[i].stream_res.stream_enc != NULL)) { + stream_enc_inst = res_ctx->pipe_ctx[i].stream_res.stream_enc->id; + break; + } + } + } - return (i >= 0 && i < enc_count) ? (link->dpia_preferred_eng_id + i) % enc_count : -1; + /* Assign dpia preferred > stream enc instance > available */ + for (i = 0; i < enc_count; i++) { + if (res_ctx->dio_link_enc_ref_cnts[i] == 0) { + if (j == -1) + j = i; + + if (link->dpia_preferred_eng_id == i) { + j = i; + break; + } + + if (stream_enc_inst == i) { + j = stream_enc_inst; + } + } + } + return j; } static inline void acquire_dio_link_enc( @@ -2781,7 +2804,7 @@ static bool add_dio_link_enc_to_ctx(const struct dc *dc, retain_dio_link_enc(res_ctx, enc_index); } else { if (stream->link->is_dig_mapping_flexible) - enc_index = find_free_dio_link_enc(res_ctx, stream->link, pool); + enc_index = find_free_dio_link_enc(res_ctx, stream->link, pool, stream); else { int link_index = 0; @@ -2791,7 +2814,7 @@ static bool add_dio_link_enc_to_ctx(const struct dc *dc, * one into the acquiring link. */ if (enc_index >= 0 && is_dio_enc_acquired_by_other_link(stream->link, enc_index, &link_index)) { - int new_enc_index = find_free_dio_link_enc(res_ctx, dc->links[link_index], pool); + int new_enc_index = find_free_dio_link_enc(res_ctx, dc->links[link_index], pool, stream); if (new_enc_index >= 0) swap_dio_link_enc_to_muxable_ctx(context, pool, new_enc_index, enc_index); @@ -5201,7 +5224,7 @@ struct link_encoder *get_temp_dio_link_enc( enc_index = link->eng_id; if (enc_index < 0) - enc_index = find_free_dio_link_enc(res_ctx, link, pool); + enc_index = find_free_dio_link_enc(res_ctx, link, pool, NULL); if (enc_index >= 0) link_enc = pool->link_encoders[enc_index]; -- cgit v1.2.3 From 9ec11bb842b6dab109f64d3aa96e53d8243c546b Mon Sep 17 00:00:00 2001 From: Dominik Kaszewski Date: Thu, 16 Oct 2025 08:44:08 +0200 Subject: drm/amd/display: Remove dc param from check_update [Why] dc_check_update_surfaces_for_stream should not have access to entire DC, especially not a mutable one. Concurrent checks should be able to run independently of one another, without risk of changing state. [How] * Replace dc and stream_status structs with new dc_check_config. * Move required fields from dc_debug and dc_caps to dc_check_config. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Dominik Kaszewski Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 51 ++++++++++------------ drivers/gpu/drm/amd/display/dc/dc.h | 17 +++++--- drivers/gpu/drm/amd/display/dc/dc_stream.h | 5 +-- .../display/dc/resource/dce100/dce100_resource.c | 7 ++- .../display/dc/resource/dce110/dce110_resource.c | 5 ++- .../display/dc/resource/dce112/dce112_resource.c | 7 ++- .../display/dc/resource/dce120/dce120_resource.c | 8 +++- .../amd/display/dc/resource/dce80/dce80_resource.c | 8 +++- .../amd/display/dc/resource/dcn10/dcn10_resource.c | 7 ++- .../amd/display/dc/resource/dcn20/dcn20_resource.c | 6 ++- .../display/dc/resource/dcn201/dcn201_resource.c | 6 ++- .../amd/display/dc/resource/dcn21/dcn21_resource.c | 6 ++- .../amd/display/dc/resource/dcn30/dcn30_resource.c | 6 ++- .../display/dc/resource/dcn301/dcn301_resource.c | 6 ++- .../display/dc/resource/dcn302/dcn302_resource.c | 6 ++- .../display/dc/resource/dcn303/dcn303_resource.c | 6 ++- .../amd/display/dc/resource/dcn31/dcn31_resource.c | 6 ++- .../display/dc/resource/dcn314/dcn314_resource.c | 6 ++- .../display/dc/resource/dcn315/dcn315_resource.c | 6 ++- .../display/dc/resource/dcn316/dcn316_resource.c | 6 ++- .../amd/display/dc/resource/dcn32/dcn32_resource.c | 6 ++- .../display/dc/resource/dcn321/dcn321_resource.c | 6 ++- .../amd/display/dc/resource/dcn35/dcn35_resource.c | 6 ++- .../display/dc/resource/dcn351/dcn351_resource.c | 6 ++- .../amd/display/dc/resource/dcn36/dcn36_resource.c | 6 ++- .../display/dc/resource/dcn401/dcn401_resource.c | 6 ++- 26 files changed, 152 insertions(+), 65 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 246ea9754810..2eb02345dadd 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1148,8 +1148,8 @@ static bool dc_construct(struct dc *dc, /* set i2c speed if not done by the respective dcnxxx__resource.c */ if (dc->caps.i2c_speed_in_khz_hdcp == 0) dc->caps.i2c_speed_in_khz_hdcp = dc->caps.i2c_speed_in_khz; - if (dc->caps.max_optimizable_video_width == 0) - dc->caps.max_optimizable_video_width = 5120; + if (dc->check_config.max_optimizable_video_width == 0) + dc->check_config.max_optimizable_video_width = 5120; dc->clk_mgr = dc_clk_mgr_create(dc->ctx, dc->res_pool->pp_smu, dc->res_pool->dccg); if (!dc->clk_mgr) goto fail; @@ -2656,7 +2656,7 @@ static bool is_surface_in_context( return false; } -static enum surface_update_type get_plane_info_update_type(const struct dc *dc, const struct dc_surface_update *u) +static enum surface_update_type get_plane_info_update_type(const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; enum surface_update_type update_type = UPDATE_TYPE_FAST; @@ -2760,7 +2760,7 @@ static enum surface_update_type get_plane_info_update_type(const struct dc *dc, } static enum surface_update_type get_scaling_info_update_type( - const struct dc *dc, + const struct dc_check_config *check_config, const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; @@ -2790,7 +2790,7 @@ static enum surface_update_type get_scaling_info_update_type( /* Making dst rect smaller requires a bandwidth change */ update_flags->bits.bandwidth_change = 1; - if (u->scaling_info->src_rect.width > dc->caps.max_optimizable_video_width && + if (u->scaling_info->src_rect.width > check_config->max_optimizable_video_width && (u->scaling_info->clip_rect.width > u->surface->clip_rect.width || u->scaling_info->clip_rect.height > u->surface->clip_rect.height)) /* Changing clip size of a large surface may result in MPC slice count change */ @@ -2817,7 +2817,8 @@ static enum surface_update_type get_scaling_info_update_type( return UPDATE_TYPE_FAST; } -static enum surface_update_type det_surface_update(const struct dc *dc, +static enum surface_update_type det_surface_update( + const struct dc_check_config *check_config, const struct dc_surface_update *u) { enum surface_update_type type; @@ -2831,10 +2832,10 @@ static enum surface_update_type det_surface_update(const struct dc *dc, update_flags->raw = 0; // Reset all flags - type = get_plane_info_update_type(dc, u); + type = get_plane_info_update_type(u); elevate_update_type(&overall_type, type); - type = get_scaling_info_update_type(dc, u); + type = get_scaling_info_update_type(check_config, u); elevate_update_type(&overall_type, type); if (u->flip_addr) { @@ -2911,7 +2912,7 @@ static enum surface_update_type det_surface_update(const struct dc *dc, elevate_update_type(&overall_type, type); } - if (dc->debug.enable_legacy_fast_update && + if (check_config->enable_legacy_fast_update && (update_flags->bits.gamma_change || update_flags->bits.gamut_remap_change || update_flags->bits.input_csc_change || @@ -2946,11 +2947,11 @@ static void force_immediate_gsl_plane_flip(struct dc *dc, struct dc_surface_upda } static enum surface_update_type check_update_surfaces_for_stream( - struct dc *dc, + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status) + struct dc_stream_update *stream_update +) { int i; enum surface_update_type overall_type = UPDATE_TYPE_FAST; @@ -2972,7 +2973,7 @@ static enum surface_update_type check_update_surfaces_for_stream( stream_update->integer_scaling_update) su_flags->bits.scaling = 1; - if (dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func) + if (check_config->enable_legacy_fast_update && stream_update->out_transfer_func) su_flags->bits.out_tf = 1; if (stream_update->abm_level) @@ -3016,13 +3017,13 @@ static enum surface_update_type check_update_surfaces_for_stream( /* Output transfer function changes do not require bandwidth recalculation, * so don't trigger a full update */ - if (!dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func) + if (!check_config->enable_legacy_fast_update && stream_update->out_transfer_func) su_flags->bits.out_tf = 1; } for (i = 0 ; i < surface_count; i++) { enum surface_update_type type = - det_surface_update(dc, &updates[i]); + det_surface_update(check_config, &updates[i]); elevate_update_type(&overall_type, type); } @@ -3036,22 +3037,17 @@ static enum surface_update_type check_update_surfaces_for_stream( * See :c:type:`enum surface_update_type ` for explanation of update types */ enum surface_update_type dc_check_update_surfaces_for_stream( - struct dc *dc, + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status) + struct dc_stream_update *stream_update) { - int i; - enum surface_update_type type; - if (stream_update) stream_update->stream->update_flags.raw = 0; - for (i = 0; i < surface_count; i++) + for (size_t i = 0; i < surface_count; i++) updates[i].surface->update_flags.raw = 0; - type = check_update_surfaces_for_stream(dc, updates, surface_count, stream_update, stream_status); - return type; + return check_update_surfaces_for_stream(check_config, updates, surface_count, stream_update); } static struct dc_stream_status *stream_get_status( @@ -3472,8 +3468,7 @@ static bool update_planes_and_stream_state(struct dc *dc, context = dc->current_state; update_type = dc_check_update_surfaces_for_stream( - dc, srf_updates, surface_count, stream_update, stream_status); - + &dc->check_config, srf_updates, surface_count, stream_update); if (full_update_required_weak(dc, srf_updates, surface_count, stream_update, stream)) update_type = UPDATE_TYPE_FULL; @@ -5208,7 +5203,7 @@ static bool update_planes_and_stream_v2(struct dc *dc, commit_minimal_transition_state_in_dc_update(dc, context, stream, srf_updates, surface_count); - if (is_fast_update_only && !dc->debug.enable_legacy_fast_update) { + if (is_fast_update_only && !dc->check_config.enable_legacy_fast_update) { commit_planes_for_stream_fast(dc, srf_updates, surface_count, @@ -5251,7 +5246,7 @@ static void commit_planes_and_stream_update_on_current_context(struct dc *dc, stream_update); if (fast_update_only(dc, fast_update, srf_updates, surface_count, stream_update, stream) && - !dc->debug.enable_legacy_fast_update) + !dc->check_config.enable_legacy_fast_update) commit_planes_for_stream_fast(dc, srf_updates, surface_count, diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 93b39710db53..0fcecf1c5deb 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -286,6 +286,15 @@ struct dc_scl_caps { bool sharpener_support; }; +struct dc_check_config { + /** + * max video plane width that can be safely assumed to be always + * supported by single DPP pipe. + */ + unsigned int max_optimizable_video_width; + bool enable_legacy_fast_update; +}; + struct dc_caps { uint32_t max_streams; uint32_t max_links; @@ -301,11 +310,6 @@ struct dc_caps { unsigned int max_cursor_size; unsigned int max_buffered_cursor_size; unsigned int max_video_width; - /* - * max video plane width that can be safely assumed to be always - * supported by single DPP pipe. - */ - unsigned int max_optimizable_video_width; unsigned int min_horizontal_blanking_period; int linear_pitch_alignment; bool dcc_const_color; @@ -1128,7 +1132,6 @@ struct dc_debug_options { uint32_t fpo_vactive_min_active_margin_us; uint32_t fpo_vactive_max_blank_us; bool enable_hpo_pg_support; - bool enable_legacy_fast_update; bool disable_dc_mode_overwrite; bool replay_skip_crtc_disabled; bool ignore_pg;/*do nothing, let pmfw control it*/ @@ -1160,7 +1163,6 @@ struct dc_debug_options { bool enable_ips_visual_confirm; unsigned int sharpen_policy; unsigned int scale_to_sharpness_policy; - bool skip_full_updated_if_possible; unsigned int enable_oled_edp_power_up_opt; bool enable_hblank_borrow; bool force_subvp_df_throttle; @@ -1711,6 +1713,7 @@ struct dc { struct dc_debug_options debug; struct dc_versions versions; struct dc_caps caps; + struct dc_check_config check_config; struct dc_cap_funcs cap_funcs; struct dc_config config; struct dc_bounding_box_overrides bb_overrides; diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index 76cf9fdedab0..c68153c77972 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -474,11 +474,10 @@ void dc_enable_stereo( void dc_trigger_sync(struct dc *dc, struct dc_state *context); enum surface_update_type dc_check_update_surfaces_for_stream( - struct dc *dc, + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status); + struct dc_stream_update *stream_update); /** * Create a new default stream for the requested sink diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index c4b4dc3ad8c9..550466de1a66 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -402,8 +402,10 @@ static const struct dc_plane_cap plane_cap = { } }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; #define CTX ctx @@ -1093,6 +1095,7 @@ static bool dce100_resource_construct( dc->caps.disable_dp_clk_share = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; for (i = 0; i < pool->base.pipe_count; i++) { pool->base.timing_generators[i] = diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c index cccde5a6f3cd..9e14ffb3c942 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c @@ -424,7 +424,9 @@ static const struct dc_plane_cap plane_cap = { 64 }; -static const struct dc_debug_options debug_defaults = { +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { .enable_legacy_fast_update = true, }; @@ -1376,6 +1378,7 @@ static bool dce110_resource_construct( dc->caps.is_apu = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c index 869a8e515fc0..62977bcdeaa0 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c @@ -429,8 +429,10 @@ static const struct dc_plane_cap plane_cap = { 64 }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; #define CTX ctx @@ -1247,6 +1249,7 @@ static bool dce112_resource_construct( dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c index 540e04ec1e2d..0770ea37183f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c @@ -526,8 +526,11 @@ static const struct dc_plane_cap plane_cap = { }; static const struct dc_debug_options debug_defaults = { - .disable_clock_gate = true, - .enable_legacy_fast_update = true, + .disable_clock_gate = true, +}; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; static struct clock_source *dce120_clock_source_create( @@ -1089,6 +1092,7 @@ static bool dce120_resource_construct( dc->caps.psp_setup_panel_mode = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index 5b7769745202..c4a56ec60b82 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -418,8 +418,10 @@ static const struct dc_plane_cap plane_cap = { } }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; static const struct dce_dmcu_registers dmcu_regs = { @@ -919,6 +921,7 @@ static bool dce80_construct( dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * @@ -1320,6 +1323,7 @@ static bool dce83_construct( dc->caps.min_horizontal_blanking_period = 80; dc->caps.is_apu = true; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c index 652c05c35494..f12367adf145 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c @@ -556,10 +556,13 @@ static const struct dc_debug_options debug_defaults_drv = { .recovery_enabled = false, /*enable this by default after testing.*/ .max_downscale_src_width = 3840, .underflow_assert_delay_us = 0xFFFFFFFF, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn10_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN10_DPP(*dpp)); @@ -1395,6 +1398,8 @@ static bool dcn10_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 0; + dc->debug = debug_defaults_drv; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c index 30f155dc54e5..6679c1a14f2f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c @@ -718,10 +718,13 @@ static const struct dc_debug_options debug_defaults_drv = { .scl_reset_length10 = true, .sanity_checks = false, .underflow_assert_delay_us = 0xFFFFFFFF, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + void dcn20_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN20_DPP(*dpp)); @@ -2471,6 +2474,7 @@ static bool dcn20_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->caps.dp_hdmi21_pcon_support = true; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c index 9a80bebcee48..055107843a70 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c @@ -614,10 +614,13 @@ static const struct dc_debug_options debug_defaults_drv = { .sanity_checks = false, .underflow_assert_delay_us = 0xFFFFFFFF, .enable_tri_buf = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn201_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN201_DPP(*dpp)); @@ -1151,6 +1154,7 @@ static bool dcn201_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->debug = debug_defaults_drv; + dc->check_config = config_defaults; /*a0 only, remove later*/ dc->work_arounds.no_connect_phy_config = true; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c index 918742a42ded..2060acd5ae09 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c @@ -626,10 +626,13 @@ static const struct dc_debug_options debug_defaults_drv = { .usbc_combo_phy_reset_wa = true, .dmub_command_table = true, .use_max_lb = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1458,6 +1461,7 @@ static bool dcn21_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->caps.dp_hdmi21_pcon_support = true; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c index ff63f59ff928..d0ebb733e802 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c @@ -727,10 +727,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -2374,6 +2377,7 @@ static bool dcn30_resource_construct( dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c index 82a205a7c25c..3ad6a3d4858e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c @@ -701,10 +701,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = false, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn301_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN20_DPP(*dpp)); @@ -1498,6 +1501,7 @@ static bool dcn301_resource_construct( bp_query_result = ctx->dc_bios->funcs->get_lttpr_interop(ctx->dc_bios, &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c index 61623cb518d9..c0d4a1dc94f8 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c @@ -98,10 +98,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1290,6 +1293,7 @@ static bool dcn302_resource_construct( &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c index 02b9a84f2db3..75e09c2c283e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c @@ -98,10 +98,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1234,6 +1237,7 @@ static bool dcn303_resource_construct( bp_query_result = ctx->dc_bios->funcs->get_lttpr_interop(ctx->dc_bios, &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c index 3ed7f50554e2..0d667b54ccf8 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c @@ -888,12 +888,15 @@ static const struct dc_debug_options debug_defaults_drv = { } }, .disable_z10 = true, - .enable_legacy_fast_update = true, .enable_z9_disable_interface = true, /* Allow support for the PMFW interface for disable Z9*/ .dml_hostvm_override = DML_HOSTVM_OVERRIDE_FALSE, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1978,6 +1981,7 @@ static bool dcn31_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c index d4917a35b991..3ccde75a4ecb 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c @@ -924,12 +924,15 @@ static const struct dc_debug_options debug_defaults_drv = { }, .seamless_boot_odm_combine = true, - .enable_legacy_fast_update = true, .using_dml2 = false, .disable_dsc_power_gate = true, .min_disp_clk_khz = 100000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1910,6 +1913,7 @@ static bool dcn314_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 82cc78c291d8..38f6328cda92 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -887,11 +887,14 @@ static const struct dc_debug_options debug_defaults_drv = { .afmt = true, } }, - .enable_legacy_fast_update = true, .psr_power_use_phy_fsm = 0, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1939,6 +1942,7 @@ static bool dcn315_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c index 636110e48d01..5a95dd54cb42 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c @@ -882,10 +882,13 @@ static const struct dc_debug_options debug_defaults_drv = { .afmt = true, } }, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1815,6 +1818,7 @@ static bool dcn316_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c index 4c2e8aa96d1f..81e64e17d0cb 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c @@ -738,10 +738,13 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_dp_plus_plus_wa = true, .fpo_vactive_min_active_margin_us = 200, .fpo_vactive_max_blank_us = 1000, - .enable_legacy_fast_update = false, .disable_stutter_for_wm_program = true }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn32_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -2294,6 +2297,7 @@ static bool dcn32_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c index ad214986f7ac..3466ca34c93f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c @@ -731,11 +731,14 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_subvp_high_refresh = false, .fpo_vactive_min_active_margin_us = 200, .fpo_vactive_max_blank_us = 1000, - .enable_legacy_fast_update = false, .disable_dc_mode_overwrite = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn321_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -1797,6 +1800,7 @@ static bool dcn321_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c index 3a83a63d2116..ef69898d2cc5 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c @@ -767,7 +767,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -788,6 +787,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1946,6 +1949,7 @@ static bool dcn35_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index d00cc4030be3..f3c614c4490c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -747,7 +747,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -768,6 +767,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1917,6 +1920,7 @@ static bool dcn351_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c index f4b28617b2de..6469d5fe2e6d 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c @@ -748,7 +748,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -769,6 +768,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1918,6 +1921,7 @@ static bool dcn36_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 8e2d3781752c..130058d7a70c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -721,7 +721,6 @@ static const struct dc_debug_options debug_defaults_drv = { .alloc_extra_way_for_cursor = true, .min_prefetch_in_strobe_ns = 60000, // 60us .disable_unbounded_requesting = false, - .enable_legacy_fast_update = false, .dcc_meta_propagation_delay_us = 10, .fams_version = { .minor = 1, @@ -737,6 +736,10 @@ static const struct dc_debug_options debug_defaults_drv = { .force_cositing = CHROMA_COSITING_NONE + 1, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn401_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -1995,6 +1998,7 @@ static bool dcn401_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; -- cgit v1.2.3 From 14bb17cc37e03e105c5e989381b1dd9769535417 Mon Sep 17 00:00:00 2001 From: Zhongwei Zhang Date: Tue, 14 Oct 2025 10:46:49 +0800 Subject: drm/amd/display: init dispclk from bootup clock for DCN315 [Why] Driver does not pick up and save vbios's clocks during init clocks, the dispclk in clk_mgr will keep 0. OS might change the timing (lower the pixel clock) after boot. Then driver will set the dispclk to lower when safe_to_lower is false, for in clk_mgr dispclk is zero, it's illegal and causes garbage. [How] Dump and save the vbios's clocks, and init the dispclk in dcn315_init_clocks. Reviewed-by: Charlene Liu Signed-off-by: Zhongwei Zhang Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- .../amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c | 87 +++++++++++++++++++++- .../amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h | 1 + .../display/dc/resource/dcn315/dcn315_resource.c | 1 + 3 files changed, 87 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c index b315ed91e010..3a881451e9da 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c @@ -40,7 +40,7 @@ #include "dm_helpers.h" #include "dc_dmub_srv.h" - +#include "reg_helper.h" #include "logger_types.h" #undef DC_LOGGER #define DC_LOGGER \ @@ -48,9 +48,43 @@ #include "link_service.h" +#define MAX_INSTANCE 7 +#define MAX_SEGMENT 8 + +struct IP_BASE_INSTANCE { + unsigned int segment[MAX_SEGMENT]; +}; + +struct IP_BASE { + struct IP_BASE_INSTANCE instance[MAX_INSTANCE]; +}; + +static const struct IP_BASE CLK_BASE = { { { { 0x00016C00, 0x02401800, 0, 0, 0, 0, 0, 0 } }, + { { 0x00016E00, 0x02401C00, 0, 0, 0, 0, 0, 0 } }, + { { 0x00017000, 0x02402000, 0, 0, 0, 0, 0, 0 } }, + { { 0x00017200, 0x02402400, 0, 0, 0, 0, 0, 0 } }, + { { 0x0001B000, 0x0242D800, 0, 0, 0, 0, 0, 0 } }, + { { 0x0001B200, 0x0242DC00, 0, 0, 0, 0, 0, 0 } } } }; + +#define regCLK1_CLK0_CURRENT_CNT 0x0314 +#define regCLK1_CLK0_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK1_CURRENT_CNT 0x0315 +#define regCLK1_CLK1_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK2_CURRENT_CNT 0x0316 +#define regCLK1_CLK2_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK3_CURRENT_CNT 0x0317 +#define regCLK1_CLK3_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK4_CURRENT_CNT 0x0318 +#define regCLK1_CLK4_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK5_CURRENT_CNT 0x0319 +#define regCLK1_CLK5_CURRENT_CNT_BASE_IDX 0 + #define TO_CLK_MGR_DCN315(clk_mgr)\ container_of(clk_mgr, struct clk_mgr_dcn315, base) +#define REG(reg_name) \ + (CLK_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name) + #define UNSUPPORTED_DCFCLK 10000000 #define MIN_DPP_DISP_CLK 100000 @@ -245,9 +279,38 @@ static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base, dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } +static void dcn315_dump_clk_registers_internal(struct dcn35_clk_internal *internal, struct clk_mgr *clk_mgr_base) +{ + struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); + + // read dtbclk + internal->CLK1_CLK4_CURRENT_CNT = REG_READ(CLK1_CLK4_CURRENT_CNT); + + // read dcfclk + internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_CLK3_CURRENT_CNT); + + // read dppclk + internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_CLK1_CURRENT_CNT); + + // read dprefclk + internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_CLK2_CURRENT_CNT); + + // read dispclk + internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_CLK0_CURRENT_CNT); +} + static void dcn315_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass, struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info) { + struct dcn35_clk_internal internal = {0}; + + dcn315_dump_clk_registers_internal(&internal, clk_mgr_base); + + regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10; + regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10; + regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10; + regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10; + regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10; return; } @@ -594,13 +657,32 @@ static struct clk_mgr_funcs dcn315_funcs = { .