diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c | 23 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h | 2 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 16 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 6 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c | 71 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c | 220 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 4 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 5 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c | 10 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c | 30 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c | 4 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_lockdep.c | 103 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 18 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c | 50 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 91 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 2 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 39 | ||||
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/ih_v6_1.c | 2 |
18 files changed, 451 insertions, 245 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 9783a3cefb04..da325863ad76 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -558,7 +558,7 @@ uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct amdgpu_device *adev) int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd, struct amdgpu_device **dmabuf_adev, - uint64_t *bo_size, void *metadata_buffer, + uint64_t *bo_size, void **metadata_buffer, size_t buffer_size, uint32_t *metadata_size, uint32_t *flags, int8_t *xcp_id) { @@ -593,9 +593,24 @@ int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd, *dmabuf_adev = adev; if (bo_size) *bo_size = amdgpu_bo_size(bo); - if (metadata_buffer) - r = amdgpu_bo_get_metadata(bo, metadata_buffer, buffer_size, - metadata_size, &metadata_flags); + if (metadata_buffer) { + /* first get metadata_size by buffer = NULL */ + r = amdgpu_bo_get_metadata(bo, NULL, 0, + metadata_size, NULL); + + /* user buf_size is bigger than bo metadata_size + * allocate a buf at kernel space and copy */ + if (*metadata_size <= buffer_size) { + *metadata_buffer = kzalloc(*metadata_size, GFP_KERNEL); + + if (!*metadata_buffer) + return -ENOMEM; + + r = amdgpu_bo_get_metadata(bo, *metadata_buffer, *metadata_size, + NULL, &metadata_flags); + } else + r = -EINVAL; + } if (flags) { *flags = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ? KFD_IOC_ALLOC_MEM_FLAGS_VRAM diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 5333e052d56d..e443a7277299 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -262,7 +262,7 @@ uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct amdgpu_device *adev); uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct amdgpu_device *adev); int amdgpu_amdkfd_get_dmabuf_info(struct amdgpu_device *adev, int dma_buf_fd, struct amdgpu_device **dmabuf_adev, - uint64_t *bo_size, void *metadata_buffer, + uint64_t *bo_size, void **metadata_buffer, size_t buffer_size, uint32_t *metadata_size, uint32_t *flags, int8_t *xcp_id); int amdgpu_amdkfd_get_pcie_bandwidth_mbytes(struct amdgpu_device *adev, bool is_min); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index d54794e5b18b..35fe2c974699 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1914,13 +1914,6 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( mutex_lock(&mem->lock); - /* Unpin MMIO/DOORBELL BO's that were pinned during allocation */ - if (mem->alloc_flags & - (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL | - KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) { - amdgpu_amdkfd_gpuvm_unpin_bo(mem->bo); - } - mapped_to_gpu_memory = mem->mapped_to_gpu_memory; is_imported = mem->is_imported; mutex_unlock(&mem->lock); @@ -1934,6 +1927,15 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( return -EBUSY; } + /* At this point the BO is guaranteed to be freed, so unpin the + * MMIO/DOORBELL BOs that were pinned during allocation. + */ + if (mem->alloc_flags & + (KFD_IOC_ALLOC_MEM_FLAGS_DOORBELL | + KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP)) { + amdgpu_amdkfd_gpuvm_unpin_bo(mem->bo); + } + /* Make sure restore workers don't access the BO any more */ mutex_lock(&process_info->lock); if (!list_empty(&mem->validate_list)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 115b134b4cd1..c2e6495a28bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -247,13 +247,17 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p, goto free_partial_kdata; break; + case AMDGPU_CHUNK_ID_CP_GFX_SHADOW: + if (size < sizeof(struct drm_amdgpu_cs_chunk_cp_gfx_shadow)) + goto free_partial_kdata; + break; + case AMDGPU_CHUNK_ID_DEPENDENCIES: case AMDGPU_CHUNK_ID_SYNCOBJ_IN: case AMDGPU_CHUNK_ID_SYNCOBJ_OUT: case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES: case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT: case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL: - case AMDGPU_CHUNK_ID_CP_GFX_SHADOW: break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index 0d7f6cd74f79..ce35b415093d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -326,7 +326,6 @@ static int amdgpu_ctx_init(struct amdgpu_ctx_mgr *mgr, int32_t priority, struct drm_file *filp, struct amdgpu_ctx *ctx) { struct amdgpu_fpriv *fpriv = filp->driver_priv; - u32 current_stable_pstate; int r; r = amdgpu_ctx_priority_permit(filp, priority); @@ -344,36 +343,21 @@ static int amdgpu_ctx_init(struct amdgpu_ctx_mgr *mgr, int32_t priority, ctx->generation = amdgpu_vm_generation(mgr->adev, &fpriv->vm); ctx->init_priority = priority; ctx->override_priority = AMDGPU_CTX_PRIORITY_UNSET; - - r = amdgpu_ctx_get_stable_pstate(ctx, ¤t_stable_pstate); - if (r) - return r; - - if (mgr->adev->pm.stable_pstate_ctx) - ctx->stable_pstate = mgr->adev->pm.stable_pstate_ctx->stable_pstate; - else - ctx->stable_pstate = current_stable_pstate; + ctx->stable_pstate = AMDGPU_CTX_STABLE_PSTATE_NONE; return 0; } -static int amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx, - u32 stable_pstate) +static int __amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx, + u32 stable_pstate) { struct amdgpu_device *adev = ctx->mgr->adev; enum amd_dpm_forced_level level; + struct amdgpu_ctx *current_ctx; u32 current_stable_pstate; - int r; + int r = 0; - mutex_lock(&adev->pm.stable_pstate_ctx_lock); - if (adev->pm.stable_pstate_ctx && adev->pm.stable_pstate_ctx != ctx) { - r = -EBUSY; - goto done; - } - - r = amdgpu_ctx_get_stable_pstate(ctx, ¤t_stable_pstate); - if (r || (stable_pstate == current_stable_pstate)) - goto done; + lockdep_assert_held(&adev->pm.stable_pstate_ctx_lock); switch (stable_pstate) { case AMDGPU_CTX_STABLE_PSTATE_NONE: @@ -392,17 +376,41 @@ static int amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx, level = AMD_DPM_FORCED_LEVEL_PROFILE_PEAK; break; default: - r = -EINVAL; - goto done; + return -EINVAL; } + current_ctx = adev->pm.stable_pstate_ctx; + if (current_ctx && current_ctx != ctx) + return -EBUSY; + + r = amdgpu_ctx_get_stable_pstate(ctx, ¤t_stable_pstate); + if (r || current_stable_pstate == stable_pstate) + return r; + r = amdgpu_dpm_force_performance_level(adev, level); + if (r) + return r; - if (level == AMD_DPM_FORCED_LEVEL_AUTO) - adev->pm.stable_pstate_ctx = NULL; - else + if (!current_ctx) { adev->pm.stable_pstate_ctx = ctx; -done: + /* + * Serialized by context taking ownership for the first time + * while holding adev->pm.stable_pstate_ctx_lock). + */ + WRITE_ONCE(ctx->stable_pstate, current_stable_pstate); + } + + return 0; +} + +static int amdgpu_ctx_set_stable_pstate(struct amdgpu_ctx *ctx, + u32 stable_pstate) +{ + struct amdgpu_device *adev = ctx->mgr->adev; + int r; + + mutex_lock(&adev->pm.stable_pstate_ctx_lock); + r = __amdgpu_ctx_set_stable_pstate(ctx, stable_pstate); mutex_unlock(&adev->pm.stable_pstate_ctx_lock); return r; @@ -428,7 +436,12 @@ static void amdgpu_ctx_fini(struct kref *ref) } if (drm_dev_enter(adev_to_drm(adev), &idx)) { - amdgpu_ctx_set_stable_pstate(ctx, ctx->stable_pstate); + mutex_lock(&adev->pm.stable_pstate_ctx_lock); + if (adev->pm.stable_pstate_ctx == ctx) { + __amdgpu_ctx_set_stable_pstate(ctx, ctx->stable_pstate); + adev->pm.stable_pstate_ctx = NULL; + } + mutex_unlock(&adev->pm.stable_pstate_ctx_lock); drm_dev_exit(idx); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c index 27830518a230..e77db76b48b8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c @@ -22,8 +22,9 @@ * */ -#include <generated/utsrelease.h> #include <linux/devcoredump.h> +#include <linux/utsname.h> +#include <drm/drm_exec.h> #include "amdgpu_dev_coredump.h" #include "atom.h" @@ -207,28 +208,143 @@ static void amdgpu_devcoredump_fw_info(struct amdgpu_device *adev, } } +static void +amdgpu_devcoredump_print_ibs(struct drm_printer *p, + struct amdgpu_coredump_info *coredump, + bool sizing_pass) +{ + struct amdgpu_device *adev = coredump->adev; + struct amdgpu_bo_va_mapping *mapping; + struct amdgpu_bo *abo; + struct drm_exec exec; + struct amdgpu_vm *vm; + u32 *ib_content; + u64 va_start, offset; + u8 *kptr; + u32 off; + int r; + + /* + * On the sizing pass there is no VM to look up and no BO to lock; the + * size estimate doesn't depend on whether the IB BOs are reachable. + * Just emit the per-IB headers (the content is not written anywhere). + */ + if (sizing_pass) { + for (int i = 0; i < coredump->num_ibs; i++) { + drm_printf(p, "\nIB #%d 0x%llx %d dw\n", i, + coredump->ibs[i].gpu_addr, + coredump->ibs[i].ib_size_dw); + } + return; + } + + /* + * Lock the VM root PD and every IB BO together in a single drm_exec + * ticket. Reserving the IB BOs one by one while the root PD is held + * would be a recursive reservation_ww_class_mutex acquire without a + * ww_acquire_ctx, which trips lockdep and self-deadlocks for IB BOs + * that share their dma_resv with the root PD (always-valid BOs). + */ + drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 1 + coredump->num_ibs); + drm_exec_until_all_locked(&exec) { + vm = amdgpu_vm_lock_by_pasid(adev, coredump->pasid, &exec); + if (!vm) + goto unlock; + + for (int i = 0; i < coredump->num_ibs; i++) { + u64 pfn = (coredump->ibs[i].gpu_addr & + AMDGPU_GMC_HOLE_MASK) / AMDGPU_GPU_PAGE_SIZE; + + mapping = amdgpu_vm_bo_lookup_mapping(vm, pfn); + if (!mapping) + continue; + + abo = mapping->bo_va->base.bo; + r = drm_exec_lock_obj(&exec, &abo->tbo.base); + drm_exec_retry_on_contention(&exec); + if (r) + goto unlock; + } + } + + for (int i = 0; i < coredump->num_ibs; i++) { + bool emit_content = false; + + ib_content = kvmalloc_array(coredump->ibs[i].ib_size_dw, 4, + GFP_KERNEL); + if (!ib_content) + continue; + + va_start = coredump->ibs[i].gpu_addr & AMDGPU_GMC_HOLE_MASK; + mapping = amdgpu_vm_bo_lookup_mapping(vm, + va_start / AMDGPU_GPU_PAGE_SIZE); + if (!mapping) + goto output_ib_content; + + abo = mapping->bo_va->base.bo; + offset = va_start - mapping->start * AMDGPU_GPU_PAGE_SIZE; + + if (abo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) { + struct amdgpu_res_cursor cursor; + + off = 0; + + if (abo->tbo.resource->mem_type != TTM_PL_VRAM) + goto output_ib_content; + + amdgpu_res_first(abo->tbo.resource, offset, + coredump->ibs[i].ib_size_dw * 4, &cursor); + while (cursor.remaining) { + amdgpu_device_mm_access(adev, cursor.start / 4, + &ib_content[off], cursor.size / 4, + false); + off += cursor.size; + amdgpu_res_next(&cursor, cursor.size); + } + emit_content = true; + } else { + r = ttm_bo_kmap(&abo->tbo, 0, PFN_UP(abo->tbo.base.size), + &abo->kmap); + if (r) + goto output_ib_content; + + kptr = amdgpu_bo_kptr(abo); + kptr += offset; + memcpy(ib_content, kptr, coredump->ibs[i].ib_size_dw * 4); + + amdgpu_bo_kunmap(abo); + emit_content = true; + } + +output_ib_content: + drm_printf(p, "\nIB #%d 0x%llx %d dw\n", i, + coredump->ibs[i].gpu_addr, coredump->ibs[i].ib_size_dw); + if (emit_content) { + for (int j = 0; j < coredump->ibs[i].ib_size_dw; j++) + drm_printf(p, "0x%08x\n", ib_content[j]); + } + kvfree(ib_content); + } + +unlock: + drm_exec_fini(&exec); +} + static ssize_t amdgpu_devcoredump_format(char *buffer, size_t count, struct amdgpu_coredump_info *coredump) { - struct amdgpu_device *adev = coredump->adev; struct drm_printer p; struct drm_print_iterator iter; struct amdgpu_vm_fault_info *fault_info; - struct amdgpu_bo_va_mapping *mapping; struct