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, .get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz, .update_clocks = dcn315_update_clocks, - .init_clocks = dcn31_init_clocks, + .init_clocks = dcn315_init_clocks, .enable_pme_wa = dcn315_enable_pme_wa, .are_clock_states_equal = dcn31_are_clock_states_equal, .notify_wm_ranges = dcn315_notify_wm_ranges }; extern struct clk_mgr_funcs dcn3_fpga_funcs; +void dcn315_init_clocks(struct clk_mgr *clk_mgr) +{ + struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr); + uint32_t ref_dtbclk = clk_mgr->clks.ref_dtbclk_khz; + struct clk_mgr_dcn315 *clk_mgr_dcn315 = TO_CLK_MGR_DCN315(clk_mgr_int); + struct clk_log_info log_info = {0}; + + memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks)); + // Assumption is that boot state always supports pstate + clk_mgr->clks.ref_dtbclk_khz = ref_dtbclk; // restore ref_dtbclk + clk_mgr->clks.p_state_change_support = true; + clk_mgr->clks.prev_p_state_change_support = true; + clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN; + clk_mgr->clks.zstate_support = DCN_ZSTATE_SUPPORT_UNKNOWN; + + dcn315_dump_clk_registers(&clk_mgr->boot_snapshot, &clk_mgr_dcn315->base.base, &log_info); + clk_mgr->clks.dispclk_khz = clk_mgr->boot_snapshot.dispclk * 1000; +} + void dcn315_clk_mgr_construct( struct dc_context *ctx, struct clk_mgr_dcn315 *clk_mgr, @@ -661,6 +743,7 @@ void dcn315_clk_mgr_construct( /* Saved clocks configured at boot for debug purposes */ dcn315_dump_clk_registers(&clk_mgr->base.base.boot_snapshot, &clk_mgr->base.base, &log_info); + clk_mgr->base.base.clks.dispclk_khz = clk_mgr->base.base.boot_snapshot.dispclk * 1000; clk_mgr->base.base.dprefclk_khz = 600000; clk_mgr->base.base.dprefclk_khz = dcn315_smu_get_dpref_clk(&clk_mgr->base); diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h index ac36ddf5dd1a..642ae3d4a790 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h @@ -44,6 +44,7 @@ void dcn315_clk_mgr_construct(struct dc_context *ctx, struct pp_smu_funcs *pp_smu, struct dccg *dccg); +void dcn315_init_clocks(struct clk_mgr *clk_mgr); void dcn315_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int); #endif //__DCN315_CLK_MGR_H__ diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 38f6328cda92..4e962f522f1b 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -889,6 +889,7 @@ static const struct dc_debug_options debug_defaults_drv = { }, .psr_power_use_phy_fsm = 0, .using_dml2 = false, + .min_disp_clk_khz = 100000, }; static const struct dc_check_config config_defaults = { -- cgit v1.2.3 From fb3896c18b224f2aa9dd8a4f73e6ba23c52b9691 Mon Sep 17 00:00:00 2001 From: Meenakshikumar Somasundaram Date: Wed, 15 Oct 2025 14:45:43 -0400 Subject: drm/amd/display: Add dc interface to log pre os firmware information [Why] Pre os firmware information is useful to debug pre os to post os fw transition issues. [How] Add dc interface dc_log_preos_dmcub_info() to log pre os firmware information. Reviewed-by: Cruise Hung Signed-off-by: Meenakshikumar Somasundaram Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 5 +++ drivers/gpu/drm/amd/display/dc/dc.h | 2 ++ drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c | 21 ++++++++++++ drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h | 7 ++++ drivers/gpu/drm/amd/display/dmub/dmub_srv.h | 25 +++++++++++++++ drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c | 39 +++++++++++++++++++++++ drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h | 2 ++ drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c | 9 ++++++ 8 files changed, 110 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 2eb02345dadd..6ef9dd9ab7a3 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -6401,3 +6401,8 @@ void dc_get_underflow_debug_data_for_otg(struct dc *dc, int primary_otg_inst, if (dc->hwss.get_underflow_debug_data) dc->hwss.get_underflow_debug_data(dc, tg, out_data); } + +void dc_log_preos_dmcub_info(const struct dc *dc) +{ + dc_dmub_srv_log_preos_dmcub_info(dc->ctx->dmub_srv); +} diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 0fcecf1c5deb..4a5aa0bd3bc4 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -2727,6 +2727,8 @@ unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context); bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index); +void dc_log_preos_dmcub_info(const struct dc *dc); + /* DSC Interfaces */ #include "dc_dsc.h" diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 4b20c01bf646..fffbf1983143 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -2344,3 +2344,24 @@ void dc_dmub_srv_release_hw(const struct dc *dc) dm_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } + +void dc_dmub_srv_log_preos_dmcub_info(struct dc_dmub_srv *dc_dmub_srv) +{ + struct dmub_srv *dmub; + + if (!dc_dmub_srv || !dc_dmub_srv->dmub) + return; + + dmub = dc_dmub_srv->dmub; + + if (dmub_srv_get_preos_info(dmub)) { + DC_LOG_DEBUG("%s: PreOS DMCUB Info", __func__); + DC_LOG_DEBUG("fw_version : 0x%08x", dmub->preos_info.fw_version); + DC_LOG_DEBUG("boot_options : 0x%08x", dmub->preos_info.boot_options); + DC_LOG_DEBUG("boot_status : 0x%08x", dmub->preos_info.boot_status); + DC_LOG_DEBUG("trace_buffer_phy_addr : 0x%016llx", dmub->preos_info.trace_buffer_phy_addr); + DC_LOG_DEBUG("trace_buffer_size_bytes : 0x%08x", dmub->preos_info.trace_buffer_size); + DC_LOG_DEBUG("fb_base : 0x%016llx", dmub->preos_info.fb_base); + DC_LOG_DEBUG("fb_offset : 0x%016llx", dmub->preos_info.fb_offset); + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h index 9bb00d48fd5e..72e0a41f39f0 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h @@ -367,4 +367,11 @@ bool dc_dmub_srv_is_cursor_offload_enabled(const struct dc *dc); * @dc - pointer to DC object */ void dc_dmub_srv_release_hw(const struct dc *dc); + +/** + * dc_dmub_srv_log_preos_dmcub_info() - Logs preos dmcub fw info. + * + * @dc - pointer to DC object + */ +void dc_dmub_srv_log_preos_dmcub_info(struct dc_dmub_srv *dc_dmub_srv); #endif /* _DMUB_DC_SRV_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h index f25c2fc2f98f..9d0168986fe7 100644 --- a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h @@ -363,6 +363,19 @@ struct dmub_diagnostic_data { uint8_t is_pwait : 1; }; +/** + * struct dmub_preos_info - preos fw info before loading post os fw. + */ +struct dmub_preos_info { + uint64_t fb_base; + uint64_t fb_offset; + uint64_t trace_buffer_phy_addr; + uint32_t trace_buffer_size; + uint32_t fw_version; + uint32_t boot_status; + uint32_t boot_options; +}; + struct dmub_srv_inbox { /* generic status */ uint64_t num_submitted; @@ -488,6 +501,7 @@ struct dmub_srv_hw_funcs { uint32_t (*get_current_time)(struct dmub_srv *dmub); void (*get_diagnostic_data)(struct dmub_srv *dmub); + bool (*get_preos_fw_info)(struct dmub_srv *dmub); bool (*should_detect)(struct dmub_srv *dmub); void (*init_reg_offsets)(struct dmub_srv *dmub, struct dc_context *ctx); @@ -588,6 +602,7 @@ struct dmub_srv { enum dmub_srv_power_state_type power_state; struct dmub_diagnostic_data debug; struct dmub_fb lsdma_rb_fb; + struct dmub_preos_info preos_info; }; /** @@ -1073,4 +1088,14 @@ enum dmub_status dmub_srv_wait_for_inbox_free(struct dmub_srv *dmub, */ enum dmub_status dmub_srv_update_inbox_status(struct dmub_srv *dmub); +/** + * dmub_srv_get_preos_info() - retrieves preos fw info + * @dmub: the dmub service + * + * Return: + * true - preos fw info retrieved successfully + * false - preos fw info not retrieved successfully + */ +bool dmub_srv_get_preos_info(struct dmub_srv *dmub); + #endif /* _DMUB_SRV_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c index b40482dbd6ad..e13557ed97be 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c @@ -521,6 +521,45 @@ void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub) dmub->debug.gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); } + +bool dmub_dcn35_get_preos_fw_info(struct dmub_srv *dmub) +{ + uint64_t region3_cw5_offset; + uint32_t top_addr, top_addr_enable, offset_low; + uint32_t offset_high, base_addr, fw_version; + bool is_vbios_fw = false; + + memset(&dmub->preos_info, 0, sizeof(dmub->preos_info)); + + fw_version = REG_READ(DMCUB_SCRATCH1); + is_vbios_fw = ((fw_version >> 6) & 0x01) ? true : false; + if (!is_vbios_fw) + return false; + + dmub->preos_info.boot_status = REG_READ(DMCUB_SCRATCH0); + dmub->preos_info.fw_version = REG_READ(DMCUB_SCRATCH1); + dmub->preos_info.boot_options = REG_READ(DMCUB_SCRATCH14); + REG_GET(DMCUB_REGION3_CW5_TOP_ADDRESS, + DMCUB_REGION3_CW5_ENABLE, &top_addr_enable); + if (top_addr_enable) { + dmub_dcn35_get_fb_base_offset(dmub, + &dmub->preos_info.fb_base, &dmub->preos_info.fb_offset); + offset_low = REG_READ(DMCUB_REGION3_CW5_OFFSET); + offset_high = REG_READ(DMCUB_REGION3_CW5_OFFSET_HIGH); + region3_cw5_offset = ((uint64_t)offset_high << 32) | offset_low; + dmub->preos_info.trace_buffer_phy_addr = region3_cw5_offset + - dmub->preos_info.fb_base + dmub->preos_info.fb_offset; + + REG_GET(DMCUB_REGION3_CW5_TOP_ADDRESS, + DMCUB_REGION3_CW5_TOP_ADDRESS, &top_addr); + base_addr = REG_READ(DMCUB_REGION3_CW5_BASE_ADDRESS) & 0x1FFFFFFF; + dmub->preos_info.trace_buffer_size = + (top_addr > base_addr) ? (top_addr - base_addr + 1) : 0; + } + + return true; +} + void dmub_dcn35_configure_dmub_in_system_memory(struct dmub_srv *dmub) { /* DMCUB_REGION3_TMR_AXI_SPACE values: diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h index 39fcb7275da5..92e6695a2c9b 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h @@ -285,4 +285,6 @@ bool dmub_dcn35_is_hw_powered_up(struct dmub_srv *dmub); void dmub_srv_dcn35_regs_init(struct dmub_srv *dmub, struct dc_context *ctx); +bool dmub_dcn35_get_preos_fw_info(struct dmub_srv *dmub); + #endif /* _DMUB_DCN35_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c index 0244c9b44ecc..a657efda89ce 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c @@ -359,6 +359,7 @@ static bool dmub_srv_hw_setup(struct dmub_srv *dmub, enum dmub_asic asic) funcs->get_current_time = dmub_dcn35_get_current_time; funcs->get_diagnostic_data = dmub_dcn35_get_diagnostic_data; + funcs->get_preos_fw_info = dmub_dcn35_get_preos_fw_info; funcs->init_reg_offsets = dmub_srv_dcn35_regs_init; if (asic == DMUB_ASIC_DCN351) @@ -1372,3 +1373,11 @@ enum dmub_status dmub_srv_update_inbox_status(struct dmub_srv *dmub) return DMUB_STATUS_OK; } + +bool dmub_srv_get_preos_info(struct dmub_srv *dmub) +{ + if (!dmub || !dmub->hw_funcs.get_preos_fw_info) + return false; + + return dmub->hw_funcs.get_preos_fw_info(dmub); +} -- cgit v1.2.3 From 518a368c57a0e63f531e471baa9499c45ab87f7c Mon Sep 17 00:00:00 2001 From: Alvin Lee Date: Thu, 2 Oct 2025 12:44:19 -0400 Subject: drm/amd/display: Update cursor offload assignments [Why & How] - Cursor lines per chunk must be assigned from hubp->att and not hubp->pos (the one in hubp->pos is unassigned) - In DCN401 DPP, cur0_enable in attribute struct must be assigned as this is the field passed to DMU - DCN401 should not program position in driver if offload is enabled Reviewed-by: Nicholas Kazlauskas Signed-off-by: Alvin Lee Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c | 1 + .../drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c | 24 ++++++++++++---------- .../drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c | 2 +- .../drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 2 +- 4 files changed, 16 insertions(+), 13 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c index 3adc17f2fc35..62bf7cea21d8 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c @@ -141,6 +141,7 @@ void dpp401_set_cursor_position( } dpp_base->pos.cur0_ctl.bits.cur0_enable = cur_en; + dpp_base->att.cur0_ctl.bits.cur0_enable = cur_en; } void dpp401_set_optional_cursor_attributes( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c index 3a2e0848173e..f01eae50d02f 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c @@ -783,21 +783,23 @@ void hubp401_cursor_set_position( if (cur_en && REG_READ(CURSOR_SURFACE_ADDRESS) == 0) hubp->funcs->set_cursor_attributes(hubp, &hubp->curs_attr); - REG_UPDATE(CURSOR_CONTROL, - CURSOR_ENABLE, cur_en); + if (!hubp->cursor_offload) + REG_UPDATE(CURSOR_CONTROL, + CURSOR_ENABLE, cur_en); } - REG_SET_2(CURSOR_POSITION, 0, - CURSOR_X_POSITION, x_pos, - CURSOR_Y_POSITION, y_pos); + if (!hubp->cursor_offload) { + REG_SET_2(CURSOR_POSITION, 0, + CURSOR_X_POSITION, x_pos, + CURSOR_Y_POSITION, y_pos); - REG_SET_2(CURSOR_HOT_SPOT, 0, - CURSOR_HOT_SPOT_X, pos->x_hotspot, - CURSOR_HOT_SPOT_Y, pos->y_hotspot); - - REG_SET(CURSOR_DST_OFFSET, 0, - CURSOR_DST_X_OFFSET, dst_x_offset); + REG_SET_2(CURSOR_HOT_SPOT, 0, + CURSOR_HOT_SPOT_X, pos->x_hotspot, + CURSOR_HOT_SPOT_Y, pos->y_hotspot); + REG_SET(CURSOR_DST_OFFSET, 0, + CURSOR_DST_X_OFFSET, dst_x_offset); + } /* Cursor Position Register Config */ hubp->pos.cur_ctl.bits.cur_enable = cur_en; hubp->pos.position.bits.x_pos = pos->x; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c index 9333b7fde3bc..7aa0f452e8f7 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c @@ -1699,7 +1699,7 @@ void dcn35_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pipe p->CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE = hubp->att.cur_ctl.bits.mode; p->CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY = hubp->pos.cur_ctl.bits.cur_2x_magnify; p->CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH = hubp->att.cur_ctl.bits.pitch; - p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->pos.cur_ctl.bits.line_per_chunk; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->att.cur_ctl.bits.line_per_chunk; p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_ENABLE = dpp->att.cur0_ctl.bits.cur0_enable; p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_MODE = dpp->att.cur0_ctl.bits.mode; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index b9c357a40707..f02edc9371b0 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -2996,7 +2996,7 @@ void dcn401_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pip p->CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE = hubp->att.cur_ctl.bits.mode; p->CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY = hubp->pos.cur_ctl.bits.cur_2x_magnify; p->CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH = hubp->att.cur_ctl.bits.pitch; - p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->pos.cur_ctl.bits.line_per_chunk; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->att.cur_ctl.bits.line_per_chunk; p->CM_CUR0_CURSOR0_CONTROL__CUR0_ENABLE = dpp->att.cur0_ctl.bits.cur0_enable; p->CM_CUR0_CURSOR0_CONTROL__CUR0_MODE = dpp->att.cur0_ctl.bits.mode; -- cgit v1.2.3 From 8e6361a9e09846ff1fd3a908345862b8dc2632fd Mon Sep 17 00:00:00 2001 From: Dominik Kaszewski Date: Wed, 30 Jul 2025 15:16:15 +0200 Subject: drm/amd/display: Add lock descriptor to check_update [Why] DM locks the global DC lock during all updates, even if multiple updates touch different resources and could be run in parallel. [How] Add extra enum specifying which kind of resources should be locked. Reviewed-by: Aric Cyr Signed-off-by: Dominik Kaszewski Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 125 ++++++++++++++++------------- drivers/gpu/drm/amd/display/dc/dc.h | 13 +++ drivers/gpu/drm/amd/display/dc/dc_stream.h | 2 +- 3 files changed, 82 insertions(+), 58 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 6ef9dd9ab7a3..1851a35f629f 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -148,10 +148,16 @@ static const char DC_BUILD_ID[] = "production-build"; /* Private functions */ -static inline void elevate_update_type(enum surface_update_type *original, enum surface_update_type new) +static inline void elevate_update_type( + struct surface_update_descriptor *descriptor, + enum surface_update_type new_type, + enum dc_lock_descriptor new_locks +) { - if (new > *original) - *original = new; + if (new_type > descriptor->update_type) + descriptor->update_type = new_type; + + descriptor->lock_descriptor |= new_locks; } static void destroy_links(struct dc *dc) @@ -2656,47 +2662,49 @@ static bool is_surface_in_context( return false; } -static enum surface_update_type get_plane_info_update_type(const struct dc_surface_update *u) +static struct surface_update_descriptor get_plane_info_update_type(const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; - enum surface_update_type update_type = UPDATE_TYPE_FAST; + struct surface_update_descriptor update_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (!u->plane_info) - return UPDATE_TYPE_FAST; + return update_type; + + elevate_update_type(&update_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_PLANE); if (u->plane_info->color_space != u->surface->color_space) { update_flags->bits.color_space_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->horizontal_mirror != u->surface->horizontal_mirror) { update_flags->bits.horizontal_mirror_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->rotation != u->surface->rotation) { update_flags->bits.rotation_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->format != u->surface->format) { update_flags->bits.pixel_format_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->stereo_format != u->surface->stereo_format) { update_flags->bits.stereo_format_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->per_pixel_alpha != u->surface->per_pixel_alpha) { update_flags->bits.per_pixel_alpha_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->global_alpha_value != u->surface->global_alpha_value) { update_flags->bits.global_alpha_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->dcc.enable != u->surface->dcc.enable @@ -2708,7 +2716,7 @@ static enum surface_update_type get_plane_info_update_type(const struct dc_surfa * recalculate stutter period. */ update_flags->bits.dcc_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (resource_pixel_format_to_bpp(u->plane_info->format) != @@ -2717,33 +2725,33 @@ static enum surface_update_type get_plane_info_update_type(const struct dc_surfa * and DML calculation */ update_flags->bits.bpp_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (u->plane_info->plane_size.surface_pitch != u->surface->plane_size.surface_pitch || u->plane_info->plane_size.chroma_pitch != u->surface->plane_size.chroma_pitch) { update_flags->bits.plane_size_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } const struct dc_tiling_info *tiling = &u->plane_info->tiling_info; if (memcmp(tiling, &u->surface->tiling_info, sizeof(*tiling)) != 0) { update_flags->bits.swizzle_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); switch (tiling->gfxversion) { case DcGfxVersion9: case DcGfxVersion10: case DcGfxVersion11: if (tiling->gfx9.swizzle != DC_SW_LINEAR) { - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); update_flags->bits.bandwidth_change = 1; } break; case DcGfxAddr3: if (tiling->gfx_addr3.swizzle != DC_ADDR3_SW_LINEAR) { - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); update_flags->bits.bandwidth_change = 1; } break; @@ -2759,14 +2767,17 @@ static enum surface_update_type get_plane_info_update_type(const struct dc_surfa return update_type; } -static enum surface_update_type get_scaling_info_update_type( +static struct surface_update_descriptor get_scaling_info_update_type( const struct dc_check_config *check_config, const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; + struct surface_update_descriptor update_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (!u->scaling_info) - return UPDATE_TYPE_FAST; + return update_type; + + elevate_update_type(&update_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_PLANE); if (u->scaling_info->src_rect.width != u->surface->src_rect.width || u->scaling_info->src_rect.height != u->surface->src_rect.height @@ -2809,40 +2820,41 @@ static enum surface_update_type get_scaling_info_update_type( if (update_flags->bits.clock_change || update_flags->bits.bandwidth_change || update_flags->bits.scaling_change) - return UPDATE_TYPE_FULL; + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); if (update_flags->bits.position_change) - return UPDATE_TYPE_MED; + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); - return UPDATE_TYPE_FAST; + return update_type; } -static enum surface_update_type det_surface_update( +static struct surface_update_descriptor det_surface_update( const struct dc_check_config *check_config, - const struct dc_surface_update *u) + struct dc_surface_update *u) { - enum surface_update_type type; - enum surface_update_type overall_type = UPDATE_TYPE_FAST; + struct surface_update_descriptor overall_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; union surface_update_flags *update_flags = &u->surface->update_flags; if (u->surface->force_full_update) { update_flags->raw = 0xFFFFFFFF; - return UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); + return overall_type; } update_flags->raw = 0; // Reset all flags - type = get_plane_info_update_type(u); - elevate_update_type(&overall_type, type); + struct surface_update_descriptor inner_type = get_plane_info_update_type(u); - type = get_scaling_info_update_type(check_config, u); - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); + + inner_type = get_scaling_info_update_type(check_config, u); + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); if (u->flip_addr) { update_flags->bits.addr_update = 1; if (u->flip_addr->address.tmz_surface != u->surface->address.tmz_surface) { update_flags->bits.tmz_changed = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_FULL); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } } if (u->in_transfer_func) @@ -2878,13 +2890,15 @@ static enum surface_update_type det_surface_update( if (u->hdr_mult.value) if (u->hdr_mult.value != u->surface->hdr_mult.value) { update_flags->bits.hdr_mult = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_MED); + // TODO: Should be fast? + elevate_update_type(&overall_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (u->sdr_white_level_nits) if (u->sdr_white_level_nits != u->surface->sdr_white_level_nits) { update_flags->bits.sdr_white_level_nits = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_FULL); + // TODO: Should be fast? + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (u->cm2_params) { @@ -2898,18 +2912,16 @@ static enum surface_update_type det_surface_update( update_flags->bits.mcm_transfer_function_enable_change = 1; } if (update_flags->bits.in_transfer_func_change) { - type = UPDATE_TYPE_MED; - elevate_update_type(&overall_type, type); + // TODO: Fast? + elevate_update_type(&overall_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STATE); } if (update_flags->bits.lut_3d && u->surface->mcm_luts.lut3d_data.lut3d_src != DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (update_flags->bits.mcm_transfer_function_enable_change) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (check_config->enable_legacy_fast_update && @@ -2917,8 +2929,7 @@ static enum surface_update_type det_surface_update( update_flags->bits.gamut_remap_change || update_flags->bits.input_csc_change || update_flags->bits.coeff_reduction_change)) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } return overall_type; } @@ -2946,28 +2957,28 @@ static void force_immediate_gsl_plane_flip(struct dc *dc, struct dc_surface_upda } } -static enum surface_update_type check_update_surfaces_for_stream( +static struct surface_update_descriptor check_update_surfaces_for_stream( const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update -) + struct dc_stream_update *stream_update) { - int i; - enum surface_update_type overall_type = UPDATE_TYPE_FAST; + struct surface_update_descriptor overall_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (stream_update && stream_update->pending_test_pattern) { - overall_type = UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } if (stream_update && stream_update->hw_cursor_req) { - overall_type = UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); } /* some stream updates require passive update */ if (stream_update) { union stream_update_flags *su_flags = &stream_update->stream->update_flags; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + if ((stream_update->src.height != 0 && stream_update->src.width != 0) || (stream_update->dst.height != 0 && stream_update->dst.width != 0) || stream_update->integer_scaling_update) @@ -3009,7 +3020,7 @@ static enum surface_update_type check_update_surfaces_for_stream( su_flags->bits.out_csc = 1; if (su_flags->raw != 0) - overall_type = UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_STATE); if (stream_update->output_csc_transform) su_flags->bits.out_csc = 1; @@ -3021,11 +3032,11 @@ static enum surface_update_type check_update_surfaces_for_stream( su_flags->bits.out_tf = 1; } - for (i = 0 ; i < surface_count; i++) { - enum surface_update_type type = + for (int i = 0 ; i < surface_count; i++) { + struct surface_update_descriptor inner_type = det_surface_update(check_config, &updates[i]); - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); } return overall_type; @@ -3036,7 +3047,7 @@ static enum surface_update_type check_update_surfaces_for_stream( * * See :c:type:`enum surface_update_type ` for explanation of update types */ -enum surface_update_type dc_check_update_surfaces_for_stream( +struct surface_update_descriptor dc_check_update_surfaces_for_stream( const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, @@ -3468,7 +3479,7 @@ static bool update_planes_and_stream_state(struct dc *dc, context = dc->current_state; update_type = dc_check_update_surfaces_for_stream( - &dc->check_config, srf_updates, surface_count, stream_update); + &dc->check_config, srf_updates, surface_count, stream_update).update_type; if (full_update_required_weak(dc, srf_updates, surface_count, stream_update, stream)) update_type = UPDATE_TYPE_FULL; diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 4a5aa0bd3bc4..142cbb7c77cd 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -467,6 +467,19 @@ enum surface_update_type { UPDATE_TYPE_FULL, /* may need to shuffle resources */ }; +enum dc_lock_descriptor { + LOCK_DESCRIPTOR_NONE = 0x0, + LOCK_DESCRIPTOR_STATE = 0x1, + LOCK_DESCRIPTOR_LINK = 0x2, + LOCK_DESCRIPTOR_STREAM = 0x4, + LOCK_DESCRIPTOR_PLANE = 0x8, +}; + +struct surface_update_descriptor { + enum surface_update_type update_type; + enum dc_lock_descriptor lock_descriptor; +}; + /* Forward declaration*/ struct dc; struct dc_plane_state; diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index c68153c77972..75869eed1d3a 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -473,7 +473,7 @@ void dc_enable_stereo( /* Triggers multi-stream synchronization. */ void dc_trigger_sync(struct dc *dc, struct dc_state *context); -enum surface_update_type dc_check_update_surfaces_for_stream( +struct surface_update_descriptor dc_check_update_surfaces_for_stream( const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, -- cgit v1.2.3 From 48d6be05f540233a72c954777ede1c60b9094386 Mon Sep 17 00:00:00 2001 From: Dominik Kaszewski Date: Thu, 16 Oct 2025 13:46:25 +0200 Subject: drm/amd/display: Make observers const-correct [Why] Observers which do not modify their pointer arguments should take them as const. This clearly signals their intent to the caller, making it clear that the function is safe to call multiple times, or remove the call if the result is no longer necessary. [How] Made const-correct all of the functions below: * full_update_required[_weak] * fast_updates_exist * fast_update_only * dc_can_clear_cursor_limit * dc_stream_get_status (added const named overload) Reviewed-by: Nicholas Kazlauskas Signed-off-by: Dominik Kaszewski Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 44 ++++++++++++++----------- drivers/gpu/drm/amd/display/dc/core/dc_stream.c | 8 +++++ drivers/gpu/drm/amd/display/dc/dc.h | 2 +- drivers/gpu/drm/amd/display/dc/dc_stream.h | 4 +-- 4 files changed, 35 insertions(+), 23 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 1851a35f629f..b720e007c654 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3427,11 +3427,12 @@ static void update_seamless_boot_flags(struct dc *dc, } } -static bool full_update_required_weak(struct dc *dc, - struct dc_surface_update *srf_updates, +static bool full_update_required_weak( + const struct dc *dc, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream); + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream); /** * update_planes_and_stream_state() - The function takes planes and stream @@ -5015,7 +5016,7 @@ void populate_fast_updates(struct dc_fast_update *fast_update, } } -static bool fast_updates_exist(struct dc_fast_update *fast_update, int surface_count) +static bool fast_updates_exist(const struct dc_fast_update *fast_update, int surface_count) { int i; @@ -5056,11 +5057,12 @@ bool fast_nonaddr_updates_exist(struct dc_fast_update *fast_update, int surface_ return false; } -static bool full_update_required_weak(struct dc *dc, - struct dc_surface_update *srf_updates, +static bool full_update_required_weak( + const struct dc *dc, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream) + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) { const struct dc_state *context = dc->current_state; if (srf_updates) @@ -5069,7 +5071,7 @@ static bool full_update_required_weak(struct dc *dc, return true; if (stream) { - const struct dc_stream_status *stream_status = dc_stream_get_status(stream); + const struct dc_stream_status *stream_status = dc_stream_get_status_const(stream); if (stream_status == NULL || stream_status->plane_count != surface_count) return true; } @@ -5082,11 +5084,12 @@ static bool full_update_required_weak(struct dc *dc, return false; } -static bool full_update_required(struct dc *dc, - struct dc_surface_update *srf_updates, +static bool full_update_required( + const struct dc *dc, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream) + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) { if (full_update_required_weak(dc, srf_updates, surface_count, stream_update, stream)) return true; @@ -5146,12 +5149,13 @@ static bool full_update_required(struct dc *dc, return false; } -static bool fast_update_only(struct dc *dc, - struct dc_fast_update *fast_update, - struct dc_surface_update *srf_updates, +static bool fast_update_only( + const struct dc *dc, + const struct dc_fast_update *fast_update, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream) + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) { return fast_updates_exist(fast_update, surface_count) && !full_update_required(dc, srf_updates, surface_count, stream_update, stream); @@ -6383,7 +6387,7 @@ bool dc_is_cursor_limit_pending(struct dc *dc) return false; } -bool dc_can_clear_cursor_limit(struct dc *dc) +bool dc_can_clear_cursor_limit(const struct dc *dc) { uint32_t i; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c index 096fbb8ad24c..6d309c320253 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c @@ -224,6 +224,14 @@ struct dc_stream_status *dc_stream_get_status( return dc_state_get_stream_status(dc->current_state, stream); } +const struct dc_stream_status *dc_stream_get_status_const( + const struct dc_stream_state *stream) +{ + struct dc *dc = stream->ctx->dc; + + return dc_state_get_stream_status(dc->current_state, stream); +} + void program_cursor_attributes( struct dc *dc, struct dc_stream_state *stream) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 142cbb7c77cd..5bfd764467a1 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -2757,7 +2757,7 @@ bool dc_is_timing_changed(struct dc_stream_state *cur_stream, struct dc_stream_state *new_stream); bool dc_is_cursor_limit_pending(struct dc *dc); -bool dc_can_clear_cursor_limit(struct dc *dc); +bool dc_can_clear_cursor_limit(const struct dc *dc); /** * dc_get_underflow_debug_data_for_otg() - Retrieve underflow debug data. diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index 75869eed1d3a..321cfe92d799 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -491,8 +491,8 @@ void update_stream_signal(struct dc_stream_state *stream, struct dc_sink *sink); void dc_stream_retain(struct dc_stream_state *dc_stream); void dc_stream_release(struct dc_stream_state *dc_stream); -struct dc_stream_status *dc_stream_get_status( - struct dc_stream_state *dc_stream); +struct dc_stream_status *dc_stream_get_status(struct dc_stream_state *dc_stream); +const struct dc_stream_status *dc_stream_get_status_const(const struct dc_stream_state *dc_stream); /******************************************************************************* * Cursor interfaces - To manages the cursor within a stream -- cgit v1.2.3 From 99441824bec63549a076cd86631d138ec9a0c71c Mon Sep 17 00:00:00 2001 From: Alex Hung Date: Thu, 16 Oct 2025 20:08:10 -0600 Subject: drm/amd/display: Add HDR workaround for a specific eDP [WHY & HOW] Some eDP panels suffer from flicking when HDR is enabled in KDE or Gnome. This add another quirk to worksaround to skip VSC that is incompatible with an eDP panel. Link: https://gitlab.freedesktop.org/drm/amd/-/issues/4452 Reviewed-by: Aurabindo Pillai Signed-off-by: Alex Hung Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index fe100e4c9801..cc21337a182f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -83,6 +83,7 @@ static void apply_edid_quirks(struct drm_device *dev, struct edid *edid, struct edid_caps->panel_patch.remove_sink_ext_caps = true; break; case drm_edid_encode_panel_id('S', 'D', 'C', 0x4154): + case drm_edid_encode_panel_id('S', 'D', 'C', 0x4171): drm_dbg_driver(dev, "Disabling VSC on monitor with panel id %X\n", panel_id); edid_caps->panel_patch.disable_colorimetry = true; break; -- cgit v1.2.3 From cb57b8cdb072dc37723b6906da1c37ff9cbc2da4 Mon Sep 17 00:00:00 2001 From: Ivan Lipski Date: Wed, 17 Sep 2025 11:00:02 -0400 Subject: drm/amd/display: Fix incorrect return of vblank enable on unconfigured crtc [Why&How] Return -EINVAL when userspace asks us to enable vblank on a crtc that is not yet enabled. Suggested-by: Aurabindo Pillai Reviewed-by: Aurabindo Pillai Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/1856 Signed-off-by: Ivan Lipski Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index f08121a2b838..38f9ea313dcb 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -308,8 +308,12 @@ static inline int amdgpu_dm_crtc_set_vblank(struct drm_crtc *crtc, bool enable) int irq_type; int rc = 0; - if (acrtc->otg_inst == -1) - goto skip; + if (enable && !acrtc->base.enabled) { + drm_dbg_vbl(crtc->dev, + "Reject vblank enable on unconfigured CRTC %d (enabled=%d)\n", + acrtc->crtc_id, acrtc->base.enabled); + return -EINVAL; + } irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, acrtc->crtc_id); @@ -394,7 +398,7 @@ static inline int amdgpu_dm_crtc_set_vblank(struct drm_crtc *crtc, bool enable) return rc; } #endif -skip: + if (amdgpu_in_reset(adev)) return 0; -- cgit v1.2.3 From 020ee49ce9a839285806cc430ca8e059fc95aade Mon Sep 17 00:00:00 2001 From: Taimur Hassan Date: Fri, 17 Oct 2025 17:20:46 -0400 Subject: drm/amd/display: [FW Promotion] Release 0.1.33.0 [Why & How] - Extend reply debug flags, define a new bit as debug_log_enabled - Replace the padding to frame_skip_number in struct dmub_cmd_replay_set_coasting_vtotal_data Acked-by: Tom Chung Signed-off-by: Taimur Hassan Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index b2d5154d1402..772e07a1a959 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -485,7 +485,13 @@ union replay_debug_flags { */ uint32_t enable_visual_confirm_debug : 1; - uint32_t reserved : 18; + /** + * 0x4000 (bit 14) + * @debug_log_enabled: Debug Log Enabled + */ + uint32_t debug_log_enabled : 1; + + uint32_t reserved : 17; } bitfields; uint32_t u32All; @@ -4608,9 +4614,9 @@ struct dmub_cmd_replay_set_coasting_vtotal_data { */ uint16_t coasting_vtotal_high; /** - * Explicit padding to 4 byte boundary. + * frame skip number. */ - uint8_t pad[2]; + uint16_t frame_skip_number; }; /** -- cgit v1.2.3 From a981cb0f37eb20f47b8fc4072602ba728783082f Mon Sep 17 00:00:00 2001 From: Taimur Hassan Date: Fri, 17 Oct 2025 19:22:02 -0500 Subject: drm/amd/display: Promote DC to 3.2.356 This version brings along following update: - Fix incorrect return of vblank enable on unconfigured crtc - Add HDR workaround for a specific eDP - Make observers const-correct - Add lock descriptor to check_update - Update cursor offload assignments - Add dc interface to log pre os firmware information - Init dispclk from bootup clock for DCN315 - Remove dc param from check_update - Update link encoder assignment - Add more DC HW state info to underflow logging - Rename dml2 to dml2_0 folder - Fix notification of vtotal to DMU for cursor offload - Fix wrong index for DCN401 cursor offload - Add opp count validation to dml2.1 - Fix DMUB reset sequence for DCN32 - Bump minimum for frame_warn_limit Acked-by: Tom Chung Signed-off-by: Taimur Hassan Signed-off-by: Wayne Lin Tested-by: Dan Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 5bfd764467a1..010d9315b96b 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -63,7 +63,7 @@ struct dcn_dsc_reg_state; struct dcn_optc_reg_state; struct dcn_dccg_reg_state; -#define DC_VER "3.2.355" +#define DC_VER "3.2.356" /** * MAX_SURFACES - representative of the upper bound of surfaces that can be piped to a single CRTC -- cgit v1.2.3 From f18719ef4bb7b0e48dee92703b01e16f6c0d6318 Mon Sep 17 00:00:00 2001 From: "Jesse.Zhang" Date: Tue, 21 Oct 2025 13:01:06 +0800 Subject: drm/amdgpu: Convert amdgpu userqueue management from IDR to XArray MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This commit refactors the AMDGPU userqueue management subsystem to replace IDR (ID Allocation) with XArray for improved performance, scalability, and maintainability. The changes address several issues with the previous IDR implementation and provide better locking semantics. Key changes: 1. **Global XArray Introduction**: - Added `userq_doorbell_xa` to `struct amdgpu_device` for global queue tracking - Uses doorbell_index as key for efficient global lookup - Replaces the previous `userq_mgr_list` linked list approach 2. **Per-process XArray Conversion**: - Replaced `userq_idr` with `userq_mgr_xa` in `struct amdgpu_userq_mgr` - Maintains per-process queue tracking with queue_id as key - Uses XA_FLAGS_ALLOC for automatic ID allocation 3. **Locking Improvements**: - Removed global `userq_mutex` from `struct amdgpu_device` - Replaced with fine-grained XArray locking using XArray's internal spinlocks 4. **Runtime Idle Check Optimization**: - Updated `amdgpu_runtime_idle_check_userq()` to use xa_empty 5. **Queue Management Functions**: - Converted all IDR operations to equivalent XArray functions: - `idr_alloc()` → `xa_alloc()` - `idr_find()` → `xa_load()` - `idr_remove()` → `xa_erase()` - `idr_for_each()` → `xa_for_each()` Benefits: - **Performance**: XArray provides better scalability for large numbers of queues - **Memory Efficiency**: Reduced memory overhead compared to IDR - **Thread Safety**: Improved locking semantics with XArray's internal spinlocks v2: rename userq_global_xa/userq_xa to userq_doorbell_xa/userq_mgr_xa Remove xa_lock and use its own lock. v3: Set queue->userq_mgr = uq_mgr in amdgpu_userq_create() v4: use xa_store_irq (Christian) hold the read side of the reset lock while creating/destroying queues and the manager data structure. (Chritian) Acked-by: Alex Deucher Suggested-by: Christian König Signed-off-by: Jesse Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 8 +- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 3 +- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 16 +-- drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 151 ++++++++++++------------ drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h | 8 +- drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c | 4 +- drivers/gpu/drm/amd/amdgpu/mes_userqueue.c | 22 ++-- 7 files changed, 101 insertions(+), 111 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index a5574e84694b..50079209c472 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1176,6 +1176,12 @@ struct amdgpu_device { * queue fence. */ struct xarray userq_xa; + /** + * @userq_doorbell_xa: Global user queue map (doorbell index → queue) + * Key: doorbell_index (unique global identifier for the queue) + * Value: struct amdgpu_usermode_queue + */ + struct xarray userq_doorbell_xa; /* df */ struct amdgpu_df df; @@ -1309,8 +1315,6 @@ struct amdgpu_device { */ bool apu_prefer_gtt; - struct list_head userq_mgr_list; - struct mutex userq_mutex; bool userq_halt_for_enforce_isolation; struct amdgpu_uid *uid_info; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 3ffb9bb1ec0b..b8d91247f51a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -4558,7 +4558,6 @@ int amdgpu_device_init(struct amdgpu_device *adev, mutex_init(&adev->gfx.userq_sch_mutex); mutex_init(&adev->gfx.workload_profile_mutex); mutex_init(&adev->vcn.workload_profile_mutex); - mutex_init(&adev->userq_mutex); amdgpu_device_init_apu_flags(adev); @@ -4586,7 +4585,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, INIT_LIST_HEAD(&adev->pm.od_kobj_list); - INIT_LIST_HEAD(&adev->userq_mgr_list); + xa_init(&adev->userq_doorbell_xa); INIT_DELAYED_WORK(&adev->delayed_init_work, amdgpu_device_delayed_init_work_handler); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index a36e15beafeb..cee90f9e58a9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2772,22 +2772,8 @@ static int amdgpu_runtime_idle_check_userq(struct device *dev) struct pci_dev *pdev = to_pci_dev(dev); struct drm_device *drm_dev = pci_get_drvdata(pdev); struct amdgpu_device *adev = drm_to_adev(drm_dev); - struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; - int ret = 0; - - mutex_lock(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - ret = -EBUSY; - goto done; - } - } -done: - mutex_unlock(&adev->userq_mutex); - return ret; + return xa_empty(&adev->userq_doorbell_xa) ? 0 : -EBUSY; } static int amdgpu_pmops_runtime_suspend(struct device *dev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index 2200e0bbf040..3b1811b110a4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -30,6 +30,7 @@ #include "amdgpu_vm.h" #include "amdgpu_userq.h" #include "amdgpu_hmm.h" +#include "amdgpu_reset.h" #include "amdgpu_userq_fence.h" u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev) @@ -278,19 +279,27 @@ amdgpu_userq_cleanup(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_device *adev = uq_mgr->adev; const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type]; + /* Wait for mode-1 reset to complete */ + down_read(&adev->reset_domain->sem); + /* Drop the userq reference. */ amdgpu_userq_buffer_vas_list_cleanup(adev, queue); uq_funcs->mqd_destroy(uq_mgr, queue); amdgpu_userq_fence_driver_free(queue); - idr_remove(&uq_mgr->userq_idr, queue_id); + /* Use interrupt-safe locking since IRQ handlers may access these XArrays */ + xa_erase_irq(&uq_mgr->userq_mgr_xa, (unsigned long)queue_id); + xa_erase_irq(&adev->userq_doorbell_xa, queue->doorbell_index); + queue->userq_mgr = NULL; list_del(&queue->userq_va_list); kfree(queue); + + up_read(&adev->reset_domain->sem); } static struct amdgpu_usermode_queue * amdgpu_userq_find(struct amdgpu_userq_mgr *uq_mgr, int qid) { - return idr_find(&uq_mgr->userq_idr, qid); + return xa_load(&uq_mgr->userq_mgr_xa, qid); } void @@ -551,8 +560,9 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) struct amdgpu_db_info db_info; char *queue_name; bool skip_map_queue; + u32 qid; uint64_t index; - int qid, r = 0; + int r = 0; int priority = (args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK) >> AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT; @@ -575,7 +585,6 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) * * This will also make sure we have a valid eviction fence ready to be used. */ - mutex_lock(&adev->userq_mutex); amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr); uq_funcs = adev->userq_funcs[args->in.ip_type]; @@ -638,15 +647,27 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) goto unlock; } - qid = idr_alloc(&uq_mgr->userq_idr, queue, 1, AMDGPU_MAX_USERQ_COUNT, GFP_KERNEL); - if (qid < 0) { + /* Wait for mode-1 reset to complete */ + down_read(&adev->reset_domain->sem); + r = xa_err(xa_store_irq(&adev->userq_doorbell_xa, index, queue, GFP_KERNEL)); + if (r) { + kfree(queue); + up_read(&adev->reset_domain->sem); + goto unlock; + } + + r = xa_alloc(&uq_mgr->userq_mgr_xa, &qid, queue, XA_LIMIT(1, AMDGPU_MAX_USERQ_COUNT), GFP_KERNEL); + if (r) { drm_file_err(uq_mgr->file, "Failed to allocate a queue id\n"); amdgpu_userq_fence_driver_free(queue); uq_funcs->mqd_destroy(uq_mgr, queue); kfree(queue); r = -ENOMEM; + up_read(&adev->reset_domain->sem); goto unlock; } + up_read(&adev->reset_domain->sem); + queue->userq_mgr = uq_mgr; /* don't map the queue if scheduling is halted */ if (adev->userq_halt_for_enforce_isolation && @@ -659,7 +680,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) r = amdgpu_userq_map_helper(uq_mgr, queue); if (r) { drm_file_err(uq_mgr->file, "Failed to map Queue\n"); - idr_remove(&uq_mgr->userq_idr, qid); + xa_erase(&uq_mgr->userq_mgr_xa, qid); amdgpu_userq_fence_driver_free(queue); uq_funcs->mqd_destroy(uq_mgr, queue); kfree(queue); @@ -684,7 +705,6 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) unlock: mutex_unlock(&uq_mgr->userq_mutex); - mutex_unlock(&adev->userq_mutex); return r; } @@ -782,11 +802,11 @@ static int amdgpu_userq_restore_all(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id; + unsigned long queue_id; int ret = 0, r; /* Resume all the queues for this process */ - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { if (!amdgpu_userq_buffer_vas_mapped(queue)) { drm_file_err(uq_mgr->file, @@ -1023,11 +1043,11 @@ static int amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id; + unsigned long queue_id; int ret = 0, r; /* Try to unmap all the queues in this process ctx */ - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { r = amdgpu_userq_preempt_helper(uq_mgr, queue); if (r) ret = r; @@ -1042,9 +1062,10 @@ static int amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id, ret; + unsigned long queue_id; + int ret; - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { struct dma_fence *f = queue->last_fence; if (!f || dma_fence_is_signaled(f)) @@ -1097,44 +1118,30 @@ int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *f struct amdgpu_device *adev) { mutex_init(&userq_mgr->userq_mutex); - idr_init_base(&userq_mgr->userq_idr, 1); + xa_init_flags(&userq_mgr->userq_mgr_xa, XA_FLAGS_ALLOC); userq_mgr->adev = adev; userq_mgr->file = file_priv; - mutex_lock(&adev->userq_mutex); - list_add(&userq_mgr->list, &adev->userq_mgr_list); - mutex_unlock(&adev->userq_mutex); - INIT_DELAYED_WORK(&userq_mgr->resume_work, amdgpu_userq_restore_worker); return 0; } void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr) { - struct amdgpu_device *adev = userq_mgr->adev; struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - uint32_t queue_id; + unsigned long queue_id; cancel_delayed_work_sync(&userq_mgr->resume_work); - mutex_lock(&adev->userq_mutex); mutex_lock(&userq_mgr->userq_mutex); - idr_for_each_entry(&userq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&userq_mgr->userq_mgr_xa, queue_id, queue) { amdgpu_userq_wait_for_last_fence(userq_mgr, queue); amdgpu_userq_unmap_helper(userq_mgr, queue); amdgpu_userq_cleanup(userq_mgr, queue, queue_id); } - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - if (uqm == userq_mgr) { - list_del(&uqm->list); - break; - } - } - idr_destroy(&userq_mgr->userq_idr); + xa_destroy(&userq_mgr->userq_mgr_xa); mutex_unlock(&userq_mgr->userq_mutex); - mutex_unlock(&adev->userq_mutex); mutex_destroy(&userq_mgr->userq_mutex); } @@ -1142,25 +1149,23 @@ int amdgpu_userq_suspend(struct amdgpu_device *adev) { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int r; if (!ip_mask) return 0; - guard(mutex)(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; cancel_delayed_work_sync(&uqm->resume_work); guard(mutex)(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (adev->in_s0ix) - r = amdgpu_userq_preempt_helper(uqm, queue); - else - r = amdgpu_userq_unmap_helper(uqm, queue); - if (r) - return r; - } + if (adev->in_s0ix) + r = amdgpu_userq_preempt_helper(uqm, queue); + else + r = amdgpu_userq_unmap_helper(uqm, queue); + if (r) + return r; } return 0; } @@ -1169,24 +1174,22 @@ int amdgpu_userq_resume(struct amdgpu_device *adev) { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int r; if (!ip_mask) return 0; - guard(mutex)(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; guard(mutex)(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (adev->in_s0ix) - r = amdgpu_userq_restore_helper(uqm, queue); - else - r = amdgpu_userq_map_helper(uqm, queue); - if (r) - return r; - } + if (adev->in_s0ix) + r = amdgpu_userq_restore_helper(uqm, queue); + else + r = amdgpu_userq_map_helper(uqm, queue); + if (r) + return r; } return 0; @@ -1197,33 +1200,31 @@ int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev, { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int ret = 0, r; /* only need to stop gfx/compute */ if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE)))) return 0; - mutex_lock(&adev->userq_mutex); if (adev->userq_halt_for_enforce_isolation) dev_warn(adev->dev, "userq scheduling already stopped!\n"); adev->userq_halt_for_enforce_isolation = true; - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; cancel_delayed_work_sync(&uqm->resume_work); mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (((queue->queue_type == AMDGPU_HW_IP_GFX) || - (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && - (queue->xcp_id == idx)) { - r = amdgpu_userq_preempt_helper(uqm, queue); - if (r) - ret = r; - } + if (((queue->queue_type == AMDGPU_HW_IP_GFX) || + (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && + (queue->xcp_id == idx)) { + r = amdgpu_userq_preempt_helper(uqm, queue); + if (r) + ret = r; } mutex_unlock(&uqm->userq_mutex); } - mutex_unlock(&adev->userq_mutex); + return ret; } @@ -1232,21 +1233,20 @@ int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev, { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int ret = 0, r; /* only need to stop gfx/compute */ if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE)))) return 0; - mutex_lock(&adev->userq_mutex); if (!adev->userq_halt_for_enforce_isolation) dev_warn(adev->dev, "userq scheduling already started!\n"); adev->userq_halt_for_enforce_isolation = false; - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { if (((queue->queue_type == AMDGPU_HW_IP_GFX) || (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && (queue->xcp_id == idx)) { @@ -1254,10 +1254,9 @@ int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev, if (r) ret = r; } - } mutex_unlock(&uqm->userq_mutex); } - mutex_unlock(&adev->userq_mutex); + return ret; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h index 036d8dd585cd..09da0617bfa2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h @@ -96,11 +96,15 @@ struct amdgpu_userq_funcs { /* Usermode queues for gfx */ struct amdgpu_userq_mgr { - struct idr userq_idr; + /** + * @userq_mgr_xa: Per-process user queue map (queue ID → queue) + * Key: queue_id (unique ID within the process's userq manager) + * Value: struct amdgpu_usermode_queue + */ + struct xarray userq_mgr_xa; struct mutex userq_mutex; struct amdgpu_device *adev; struct delayed_work resume_work; - struct list_head list; struct drm_file *file; }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c index 761bad98da3e..2aeeaa954882 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c @@ -537,7 +537,7 @@ int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data, } /* Retrieve the user queue */ - queue = idr_find(&userq_mgr->userq_idr, args->queue_id); + queue = xa_load(&userq_mgr->userq_mgr_xa, args->queue_id); if (!queue) { r = -ENOENT; goto put_gobj_write; @@ -899,7 +899,7 @@ int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data, */ num_fences = dma_fence_dedup_array(fences, num_fences); - waitq = idr_find(&userq_mgr->userq_idr, wait_info->waitq_id); + waitq = xa_load(&userq_mgr->userq_mgr_xa, wait_info->waitq_id); if (!waitq) { r = -EINVAL; goto free_fences; diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c index 5c63480dda9c..9894a3eed215 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c @@ -205,10 +205,10 @@ static int mes_userq_detect_and_reset(struct amdgpu_device *adev, int db_array_size = amdgpu_mes_get_hung_queue_db_array_size(adev); struct mes_detect_and_reset_queue_input input; struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; unsigned int hung_db_num = 0; - int queue_id, r, i; + unsigned long queue_id; u32 db_array[8]; + int r, i; if (db_array_size > 8) { dev_err(adev->dev, "DB array size (%d vs 8) too small\n", @@ -227,16 +227,14 @@ static int mes_userq_detect_and_reset(struct amdgpu_device *adev, if (r) { dev_err(adev->dev, "Failed to detect and reset queues, err (%d)\n", r); } else if (hung_db_num) { - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (queue->queue_type == queue_type) { - for (i = 0; i < hung_db_num; i++) { - if (queue->doorbell_index == db_array[i]) { - queue->state = AMDGPU_USERQ_STATE_HUNG; - atomic_inc(&adev->gpu_reset_counter); - amdgpu_userq_fence_driver_force_completion(queue); - drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL); - } + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + if (queue->queue_type == queue_type) { + for (i = 0; i < hung_db_num; i++) { + if (queue->doorbell_index == db_array[i]) { + queue->state = AMDGPU_USERQ_STATE_HUNG; + atomic_inc(&adev->gpu_reset_counter); + amdgpu_userq_fence_driver_force_completion(queue); + drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL); } } } -- cgit v1.2.3 From 97010d773208c4b800e4af1063e58a97c056f40b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:41 +0200 Subject: drm/amd/display: Determine DVI-I connector type (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit DC determines the DRM connector type based on the signal type, which becomes problematic when a connector may support different signal types, such as DVI-I. With this patch, it is now determined according to the actual connector type for DVI-D and DVI-I connectors. Also set the HPD (hotplug detection) flag for DVI-I connectors to prevent regressing their digital functionality, which has been already working. A subsequent commit will also implement polling for DVI-I. v2: Only use connector type for DVI to prevent regressions for other signal types. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 5b446e4000ef..9e260ca04e1e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -8263,7 +8263,7 @@ static int dm_update_mst_vcpi_slots_for_dsc(struct drm_atomic_state *state, return 0; } -static int to_drm_connector_type(enum signal_type st) +static int to_drm_connector_type(enum signal_type st, uint32_t connector_id) { switch (st) { case SIGNAL_TYPE_HDMI_TYPE_A: @@ -8279,6 +8279,10 @@ static int to_drm_connector_type(enum signal_type st) return DRM_MODE_CONNECTOR_DisplayPort; case SIGNAL_TYPE_DVI_DUAL_LINK: case SIGNAL_TYPE_DVI_SINGLE_LINK: + if (connector_id == CONNECTOR_ID_SINGLE_LINK_DVII || + connector_id == CONNECTOR_ID_DUAL_LINK_DVII) + return DRM_MODE_CONNECTOR_DVII; + return DRM_MODE_CONNECTOR_DVID; case SIGNAL_TYPE_VIRTUAL: return DRM_MODE_CONNECTOR_VIRTUAL; @@ -8668,6 +8672,7 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm, link->link_enc->features.dp_ycbcr420_supported ? true : false; break; case DRM_MODE_CONNECTOR_DVID: + case DRM_MODE_CONNECTOR_DVII: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; break; default: @@ -8871,7 +8876,7 @@ static int amdgpu_dm_connector_init(struct amdgpu_display_manager *dm, goto out_free; } - connector_type = to_drm_connector_type(link->connector_signal); + connector_type = to_drm_connector_type(link->connector_signal, link->link_id.id); res = drm_connector_init_with_ddc( dm->ddev, -- cgit v1.2.3 From 7bb02a34c2ba69bbe0329fc3049c795f7ba66b9a Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Fri, 24 Oct 2025 10:14:56 +0530 Subject: drm/amdkfd: add missing return value check for range amdgpu_hmm_range_alloc could fails in case of low memory condition and hence we should have a check for the return value. Signed-off-by: Sunil Khatri Reviewed-by: Shirish S Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index f041643308ca..7f0ab73e2396 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1738,6 +1738,11 @@ static int svm_range_validate_and_map(struct mm_struct *mm, WRITE_ONCE(p->svms.faulting_task, current); range = amdgpu_hmm_range_alloc(NULL); + if (unlikely(!range)) { + r = -ENOMEM; + goto free_ctx; + } + r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, readonly, owner, range); -- cgit v1.2.3 From 22d9e071e5028815accf26956ed0b9634a4c0b8d Mon Sep 17 00:00:00 2001 From: Philip Yang Date: Wed, 15 Oct 2025 15:17:54 -0400 Subject: drm/amdkfd: Dequeue user queues when process mm released Move dequeue user queues and destroy user queues from kfd_process_wq_release to mmu notifier release callback, to ensure no system memory access from GPU because the process memory is going to free from CPU after mmu release notifier callback returns. Destroy queue releases the svm prange queue_refcount, this also removes fake flase positive warning message "Freeing queue vital buffer" message if application crash or killed. Suggested-by: Felix Kuehling Signed-off-by: Philip Yang Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_process.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index ddfe30c13e9d..f5d173f1ca3b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1162,9 +1162,6 @@ static void kfd_process_wq_release(struct work_struct *work) release_work); struct dma_fence *ef; - kfd_process_dequeue_from_all_devices(p); - pqm_uninit(&p->pqm); - /* * If GPU in reset, user queues may still running, wait for reset complete. */ @@ -1226,6 +1223,14 @@ static void kfd_process_notifier_release_internal(struct kfd_process *p) cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); + /* + * Dequeue and destroy user queues, it is not safe for GPU to access + * system memory after mmu release notifier callback returns because + * exit_mmap free process memory afterwards. + */ + kfd_process_dequeue_from_all_devices(p); + pqm_uninit(&p->pqm); + for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; -- cgit v1.2.3 From a0559012a18a5a6ad87516e982892765a403b8ab Mon Sep 17 00:00:00 2001 From: Alex Deucher Date: Fri, 10 Oct 2025 15:21:02 -0400 Subject: drm/amdgpu/userq: fix SDMA and compute validation The CSA and EOP buffers have different alignement requirements. Hardcode them for now as a bug fix. A proper query will be added in a subsequent patch. v2: verify gfx shadow helper callback (Prike) Fixes: 9e46b8bb0539 ("drm/amdgpu: validate userq buffer virtual address and size") Reviewed-by: Prike Liang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/mes_userqueue.c | 19 ++++++++++++++----- 1 file changed, 14 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c index 9894a3eed215..b1ee9473d628 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c @@ -252,7 +252,6 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_mqd *mqd_hw_default = &adev->mqds[queue->queue_type]; struct drm_amdgpu_userq_in *mqd_user = args_in; struct amdgpu_mqd_prop *userq_props; - struct amdgpu_gfx_shadow_info shadow_info; int r; /* Structure to initialize MQD for userqueue using generic MQD init function */ @@ -278,8 +277,6 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, userq_props->doorbell_index = queue->doorbell_index; userq_props->fence_address = queue->fence_drv->gpu_addr; - if (adev->gfx.funcs->get_gfx_shadow_info) - adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true); if (queue->queue_type == AMDGPU_HW_IP_COMPUTE) { struct drm_amdgpu_userq_mqd_compute_gfx11 *compute_mqd; @@ -297,7 +294,7 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, } r = amdgpu_userq_input_va_validate(queue, compute_mqd->eop_va, - max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE)); + 2048); if (r) goto free_mqd; @@ -310,6 +307,14 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, kfree(compute_mqd); } else if (queue->queue_type == AMDGPU_HW_IP_GFX) { struct drm_amdgpu_userq_mqd_gfx11 *mqd_gfx_v11; + struct amdgpu_gfx_shadow_info shadow_info; + + if (adev->gfx.funcs->get_gfx_shadow_info) { + adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true); + } else { + r = -EINVAL; + goto free_mqd; + } if (mqd_user->mqd_size != sizeof(*mqd_gfx_v11) || !mqd_user->mqd) { DRM_ERROR("Invalid GFX MQD\n"); @@ -333,6 +338,10 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, shadow_info.shadow_size); if (r) goto free_mqd; + r = amdgpu_userq_input_va_validate(queue, mqd_gfx_v11->csa_va, + shadow_info.csa_size); + if (r) + goto free_mqd; kfree(mqd_gfx_v11); } else if (queue->queue_type == AMDGPU_HW_IP_DMA) { @@ -351,7 +360,7 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, goto free_mqd; } r = amdgpu_userq_input_va_validate(queue, mqd_sdma_v11->csa_va, - shadow_info.csa_size); + 32); if (r) goto free_mqd; -- cgit v1.2.3 From 1017e393ad6d61756f7f35a6eb059a42a40818e6 Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Fri, 24 Oct 2025 22:29:00 +0530 Subject: drm/amdkfd: fix the clean up when amdgpu_hmm_range_alloc fails we need to unreserve the bo's too during clean up along with freeing the memory of context. Fixes: 7bb02a34c2ba ("drm/amdkfd: add missing return value check for range") Signed-off-by: Sunil Khatri Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 7f0ab73e2396..729aac81563c 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1738,14 +1738,11 @@ static int svm_range_validate_and_map(struct mm_struct *mm, WRITE_ONCE(p->svms.faulting_task, current); range = amdgpu_hmm_range_alloc(NULL); - if (unlikely(!range)) { + if (likely(range)) + r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, + readonly, owner, range); + else r = -ENOMEM; - goto free_ctx; - } - - r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, - readonly, owner, - range); WRITE_ONCE(p->svms.faulting_task, NULL); if (r) { amdgpu_hmm_range_free(range); -- cgit v1.2.3 From dfc74e37bdb487eed5ad90d0eac4055f60217fb0 Mon Sep 17 00:00:00 2001 From: Srinivasan Shanmugam Date: Thu, 23 Oct 2025 19:54:16 +0530 Subject: drm/amdkfd: Fix use-after-free of HMM range in svm_range_validate_and_map() MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The function svm_range_validate_and_map() was freeing `range` when amdgpu_hmm_range_get_pages() failed. But later, the code still used the same `range` pointer and freed it again. This could cause a use-after-free and double-free issue. The fix sets `range = NULL` right after it is freed and checks for `range` before using or freeing it again. v2: Removed duplicate !r check in the condition for clarity. v3: In amdgpu_hmm_range_get_pages(), when hmm_range_fault() fails, we kvfree(pfns) but leave the pointer in hmm_range->hmm_pfns still pointing to freed memory. The caller (or amdgpu_hmm_range_free(range)) may try to free range->hmm_range.hmm_pfns again, causing a double free, Setting hmm_range->hmm_pfns = NULL immediately after kvfree(pfns) prevents both double free. (Philip) In svm_range_validate_and_map(), When r == 0, it means success → range is not NULL. When r != 0, it means failure → already made range = NULL. So checking both (!r && range) is unnecessary because the moment r == 0, we automatically know range exists and is safe to use. (Philip) Fixes: 737da5363cc0 ("drm/amdgpu: update the functions to use amdgpu version of hmm") Reported by: Dan Carpenter Cc: Philip Yang Cc: Sunil Khatri Cc: Christian König Cc: Alex Deucher Signed-off-by: Srinivasan Shanmugam Reviewed-by: Philip Yang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c | 1 + drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 6 ++++-- 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c index 7e5a09b0bc78..518ca3f4db2b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c @@ -221,6 +221,7 @@ retry: out_free_pfns: kvfree(pfns); + hmm_range->hmm_pfns = NULL; out_free_range: if (r == -EBUSY) r = -EAGAIN; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 729aac81563c..ffb7b36e577c 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1746,6 +1746,7 @@ static int svm_range_validate_and_map(struct mm_struct *mm, WRITE_ONCE(p->svms.faulting_task, NULL); if (r) { amdgpu_hmm_range_free(range); + range = NULL; pr_debug("failed %d to get svm range pages\n", r); } } else { @@ -1763,7 +1764,7 @@ static int svm_range_validate_and_map(struct mm_struct *mm, svm_range_lock(prange); /* Free backing memory of hmm_range if it was initialized - * Overrride return value to TRY AGAIN only if prior returns + * Override return value to TRY AGAIN only if prior returns * were successful */ if (range && !amdgpu_hmm_range_valid(range) && !r) { @@ -1771,7 +1772,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm, r = -EAGAIN; } /* Free the hmm range */ - amdgpu_hmm_range_free(range); + if (range) + amdgpu_hmm_range_free(range); if (!r && !list_empty(&prange->child_list)) { -- cgit v1.2.3 From d64fd9da7ee95a344dc7e076e11e3e4def715510 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:42 +0200 Subject: drm/amd/display: Add analog bit to edid_caps (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The new analog bit will be used with DVI-I connectors. DVI-I connectors can connect to both digital and analog monitors and this bit will help distinguish between those. v2: Sanitize analog bit based on connector type. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c | 1 + drivers/gpu/drm/amd/display/dc/dc_types.h | 5 +++++ drivers/gpu/drm/amd/display/dc/link/link_detection.c | 2 ++ drivers/gpu/drm/amd/display/include/grph_object_id.h | 7 +++++++ 4 files changed, 15 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index cc21337a182f..eb2c587b0b9b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -131,6 +131,7 @@ enum dc_edid_status dm_helpers_parse_edid_caps( edid_caps->serial_number = edid_buf->serial; edid_caps->manufacture_week = edid_buf->mfg_week; edid_caps->manufacture_year = edid_buf->mfg_year; + edid_caps->analog = !(edid_buf->input & DRM_EDID_INPUT_DIGITAL); drm_edid_get_monitor_name(edid_buf, edid_caps->display_name, diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index 8f8ccde7ad94..9989c8af5174 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -185,6 +185,10 @@ struct dc_panel_patch { unsigned int wait_after_dpcd_poweroff_ms; }; +/** + * struct dc_edid_caps - Capabilities read from EDID. + * @analog: Whether the monitor is analog. Used by DVI-I handling. + */ struct dc_edid_caps { /* sink identification */ uint16_t manufacturer_id; @@ -213,6 +217,7 @@ struct dc_edid_caps { bool hdr_supported; bool rr_capable; bool scdc_present; + bool analog; struct dc_panel_patch panel_patch; }; diff --git a/drivers/gpu/drm/amd/display/dc/link/link_detection.c b/drivers/gpu/drm/amd/display/dc/link/link_detection.c index 82a9e52d5ae5..801c94efa7fc 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -1107,6 +1107,8 @@ static bool detect_link_and_local_sink(struct dc_link *link, break; } + sink->edid_caps.analog &= dc_connector_supports_analog(link->link_id.id); + // Check if edid is the same if ((prev_sink) && (edid_status == EDID_THE_SAME || edid_status == EDID_OK)) diff --git a/drivers/gpu/drm/amd/display/include/grph_object_id.h b/drivers/gpu/drm/amd/display/include/grph_object_id.h index 54e33062b3c0..1386fa124e85 100644 --- a/drivers/gpu/drm/amd/display/include/grph_object_id.h +++ b/drivers/gpu/drm/amd/display/include/grph_object_id.h @@ -310,4 +310,11 @@ static inline bool dal_graphics_object_id_equal( } return false; } + +static inline bool dc_connector_supports_analog(const enum connector_id conn) +{ + return conn == CONNECTOR_ID_VGA || + conn == CONNECTOR_ID_SINGLE_LINK_DVII || + conn == CONNECTOR_ID_DUAL_LINK_DVII; +} #endif -- cgit v1.2.3 From 2394736e70665117ad78f208e794ea7df03c73f4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:43 +0200 Subject: drm/amd/display: Introduce MAX_LINK_ENCODERS (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We are going to support analog encoders as well, not just digital, so we need to make space for them in various arrays. v2: Fix typo. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c | 4 ++-- drivers/gpu/drm/amd/display/dc/inc/core_types.h | 8 ++++---- drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h | 24 ++++++++++++++++++++++ 3 files changed, 30 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c index a180f68f711c..deb23d20bca6 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c @@ -522,10 +522,10 @@ struct link_encoder *link_enc_cfg_get_link_enc_used_by_link( struct link_encoder *link_enc_cfg_get_next_avail_link_enc(struct dc *dc) { struct link_encoder *link_enc = NULL; - enum engine_id encs_assigned[MAX_DIG_LINK_ENCODERS]; + enum engine_id encs_assigned[MAX_LINK_ENCODERS]; int i; - for (i = 0; i < MAX_DIG_LINK_ENCODERS; i++) + for (i = 0; i < MAX_LINK_ENCODERS; i++) encs_assigned[i] = ENGINE_ID_UNKNOWN; /* Add assigned encoders to list. */ diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index 3fe590436e74..5ed2cd344804 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -274,7 +274,7 @@ struct resource_pool { /* An array for accessing the link encoder objects that have been created. * Index in array corresponds to engine ID - viz. 0: ENGINE_ID_DIGA */ - struct link_encoder *link_encoders[MAX_DIG_LINK_ENCODERS]; + struct link_encoder *link_encoders[MAX_LINK_ENCODERS]; /* Number of DIG link encoder objects created - i.e. number of valid * entries in link_encoders array. */ @@ -514,7 +514,7 @@ struct pipe_ctx { struct link_enc_cfg_context { enum link_enc_cfg_mode mode; struct link_enc_assignment link_enc_assignments[MAX_PIPES]; - enum engine_id link_enc_avail[MAX_DIG_LINK_ENCODERS]; + enum engine_id link_enc_avail[MAX_LINK_ENCODERS]; struct link_enc_assignment transient_assignments[MAX_PIPES]; }; @@ -526,8 +526,8 @@ struct resource_context { uint8_t dp_clock_source_ref_count; bool is_dsc_acquired[MAX_PIPES]; struct link_enc_cfg_context link_enc_cfg_ctx; - unsigned int dio_link_enc_to_link_idx[MAX_DIG_LINK_ENCODERS]; - int dio_link_enc_ref_cnts[MAX_DIG_LINK_ENCODERS]; + unsigned int dio_link_enc_to_link_idx[MAX_LINK_ENCODERS]; + int dio_link_enc_ref_cnts[MAX_LINK_ENCODERS]; bool is_hpo_dp_stream_enc_acquired[MAX_HPO_DP2_ENCODERS]; unsigned int hpo_dp_link_enc_to_link_idx[MAX_HPO_DP2_LINK_ENCODERS]; int hpo_dp_link_enc_ref_cnts[MAX_HPO_DP2_LINK_ENCODERS]; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h index 62a39204fe0b..5e2813e9ae2f 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h @@ -51,6 +51,30 @@ #define MAX_LINKS (MAX_DPIA + MAX_CONNECTOR + MAX_VIRTUAL_LINKS) +/** + * define MAX_DIG_LINK_ENCODERS - maximum number of digital encoders + * + * Digital encoders are ENGINE_ID_DIGA...G, there are at most 7, + * although not every GPU may have that many. + */ +#define MAX_DIG_LINK_ENCODERS 7 + +/** + * define MAX_DAC_LINK_ENCODERS - maximum number of analog link encoders + * + * Analog encoders are ENGINE_ID_DACA/B, there are at most 2, + * although not every GPU may have that many. Modern GPUs typically + * don't have analog encoders. + */ +#define MAX_DAC_LINK_ENCODERS 2 + +/** + * define MAX_LINK_ENCODERS - maximum number link encoders in total + * + * This includes both analog and digital encoders. + */ +#define MAX_LINK_ENCODERS (MAX_DIG_LINK_ENCODERS + MAX_DAC_LINK_ENCODERS) + #define MAX_DIG_LINK_ENCODERS 7 #define MAX_DWB_PIPES 1 #define MAX_HPO_DP2_ENCODERS 4 -- cgit v1.2.3 From 69b29b894660429280dfff0daed7d183c2fae7a7 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:44 +0200 Subject: drm/amd/display: Hook up DAC to bios_parser_encoder_control MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Enable the codebase to use encoder_control() when the encoder engine is one of the DACs. The BIOS parser already supports calling the DAC1EncoderControl function from the VBIOS, but it was not exposed anywhere. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 154fd2c18e88..318be0bb5549 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -742,6 +742,22 @@ static enum bp_result bios_parser_encoder_control( { struct bios_parser *bp = BP_FROM_DCB(dcb); + if (cntl->engine_id == ENGINE_ID_DACA) { + if (!bp->cmd_tbl.dac1_encoder_control) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.dac1_encoder_control( + bp, cntl->action == ENCODER_CONTROL_ENABLE, + cntl->pixel_clock, ATOM_DAC1_PS2); + } else if (cntl->engine_id == ENGINE_ID_DACB) { + if (!bp->cmd_tbl.dac2_encoder_control) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.dac2_encoder_control( + bp, cntl->action == ENCODER_CONTROL_ENABLE, + cntl->pixel_clock, ATOM_DAC1_PS2); + } + if (!bp->cmd_tbl.dig_encoder_control) return BP_RESULT_FAILURE; -- cgit v1.2.3 From 7fb4f254c8ebcd89e32c13c2a5758b0664f9fcc4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:45 +0200 Subject: drm/amd/display: Add SelectCRTC_Source to BIOS parser MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The SelectCRTC_Source command will be used to change which CRTC should be connected to which encoder. For reference, see the legacy display code: amdgpu_atombios_encoder_set_crtc_source Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 14 ++ .../gpu/drm/amd/display/dc/bios/command_table.c | 194 +++++++++++++++++++++ .../gpu/drm/amd/display/dc/bios/command_table.h | 3 + drivers/gpu/drm/amd/display/dc/dc_bios_types.h | 3 + .../drm/amd/display/include/bios_parser_types.h | 6 +- 5 files changed, 215 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 318be0bb5549..c7875edea321 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -736,6 +736,18 @@ static enum bp_result bios_parser_transmitter_control( return bp->cmd_tbl.transmitter_control(bp, cntl); } +static enum bp_result bios_parser_select_crtc_source( + struct dc_bios *dcb, + struct bp_crtc_source_select *bp_params) +{ + struct bios_parser *bp = BP_FROM_DCB(dcb); + + if (!bp->cmd_tbl.select_crtc_source) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.select_crtc_source(bp, bp_params); +} + static enum bp_result bios_parser_encoder_control( struct dc_bios *dcb, struct bp_encoder_control *cntl) @@ -2845,6 +2857,8 @@ static const struct dc_vbios_funcs vbios_funcs = { .is_device_id_supported = bios_parser_is_device_id_supported, /* COMMANDS */ + .select_crtc_source = bios_parser_select_crtc_source, + .encoder_control = bios_parser_encoder_control, .transmitter_control = bios_parser_transmitter_control, diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.c b/drivers/gpu/drm/amd/display/dc/bios/command_table.c index 58e88778da7f..dbd84477ceb7 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.c @@ -52,6 +52,7 @@ static void init_transmitter_control(struct bios_parser *bp); static void init_set_pixel_clock(struct bios_parser *bp); static void init_enable_spread_spectrum_on_ppll(struct bios_parser *bp); static void init_adjust_display_pll(struct bios_parser *bp); +static void init_select_crtc_source(struct bios_parser *bp); static void init_dac_encoder_control(struct bios_parser *bp); static void init_dac_output_control(struct bios_parser *bp); static void init_set_crtc_timing(struct bios_parser *bp); @@ -69,6 +70,7 @@ void dal_bios_parser_init_cmd_tbl(struct bios_parser *bp) init_set_pixel_clock(bp); init_enable_spread_spectrum_on_ppll(bp); init_adjust_display_pll(bp); + init_select_crtc_source(bp); init_dac_encoder_control(bp); init_dac_output_control(bp); init_set_crtc_timing(bp); @@ -1609,6 +1611,198 @@ static enum bp_result adjust_display_pll_v3( return result; } +/******************************************************************************* + ******************************************************************************** + ** + ** SELECT CRTC SOURCE + ** + ******************************************************************************** + *******************************************************************************/ + +static enum bp_result select_crtc_source_v1( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); +static enum bp_result select_crtc_source_v2( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); +static enum bp_result select_crtc_source_v3( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); + +static void init_select_crtc_source(struct bios_parser *bp) +{ + switch (BIOS_CMD_TABLE_PARA_REVISION(SelectCRTC_Source)) { + case 1: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v1; + break; + case 2: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v2; + break; + case 3: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v3; + break; + default: + bp->cmd_tbl.select_crtc_source = NULL; + break; + } +} + +static enum bp_result select_crtc_source_v1( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PS_ALLOCATION params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + switch (bp_params->engine_id) { + case ENGINE_ID_DACA: + params.ucDevice = ATOM_DEVICE_CRT1_INDEX; + break; + case ENGINE_ID_DACB: + params.ucDevice = ATOM_DEVICE_CRT2_INDEX; + break; + default: + return BP_RESULT_BADINPUT; + } + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + +static bool select_crtc_source_v2_encoder_id( + enum engine_id engine_id, uint8_t *out_encoder_id) +{ + uint8_t encoder_id = 0; + + switch (engine_id) { + case ENGINE_ID_DIGA: + encoder_id = ASIC_INT_DIG1_ENCODER_ID; + break; + case ENGINE_ID_DIGB: + encoder_id = ASIC_INT_DIG2_ENCODER_ID; + break; + case ENGINE_ID_DIGC: + encoder_id = ASIC_INT_DIG3_ENCODER_ID; + break; + case ENGINE_ID_DIGD: + encoder_id = ASIC_INT_DIG4_ENCODER_ID; + break; + case ENGINE_ID_DIGE: + encoder_id = ASIC_INT_DIG5_ENCODER_ID; + break; + case ENGINE_ID_DIGF: + encoder_id = ASIC_INT_DIG6_ENCODER_ID; + break; + case ENGINE_ID_DIGG: + encoder_id = ASIC_INT_DIG7_ENCODER_ID; + break; + case ENGINE_ID_DACA: + encoder_id = ASIC_INT_DAC1_ENCODER_ID; + break; + case ENGINE_ID_DACB: + encoder_id = ASIC_INT_DAC2_ENCODER_ID; + break; + default: + return false; + } + + *out_encoder_id = encoder_id; + return true; +} + +static bool select_crtc_source_v2_encoder_mode( + enum signal_type signal_type, uint8_t *out_encoder_mode) +{ + uint8_t encoder_mode = 0; + + switch (signal_type) { + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + encoder_mode = ATOM_ENCODER_MODE_DVI; + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + encoder_mode = ATOM_ENCODER_MODE_HDMI; + break; + case SIGNAL_TYPE_LVDS: + encoder_mode = ATOM_ENCODER_MODE_LVDS; + break; + case SIGNAL_TYPE_RGB: + encoder_mode = ATOM_ENCODER_MODE_CRT; + break; + case SIGNAL_TYPE_DISPLAY_PORT: + encoder_mode = ATOM_ENCODER_MODE_DP; + break; + case SIGNAL_TYPE_DISPLAY_PORT_MST: + encoder_mode = ATOM_ENCODER_MODE_DP_MST; + break; + case SIGNAL_TYPE_EDP: + encoder_mode = ATOM_ENCODER_MODE_DP; + break; + default: + return false; + } + + *out_encoder_mode = encoder_mode; + return true; +} + +static enum bp_result select_crtc_source_v2( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PARAMETERS_V3 params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + if (!select_crtc_source_v2_encoder_id( + bp_params->engine_id, + ¶ms.ucEncoderID)) + return BP_RESULT_BADINPUT; + if (!select_crtc_source_v2_encoder_mode( + bp_params->sink_signal, + ¶ms.ucEncodeMode)) + return BP_RESULT_BADINPUT; + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + +static enum bp_result select_crtc_source_v3( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PARAMETERS_V3 params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + if (!select_crtc_source_v2_encoder_id( + bp_params->engine_id, + ¶ms.ucEncoderID)) + return BP_RESULT_BADINPUT; + if (!select_crtc_source_v2_encoder_mode( + bp_params->sink_signal, + ¶ms.ucEncodeMode)) + return BP_RESULT_BADINPUT; + + params.ucDstBpc = bp_params->bit_depth; + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + /******************************************************************************* ******************************************************************************** ** diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.h b/drivers/gpu/drm/amd/display/dc/bios/command_table.h index ad533775e724..8b04b903e93d 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.h +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.h @@ -52,6 +52,9 @@ struct cmd_tbl { enum bp_result (*adjust_display_pll)( struct bios_parser *bp, struct bp_adjust_pixel_clock_parameters *bp_params); + enum bp_result (*select_crtc_source)( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); enum bp_result (*dac1_encoder_control)( struct bios_parser *bp, bool enable, diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h index 5fa5e2b63fb7..545ce1e15eae 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h @@ -91,6 +91,9 @@ struct dc_vbios_funcs { struct device_id id); /* COMMANDS */ + enum bp_result (*select_crtc_source)( + struct dc_bios *bios, + struct bp_crtc_source_select *bp_params); enum bp_result (*encoder_control)( struct dc_bios *bios, struct bp_encoder_control *cntl); diff --git a/drivers/gpu/drm/amd/display/include/bios_parser_types.h b/drivers/gpu/drm/amd/display/include/bios_parser_types.h index 812377d9e48f..d9e58a6a0d36 100644 --- a/drivers/gpu/drm/amd/display/include/bios_parser_types.h +++ b/drivers/gpu/drm/amd/display/include/bios_parser_types.h @@ -135,12 +135,8 @@ struct bp_external_encoder_control { struct bp_crtc_source_select { enum engine_id engine_id; enum controller_id controller_id; - /* from GPU Tx aka asic_signal */ - enum signal_type signal; - /* sink_signal may differ from asicSignal if Translator encoder */ enum signal_type sink_signal; - enum display_output_bit_depth display_output_bit_depth; - bool enable_dp_audio; + uint8_t bit_depth; }; struct bp_transmitter_control { -- cgit v1.2.3 From 727b179a8032c256b02b816661b2d96713f11e41 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:46 +0200 Subject: drm/amd/display: Get maximum pixel clock from VBIOS MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We will use this for validating the pixel clock when an analog monitor is connected to VGA or DVI-I connectors. For reference, see the legacy display code: amdgpu_connector_vga_mode_valid Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 2 ++ drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h | 1 + 2 files changed, 3 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index c7875edea321..33d0ec38ded7 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -441,6 +441,7 @@ static enum bp_result get_firmware_info_v1_4( le32_to_cpu(firmware_info->ulMinPixelClockPLL_Output) * 10; info->pll_info.max_output_pxl_clk_pll_frequency = le32_to_cpu(firmware_info->ulMaxPixelClockPLL_Output) * 10; + info->max_pixel_clock = le16_to_cpu(firmware_info->usMaxPixelClock) * 10; if (firmware_info->usFirmwareCapability.sbfAccess.MemoryClockSS_Support) /* Since there is no information on the SS, report conservative @@ -497,6 +498,7 @@ static enum bp_result get_firmware_info_v2_1( info->external_clock_source_frequency_for_dp = le16_to_cpu(firmwareInfo->usUniphyDPModeExtClkFreq) * 10; info->min_allowed_bl_level = firmwareInfo->ucMinAllowedBL_Level; + info->max_pixel_clock = le16_to_cpu(firmwareInfo->usMaxPixelClock) * 10; /* There should be only one entry in the SS info table for Memory Clock */ diff --git a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h index cc467031651d..38a77fa9b4af 100644 --- a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h +++ b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h @@ -169,6 +169,7 @@ struct dc_firmware_info { uint32_t engine_clk_ss_percentage; } feature; + uint32_t max_pixel_clock; /* in KHz */ uint32_t default_display_engine_pll_frequency; /* in KHz */ uint32_t external_clock_source_frequency_for_dp; /* in KHz */ uint32_t smu_gpu_pll_output_freq; /* in KHz */ -- cgit v1.2.3 From 8c8df54a2b4b6b3a229fa40ab7a1b0ce3edb5896 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:47 +0200 Subject: drm/amd/display: Don't use stereo sync and audio on RGB signals (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Analog video signals on VGA or DVI-A (analog part of DVI-I) don't support audio, so avoid calling any audio related functions on analog signals. Stereo sync was not set up for analog signals in the legacy display code either, so there is no loss of functionality if we omit it from DC for now. Also add a dc_is_rgb_signal similar to other dc_is_*_signal. v2: Added comment to clarify what we mean by RGB in this context. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c | 3 ++- drivers/gpu/drm/amd/display/dc/link/link_dpms.c | 6 ++++-- drivers/gpu/drm/amd/display/include/signal_types.h | 12 ++++++++++++ 3 files changed, 18 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c index 892907991f91..80344cbd1f99 100644 --- a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c +++ b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c @@ -115,7 +115,8 @@ void setup_dio_stream_attribute(struct pipe_ctx *pipe_ctx) struct dc_stream_state *stream = pipe_ctx->stream; struct dc_link *link = stream->link; - if (!dc_is_virtual_signal(stream->signal)) + if (!dc_is_virtual_signal(stream->signal) && + !dc_is_rgb_signal(stream->signal)) stream_encoder->funcs->setup_stereo_sync( stream_encoder, pipe_ctx->stream_res.tg->inst, diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c index dba8ec0988a1..fdefe27f4ba3 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c @@ -2371,7 +2371,8 @@ void link_set_dpms_off(struct pipe_ctx *pipe_ctx) set_avmute(pipe_ctx, true); } - dc->hwss.disable_audio_stream(pipe_ctx); + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + dc->hwss.disable_audio_stream(pipe_ctx); update_psp_stream_config(pipe_ctx, true); dc->hwss.blank_stream(pipe_ctx); @@ -2656,7 +2657,8 @@ void link_set_dpms_on( enable_stream_features(pipe_ctx); update_psp_stream_config(pipe_ctx, false); - dc->hwss.enable_audio_stream(pipe_ctx); + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + dc->hwss.enable_audio_stream(pipe_ctx); if (dc_is_hdmi_signal(pipe_ctx->stream->signal)) { set_avmute(pipe_ctx, false); diff --git a/drivers/gpu/drm/amd/display/include/signal_types.h b/drivers/gpu/drm/amd/display/include/signal_types.h index a10d6b988aab..3a2c2d2fb629 100644 --- a/drivers/gpu/drm/amd/display/include/signal_types.h +++ b/drivers/gpu/drm/amd/display/include/signal_types.h @@ -118,6 +118,18 @@ static inline bool dc_is_dvi_signal(enum signal_type signal) } } +/** + * dc_is_rgb_signal() - Whether the signal is analog RGB. + * + * Returns whether the given signal type is an analog RGB signal + * that is used with a DAC on VGA or DVI-I connectors. + * Not to be confused with other uses of "RGB", such as RGB color space. + */ +static inline bool dc_is_rgb_signal(enum signal_type signal) +{ + return (signal == SIGNAL_TYPE_RGB); +} + static inline bool dc_is_tmds_signal(enum signal_type signal) { switch (signal) { -- cgit v1.2.3 From e3ea56e995a425a44870dfa357b8608101a67034 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:48 +0200 Subject: drm/amd/display: Don't try to enable/disable HPD when unavailable MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit VGA connectors don't have HPD (hotplug detection), so don't touch any HPD related registers for VGA. Determine whether hotplug detection is available by checking that the interrupt source is invalid. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c index f263e1a4537e..cb4bb67289a4 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c @@ -759,6 +759,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us int max_param_num = 11; enum dp_test_pattern test_pattern = DP_TEST_PATTERN_UNSUPPORTED; bool disable_hpd = false; + bool supports_hpd = link->irq_source_hpd != DC_IRQ_SOURCE_INVALID; bool valid_test_pattern = false; uint8_t param_nums = 0; /* init with default 80bit custom pattern */ @@ -850,7 +851,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us * because it might have been disabled after a test pattern was set. * AUX depends on HPD * sequence dependent, do not move! */ - if (!disable_hpd) + if (supports_hpd && !disable_hpd) dc_link_enable_hpd(link); prefer_link_settings.lane_count = link->verified_link_cap.lane_count; @@ -888,7 +889,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us * Need disable interrupt to avoid SW driver disable DP output. This is * done after the test pattern is set. */ - if (valid_test_pattern && disable_hpd) + if (valid_test_pattern && supports_hpd && disable_hpd) dc_link_disable_hpd(link); kfree(wr_buf); -- cgit v1.2.3 From c2192872d99b8bc7d1ac2494e850fb6fb6c915eb Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:49 +0200 Subject: drm/amd/display: Determine early if a link has supported encoders (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Avoid initializing DDC, HPD, etc. when we know that the link is not going to be constructed because it has no supported encoders. This is mainly useful for old GPUs which may have encoders such as TRAVIS and NUTMEG that are not yet supported by DC. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/link/link_factory.c | 23 ++++++++++++++++++---- 1 file changed, 19 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/link_factory.c b/drivers/gpu/drm/amd/display/dc/link/link_factory.c index f06af98d46ee..8899797f64a9 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_factory.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_factory.c @@ -451,6 +451,14 @@ static enum channel_id get_ddc_line(struct dc_link *link) return channel; } +static bool transmitter_supported(const enum transmitter transmitter) +{ + return transmitter != TRANSMITTER_UNKNOWN && + transmitter != TRANSMITTER_NUTMEG_CRT && + transmitter != TRANSMITTER_TRAVIS_CRT && + transmitter != TRANSMITTER_TRAVIS_LCD; +} + static bool construct_phy(struct dc_link *link, const struct link_init_data *init_params) { @@ -482,6 +490,17 @@ static bool construct_phy(struct dc_link *link, link->link_id = bios->funcs->get_connector_id(bios, init_params->connector_index); + /* Determine early if the link has any supported encoders, + * so that we avoid initializing DDC and HPD, etc. + */ + bp_funcs->get_src_obj(bios, link->link_id, 0, &enc_init_data.encoder); + enc_init_data.transmitter = translate_encoder_to_transmitter(enc_init_data.encoder); + + if (!transmitter_supported(enc_init_data.transmitter)) { + DC_LOG_WARNING("link_id %d has unsupported encoder\n", link->link_id.id); + return false; + } + link->ep_type = DISPLAY_ENDPOINT_PHY; DC_LOG_DC("BIOS object table - link_id: %d", link->link_id.id); @@ -611,16 +630,12 @@ static bool construct_phy(struct dc_link *link, dal_ddc_get_line(get_ddc_pin(link->ddc)); enc_init_data.ctx = dc_ctx; - bp_funcs->get_src_obj(dc_ctx->dc_bios, link->link_id, 0, - &enc_init_data.encoder); enc_init_data.connector = link->link_id; enc_init_data.channel = get_ddc_line(link); enc_init_data.hpd_source = get_hpd_line(link); link->hpd_src = enc_init_data.hpd_source; - enc_init_data.transmitter = - translate_encoder_to_transmitter(enc_init_data.encoder); link->link_enc = link->dc->res_pool->funcs->link_enc_create(dc_ctx, &enc_init_data); -- cgit v1.2.3 From 5834c33fd3f6f2a26dd4d6d4bbc7b8ed1c2ac4aa Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:50 +0200 Subject: drm/amd/display: Add concept of analog encoders (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add a num_analog_stream_encoders field to indicate how many analog stream encoders are present. When analog stream encoders are present, create them. Additionally, add an analog_engine field to link encoders and search for supported analog encoders in the BIOS for each link. When connecting an RGB signal, search for analog stream encoders. The actual DCE analog link and stream encoder is going to be added in a subsequent commit. v2: Add check to see if an analog engine is really supported. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 8 +++++ .../gpu/drm/amd/display/dc/inc/hw/link_encoder.h | 2 ++ drivers/gpu/drm/amd/display/dc/inc/resource.h | 1 + drivers/gpu/drm/amd/display/dc/link/link_factory.c | 36 +++++++++++++++++++++- .../display/dc/resource/dce100/dce100_resource.c | 7 +++-- 5 files changed, 51 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 89fe72820f6a..dc0c4065a92c 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -446,6 +446,14 @@ bool resource_construct( DC_ERR("DC: failed to create stream_encoder!\n"); pool->stream_enc_count++; } + + for (i = 0; i < caps->num_analog_stream_encoder; i++) { + pool->stream_enc[caps->num_stream_encoder + i] = + create_funcs->create_stream_encoder(ENGINE_ID_DACA + i, ctx); + if (pool->stream_enc[caps->num_stream_encoder + i] == NULL) + DC_ERR("DC: failed to create analog stream_encoder %d!\n", i); + pool->stream_enc_count++; + } } pool->hpo_dp_stream_enc_count = 0; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h index 08c16ba52a51..df512920a9fa 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h @@ -47,6 +47,7 @@ struct encoder_init_data { enum hpd_source_id hpd_source; /* TODO: in DAL2, here was pointer to EventManagerInterface */ struct graphics_object_id encoder; + enum engine_id analog_engine; struct dc_context *ctx; enum transmitter transmitter; }; @@ -83,6 +84,7 @@ struct link_encoder { struct graphics_object_id connector; uint32_t output_signals; enum engine_id preferred_engine; + enum engine_id analog_engine; struct encoder_feature_support features; enum transmitter transmitter; enum hpd_source_id hpd_source; diff --git a/drivers/gpu/drm/amd/display/dc/inc/resource.h b/drivers/gpu/drm/amd/display/dc/inc/resource.h index 4e26a16a8743..79746d931471 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/resource.h +++ b/drivers/gpu/drm/amd/display/dc/inc/resource.h @@ -49,6 +49,7 @@ struct resource_caps { int num_video_plane; int num_audio; int num_stream_encoder; + int num_analog_stream_encoder; int num_pll; int num_dwb; int num_ddc; diff --git a/drivers/gpu/drm/amd/display/dc/link/link_factory.c b/drivers/gpu/drm/amd/display/dc/link/link_factory.c index 8899797f64a9..ebc3798e7d25 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_factory.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_factory.c @@ -451,6 +451,32 @@ static enum channel_id get_ddc_line(struct dc_link *link) return channel; } +static enum engine_id find_analog_engine(struct dc_link *link) +{ + struct dc_bios *bp = link->ctx->dc_bios; + struct graphics_object_id encoder = {0}; + enum bp_result bp_result = BP_RESULT_OK; + int i; + + for (i = 0; i < 3; i++) { + bp_result = bp->funcs->get_src_obj(bp, link->link_id, i, &encoder); + + if (bp_result != BP_RESULT_OK) + return ENGINE_ID_UNKNOWN; + + switch (encoder.id) { + case ENCODER_ID_INTERNAL_DAC1: + case ENCODER_ID_INTERNAL_KLDSCP_DAC1: + return ENGINE_ID_DACA; + case ENCODER_ID_INTERNAL_DAC2: + case ENCODER_ID_INTERNAL_KLDSCP_DAC2: + return ENGINE_ID_DACB; + } + } + + return ENGINE_ID_UNKNOWN; +} + static bool transmitter_supported(const enum transmitter transmitter) { return transmitter != TRANSMITTER_UNKNOWN && @@ -459,6 +485,12 @@ static bool transmitter_supported(const enum transmitter transmitter) transmitter != TRANSMITTER_TRAVIS_LCD; } +static bool analog_engine_supported(const enum engine_id engine_id) +{ + return engine_id == ENGINE_ID_DACA || + engine_id == ENGINE_ID_DACB; +} + static bool construct_phy(struct dc_link *link, const struct link_init_data *init_params) { @@ -495,8 +527,10 @@ static bool construct_phy(struct dc_link *link, */ bp_funcs->get_src_obj(bios, link->link_id, 0, &enc_init_data.encoder); enc_init_data.transmitter = translate_encoder_to_transmitter(enc_init_data.encoder); + enc_init_data.analog_engine = find_analog_engine(link); - if (!transmitter_supported(enc_init_data.transmitter)) { + if (!transmitter_supported(enc_init_data.transmitter) && + !analog_engine_supported(enc_init_data.analog_engine)) { DC_LOG_WARNING("link_id %d has unsupported encoder\n", link->link_id.id); return false; } diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index 550466de1a66..7573f5130ea9 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -954,6 +954,10 @@ struct stream_encoder *dce100_find_first_free_match_stream_enc_for_link( int i; int j = -1; struct dc_link *link = stream->link; + enum engine_id preferred_engine = link->link_enc->preferred_engine; + + if (dc_is_rgb_signal(stream->signal)) + preferred_engine = link->link_enc->analog_engine; for (i = 0; i < pool->stream_enc_count; i++) { if (!res_ctx->is_stream_enc_acquired[i] && @@ -962,8 +966,7 @@ struct stream_encoder *dce100_find_first_free_match_stream_enc_for_link( * in daisy chain use case */ j = i; - if (pool->stream_enc[i]->id == - link->link_enc->preferred_engine) + if (pool->stream_enc[i]->id == preferred_engine) return pool->stream_enc[i]; } } -- cgit v1.2.3 From deb072d241c906ae378bc88ff91dfcdc34ebe71a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:51 +0200 Subject: drm/amd/display: Implement DCE analog stream encoders MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add analog stream encoders for DCE which will be used when connecting an analog display through VGA or DVI-I. Considering that all stream encoder functions currently deal with digital streams, there is nothing for an analog stream encoder to do, making them basically a no-op. That being said, we still need some kind of stream encoder to represent an analog stream, and it is beneficial to split them from digital stream encoders in the code to make sure they don't accidentally write any DIG* registers. On supported chips there is currently up to 1 analog encoder, which is DACA. There are references to DACB in some code such as VBIOS commands and register files but it seems to be not present on DCE 6 and newer. Set num_analog_stream_encoder = 1 so that we can support the analog connectors on DCE 6-10, for now. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c | 14 ++++++++++++++ drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h | 5 +++++ .../drm/amd/display/dc/resource/dce100/dce100_resource.c | 6 ++++++ .../gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c | 8 ++++++++ .../gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c | 8 ++++++++ 5 files changed, 41 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c index 1130d7619b26..f8996ee2856b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c @@ -1567,3 +1567,17 @@ void dce110_stream_encoder_construct( enc110->se_shift = se_shift; enc110->se_mask = se_mask; } + +static const struct stream_encoder_funcs dce110_an_str_enc_funcs = {0}; + +void dce110_analog_stream_encoder_construct( + struct dce110_stream_encoder *enc110, + struct dc_context *ctx, + struct dc_bios *bp, + enum engine_id eng_id) +{ + enc110->base.funcs = &dce110_an_str_enc_funcs; + enc110->base.ctx = ctx; + enc110->base.id = eng_id; + enc110->base.bp = bp; +} diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h index cc5020a8e1e1..068de1392121 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h @@ -708,6 +708,11 @@ void dce110_stream_encoder_construct( const struct dce_stream_encoder_shift *se_shift, const struct dce_stream_encoder_mask *se_mask); +void dce110_analog_stream_encoder_construct( + struct dce110_stream_encoder *enc110, + struct dc_context *ctx, + struct dc_bios *bp, + enum engine_id eng_id); void dce110_se_audio_mute_control( struct stream_encoder *enc, bool mute); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index 7573f5130ea9..3856477981c6 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -375,6 +375,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -486,6 +487,11 @@ static struct stream_encoder *dce100_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); return &enc110->base; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c index b75be6ad64f6..7195f435e3f7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c @@ -373,6 +373,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -382,6 +383,7 @@ static const struct resource_caps res_cap_61 = { .num_timing_generator = 4, .num_audio = 6, .num_stream_encoder = 6, + .num_analog_stream_encoder = 1, .num_pll = 3, .num_ddc = 6, }; @@ -389,6 +391,7 @@ static const struct resource_caps res_cap_61 = { static const struct resource_caps res_cap_64 = { .num_timing_generator = 2, .num_audio = 2, + .num_analog_stream_encoder = 1, .num_stream_encoder = 2, .num_pll = 3, .num_ddc = 2, @@ -599,6 +602,11 @@ static struct stream_encoder *dce60_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index c4a56ec60b82..e9b79db37d1b 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -375,6 +375,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -383,6 +384,7 @@ static const struct resource_caps res_cap = { static const struct resource_caps res_cap_81 = { .num_timing_generator = 4, .num_audio = 7, + .num_analog_stream_encoder = 1, .num_stream_encoder = 7, .num_pll = 3, .num_ddc = 6, @@ -391,6 +393,7 @@ static const struct resource_caps res_cap_81 = { static const struct resource_caps res_cap_83 = { .num_timing_generator = 2, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 2, .num_ddc = 2, @@ -607,6 +610,11 @@ static struct stream_encoder *dce80_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); -- cgit v1.2.3 From 0fbe321a93ce995da90483f716f9372e8eaa6797 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:52 +0200 Subject: drm/amd/display: Implement DCE analog link encoders (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit We support two kinds of analog connections: 1. DVI-I, which allows both digital and analog signals: The DC code base only allows 1 encoder per connector, and the preferred engine type is still going to be digital. So, for DVI-I to work, we need to make sure the pre-existing link encoder can also work with analog signals. 