amdgpu_ip_block *ip_block; - struct amdgpu_res_cursor cursor; - struct amdgpu_bo *abo, *root; - uint64_t va_start, offset; struct amdgpu_ring *ring; - struct amdgpu_vm *vm; - u32 *ib_content; - uint8_t *kptr; - int ver, i, j, r; + int ver, i, j; u32 ring_idx, off; bool sizing_pass; sizing_pass = buffer == NULL; iter.data = buffer; + iter.start = 0; iter.offset = 0; iter.remain = count; @@ -236,7 +352,7 @@ amdgpu_devcoredump_format(char *buffer, size_t count, struct amdgpu_coredump_inf drm_printf(&p, "**** AMDGPU Device Coredump ****\n"); drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n"); - drm_printf(&p, "kernel: " UTS_RELEASE "\n"); + drm_printf(&p, "kernel: %s\n", init_utsname()->release); drm_printf(&p, "module: " KBUILD_MODNAME "\n"); drm_printf(&p, "time: %ptSp\n", &coredump->reset_time); @@ -342,86 +458,8 @@ amdgpu_devcoredump_format(char *buffer, size_t count, struct amdgpu_coredump_inf else if (coredump->reset_vram_lost) drm_printf(&p, "VRAM is lost due to GPU reset!\n"); - if (coredump->num_ibs) { - /* Don't try to lookup the VM or map the BOs when calculating the - * size required to store the devcoredump. - */ - if (sizing_pass) - vm = NULL; - else - vm = amdgpu_vm_lock_by_pasid(adev, &root, coredump->pasid); - - for (int i = 0; i < coredump->num_ibs && (sizing_pass || vm); i++) { - ib_content = kvmalloc_array(coredump->ibs[i].ib_size_dw, 4, - GFP_KERNEL); - if (!ib_content) - continue; - - /* vm=NULL can only happen when 'sizing_pass' is true. Skip to the - * drm_printf() calls (ib_content doesn't need to be initialized - * as its content won't be written anywhere). - */ - if (!vm) - goto output_ib_content; - - va_start = coredump->ibs[i].gpu_addr & AMDGPU_GMC_HOLE_MASK; - mapping = amdgpu_vm_bo_lookup_mapping(vm, va_start / AMDGPU_GPU_PAGE_SIZE); - if (!mapping) - goto free_ib_content; - - offset = va_start - (mapping->start * AMDGPU_GPU_PAGE_SIZE); - abo = amdgpu_bo_ref(mapping->bo_va->base.bo); - r = amdgpu_bo_reserve(abo, false); - if (r) - goto free_ib_content; - - if (abo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) { - off = 0; - - if (abo->tbo.resource->mem_type != TTM_PL_VRAM) - goto unreserve_abo; - - amdgpu_res_first(abo->tbo.resource, offset, - coredump->ibs[i].ib_size_dw * 4, - &cursor); - while (cursor.remaining) { - amdgpu_device_mm_access(adev, cursor.start / 4, - &ib_content[off], cursor.size / 4, - false); - off += cursor.size; - amdgpu_res_next(&cursor, cursor.size); - } - } else { - r = ttm_bo_kmap(&abo->tbo, 0, - PFN_UP(abo->tbo.base.size), - &abo->kmap); - if (r) - goto unreserve_abo; - - kptr = amdgpu_bo_kptr(abo); - kptr += offset; - memcpy(ib_content, kptr, - coredump->ibs[i].ib_size_dw * 4); - - amdgpu_bo_kunmap(abo); - } - -output_ib_content: - drm_printf(&p, "\nIB #%d 0x%llx %d dw\n", - i, coredump->ibs[i].gpu_addr, coredump->ibs[i].ib_size_dw); - for (int j = 0; j < coredump->ibs[i].ib_size_dw; j++) - drm_printf(&p, "0x%08x\n", ib_content[j]); -unreserve_abo: - if (vm) - amdgpu_bo_unreserve(abo); -free_ib_content: - kvfree(ib_content); - } - if (vm) { - amdgpu_bo_unreserve(root); - amdgpu_bo_unref(&root); - } - } + if (coredump->num_ibs) + amdgpu_devcoredump_print_ibs(&p, coredump, sizing_pass); return count - iter.remain; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 942f0251c748..211d30f03d25 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -3043,7 +3043,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry); for (i = adev->num_ip_blocks - 1; i >= 0; i--) { - if (!adev->ip_blocks[i].status.valid) + if (!adev->ip_blocks[i].status.valid || !adev->ip_blocks[i].status.hw) continue; /* displays are handled in phase1 */ if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) @@ -3771,6 +3771,8 @@ int amdgpu_device_init(struct amdgpu_device *adev, mutex_init(&adev->gfx.workload_profile_mutex); mutex_init(&adev->vcn.workload_profile_mutex); + spin_lock_init(&adev->irq.lock); + amdgpu_device_init_apu_flags(adev); r = amdgpu_device_check_arguments(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 212c14d99f6b..76da3f932f24 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -1094,6 +1094,11 @@ int amdgpu_gem_op_ioctl(struct drm_device *dev, void *data, * If that number is larger than the size of the array, the ioctl must * be retried. */ + if (!bo_va) { + r = -ENOENT; + goto out_exec; + } + if (args->num_entries > INT_MAX / sizeof(*vm_entries)) { r = -EINVAL; goto out_exec; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 1e190fb54a97..85372af1216d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -1664,12 +1664,13 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; struct drm_gpu_scheduler *sched = &ring->sched; struct drm_sched_entity entity; + unsigned int ib_size_dw = 16; static atomic_t counter; struct dma_fence *f; struct amdgpu_job *job; struct amdgpu_ib *ib; void *owner; - int i, r; + int r; /* Initialize the scheduler entity */ r = drm_sched_entity_init(&entity, DRM_SCHED_PRIORITY_NORMAL, @@ -1687,7 +1688,7 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring) owner = (void *)(unsigned long)atomic_inc_return(&counter); r = amdgpu_job_alloc_with_ib(ring->adev, &entity, owner, - 64, 0, &job, + ib_size_dw * sizeof(uint32_t), 0, &job, AMDGPU_KERNEL_JOB_ID_CLEANER_SHADER); if (r) goto err; @@ -1697,9 +1698,8 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring) job->run_cleaner_shader = true; ib = &job->ibs[0]; - for (i = 0; i <= ring->funcs->align_mask; ++i) - ib->ptr[i] = ring->funcs->nop; - ib->length_dw = ring->funcs->align_mask + 1; + memset32(ib->ptr, ring->funcs->nop, ib_size_dw); + ib->length_dw = ib_size_dw; f = amdgpu_job_submit(job); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index d23a91d029aa..0ea32561c4bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -272,7 +272,20 @@ static bool amdgpu_gtt_mgr_intersects(struct ttm_resource_manager *man, const struct ttm_place *place, size_t size) { - return !place->lpfn || amdgpu_gtt_mgr_has_gart_addr(res); + const struct drm_mm_node *const node = &to_ttm_range_mgr_node(res)->mm_nodes[0]; + const u32 num_pages = PFN_UP(size); + + if (!place->lpfn) + return true; + + if (!amdgpu_gtt_mgr_has_gart_addr(res)) + return false; + + if (place->fpfn >= (node->start + num_pages) || + (place->lpfn && place->lpfn <= node->start)) + return false; + + return true; } /** @@ -290,7 +303,20 @@ static bool amdgpu_gtt_mgr_compatible(struct ttm_resource_manager *man, const struct ttm_place *place, size_t size) { - return !place->lpfn || amdgpu_gtt_mgr_has_gart_addr(res); + const struct drm_mm_node *const node = &to_ttm_range_mgr_node(res)->mm_nodes[0]; + const u32 num_pages = PFN_UP(size); + + if (!place->lpfn) + return true; + + if (!amdgpu_gtt_mgr_has_gart_addr(res)) + return false; + + if (node->start < place->fpfn || + (place->lpfn && (node->start + num_pages) > place->lpfn)) + return false; + + return true; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 254a4e983f40..53be764968e4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -309,8 +309,6 @@ int amdgpu_irq_init(struct amdgpu_device *adev) unsigned int irq, flags; int r; - spin_lock_init(&adev->irq.lock); - /* Enable MSI if not disabled by module parameter */ adev->irq.msi_enabled = false; @@ -547,7 +545,7 @@ void amdgpu_irq_delegate(struct amdgpu_device *adev, unsigned int num_dw) { amdgpu_ih_ring_write(adev, &adev->irq.ih_soft, entry->iv_entry, num_dw); - schedule_work(&adev->irq.ih_soft_work); + queue_work(system_unbound_wq, &adev->irq.ih_soft_work); } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_lockdep.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_lockdep.c index d5d71fd7c70d..61450af539a6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_lockdep.