1. VGA, which only supports analog signals: For VGA, we need to create a link encoder that only works with the DAC without perturbing any digital transmitter functionality. Since dce110_link_encoder already supports analog DVI-I, just reuse that code for VGA as well. v2: Reduce code churn by reusing same link encoder for VGA and DVI-I. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/dc/dce/dce_link_encoder.c | 85 ++++++++++++++++++++++ .../gpu/drm/amd/display/dc/dce/dce_link_encoder.h | 16 ++-- .../display/dc/resource/dce100/dce100_resource.c | 16 +++- .../amd/display/dc/resource/dce60/dce60_resource.c | 19 ++++- .../amd/display/dc/resource/dce80/dce80_resource.c | 16 +++- 5 files changed, 141 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c index 0c50fe266c8a..87dbb8d7ed27 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c @@ -302,6 +302,10 @@ static void setup_panel_mode( if (ctx->dc->caps.psp_setup_panel_mode) return; + /* The code below is only applicable to encoders with a digital transmitter. */ + if (enc110->base.transmitter == TRANSMITTER_UNKNOWN) + return; + ASSERT(REG(DP_DPHY_INTERNAL_CTRL)); value = REG_READ(DP_DPHY_INTERNAL_CTRL); @@ -804,6 +808,33 @@ bool dce110_link_encoder_validate_dp_output( return true; } +static bool dce110_link_encoder_validate_rgb_output( + const struct dce110_link_encoder *enc110, + const struct dc_crtc_timing *crtc_timing) +{ + /* When the VBIOS doesn't specify any limits, use 400 MHz. + * The value comes from amdgpu_atombios_get_clock_info. + */ + uint32_t max_pixel_clock_khz = 400000; + + if (enc110->base.ctx->dc_bios->fw_info_valid && + enc110->base.ctx->dc_bios->fw_info.max_pixel_clock) { + max_pixel_clock_khz = + enc110->base.ctx->dc_bios->fw_info.max_pixel_clock; + } + + if (crtc_timing->pix_clk_100hz > max_pixel_clock_khz * 10) + return false; + + if (crtc_timing->display_color_depth != COLOR_DEPTH_888) + return false; + + if (crtc_timing->pixel_encoding != PIXEL_ENCODING_RGB) + return false; + + return true; +} + void dce110_link_encoder_construct( struct dce110_link_encoder *enc110, const struct encoder_init_data *init_data, @@ -824,6 +855,7 @@ void dce110_link_encoder_construct( enc110->base.connector = init_data->connector; enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; + enc110->base.analog_engine = init_data->analog_engine; enc110->base.features = *enc_features; @@ -847,6 +879,11 @@ void dce110_link_encoder_construct( SIGNAL_TYPE_EDP | SIGNAL_TYPE_HDMI_TYPE_A; + if ((enc110->base.connector.id == CONNECTOR_ID_DUAL_LINK_DVII || + enc110->base.connector.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + enc110->base.analog_engine != ENGINE_ID_UNKNOWN) + enc110->base.output_signals |= SIGNAL_TYPE_RGB; + /* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE. * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY. * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer @@ -885,6 +922,13 @@ void dce110_link_encoder_construct( enc110->base.preferred_engine = ENGINE_ID_DIGG; break; default: + if (init_data->analog_engine != ENGINE_ID_UNKNOWN) { + /* The connector is analog-only, ie. VGA */ + enc110->base.preferred_engine = init_data->analog_engine; + enc110->base.output_signals = SIGNAL_TYPE_RGB; + enc110->base.transmitter = TRANSMITTER_UNKNOWN; + break; + } ASSERT_CRITICAL(false); enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; } @@ -939,6 +983,10 @@ bool dce110_link_encoder_validate_output_with_stream( is_valid = dce110_link_encoder_validate_dp_output( enc110, &stream->timing); break; + case SIGNAL_TYPE_RGB: + is_valid = dce110_link_encoder_validate_rgb_output( + enc110, &stream->timing); + break; case SIGNAL_TYPE_EDP: case SIGNAL_TYPE_LVDS: is_valid = stream->timing.pixel_encoding == PIXEL_ENCODING_RGB; @@ -969,6 +1017,10 @@ void dce110_link_encoder_hw_init( cntl.coherent = false; cntl.hpd_sel = enc110->base.hpd_source; + /* The code below is only applicable to encoders with a digital transmitter. */ + if (enc110->base.transmitter == TRANSMITTER_UNKNOWN) + return; + if (enc110->base.connector.id == CONNECTOR_ID_EDP) cntl.signal = SIGNAL_TYPE_EDP; @@ -1034,6 +1086,8 @@ void dce110_link_encoder_setup( /* DP MST */ REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 5); break; + case SIGNAL_TYPE_RGB: + break; default: ASSERT_CRITICAL(false); /* invalid mode ! */ @@ -1282,6 +1336,24 @@ void dce110_link_encoder_disable_output( struct bp_transmitter_control cntl = { 0 }; enum bp_result result; + switch (enc->analog_engine) { + case ENGINE_ID_DACA: + REG_UPDATE(DAC_ENABLE, DAC_ENABLE, 0); + break; + case ENGINE_ID_DACB: + /* DACB doesn't seem to be present on DCE6+, + * although there are references to it in the register file. + */ + DC_LOG_ERROR("%s DACB is unsupported\n", __func__); + break; + default: + break; + } + + /* The code below only applies to connectors that support digital signals. */ + if (enc->transmitter == TRANSMITTER_UNKNOWN) + return; + if (!dce110_is_dig_enabled(enc)) { /* OF_SKIP_POWER_DOWN_INACTIVE_ENCODER */ return; @@ -1726,6 +1798,7 @@ void dce60_link_encoder_construct( enc110->base.connector = init_data->connector; enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; + enc110->base.analog_engine = init_data->analog_engine; enc110->base.features = *enc_features; @@ -1749,6 +1822,11 @@ void dce60_link_encoder_construct( SIGNAL_TYPE_EDP | SIGNAL_TYPE_HDMI_TYPE_A; + if ((enc110->base.connector.id == CONNECTOR_ID_DUAL_LINK_DVII || + enc110->base.connector.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + enc110->base.analog_engine != ENGINE_ID_UNKNOWN) + enc110->base.output_signals |= SIGNAL_TYPE_RGB; + /* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE. * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY. * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer @@ -1787,6 +1865,13 @@ void dce60_link_encoder_construct( enc110->base.preferred_engine = ENGINE_ID_DIGG; break; default: + if (init_data->analog_engine != ENGINE_ID_UNKNOWN) { + /* The connector is analog-only, ie. VGA */ + enc110->base.preferred_engine = init_data->analog_engine; + enc110->base.output_signals = SIGNAL_TYPE_RGB; + enc110->base.transmitter = TRANSMITTER_UNKNOWN; + break; + } ASSERT_CRITICAL(false); enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; } diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h index 261c70e01e33..c58b69bc319b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h @@ -101,18 +101,21 @@ SRI(DP_SEC_CNTL, DP, id), \ SRI(DP_VID_STREAM_CNTL, DP, id), \ SRI(DP_DPHY_FAST_TRAINING, DP, id), \ - SRI(DP_SEC_CNTL1, DP, id) + SRI(DP_SEC_CNTL1, DP, id), \ + SR(DAC_ENABLE) #endif #define LE_DCE80_REG_LIST(id)\ SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \ - LE_COMMON_REG_LIST_BASE(id) + LE_COMMON_REG_LIST_BASE(id), \ + SR(DAC_ENABLE) #define LE_DCE100_REG_LIST(id)\ LE_COMMON_REG_LIST_BASE(id), \ SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \ SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \ - SR(DCI_MEM_PWR_STATUS) + SR(DCI_MEM_PWR_STATUS), \ + SR(DAC_ENABLE) #define LE_DCE110_REG_LIST(id)\ LE_COMMON_REG_LIST_BASE(id), \ @@ -181,6 +184,9 @@ struct dce110_link_enc_registers { uint32_t DP_DPHY_BS_SR_SWAP_CNTL; uint32_t DP_DPHY_HBR2_PATTERN_CONTROL; uint32_t DP_SEC_CNTL1; + + /* DAC registers */ + uint32_t DAC_ENABLE; }; struct dce110_link_encoder { @@ -215,10 +221,6 @@ bool dce110_link_encoder_validate_dvi_output( enum signal_type signal, const struct dc_crtc_timing *crtc_timing); -bool dce110_link_encoder_validate_rgb_output( - const struct dce110_link_encoder *enc110, - const struct dc_crtc_timing *crtc_timing); - bool dce110_link_encoder_validate_dp_output( const struct dce110_link_encoder *enc110, const struct dc_crtc_timing *crtc_timing); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index 3856477981c6..85ea219e4fd4 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -225,6 +225,7 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(4), link_regs(5), link_regs(6), + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -632,7 +633,20 @@ static struct link_encoder *dce100_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c index 7195f435e3f7..6e89b7020b61 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c @@ -240,7 +240,9 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(2), link_regs(3), link_regs(4), - link_regs(5) + link_regs(5), + {0}, + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -726,7 +728,20 @@ static struct link_encoder *dce60_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index e9b79db37d1b..a1ad7398e39f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -241,6 +241,7 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(4), link_regs(5), link_regs(6), + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -734,7 +735,20 @@ static struct link_encoder *dce80_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = -- cgit v1.2.3 From 3be26d81b150c8bd4ee24f58b5bb558d9baaf064 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:53 +0200 Subject: drm/amd/display: Support DAC in dce110_hwseq MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The dce110_hwseq is used by all DCE hardware, so add the DAC support here. When enabling/disabling a stream for a RGB signal, this will call the VBIOS to enable/disable the DAC. Additionally, when applying the controller context, call SelectCRTC_Source from VBIOS in order to direct the CRTC output to the DAC. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/hwss/dce110/dce110_hwseq.c | 75 +++++++++++++++++++++- 1 file changed, 73 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index 6b0566baa2f2..3005115c8505 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -659,6 +659,20 @@ void dce110_update_info_frame(struct pipe_ctx *pipe_ctx) } } +static void +dce110_dac_encoder_control(struct pipe_ctx *pipe_ctx, bool enable) +{ + struct dc_link *link = pipe_ctx->stream->link; + struct dc_bios *bios = link->ctx->dc_bios; + struct bp_encoder_control encoder_control = {0}; + + encoder_control.action = enable ? ENCODER_CONTROL_ENABLE : ENCODER_CONTROL_DISABLE; + encoder_control.engine_id = link->link_enc->analog_engine; + encoder_control.pixel_clock = pipe_ctx->stream->timing.pix_clk_100hz / 10; + + bios->funcs->encoder_control(bios, &encoder_control); +} + void dce110_enable_stream(struct pipe_ctx *pipe_ctx) { enum dc_lane_count lane_count = @@ -689,6 +703,9 @@ void dce110_enable_stream(struct pipe_ctx *pipe_ctx) early_control = lane_count; tg->funcs->set_early_control(tg, early_control); + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + dce110_dac_encoder_control(pipe_ctx, true); } static enum bp_result link_transmitter_control( @@ -1176,7 +1193,8 @@ void dce110_disable_stream(struct pipe_ctx *pipe_ctx) pipe_ctx->stream_res.stream_enc->funcs->stop_dp_info_packets( pipe_ctx->stream_res.stream_enc); - dc->hwss.disable_audio_stream(pipe_ctx); + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + dc->hwss.disable_audio_stream(pipe_ctx); link_hwss->reset_stream_encoder(pipe_ctx); @@ -1196,6 +1214,9 @@ void dce110_disable_stream(struct pipe_ctx *pipe_ctx) dccg->funcs->disable_symclk_se(dccg, stream_enc->stream_enc_inst, link_enc->transmitter - TRANSMITTER_UNIPHY_A); } + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + dce110_dac_encoder_control(pipe_ctx, false); } void dce110_unblank_stream(struct pipe_ctx *pipe_ctx, @@ -1581,6 +1602,51 @@ static enum dc_status dce110_enable_stream_timing( return DC_OK; } +static void +dce110_select_crtc_source(struct pipe_ctx *pipe_ctx) +{ + struct dc_link *link = pipe_ctx->stream->link; + struct dc_bios *bios = link->ctx->dc_bios; + struct bp_crtc_source_select crtc_source_select = {0}; + enum engine_id engine_id = link->link_enc->preferred_engine; + uint8_t bit_depth; + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + engine_id = link->link_enc->analog_engine; + + switch (pipe_ctx->stream->timing.display_color_depth) { + case COLOR_DEPTH_UNDEFINED: + bit_depth = 0; + break; + case COLOR_DEPTH_666: + bit_depth = 6; + break; + default: + case COLOR_DEPTH_888: + bit_depth = 8; + break; + case COLOR_DEPTH_101010: + bit_depth = 10; + break; + case COLOR_DEPTH_121212: + bit_depth = 12; + break; + case COLOR_DEPTH_141414: + bit_depth = 14; + break; + case COLOR_DEPTH_161616: + bit_depth = 16; + break; + } + + crtc_source_select.controller_id = CONTROLLER_ID_D0 + pipe_ctx->stream_res.tg->inst; + crtc_source_select.bit_depth = bit_depth; + crtc_source_select.engine_id = engine_id; + crtc_source_select.sink_signal = pipe_ctx->stream->signal; + + bios->funcs->select_crtc_source(bios, &crtc_source_select); +} + enum dc_status dce110_apply_single_controller_ctx_to_hw( struct pipe_ctx *pipe_ctx, struct dc_state *context, @@ -1600,6 +1666,10 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( hws->funcs.disable_stream_gating(dc, pipe_ctx); } + if (pipe_ctx->stream->signal == SIGNAL_TYPE_RGB) { + dce110_select_crtc_source(pipe_ctx); + } + if (pipe_ctx->stream_res.audio != NULL) { struct audio_output audio_output = {0}; @@ -1679,7 +1749,8 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( pipe_ctx->stream_res.tg->funcs->set_static_screen_control( pipe_ctx->stream_res.tg, event_triggers, 2); - if (!dc_is_virtual_signal(pipe_ctx->stream->signal)) + if (!dc_is_virtual_signal(pipe_ctx->stream->signal) && + !dc_is_rgb_signal(pipe_ctx->stream->signal)) pipe_ctx->stream_res.stream_enc->funcs->dig_connect_to_otg( pipe_ctx->stream_res.stream_enc, pipe_ctx->stream_res.tg->inst); -- cgit v1.2.3 From e45d60a76999a2128bd1bd3d2e1597eec8b49c78 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:54 +0200 Subject: drm/amd/display: Add analog link detection (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Analog displays typically have a DDC connection which can be used by the GPU to read EDID. This commit adds the capability to probe analog displays using DDC, reading the EDID header and deciding whether the analog link is connected based on the data that was read. Note that VGA has no HPD (hotplug detection), so we need to to do analog link detection for VGA before checking HPD. In case of DVI-I, while the connector supports HPD, not all analog cables connect the HPD pins, so we can't rely on HPD either. For reference, see the legacy display code: amdgpu_connector_vga_detect amdgpu_display_ddc_probe DAC load detection will be implemented in a separate commit. v2: Fix crash / black screen on newer GPUs during link detection. Ignore HPD pin for analog connectors. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/link/hwss/link_hwss_dio.c | 16 ++-- .../gpu/drm/amd/display/dc/link/link_detection.c | 86 +++++++++++++++++++++- drivers/gpu/drm/amd/display/dc/link/link_dpms.c | 3 + drivers/gpu/drm/amd/display/dc/link/link_factory.c | 3 + 4 files changed, 101 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c index 80344cbd1f99..befa67b2b2ae 100644 --- a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c +++ b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c @@ -58,8 +58,9 @@ void setup_dio_stream_encoder(struct pipe_ctx *pipe_ctx) return; } - link_enc->funcs->connect_dig_be_to_fe(link_enc, - pipe_ctx->stream_res.stream_enc->id, true); + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + link_enc->funcs->connect_dig_be_to_fe(link_enc, + pipe_ctx->stream_res.stream_enc->id, true); if (dc_is_dp_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->ctx->dc->link_srv->dp_trace_source_sequence(pipe_ctx->stream->link, DPCD_SOURCE_SEQ_AFTER_CONNECT_DIG_FE_BE); @@ -98,10 +99,13 @@ void reset_dio_stream_encoder(struct pipe_ctx *pipe_ctx) if (stream_enc->funcs->enable_stream) stream_enc->funcs->enable_stream(stream_enc, pipe_ctx->stream->signal, false); - link_enc->funcs->connect_dig_be_to_fe( - link_enc, - pipe_ctx->stream_res.stream_enc->id, - false); + + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + link_enc->funcs->connect_dig_be_to_fe( + link_enc, + pipe_ctx->stream_res.stream_enc->id, + false); + if (dc_is_dp_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->ctx->dc->link_srv->dp_trace_source_sequence( pipe_ctx->stream->link, diff --git a/drivers/gpu/drm/amd/display/dc/link/link_detection.c b/drivers/gpu/drm/amd/display/dc/link/link_detection.c index 801c94efa7fc..d83fd52cb1b0 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -862,6 +862,48 @@ static void verify_link_capability(struct dc_link *link, struct dc_sink *sink, verify_link_capability_non_destructive(link); } +/** + * link_detect_evaluate_edid_header() - Evaluate if an EDID header is acceptable. + * + * Evaluates an 8-byte EDID header to check if it's good enough + * for the purpose of determining whether a display is connected + * without reading the full EDID. + */ +static bool link_detect_evaluate_edid_header(uint8_t edid_header[8]) +{ + int edid_header_score = 0; + int i; + + for (i = 0; i < 8; ++i) + edid_header_score += edid_header[i] == ((i == 0 || i == 7) ? 0x00 : 0xff); + + return edid_header_score >= 6; +} + +/** + * link_detect_ddc_probe() - Probe the DDC to see if a display is connected. + * + * Detect whether a display is connected to DDC without reading full EDID. + * Reads only the EDID header (the first 8 bytes of EDID) from DDC and + * evaluates whether that matches. + */ +static bool link_detect_ddc_probe(struct dc_link *link) +{ + if (!link->ddc) + return false; + + uint8_t edid_header[8] = {0}; + bool ddc_probed = i2c_read(link->ddc, 0x50, edid_header, sizeof(edid_header)); + + if (!ddc_probed) + return false; + + if (!link_detect_evaluate_edid_header(edid_header)) + return false; + + return true; +} + /* * detect_link_and_local_sink() - Detect if a sink is attached to a given link * @@ -946,6 +988,12 @@ static bool detect_link_and_local_sink(struct dc_link *link, break; } + case SIGNAL_TYPE_RGB: { + sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C; + sink_caps.signal = SIGNAL_TYPE_RGB; + break; + } + case SIGNAL_TYPE_LVDS: { sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C; sink_caps.signal = SIGNAL_TYPE_LVDS; @@ -1139,9 +1187,17 @@ static bool detect_link_and_local_sink(struct dc_link *link, sink = prev_sink; prev_sink = NULL; } - query_hdcp_capability(sink->sink_signal, link); + + if (!sink->edid_caps.analog) + query_hdcp_capability(sink->sink_signal, link); } + /* DVI-I connector connected to analog display. */ + if ((link->link_id.id == CONNECTOR_ID_DUAL_LINK_DVII || + link->link_id.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + sink->edid_caps.analog) + sink->sink_signal = SIGNAL_TYPE_RGB; + /* HDMI-DVI Dongle */ if (sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A && !sink->edid_caps.edid_hdmi) @@ -1238,6 +1294,23 @@ static bool detect_link_and_local_sink(struct dc_link *link, return true; } +/** + * link_detect_analog() - Determines if an analog sink is connected. + */ +static bool link_detect_analog(struct dc_link *link, enum dc_connection_type *type) +{ + /* Don't care about connectors that don't support an analog signal. */ + ASSERT(dc_connector_supports_analog(link->link_id.id)); + + if (link_detect_ddc_probe(link)) { + *type = dc_connection_single; + return true; + } + + *type = dc_connection_none; + return true; +} + /* * link_detect_connection_type() - Determine if there is a sink connected * @@ -1254,6 +1327,17 @@ bool link_detect_connection_type(struct dc_link *link, enum dc_connection_type * return true; } + /* Ignore the HPD pin (if any) for analog connectors. + * Instead rely on DDC. + * + * - VGA connectors don't have any HPD at all. + * - Some DVI-A cables don't connect the HPD pin. + * - Some DVI-A cables pull up the HPD pin. + * (So it's high even when no display is connected.) + */ + if (dc_connector_supports_analog(link->link_id.id)) + return link_detect_analog(link, type); + if (link->connector_signal == SIGNAL_TYPE_EDP) { /*in case it is not on*/ if (!link->dc->config.edp_no_power_sequencing) diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c index fdefe27f4ba3..4ddcdc222913 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c @@ -2258,6 +2258,9 @@ static enum dc_status enable_link( enable_link_lvds(pipe_ctx); status = DC_OK; break; + case SIGNAL_TYPE_RGB: + status = DC_OK; + break; case SIGNAL_TYPE_VIRTUAL: status = enable_link_virtual(pipe_ctx); break; diff --git a/drivers/gpu/drm/amd/display/dc/link/link_factory.c b/drivers/gpu/drm/amd/display/dc/link/link_factory.c index ebc3798e7d25..7989baf3843c 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_factory.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_factory.c @@ -583,6 +583,9 @@ static bool construct_phy(struct dc_link *link, case CONNECTOR_ID_DUAL_LINK_DVII: link->connector_signal = SIGNAL_TYPE_DVI_DUAL_LINK; break; + case CONNECTOR_ID_VGA: + link->connector_signal = SIGNAL_TYPE_RGB; + break; case CONNECTOR_ID_DISPLAY_PORT: case CONNECTOR_ID_MXM: case CONNECTOR_ID_USBC: -- cgit v1.2.3 From 8223a605744bb471f31018eac9075a539415b16f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:55 +0200 Subject: drm/amd/display: Refactor amdgpu_dm_connector_detect (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Prepare for polling analog connectors. Document the function better. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 35 ++++++++++++----------- 1 file changed, 19 insertions(+), 16 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 9e260ca04e1e..3bd404b727b6 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -7226,29 +7226,32 @@ finish: return stream; } +/** + * amdgpu_dm_connector_detect() - Detect whether a DRM connector is connected to a display + * + * A connector is considered connected when it has a sink that is not NULL. + * For connectors that support HPD (hotplug detection), the connection is + * handled in the HPD interrupt. + * + * Notes: + * 1. This interface is NOT called in context of HPD irq. + * 2. This interface *is called* in context of user-mode ioctl. Which + * makes it a bad place for *any* MST-related activity. + */ static enum drm_connector_status amdgpu_dm_connector_detect(struct drm_connector *connector, bool force) { - bool connected; struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector); - /* - * Notes: - * 1. This interface is NOT called in context of HPD irq. - * 2. This interface *is called* in context of user-mode ioctl. Which - * makes it a bad place for *any* MST-related activity. - */ - - if (aconnector->base.force == DRM_FORCE_UNSPECIFIED && - !aconnector->fake_enable) - connected = (aconnector->dc_sink != NULL); - else - connected = (aconnector->base.force == DRM_FORCE_ON || - aconnector->base.force == DRM_FORCE_ON_DIGITAL); - update_subconnector_property(aconnector); - return (connected ? connector_status_connected : + if (aconnector->base.force == DRM_FORCE_ON || + aconnector->base.force == DRM_FORCE_ON_DIGITAL) + return connector_status_connected; + else if (aconnector->base.force == DRM_FORCE_OFF) + return connector_status_disconnected; + + return (aconnector->dc_sink ? connector_status_connected : connector_status_disconnected); } -- cgit v1.2.3 From c4f3f114e73c331f02c7188af650901ee3bad2c1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:56 +0200 Subject: drm/amd/display: Poll analog connectors (v3) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit VGA connectors don't support any hotplug detection, so the kernel needs to periodically poll them to see if a display is connected. DVI-I connectors have hotplug detection for digital signals, and some analog DVI cables pull up that pin to work with that. However, in general not all DVI cables do this so we can't rely on this feature, therefore we need to poll DVI-I connectors as well. v2: Call drm_kms_helper_poll_fini in amdgpu_dm_hpd_fini. Disable/enable polling on suspend/resume. Don't call full link detection when already connected. v3: Encounter CLANG build failure. Remove unused variable: drivers/gpu/drm/amd/amdgpu/../display/amdgpu_dm/amdgpu_dm_irq.c:980:7: error: variable 'use_polling' set but not used [-Werror,-Wunused-but- set-variable] 980 | bool use_polling = false; Signed-off-by: Timur Kristóf Signed-off-by: Wayne Lin Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 73 +++++++++++++++++++++- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c | 17 +++++ 2 files changed, 88 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 3bd404b727b6..d18d72a520f2 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -3853,7 +3853,9 @@ void amdgpu_dm_update_connector_after_detect( drm_dbg_kms(dev, "DCHPD: connector_id=%d: Old sink=%p New sink=%p\n", aconnector->connector_id, aconnector->dc_sink, sink); - guard(mutex)(&dev->mode_config.mutex); + /* When polling, DRM has already locked the mutex for us. */ + if (!drm_kms_helper_is_poll_worker()) + mutex_lock(&dev->mode_config.mutex); /* * 1. Update status of the drm connector @@ -3916,6 +3918,10 @@ void amdgpu_dm_update_connector_after_detect( } update_subconnector_property(aconnector); + + /* When polling, the mutex will be unlocked for us by DRM. */ + if (!drm_kms_helper_is_poll_worker()) + mutex_unlock(&dev->mode_config.mutex); } static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector) @@ -7226,12 +7232,63 @@ finish: return stream; } +/** + * amdgpu_dm_connector_poll() - Poll a connector to see if it's connected to a display + * + * Used for connectors that don't support HPD (hotplug detection) + * to periodically checked whether the connector is connected to a display. + */ +static enum drm_connector_status +amdgpu_dm_connector_poll(struct amdgpu_dm_connector *aconnector, bool force) +{ + struct drm_connector *connector = &aconnector->base; + struct drm_device *dev = connector->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + struct dc_link *link = aconnector->dc_link; + enum dc_connection_type conn_type = dc_connection_none; + enum drm_connector_status status = connector_status_disconnected; + + mutex_lock(&aconnector->hpd_lock); + + if (dc_link_detect_connection_type(aconnector->dc_link, &conn_type) && + conn_type != dc_connection_none) { + mutex_lock(&adev->dm.dc_lock); + + /* Only call full link detection when a sink isn't created yet, + * ie. just when the display is plugged in, otherwise we risk flickering. + */ + if (link->local_sink || + dc_link_detect(link, DETECT_REASON_HPD)) + status = connector_status_connected; + + mutex_unlock(&adev->dm.dc_lock); + } + + if (connector->status != status) { + if (status == connector_status_disconnected) { + if (link->local_sink) + dc_sink_release(link->local_sink); + + link->local_sink = NULL; + link->dpcd_sink_count = 0; + link->type = dc_connection_none; + } + + amdgpu_dm_update_connector_after_detect(aconnector); + } + + mutex_unlock(&aconnector->hpd_lock); + return status; +} + /** * amdgpu_dm_connector_detect() - Detect whether a DRM connector is connected to a display * * A connector is considered connected when it has a sink that is not NULL. * For connectors that support HPD (hotplug detection), the connection is * handled in the HPD interrupt. + * For connectors that may not support HPD, such as analog connectors, + * DRM will call this function repeatedly to poll them. * * Notes: * 1. This interface is NOT called in context of HPD irq. @@ -7251,6 +7308,14 @@ amdgpu_dm_connector_detect(struct drm_connector *connector, bool force) else if (aconnector->base.force == DRM_FORCE_OFF) return connector_status_disconnected; + /* Poll analog connectors and only when either + * disconnected or connected to an analog display. + */ + if (drm_kms_helper_is_poll_worker() && + dc_connector_supports_analog(aconnector->dc_link->link_id.id) && + (!aconnector->dc_sink || aconnector->dc_sink->edid_caps.analog)) + return amdgpu_dm_connector_poll(aconnector, force); + return (aconnector->dc_sink ? connector_status_connected : connector_status_disconnected); } @@ -8675,9 +8740,13 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm, link->link_enc->features.dp_ycbcr420_supported ? true : false; break; case DRM_MODE_CONNECTOR_DVID: - case DRM_MODE_CONNECTOR_DVII: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_VGA: + aconnector->base.polled = + DRM_CONNECTOR_POLL_CONNECT | DRM_CONNECTOR_POLL_DISCONNECT; + break; default: break; } diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c index a1c722112c22..0a2a3f233a0e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c @@ -476,6 +476,7 @@ void amdgpu_dm_irq_fini(struct amdgpu_device *adev) void amdgpu_dm_irq_suspend(struct amdgpu_device *adev) { + struct drm_device *dev = adev_to_drm(adev); int src; struct list_head *hnd_list_h; struct list_head *hnd_list_l; @@ -512,6 +513,9 @@ void amdgpu_dm_irq_suspend(struct amdgpu_device *adev) } DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_disable(dev); } void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev) @@ -537,6 +541,7 @@ void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev) void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev) { + struct drm_device *dev = adev_to_drm(adev); int src; struct list_head *hnd_list_h, *hnd_list_l; unsigned long irq_table_flags; @@ -557,6 +562,9 @@ void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev) } DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_enable(dev); } /* @@ -893,6 +901,7 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) struct drm_connector_list_iter iter; int irq_type; int i; + bool use_polling = false; /* First, clear all hpd and hpdrx interrupts */ for (i = DC_IRQ_SOURCE_HPD1; i <= DC_IRQ_SOURCE_HPD6RX; i++) { @@ -906,6 +915,8 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) struct amdgpu_dm_connector *amdgpu_dm_connector; const struct dc_link *dc_link; + use_polling |= connector->polled != DRM_CONNECTOR_POLL_HPD; + if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK) continue; @@ -947,6 +958,9 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); + + if (use_polling) + drm_kms_helper_poll_init(dev); } /** @@ -997,4 +1011,7 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_fini(dev); } -- cgit v1.2.3 From 2debe0ecb53d24b5150334fa87fc02351c030046 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:57 +0200 Subject: drm/amd/display: Add DCE BIOS_SCRATCH_0 register MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The BIOS uses this register to write the results of the DAC_LoadDetection command, so we'll need to read this in order to make DAC load detection work. As a reference, I used the mmBIOS_SCRATCH_0 definition from the amdgpu legacy display code. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_bios_types.h | 1 + drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c | 2 ++ drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c | 2 ++ drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c | 2 ++ drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c | 1 + drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c | 2 ++ drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c | 2 ++ 7 files changed, 12 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h index 545ce1e15eae..50c8906b74c5 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h @@ -168,6 +168,7 @@ struct dc_vbios_funcs { }; struct bios_registers { + uint32_t BIOS_SCRATCH_0; uint32_t BIOS_SCRATCH_3; uint32_t BIOS_SCRATCH_6; }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index 85ea219e4fd4..d40d91ec2035 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -78,6 +78,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -369,6 +370,7 @@ static const struct dce_abm_mask abm_mask = { #define DCFE_MEM_PWR_CTRL_REG_BASE 0x1b03 static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c index 9e14ffb3c942..cd54382c0af3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c @@ -82,6 +82,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -377,6 +378,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c index 62977bcdeaa0..3f0a6bc4dcc2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c @@ -76,6 +76,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -385,6 +386,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c index 0770ea37183f..b1570b6b1af3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c @@ -491,6 +491,7 @@ static struct dce_i2c_hw *dce120_i2c_hw_create( return dce_i2c_hw; } static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0 + NBIO_BASE(mmBIOS_SCRATCH_0_BASE_IDX), .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3 + NBIO_BASE(mmBIOS_SCRATCH_3_BASE_IDX), .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 + NBIO_BASE(mmBIOS_SCRATCH_6_BASE_IDX) }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c index 6e89b7020b61..f0152933bee2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c @@ -80,6 +80,7 @@ #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -368,6 +369,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index a1ad7398e39f..8687104cabb7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -78,6 +78,7 @@ #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -369,6 +370,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; -- cgit v1.2.3 From 7f1d1c2f4da0640d6e54496245cbac3ed1d2e8b9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:58 +0200 Subject: drm/amd/display: Make get_support_mask_for_device_id reusable MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This will be reused by DAC load detection. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 33d0ec38ded7..a126ca3a53fb 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -67,7 +67,9 @@ static ATOM_HPD_INT_RECORD *get_hpd_record(struct bios_parser *bp, ATOM_OBJECT *object); static struct device_id device_type_from_device_id(uint16_t device_id); static uint32_t signal_to_ss_id(enum as_signal_type signal); -static uint32_t get_support_mask_for_device_id(struct device_id device_id); +static uint32_t get_support_mask_for_device_id( + enum dal_device_type device_type, + uint32_t enum_id); static ATOM_ENCODER_CAP_RECORD_V2 *get_encoder_cap_record( struct bios_parser *bp, ATOM_OBJECT *object); @@ -888,7 +890,7 @@ static bool bios_parser_is_device_id_supported( { struct bios_parser *bp = BP_FROM_DCB(dcb); - uint32_t mask = get_support_mask_for_device_id(id); + uint32_t mask = get_support_mask_for_device_id(id.device_type, id.enum_id); return (le16_to_cpu(bp->object_info_tbl.v1_1->usDeviceSupport) & mask) != 0; } @@ -2179,11 +2181,10 @@ static uint32_t signal_to_ss_id(enum as_signal_type signal) return clk_id_ss; } -static uint32_t get_support_mask_for_device_id(struct device_id device_id) +static uint32_t get_support_mask_for_device_id( + enum dal_device_type device_type, + uint32_t enum_id) { - enum dal_device_type device_type = device_id.device_type; - uint32_t enum_id = device_id.enum_id; - switch (device_type) { case DEVICE_TYPE_LCD: switch (enum_id) { -- cgit v1.2.3 From d75e45b853630dbcd2b5c1a1aea7e2f3dc12effc Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:01:59 +0200 Subject: drm/amd/display: Add DAC_LoadDetection to BIOS parser (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit DAC_LoadDetection can be used to determine whether something is connected to an analog connector by determining if there is an analog load. This causes visible flickering on displays, so we only resort to using this when the connected display doesn't have an EDID. For reference, see the legacy display code: amdgpu_atombios_encoder_dac_load_detect v2: Only clear corresponding bit from BIOS_SCRATCH_0. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/bios/bios_parser.c | 50 ++++++++++++ .../gpu/drm/amd/display/dc/bios/command_table.c | 92 ++++++++++++++++++++++ .../gpu/drm/amd/display/dc/bios/command_table.h | 3 + drivers/gpu/drm/amd/display/dc/dc_bios_types.h | 5 ++ .../drm/amd/display/include/bios_parser_types.h | 5 ++ 5 files changed, 155 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index a126ca3a53fb..4120d6c4c5e4 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -780,6 +780,54 @@ static enum bp_result bios_parser_encoder_control( return bp->cmd_tbl.dig_encoder_control(bp, cntl); } +static enum bp_result bios_parser_dac_load_detection( + struct dc_bios *dcb, + enum engine_id engine_id, + enum dal_device_type device_type, + uint32_t enum_id) +{ + struct bios_parser *bp = BP_FROM_DCB(dcb); + struct dc_context *ctx = dcb->ctx; + struct bp_load_detection_parameters bp_params = {0}; + enum bp_result bp_result; + uint32_t bios_0_scratch; + uint32_t device_id_mask = 0; + + bp_params.engine_id = engine_id; + bp_params.device_id = get_support_mask_for_device_id(device_type, enum_id); + + if (engine_id != ENGINE_ID_DACA && + engine_id != ENGINE_ID_DACB) + return BP_RESULT_UNSUPPORTED; + + if (!bp->cmd_tbl.dac_load_detection) + return BP_RESULT_UNSUPPORTED; + + if (bp_params.device_id == ATOM_DEVICE_CRT1_SUPPORT) + device_id_mask = ATOM_S0_CRT1_MASK; + else if (bp_params.device_id == ATOM_DEVICE_CRT1_SUPPORT) + device_id_mask = ATOM_S0_CRT2_MASK; + else + return BP_RESULT_UNSUPPORTED; + + /* BIOS will write the detected devices to BIOS_SCRATCH_0, clear corresponding bit */ + bios_0_scratch = dm_read_reg(ctx, bp->base.regs->BIOS_SCRATCH_0); + bios_0_scratch &= ~device_id_mask; + dm_write_reg(ctx, bp->base.regs->BIOS_SCRATCH_0, bios_0_scratch); + + bp_result = bp->cmd_tbl.dac_load_detection(bp, &bp_params); + + if (bp_result != BP_RESULT_OK) + return bp_result; + + bios_0_scratch = dm_read_reg(ctx, bp->base.regs->BIOS_SCRATCH_0); + + if (bios_0_scratch & device_id_mask) + return BP_RESULT_OK; + + return BP_RESULT_FAILURE; +} + static enum bp_result bios_parser_adjust_pixel_clock( struct dc_bios *dcb, struct bp_adjust_pixel_clock_parameters *bp_params) @@ -2864,6 +2912,8 @@ static const struct dc_vbios_funcs vbios_funcs = { .encoder_control = bios_parser_encoder_control, + .dac_load_detection = bios_parser_dac_load_detection, + .transmitter_control = bios_parser_transmitter_control, .enable_crtc = bios_parser_enable_crtc, diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.c b/drivers/gpu/drm/amd/display/dc/bios/command_table.c index dbd84477ceb7..22457f417e65 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.c @@ -54,6 +54,7 @@ static void init_enable_spread_spectrum_on_ppll(struct bios_parser *bp); static void init_adjust_display_pll(struct bios_parser *bp); static void init_select_crtc_source(struct bios_parser *bp); static void init_dac_encoder_control(struct bios_parser *bp); +static void init_dac_load_detection(struct bios_parser *bp); static void init_dac_output_control(struct bios_parser *bp); static void init_set_crtc_timing(struct bios_parser *bp); static void init_enable_crtc(struct bios_parser *bp); @@ -72,6 +73,7 @@ void dal_bios_parser_init_cmd_tbl(struct bios_parser *bp) init_adjust_display_pll(bp); init_select_crtc_source(bp); init_dac_encoder_control(bp); + init_dac_load_detection(bp); init_dac_output_control(bp); init_set_crtc_timing(bp); init_enable_crtc(bp); @@ -1902,6 +1904,96 @@ static enum bp_result dac2_encoder_control_v1( return result; } +/******************************************************************************* + ******************************************************************************** + ** + ** DAC LOAD DETECTION + ** + ******************************************************************************** + *******************************************************************************/ + +static enum bp_result dac_load_detection_v1( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); + +static enum bp_result dac_load_detection_v3( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); + +static void init_dac_load_detection(struct bios_parser *bp) +{ + switch (BIOS_CMD_TABLE_PARA_REVISION(DAC_LoadDetection)) { + case 1: + case 2: + bp->cmd_tbl.dac_load_detection = dac_load_detection_v1; + break; + case 3: + default: + bp->cmd_tbl.dac_load_detection = dac_load_detection_v3; + break; + } +} + +static void dac_load_detect_prepare_params( + struct _DAC_LOAD_DETECTION_PS_ALLOCATION *params, + enum engine_id engine_id, + uint16_t device_id, + uint8_t misc) +{ + uint8_t dac_type = ENGINE_ID_DACA; + + if (engine_id == ENGINE_ID_DACB) + dac_type = ATOM_DAC_B; + + params->sDacload.usDeviceID = cpu_to_le16(device_id); + params->sDacload.ucDacType = dac_type; + params->sDacload.ucMisc = misc; +} + +static enum bp_result dac_load_detection_v1( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + DAC_LOAD_DETECTION_PS_ALLOCATION params; + + dac_load_detect_prepare_params( + ¶ms, + bp_params->engine_id, + bp_params->device_id, + 0); + + if (EXEC_BIOS_CMD_TABLE(DAC_LoadDetection, params)) + result = BP_RESULT_OK; + + return result; +} + +static enum bp_result dac_load_detection_v3( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + DAC_LOAD_DETECTION_PS_ALLOCATION params; + + uint8_t misc = 0; + + if (bp_params->device_id == ATOM_DEVICE_CV_SUPPORT || + bp_params->device_id == ATOM_DEVICE_TV1_SUPPORT) + misc = DAC_LOAD_MISC_YPrPb; + + dac_load_detect_prepare_params( + ¶ms, + bp_params->engine_id, + bp_params->device_id, + misc); + + if (EXEC_BIOS_CMD_TABLE(DAC_LoadDetection, params)) + result = BP_RESULT_OK; + + return result; +} + /******************************************************************************* ******************************************************************************** ** diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.h b/drivers/gpu/drm/amd/display/dc/bios/command_table.h index 8b04b903e93d..e89b1ba0048b 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.h +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.h @@ -71,6 +71,9 @@ struct cmd_tbl { enum bp_result (*dac2_output_control)( struct bios_parser *bp, bool enable); + enum bp_result (*dac_load_detection)( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); enum bp_result (*set_crtc_timing)( struct bios_parser *bp, struct bp_hw_crtc_timing_parameters *bp_params); diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h index 50c8906b74c5..40d7a7d83c40 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h @@ -97,6 +97,11 @@ struct dc_vbios_funcs { enum bp_result (*encoder_control)( struct dc_bios *bios, struct bp_encoder_control *cntl); + enum bp_result (*dac_load_detection)( + struct dc_bios *bios, + enum engine_id engine_id, + enum dal_device_type device_type, + uint32_t enum_id); enum bp_result (*transmitter_control)( struct dc_bios *bios, struct bp_transmitter_control *cntl); diff --git a/drivers/gpu/drm/amd/display/include/bios_parser_types.h b/drivers/gpu/drm/amd/display/include/bios_parser_types.h index d9e58a6a0d36..973b6bdbac63 100644 --- a/drivers/gpu/drm/amd/display/include/bios_parser_types.h +++ b/drivers/gpu/drm/amd/display/include/bios_parser_types.h @@ -162,6 +162,11 @@ struct bp_transmitter_control { bool single_pll_mode; }; +struct bp_load_detection_parameters { + enum engine_id engine_id; + uint16_t device_id; +}; + struct bp_hw_crtc_timing_parameters { enum controller_id controller_id; /* horizontal part */ -- cgit v1.2.3 From ac1bb49522676dfbd83c323376f8d5daa5b1ba84 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:02:00 +0200 Subject: drm/amd/display: Use DAC load detection on analog connectors (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This feature is useful for analog connections without EDID: - Really old monitors with a VGA connector - Cheap DVI/VGA adapters that don't connect DDC pins When a connection is established through DAC load detection, the driver is supposed to fill in the supported modes for the display, which we already do in amdgpu_dm_connector_get_modes. Also, because the load detection causes visible glitches, do not attempt to poll the connector again after it was detected this way. Note that it will still be polled after sleep/resume or when force is enabled, which is okay. v2: Add dc_connection_dac_load connection type. Properly release sink when no display is connected. Don't print error when EDID isn't read from an analog display. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 10 ++++ drivers/gpu/drm/amd/display/dc/dc_types.h | 3 +- .../gpu/drm/amd/display/dc/link/link_detection.c | 61 +++++++++++++++++++++- 3 files changed, 72 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index d18d72a520f2..a235a5b61186 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -7248,6 +7248,16 @@ amdgpu_dm_connector_poll(struct amdgpu_dm_connector *aconnector, bool force) enum dc_connection_type conn_type = dc_connection_none; enum drm_connector_status status = connector_status_disconnected; + /* When we determined the connection using DAC load detection, + * do NOT poll the connector do detect disconnect because + * that would run DAC load detection again which can cause + * visible visual glitches. + * + * Only allow to poll such a connector again when forcing. + */ + if (!force && link->local_sink && link->type == dc_connection_dac_load) + return connector->status; + mutex_lock(&aconnector->hpd_lock); if (dc_link_detect_connection_type(aconnector->dc_link, &conn_type) && diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index 9989c8af5174..ea6b71c43d2c 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -353,7 +353,8 @@ enum dc_connection_type { dc_connection_none, dc_connection_single, dc_connection_mst_branch, - dc_connection_sst_branch + dc_connection_sst_branch, + dc_connection_dac_load }; struct dc_csc_adjustments { diff --git a/drivers/gpu/drm/amd/display/dc/link/link_detection.c b/drivers/gpu/drm/amd/display/dc/link/link_detection.c index d83fd52cb1b0..c417780f37bc 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -904,6 +904,37 @@ static bool link_detect_ddc_probe(struct dc_link *link) return true; } +/** + * link_detect_dac_load_detect() - Performs DAC load detection. + * + * Load detection can be used to detect the presence of an + * analog display when we can't read DDC. This causes a visible + * visual glitch so it should be used sparingly. + */ +static bool link_detect_dac_load_detect(struct dc_link *link) +{ + struct dc_bios *bios = link->ctx->dc_bios; + struct link_encoder *link_enc = link->link_enc; + enum engine_id engine_id = link_enc->preferred_engine; + enum dal_device_type device_type = DEVICE_TYPE_CRT; + enum bp_result bp_result; + uint32_t enum_id; + + switch (engine_id) { + case ENGINE_ID_DACB: + enum_id = 2; + break; + case ENGINE_ID_DACA: + default: + engine_id = ENGINE_ID_DACA; + enum_id = 1; + break; + } + + bp_result = bios->funcs->dac_load_detection(bios, engine_id, device_type, enum_id); + return bp_result == BP_RESULT_OK; +} + /* * detect_link_and_local_sink() - Detect if a sink is attached to a given link * @@ -1118,7 +1149,30 @@ static bool detect_link_and_local_sink(struct dc_link *link, DC_LOG_ERROR("Partial EDID valid, abandon invalid blocks.\n"); break; case EDID_NO_RESPONSE: + /* Analog connectors without EDID: + * - old monitor that actually doesn't have EDID + * - cheap DVI-A cable or adapter that doesn't connect DDC + */ + if (dc_connector_supports_analog(link->link_id.id)) { + /* If we didn't do DAC load detection yet, do it now + * to verify there really is a display connected. + */ + if (link->type != dc_connection_dac_load && + !link_detect_dac_load_detect(link)) { + if (prev_sink) + dc_sink_release(prev_sink); + link_disconnect_sink(link); + return false; + } + + DC_LOG_INFO("%s detected analog display without EDID\n", __func__); + link->type = dc_connection_dac_load; + sink->edid_caps.analog = true; + break; + } + DC_LOG_ERROR("No EDID read.\n"); + /* * Abort detection for non-DP connectors if we have * no EDID @@ -1307,6 +1361,11 @@ static bool link_detect_analog(struct dc_link *link, enum dc_connection_type *ty return true; } + if (link_detect_dac_load_detect(link)) { + *type = dc_connection_dac_load; + return true; + } + *type = dc_connection_none; return true; } @@ -1328,7 +1387,7 @@ bool link_detect_connection_type(struct dc_link *link, enum dc_connection_type * } /* Ignore the HPD pin (if any) for analog connectors. - * Instead rely on DDC. + * Instead rely on DDC and DAC. * * - VGA connectors don't have any HPD at all. * - Some DVI-A cables don't connect the HPD pin. -- cgit v1.2.3 From 70181ad96ec22c7ade7e373578b7a1a298035acd Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:02:01 +0200 Subject: drm/amd/display: Add common modes to analog displays without EDID MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit When the EDID of an analog display is not available, we can't know the possible modes supported by the display. However, we still need to offer the user to select from a variety of common modes. It will be up to the user to select the best one, though. This is how it works on other operating systems as well as the legacy display code path in amdgpu. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 46 ++++++++++++++--------- 1 file changed, 28 insertions(+), 18 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index a235a5b61186..dc091ccab46e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -8412,7 +8412,7 @@ static void amdgpu_dm_get_native_mode(struct drm_connector *connector) static struct drm_display_mode * amdgpu_dm_create_common_mode(struct drm_encoder *encoder, - char *name, + const char *name, int hdisplay, int vdisplay) { struct drm_device *dev = encoder->dev; @@ -8434,6 +8434,24 @@ amdgpu_dm_create_common_mode(struct drm_encoder *encoder, } +static const struct amdgpu_dm_mode_size { + char name[DRM_DISPLAY_MODE_LEN]; + int w; + int h; +} common_modes[] = { + { "640x480", 640, 480}, + { "800x600", 800, 600}, + { "1024x768", 1024, 768}, + { "1280x720", 1280, 720}, + { "1280x800", 1280, 800}, + {"1280x1024", 1280, 1024}, + { "1440x900", 1440, 900}, + {"1680x1050", 1680, 1050}, + {"1600x1200", 1600, 1200}, + {"1920x1080", 1920, 1080}, + {"1920x1200", 1920, 1200} +}; + static void amdgpu_dm_connector_add_common_modes(struct drm_encoder *encoder, struct drm_connector *connector) { @@ -8444,23 +8462,6 @@ static void amdgpu_dm_connector_add_common_modes(struct drm_encoder *encoder, to_amdgpu_dm_connector(connector); int i; int n; - struct mode_size { - char name[DRM_DISPLAY_MODE_LEN]; - int w; - int h; - } common_modes[] = { - { "640x480", 640, 480}, - { "800x600", 800, 600}, - { "1024x768", 1024, 768}, - { "1280x720", 1280, 720}, - { "1280x800", 1280, 800}, - {"1280x1024", 1280, 1024}, - { "1440x900", 1440, 900}, - {"1680x1050", 1680, 1050}, - {"1600x1200", 1600, 1200}, - {"1920x1080", 1920, 1080}, - {"1920x1200", 1920, 1200} - }; if ((connector->connector_type != DRM_MODE_CONNECTOR_eDP) && (connector->connector_type != DRM_MODE_CONNECTOR_LVDS)) @@ -8684,6 +8685,15 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector) if (dc->link_srv->dp_get_encoding_format(verified_link_cap) == DP_128b_132b_ENCODING) amdgpu_dm_connector->num_modes += drm_add_modes_noedid(connector, 1920, 1080); + + if (amdgpu_dm_connector->dc_sink->edid_caps.analog) { + /* Analog monitor connected by DAC load detection. + * Add common modes. It will be up to the user to select one that works. + */ + for (int i = 0; i < ARRAY_SIZE(common_modes); i++) + amdgpu_dm_connector->num_modes += drm_add_modes_noedid( + connector, common_modes[i].w, common_modes[i].h); + } } else { amdgpu_dm_connector_ddc_get_modes(connector, drm_edid); if (encoder) -- cgit v1.2.3 From a9466f63ef36aa42fa989cfe50e30c07992b7d25 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:02:02 +0200 Subject: drm/amd/display: Don't add freesync modes to analog displays (v2) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit VRR is not supported on analog signals. Don't add freesync modes to analog displays or when VRR is unsupported by DC. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index dc091ccab46e..cce82deff51e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -8662,6 +8662,10 @@ static void amdgpu_dm_connector_add_freesync_modes(struct drm_connector *connect if (!(amdgpu_freesync_vid_mode && drm_edid)) return; + if (!amdgpu_dm_connector->dc_sink || amdgpu_dm_connector->dc_sink->edid_caps.analog || + !dc_supports_vrr(amdgpu_dm_connector->dc_sink->ctx->dce_version)) + return; + if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10) amdgpu_dm_connector->num_modes += add_fs_modes(amdgpu_dm_connector); -- cgit v1.2.3 From f44aad39b662eb9058dc5123168d2502d31d0ab3 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Timur=20Krist=C3=B3f?= Date: Fri, 26 Sep 2025 20:02:03 +0200 Subject: drm/amdgpu: Use DC by default for Bonaire MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Now that DC supports analog connectors, there is nothing stopping us from using it by default on Bonaire. Signed-off-by: Timur Kristóf Reviewed-by: Harry Wentland Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index b8d91247f51a..b92234d63562 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -4215,7 +4215,6 @@ bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev, #else return false; #endif - case CHIP_BONAIRE: case CHIP_KAVERI: case CHIP_KABINI: case CHIP_MULLINS: -- cgit v1.2.3 From fea8f13f4f9f1f824a998624d6ec20cab725b5f5 Mon Sep 17 00:00:00 2001 From: Sunday Clement Date: Fri, 17 Oct 2025 10:15:50 -0400 Subject: drm/amdkfd: Fix Unchecked Return Value Properly check the return values for function, as done elsewhere. Signed-off-by: Sunday Clement Reviewed-by: Harish Kasiviswanathan Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index 6e7bc983fc0b..4fbe865ff279 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -1897,6 +1897,8 @@ fail_packet_manager_init: static int stop_cpsch(struct device_queue_manager *dqm) { + int ret = 0; + dqm_lock(dqm); if (!dqm->sched_running) { dqm_unlock(dqm); @@ -1904,9 +1906,10 @@ static int stop_cpsch(struct device_queue_manager *dqm) } if (!dqm->dev->kfd->shared_resources.enable_mes) - unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD, false); + ret = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, + 0, USE_DEFAULT_GRACE_PERIOD, false); else - remove_all_kfd_queues_mes(dqm); + ret = remove_all_kfd_queues_mes(dqm); dqm->sched_running = false; @@ -1920,7 +1923,7 @@ static int stop_cpsch(struct device_queue_manager *dqm) dqm->detect_hang_info = NULL; dqm_unlock(dqm); - return 0; + return ret; } static int create_kernel_queue_cpsch(struct device_queue_manager *dqm, -- cgit v1.2.3 From 849ad2a3008957a4e1d846507799bd169f0e5d9c Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Mon, 11 Aug 2025 17:30:40 +0530 Subject: drm/amdgpu/pm: Add definition for gpu_metrics v1.9 Add gpu metrics definition which is only a set of gpu metrics attributes. A field is encoded by its id, type and number of instances. Signed-off-by: Lijo Lazar Reviewed-by: Asad Kamal Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/include/kgd_pp_interface.h | 117 +++++++++++++++++++++++++ 1 file changed, 117 insertions(+) diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index 87814c2b526e..f92f78d5d330 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -532,6 +532,110 @@ struct metrics_table_header { uint8_t content_revision; }; +enum amdgpu_metrics_attr_id { + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_HOTSPOT, + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_MEM, + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_VRSOC, + AMDGPU_METRICS_ATTR_ID_CURR_SOCKET_POWER, + AMDGPU_METRICS_ATTR_ID_AVERAGE_GFX_ACTIVITY, + AMDGPU_METRICS_ATTR_ID_AVERAGE_UMC_ACTIVITY, + AMDGPU_METRICS_ATTR_ID_MEM_MAX_BANDWIDTH, + AMDGPU_METRICS_ATTR_ID_ENERGY_ACCUMULATOR, + AMDGPU_METRICS_ATTR_ID_SYSTEM_CLOCK_COUNTER, + AMDGPU_METRICS_ATTR_ID_ACCUMULATION_COUNTER, + AMDGPU_METRICS_ATTR_ID_PROCHOT_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_PPT_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_SOCKET_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_VR_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_HBM_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_GFXCLK_LOCK_STATUS, + AMDGPU_METRICS_ATTR_ID_PCIE_LINK_WIDTH, + AMDGPU_METRICS_ATTR_ID_PCIE_LINK_SPEED, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_WIDTH, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_SPEED, + AMDGPU_METRICS_ATTR_ID_GFX_ACTIVITY_ACC, + AMDGPU_METRICS_ATTR_ID_MEM_ACTIVITY_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_BANDWIDTH_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_BANDWIDTH_INST, + AMDGPU_METRICS_ATTR_ID_PCIE_L0_TO_RECOV_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_REPLAY_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_REPLAY_ROVER_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_NAK_SENT_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_NAK_RCVD_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_READ_DATA_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_WRITE_DATA_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_STATUS, + AMDGPU_METRICS_ATTR_ID_FIRMWARE_TIMESTAMP, + AMDGPU_METRICS_ATTR_ID_CURRENT_GFXCLK, + AMDGPU_METRICS_ATTR_ID_CURRENT_SOCCLK, + AMDGPU_METRICS_ATTR_ID_CURRENT_VCLK0, + AMDGPU_METRICS_ATTR_ID_CURRENT_DCLK0, + AMDGPU_METRICS_ATTR_ID_CURRENT_UCLK, + AMDGPU_METRICS_ATTR_ID_NUM_PARTITION, + AMDGPU_METRICS_ATTR_ID_PCIE_LC_PERF_OTHER_END_RECOVERY, + AMDGPU_METRICS_ATTR_ID_GFX_BUSY_INST, + AMDGPU_METRICS_ATTR_ID_JPEG_BUSY, + AMDGPU_METRICS_ATTR_ID_VCN_BUSY, + AMDGPU_METRICS_ATTR_ID_GFX_BUSY_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_PPT_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_THM_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_LOW_UTILIZATION_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_TOTAL_ACC, + AMDGPU_METRICS_ATTR_ID_MAX, +}; + +enum amdgpu_metrics_attr_type { + AMDGPU_METRICS_TYPE_U8, + AMDGPU_METRICS_TYPE_S8, + AMDGPU_METRICS_TYPE_U16, + AMDGPU_METRICS_TYPE_S16, + AMDGPU_METRICS_TYPE_U32, + AMDGPU_METRICS_TYPE_S32, + AMDGPU_METRICS_TYPE_U64, + AMDGPU_METRICS_TYPE_S64, + AMDGPU_METRICS_TYPE_MAX, +}; + +enum amdgpu_metrics_attr_unit { + /* None */ + AMDGPU_METRICS_UNIT_NONE, + /* MHz*/ + AMDGPU_METRICS_UNIT_CLOCK_1, + /* Degree Celsius*/ + AMDGPU_METRICS_UNIT_TEMP_1, + /* Watts*/ + AMDGPU_METRICS_UNIT_POWER_1, + /* In nanoseconds*/ + AMDGPU_METRICS_UNIT_TIME_1, + /* In 10 nanoseconds*/ + AMDGPU_METRICS_UNIT_TIME_2, + /* Speed in GT/s */ + AMDGPU_METRICS_UNIT_SPEED_1, + /* Speed in 0.1 GT/s */ + AMDGPU_METRICS_UNIT_SPEED_2, + /* Bandwidth GB/s */ + AMDGPU_METRICS_UNIT_BW_1, + /* Data in KB */ + AMDGPU_METRICS_UNIT_DATA_1, + /* Percentage */ + AMDGPU_METRICS_UNIT_PERCENT, + AMDGPU_METRICS_UNIT_MAX, +}; + +#define AMDGPU_METRICS_ATTR_UNIT_MASK 0xFF000000 +#define AMDGPU_METRICS_ATTR_UNIT_SHIFT 24 +#define AMDGPU_METRICS_ATTR_TYPE_MASK 0x00F00000 +#define AMDGPU_METRICS_ATTR_TYPE_SHIFT 20 +#define AMDGPU_METRICS_ATTR_ID_MASK 0x000FFC00 +#define AMDGPU_METRICS_ATTR_ID_SHIFT 10 +#define AMDGPU_METRICS_ATTR_INST_MASK 0x000003FF +#define AMDGPU_METRICS_ATTR_INST_SHIFT 0 + +#define AMDGPU_METRICS_ENC_ATTR(unit, type, id, inst) \ + (((u64)(unit) << AMDGPU_METRICS_ATTR_UNIT_SHIFT) | \ + ((u64)(type) << AMDGPU_METRICS_ATTR_TYPE_SHIFT) | \ + ((u64)(id) << AMDGPU_METRICS_ATTR_ID_SHIFT) | (inst)) + /* * gpu_metrics_v1_0 is not recommended as it's not naturally aligned. * Use gpu_metrics_v1_1 or later instead. @@ -1221,6 +1325,19 @@ struct gpu_metrics_v1_8 { uint32_t pcie_lc_perf_other_end_recovery; }; +struct gpu_metrics_attr { + /* Field type encoded with AMDGPU_METRICS_ENC_ATTR */ + uint64_t attr_encoding; + /* Attribute value, depends on attr_encoding */ + void *attr_value; +}; + +struct gpu_metrics_v1_9 { + struct metrics_table_header common_header; + int attr_count; + struct gpu_metrics_attr metrics_attrs[]; +}; + /* * gpu_metrics_v2_0 is not recommended as it's not naturally aligned. * Use gpu_metrics_v2_1 or later instead. -- cgit v1.2.3 From ef4a4b878124ffe6afac5c7b1716308c6964bbdd Mon Sep 17 00:00:00 2001 From: Sakari Ailus Date: Mon, 27 Oct 2025 15:14:38 +0200 Subject: drm/amd: Remove redundant pm_runtime_mark_last_busy() calls pm_runtime_put_autosuspend(), pm_runtime_put_sync_autosuspend(), pm_runtime_autosuspend() and pm_request_autosuspend() now include a call to pm_runtime_mark_last_busy(). Remove the now-redundant explicit call to pm_runtime_mark_last_busy(). Acked-by: Alex Deucher Signed-off-by: Sakari Ailus Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c | 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c | 16 ++++----------- drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c | 25 ----------------------- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 2 -- drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 4 ---- drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c | 2 -- drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 2 -- drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c | 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c | 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c | 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 1 - drivers/gpu/drm/amd/amdkfd/kfd_process.c | 1 - drivers/gpu/drm/amd/pm/amdgpu_pm.c | 1 - 14 files changed, 4 insertions(+), 55 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 6c62e27b9800..d31460a9e958 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -507,7 +507,6 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev, pm_runtime_get_sync(adev_to_drm(adev)->dev); /* Just fire off a uevent and let userspace tell us what to do */ drm_helper_hpd_irq_event(adev_to_drm(adev)); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c index 47e9bfba0642..9f96d568acf2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c @@ -734,10 +734,8 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -919,10 +917,8 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1146,10 +1142,8 @@ out: amdgpu_connector_update_scratch_regs(connector, ret); exit: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1486,10 +1480,8 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort || connector->connector_type == DRM_MODE_CONNECTOR_eDP) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index d3a5189bd512..62d43b8cbe58 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -129,7 +129,6 @@ static int amdgpu_debugfs_process_reg_op(bool read, struct file *f, if (use_bank) { if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) || (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines)) { - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return -EINVAL; @@ -179,7 +178,6 @@ end: if (pm_pg_lock) mutex_unlock(&adev->pm.mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); @@ -255,7 +253,6 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off if (rd->id.use_grbm) { if ((rd->id.grbm.sh != 0xFFFFFFFF && rd->id.grbm.sh >= adev->gfx.config.max_sh_per_se) || (rd->id.grbm.se != 0xFFFFFFFF && rd->id.grbm.se >= adev->gfx.config.max_shader_engines)) { - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); mutex_unlock(&rd->lock); @@ -310,7 +307,6 @@ end: mutex_unlock(&rd->lock); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); @@ -446,7 +442,6 @@ static ssize_t amdgpu_debugfs_gprwave_read(struct file *f, char __user *buf, siz amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, rd->id.xcc_id); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (!x) { @@ -557,7 +552,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -617,7 +611,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -676,7 +669,6 @@ static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -736,7 +728,6 @@ static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -795,7 +786,6 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -855,7 +845,6 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user * r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -1003,7 +992,6 @@ static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf, r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (r) { @@ -1094,7 +1082,6 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf, amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (!x) { @@ -1192,7 +1179,6 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf, amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); while (size) { @@ -1266,7 +1252,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_read(struct file *f, char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1315,7 +1300,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_write(struct file *f, const char r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1365,7 +1349,6 @@ static ssize_t amdgpu_debugfs_gfxoff_count_read(struct file *f, char __user *buf r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1414,7 +1397,6 @@ static ssize_t amdgpu_debugfs_gfxoff_write(struct file *f, const char __user *bu r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1460,7 +1442,6 @@ static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1501,7 +1482,6 @@ static ssize_t amdgpu_debugfs_gfxoff_status_read(struct file *f, char __user *bu r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1701,7 +1681,6 @@ static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused) up_write(&adev->reset_domain->sem); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1721,7 +1700,6 @@ static int amdgpu_debugfs_evict_vram(void *data, u64 *val) *val = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1742,7 +1720,6 @@ static int amdgpu_debugfs_evict_gtt(void *data, u64 *val) *val = amdgpu_ttm_evict_resources(adev, TTM_PL_TT); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1762,7 +1739,6 @@ static int amdgpu_debugfs_benchmark(void *data, u64 val) r = amdgpu_benchmark(adev, val); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return r; @@ -2014,7 +1990,6 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val) ret = -EINVAL; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index c74b95bd41b5..b5d34797d606 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -332,8 +332,6 @@ int amdgpu_display_crtc_set_config(struct drm_mode_set *set, if (crtc->enabled) active = true; - pm_runtime_mark_last_busy(dev->dev); - adev = drm_to_adev(dev); /* if we have active crtcs and we don't have a power ref, * take the current one diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index cee90f9e58a9..3776901bbb1b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2228,7 +2228,6 @@ static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev) adev->pdev->bus->number, i); if (p) { pm_runtime_get_sync(&p->dev); - pm_runtime_mark_last_busy(&p->dev); pm_runtime_put_autosuspend(&p->dev); pci_dev_put(p); } @@ -2474,7 +2473,6 @@ retry_init: pm_runtime_allow(ddev->dev); - pm_runtime_mark_last_busy(ddev->dev); pm_runtime_put_autosuspend(ddev->dev); pci_wake_from_d3(pdev, TRUE); @@ -2920,7 +2918,6 @@ static int amdgpu_pmops_runtime_idle(struct device *dev) ret = amdgpu_runtime_idle_check_userq(dev); done: - pm_runtime_mark_last_busy(dev); pm_runtime_autosuspend(dev); return ret; } @@ -2956,7 +2953,6 @@ long amdgpu_drm_ioctl(struct file *filp, ret = drm_ioctl(filp, cmd, arg); - pm_runtime_mark_last_busy(dev->dev); out: pm_runtime_put_autosuspend(dev->dev); return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index 1fe31d2f2706..c7843e336310 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -250,7 +250,6 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring) drv->signalled_wptr = am_fence->wptr; dma_fence_signal(fence); dma_fence_put(fence); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); } while (last_seq != seq); @@ -928,7 +927,6 @@ static int gpu_recover_get(void *data, u64 *val) *val = atomic_read(&adev->reset_domain->reset_res); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 29c927f4d6df..8b118c53f351 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -1670,7 +1670,6 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev, ret = amdgpu_gfx_run_cleaner_shader(adev, value); - pm_runtime_mark_last_busy(ddev->dev); pm_runtime_put_autosuspend(ddev->dev); if (ret) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index b3e6b3fcdf2c..6ee77f431d56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -1471,7 +1471,6 @@ error_pasid: kfree(fpriv); out_suspend: - pm_runtime_mark_last_busy(dev->dev); pm_put: pm_runtime_put_autosuspend(dev->dev); @@ -1539,7 +1538,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, kfree(fpriv); file_priv->driver_priv = NULL; - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c index 123bcf5c2bb1..bacf888735db 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c @@ -101,7 +101,6 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf, } amdgpu_gfx_off_ctrl(adev, true); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index 7f899d3e1bfa..c8b4dd3ea5c3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -4151,7 +4151,6 @@ static void amdgpu_ras_counte_dw(struct work_struct *work) atomic_set(&con->ras_ue_count, ue_count); } - pm_runtime_mark_last_busy(dev->dev); Out: pm_runtime_put_autosuspend(dev->dev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c index 41ebe690eeff..3739be1b71e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c @@ -159,7 +159,6 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u dev_err(adev->dev, "Invalid input: %s\n", str); } - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index 3b1811b110a4..13cc5a686dfd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -488,7 +488,6 @@ amdgpu_userq_destroy(struct drm_file *filp, int queue_id) amdgpu_userq_cleanup(uq_mgr, queue, queue_id); mutex_unlock(&uq_mgr->userq_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index f5d173f1ca3b..a085faac9fe1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1083,7 +1083,6 @@ static void kfd_process_destroy_pdds(struct kfd_process *p) * for auto suspend */ if (pdd->runtime_inuse) { - pm_runtime_mark_last_busy(adev_to_drm(pdd->dev->adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(pdd->dev->adev)->dev); pdd->runtime_inuse = false; } diff --git a/drivers/gpu/drm/amd/pm/amdgpu_pm.c b/drivers/gpu/drm/amd/pm/amdgpu_pm.c index c83d69994380..c88a76cce401 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c @@ -174,7 +174,6 @@ static int amdgpu_pm_get_access_if_active(struct amdgpu_device *adev) */ static inline void amdgpu_pm_put_access(struct amdgpu_device *adev) { - pm_runtime_mark_last_busy(adev->dev); pm_runtime_put_autosuspend(adev->dev); } -- cgit v1.2.3 From 1bc9d39275e08853ff15410b4d530b46b546affb Mon Sep 17 00:00:00 2001 From: Sakari Ailus Date: Mon, 27 Oct 2025 15:14:40 +0200 Subject: drm/radeon: Remove redundant pm_runtime_mark_last_busy() calls pm_runtime_put_autosuspend(), pm_runtime_put_sync_autosuspend(), pm_runtime_autosuspend() and pm_request_autosuspend() now include a call to pm_runtime_mark_last_busy(). Remove the now-redundant explicit call to pm_runtime_mark_last_busy(). Acked-by: Alex Deucher Signed-off-by: Sakari Ailus Signed-off-by: Alex Deucher --- drivers/gpu/drm/radeon/radeon_acpi.c | 1 - drivers/gpu/drm/radeon/radeon_connectors.c | 20 +++++--------------- drivers/gpu/drm/radeon/radeon_display.c | 2 -- drivers/gpu/drm/radeon/radeon_drv.c | 2 -- drivers/gpu/drm/radeon/radeon_fbdev.c | 2 -- drivers/gpu/drm/radeon/radeon_kms.c | 4 ---- 6 files changed, 5 insertions(+), 26 deletions(-) diff --git a/drivers/gpu/drm/radeon/radeon_acpi.c b/drivers/gpu/drm/radeon/radeon_acpi.c index 22ce61bdfc06..08f8ba4fd148 100644 --- a/drivers/gpu/drm/radeon/radeon_acpi.c +++ b/drivers/gpu/drm/radeon/radeon_acpi.c @@ -408,7 +408,6 @@ static int radeon_atif_handler(struct radeon_device *rdev, pm_runtime_get_sync(rdev_to_drm(rdev)->dev); /* Just fire off a uevent and let userspace tell us what to do */ drm_helper_hpd_irq_event(rdev_to_drm(rdev)); - pm_runtime_mark_last_busy(rdev_to_drm(rdev)->dev); pm_runtime_put_autosuspend(rdev_to_drm(rdev)->dev); } } diff --git a/drivers/gpu/drm/radeon/radeon_connectors.c b/drivers/gpu/drm/radeon/radeon_connectors.c index 9f6a3df951ba..012d8b2295b8 100644 --- a/drivers/gpu/drm/radeon/radeon_connectors.c +++ b/drivers/gpu/drm/radeon/radeon_connectors.c @@ -875,10 +875,8 @@ radeon_lvds_detect(struct drm_connector *connector, bool force) radeon_connector_update_scratch_regs(connector, ret); - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1066,10 +1064,8 @@ radeon_vga_detect(struct drm_connector *connector, bool force) radeon_connector_update_scratch_regs(connector, ret); out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1154,10 +1150,8 @@ radeon_tv_detect(struct drm_connector *connector, bool force) ret = radeon_connector_analog_encoder_conflict_solve(connector, encoder, ret, false); radeon_connector_update_scratch_regs(connector, ret); - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1402,10 +1396,8 @@ out: } exit: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1714,10 +1706,8 @@ radeon_dp_detect(struct drm_connector *connector, bool force) } out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c index 351b9dfcdad8..35fb99bcd9a7 100644 --- a/drivers/gpu/drm/radeon/radeon_display.c +++ b/drivers/gpu/drm/radeon/radeon_display.c @@ -644,8 +644,6 @@ radeon_crtc_set_config(struct drm_mode_set *set, if (crtc->enabled) active = true; - pm_runtime_mark_last_busy(dev->dev); - rdev = dev->dev_private; /* if we have active crtcs and we don't have a power ref, take the current one */ diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index 350f88af888d..26ad9adc5d8c 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -462,7 +462,6 @@ static int radeon_pmops_runtime_idle(struct device *dev) } } - pm_runtime_mark_last_busy(dev); pm_runtime_autosuspend(dev); /* we don't want the main rpm_idle to call suspend - we want to autosuspend */ return 1; @@ -484,7 +483,6 @@ long radeon_drm_ioctl(struct file *filp, ret = drm_ioctl(filp, cmd, arg); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return ret; } diff --git a/drivers/gpu/drm/radeon/radeon_fbdev.c b/drivers/gpu/drm/radeon/radeon_fbdev.c index dc81b0c2dbff..4734dbd79c60 100644 --- a/drivers/gpu/drm/radeon/radeon_fbdev.c +++ b/drivers/gpu/drm/radeon/radeon_fbdev.c @@ -154,7 +154,6 @@ static int radeon_fbdev_fb_open(struct fb_info *info, int user) return 0; err_pm_runtime_mark_last_busy: - pm_runtime_mark_last_busy(rdev_to_drm(rdev)->dev); pm_runtime_put_autosuspend(rdev_to_drm(rdev)->dev); return ret; } @@ -164,7 +163,6 @@ static int radeon_fbdev_fb_release(struct fb_info *info, int user) struct drm_fb_helper *fb_helper = info->par; struct radeon_device *rdev = fb_helper->dev->dev_private; - pm_runtime_mark_last_busy(rdev_to_drm(rdev)->dev); pm_runtime_put_autosuspend(rdev_to_drm(rdev)->dev); return 0; diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index ba1446acd703..7cbe02ffb193 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -169,7 +169,6 @@ int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags) pm_runtime_set_autosuspend_delay(dev->dev, 5000); pm_runtime_set_active(dev->dev); pm_runtime_allow(dev->dev); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); } @@ -676,7 +675,6 @@ int radeon_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) file_priv->driver_priv = fpriv; } - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -686,7 +684,6 @@ err_fpriv: kfree(fpriv); err_suspend: - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return r; } @@ -736,7 +733,6 @@ void radeon_driver_postclose_kms(struct drm_device *dev, kfree(fpriv); file_priv->driver_priv = NULL; } - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); } -- cgit v1.2.3