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_lockdep.c @@ -16,6 +16,17 @@ #ifdef CONFIG_LOCKDEP +struct amdgpu_lockdep_dummy_locks { + struct mutex reset_lock; + struct mutex userq_sch_mutex; + struct mutex userq_mutex; + struct mutex notifier_lock; + struct mutex vram_lock; + struct mutex srbm_mutex; + struct mutex grbm_idx_mutex; + spinlock_t mmio_idx_lock; +}; + /* Lock class keys for associating with real driver locks */ static struct lock_class_key amdgpu_userq_sch_mutex_key; static struct lock_class_key amdgpu_userq_mutex_key; @@ -84,72 +95,65 @@ void amdgpu_lockdep_set_class(struct amdgpu_device *adev) int amdgpu_lockdep_init(void) { struct amdgpu_reset_domain *reset_domain = NULL; - struct amdgpu_reset_control reset_ctl; - struct mutex userq_sch_mutex; - struct mutex userq_mutex; - struct mutex notifier_lock; - struct mutex vram_lock; - struct mutex srbm_mutex; - struct mutex grbm_idx_mutex; - spinlock_t mmio_idx_lock; + struct amdgpu_lockdep_dummy_locks *locks; unsigned long flags; + locks = kzalloc(sizeof(*locks), GFP_KERNEL); + if (!locks) + return -ENOMEM; + /* * Initialize dummy reset domain */ reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE, "lockdep_test"); - if (!reset_domain) + if (!reset_domain) { + kfree(locks); return -ENOMEM; - + } /* Initialize dummy locks */ - mutex_init(&userq_sch_mutex); - mutex_init(&userq_mutex); - mutex_init(¬ifier_lock); - mutex_init(&vram_lock); - mutex_init(&reset_ctl.reset_lock); - mutex_init(&srbm_mutex); - mutex_init(&grbm_idx_mutex); - spin_lock_init(&mmio_idx_lock); + mutex_init(&locks->userq_sch_mutex); + mutex_init(&locks->userq_mutex); + mutex_init(&locks->notifier_lock); + mutex_init(&locks->vram_lock); + mutex_init(&locks->reset_lock); + mutex_init(&locks->srbm_mutex); + mutex_init(&locks->grbm_idx_mutex); + spin_lock_init(&locks->mmio_idx_lock); /* * Associate dummy locks with the same class keys used for real * driver locks. This ensures lockdep connects the ordering learned * here with the actual locks used at runtime. */ - lockdep_set_class(&userq_sch_mutex, &amdgpu_userq_sch_mutex_key); - lockdep_set_class(&userq_mutex, &amdgpu_userq_mutex_key); - lockdep_set_class(¬ifier_lock, &amdgpu_notifier_lock_key); - lockdep_set_class(&vram_lock, &amdgpu_vram_lock_key); + lockdep_set_class(&locks->userq_sch_mutex, &amdgpu_userq_sch_mutex_key); + lockdep_set_class(&locks->userq_mutex, &amdgpu_userq_mutex_key); + lockdep_set_class(&locks->notifier_lock, &amdgpu_notifier_lock_key); + lockdep_set_class(&locks->vram_lock, &amdgpu_vram_lock_key); lockdep_set_class(&reset_domain->sem, &amdgpu_reset_sem_key); - lockdep_set_class(&reset_ctl.reset_lock, &amdgpu_reset_lock_key); - lockdep_set_class(&srbm_mutex, &amdgpu_srbm_lock_key); - lockdep_set_class(&grbm_idx_mutex, &amdgpu_grbm_lock_key); - lockdep_set_class(&mmio_idx_lock, &amdgpu_mmio_lock_key); - + lockdep_set_class(&locks->reset_lock, &amdgpu_reset_lock_key); + lockdep_set_class(&locks->srbm_mutex, &amdgpu_srbm_lock_key); + lockdep_set_class(&locks->grbm_idx_mutex, &amdgpu_grbm_lock_key); + lockdep_set_class(&locks->mmio_idx_lock, &amdgpu_mmio_lock_key); /* * Take locks in the correct order to train lockdep. * This establishes the dependency chain. */ /* Level 1: Global userq scheduler mutex (outermost) */ - mutex_lock(&userq_sch_mutex); + mutex_lock(&locks->userq_sch_mutex); /* Level 2: Per-context userq mutex */ - mutex_lock(&userq_mutex); - + mutex_lock(&locks->userq_mutex); /* Level 3: MMU notifier lock */ - mutex_lock(¬ifier_lock); - + mutex_lock(&locks->notifier_lock); /* Level 4: VRAM allocator lock */ - mutex_lock(&vram_lock); - + mutex_lock(&locks->vram_lock); /* Level 5: Reset domain semaphore */ down_read(&reset_domain->sem); /* Level 6: Reset control lock */ - mutex_lock(&reset_ctl.reset_lock); - + mutex_lock(&locks->reset_lock); /* * Mark potential memory reclaim boundary. * GPU operations might trigger memory allocation/reclaim. @@ -157,36 +161,35 @@ int amdgpu_lockdep_init(void) fs_reclaim_acquire(GFP_KERNEL); /* Level 7: SRBM register access */ - mutex_lock(&srbm_mutex); - + mutex_lock(&locks->srbm_mutex); /* Level 8: GRBM index access */ - mutex_lock(&grbm_idx_mutex); + mutex_lock(&locks->grbm_idx_mutex); /* Level 9: MMIO index access (innermost lock, spinlock) */ - spin_lock_irqsave(&mmio_idx_lock, flags); - + spin_lock_irqsave(&locks->mmio_idx_lock, flags); /* * All locks acquired in order. * Lockdep has now learned the valid dependency chain. */ /* Release in reverse order */ - spin_unlock_irqrestore(&mmio_idx_lock, flags); - mutex_unlock(&grbm_idx_mutex); - mutex_unlock(&srbm_mutex); - + spin_unlock_irqrestore(&locks->mmio_idx_lock, flags); + mutex_unlock(&locks->grbm_idx_mutex); + mutex_unlock(&locks->srbm_mutex); fs_reclaim_release(GFP_KERNEL); - mutex_unlock(&reset_ctl.reset_lock); + mutex_unlock(&locks->reset_lock); up_read(&reset_domain->sem); - mutex_unlock(&vram_lock); - mutex_unlock(¬ifier_lock); - mutex_unlock(&userq_mutex); - mutex_unlock(&userq_sch_mutex); + + mutex_unlock(&locks->vram_lock); + mutex_unlock(&locks->notifier_lock); + mutex_unlock(&locks->userq_mutex); + mutex_unlock(&locks->userq_sch_mutex); /* Cleanup */ amdgpu_reset_put_reset_domain(reset_domain); + kfree(locks); pr_info("AMDGPU: Lockdep annotations initialized (9 lock levels)\n"); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 2740de94e93c..16c060badaee 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -515,6 +515,15 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict, if (new_mem->mem_type == TTM_PL_TT || new_mem->mem_type == AMDGPU_PL_PREEMPT) { + if (old_mem && (old_mem->mem_type == TTM_PL_TT || + old_mem->mem_type == AMDGPU_PL_PREEMPT)) { + r = ttm_bo_wait_ctx(bo, ctx); + if (r) + return r; + + amdgpu_ttm_backend_unbind(bo->bdev, bo->ttm); + } + r = amdgpu_ttm_backend_bind(bo->bdev, bo->ttm, new_mem); if (r) return r; @@ -549,6 +558,15 @@ static int amdgpu_bo_move(struct ttm_buffer_object *bo, bool evict, ttm_bo_assign_mem(bo, new_mem); return 0; } + if ((old_mem->mem_type == TTM_PL_TT || + old_mem->mem_type == AMDGPU_PL_PREEMPT) && + (new_mem->mem_type == TTM_PL_TT || + new_mem->mem_type == AMDGPU_PL_PREEMPT)) { + amdgpu_bo_move_notify(bo, evict, new_mem); + ttm_resource_free(bo, &bo->resource); + ttm_bo_assign_mem(bo, new_mem); + return 0; + } if (old_mem->mem_type == AMDGPU_PL_GDS || old_mem->mem_type == AMDGPU_PL_GWS || diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c index 3a3bc0d370fa..480bf88def46 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c @@ -135,7 +135,7 @@ MODULE_FIRMWARE(FIRMWARE_VEGA12); MODULE_FIRMWARE(FIRMWARE_VEGA20); static void amdgpu_uvd_idle_work_handler(struct work_struct *work); -static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *abo); +static void amdgpu_uvd_force_into_vcpu_segment(struct amdgpu_bo *abo); static int amdgpu_uvd_create_msg_bo_helper(struct amdgpu_device *adev, uint32_t size, @@ -158,7 +158,7 @@ static int amdgpu_uvd_create_msg_bo_helper(struct amdgpu_device *adev, amdgpu_bo_kunmap(bo); amdgpu_bo_unpin(bo); amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_VRAM); - amdgpu_uvd_force_into_uvd_segment(bo); + amdgpu_uvd_force_into_vcpu_segment(bo); r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx); if (r) goto err; @@ -188,6 +188,7 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) const struct common_firmware_header *hdr; unsigned int family_id; int i, j, r; + u32 vcpu_bo_domain; INIT_DELAYED_WORK(&adev->uvd.idle_work, amdgpu_uvd_idle_work_handler); @@ -319,12 +320,20 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) bo_size += AMDGPU_GPU_PAGE_ALIGN(le32_to_cpu(hdr->ucode_size_bytes) + 8); + /* UVD 5.0 and newer HW can use 64 bit addressing. */ + adev->uvd.address_64_bit = + !amdgpu_device_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0); + + vcpu_bo_domain = AMDGPU_GEM_DOMAIN_VRAM; + if (adev->uvd.address_64_bit) + vcpu_bo_domain |= AMDGPU_GEM_DOMAIN_GTT; + for (j = 0; j < adev->uvd.num_uvd_inst; j++) { if (adev->uvd.harvest_config & (1 << j)) continue; + r = amdgpu_bo_create_kernel(adev, bo_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM | - AMDGPU_GEM_DOMAIN_GTT, + vcpu_bo_domain, &adev->uvd.inst[j].vcpu_bo, &adev->uvd.inst[j].gpu_addr, &adev->uvd.inst[j].cpu_addr); @@ -339,10 +348,6 @@ int amdgpu_uvd_sw_init(struct amdgpu_device *adev) adev->uvd.filp[i] = NULL; } - /* from uvd v5.0 HW addressing capacity increased to 64 bits */ - if (!amdgpu_device_ip_block_version_cmp(adev, AMD_IP_BLOCK_TYPE_UVD, 5, 0)) - adev->uvd.address_64_bit = true; - r = amdgpu_uvd_create_msg_bo_helper(adev, 128 << 10, &adev->uvd.ib_bo); if (r) return r; @@ -545,6 +550,24 @@ void amdgpu_uvd_free_handles(struct amdgpu_device *adev, struct drm_file *filp) } } +static void amdgpu_uvd_force_into_vcpu_segment(struct amdgpu_bo *bo) +{ + struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + struct amdgpu_bo *vcpu_bo = adev->uvd.inst[0].vcpu_bo; + struct amdgpu_res_cursor vcpu_cur; + + amdgpu_res_first(vcpu_bo->tbo.resource, 0, + amdgpu_bo_size(vcpu_bo), &vcpu_cur); + + bo->placement.num_placement = 1; + bo->placement.placement = &bo->placements[0]; + bo->placements[0].fpfn = ALIGN_DOWN(vcpu_cur.start, SZ_256M) >> PAGE_SHIFT; + bo->placements[0].lpfn = bo->placements[0].fpfn + (SZ_256M >> PAGE_SHIFT); + bo->placements[0].mem_type = vcpu_bo->tbo.resource->mem_type; + if (bo->placements[0].mem_type == TTM_PL_VRAM) + bo->placements[0].flags |= TTM_PL_FLAG_CONTIGUOUS; +} + static void amdgpu_uvd_force_into_uvd_segment(struct amdgpu_bo *abo) { int i; @@ -595,13 +618,10 @@ static int amdgpu_uvd_cs_pass1(struct amdgpu_uvd_cs_ctx *ctx) if (!ctx->parser->adev->uvd.address_64_bit) { /* check if it's a message or feedback command */ cmd = amdgpu_ib_get_value(ctx->ib, ctx->idx) >> 1; - if (cmd == 0x0 || cmd == 0x3) { - /* yes, force it into VRAM */ - uint32_t domain = AMDGPU_GEM_DOMAIN_VRAM; - - amdgpu_bo_placement_from_domain(bo, domain); - } - amdgpu_uvd_force_into_uvd_segment(bo); + if (cmd == 0x0 || cmd == 0x3) + amdgpu_uvd_force_into_vcpu_segment(bo); + else + amdgpu_uvd_force_into_uvd_segment(bo); r = ttm_bo_validate(&bo->tbo, &bo->placement, &tctx); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 7d51880b4860..fee4c94c2585 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -2920,47 +2920,56 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) } /** - * amdgpu_vm_lock_by_pasid - return an amdgpu_vm and its root bo from a pasid, if possible. + * amdgpu_vm_lock_by_pasid - look up a VM by PASID and lock its root PD * @adev: amdgpu device pointer - * @root: root BO of the VM * @pasid: PASID of the VM - * The caller needs to unreserve and unref the root bo on success. + * @exec: drm_exec context to lock the root PD in + * + * Must be called from within a drm_exec_until_all_locked() loop; the caller + * runs drm_exec_retry_on_contention() afterwards. The drm_exec context holds + * a reference on the root BO until it is finalised. + * + * Return: the VM on success, or NULL if the PASID has no VM, the VM is being + * torn down, or locking the root PD failed. */ struct amdgpu_vm *amdgpu_vm_lock_by_pasid(struct amdgpu_device *adev, - struct amdgpu_bo **root, u32 pasid) + u32 pasid, struct drm_exec *exec) { unsigned long irqflags; + struct amdgpu_bo *root; struct amdgpu_vm *vm; int r; xa_lock_irqsave(&adev->vm_manager.pasids, irqflags); vm = xa_load(&adev->vm_manager.pasids, pasid); - *root = vm ? amdgpu_bo_ref(vm->root.bo) : NULL; + root = vm ? amdgpu_bo_ref(vm->root.bo) : NULL; xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags); - if (!*root) + if (!root) return NULL; - r = amdgpu_bo_reserve(*root, true); - if (r) - goto error_unref; + r = drm_exec_lock_obj(exec, &root->tbo.base); + if (r) { + amdgpu_bo_unref(&root); + return NULL; + } /* Double check that the VM still exists */ xa_lock_irqsave(&adev->vm_manager.pasids, irqflags); vm = xa_load(&adev->vm_manager.pasids, pasid); - if (vm && vm->root.bo != *root) + if (vm && vm->root.bo != root) vm = NULL; xa_unlock_irqrestore(&adev->vm_manager.pasids, irqflags); - if (!vm) - goto error_unlock; + if (!vm) { + drm_exec_unlock_obj(exec, &root->tbo.base); + amdgpu_bo_unref(&root); + return NULL; + } - return vm; -error_unlock: - amdgpu_bo_unreserve(*root); + /* The drm_exec context holds its own reference on the root BO. */ + amdgpu_bo_unref(&root); -error_unref: - amdgpu_bo_unref(root); - return NULL; + return vm; } /** @@ -2982,33 +2991,49 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, uint64_t ts, bool write_fault) { bool is_compute_context = false; - struct amdgpu_bo *root; + struct drm_exec exec; uint64_t value, flags; struct amdgpu_vm *vm; int r; - vm = amdgpu_vm_lock_by_pasid(adev, &root, pasid); - if (!vm) + drm_exec_init(&exec, 0, 1); + drm_exec_until_all_locked(&exec) { + vm = amdgpu_vm_lock_by_pasid(adev, pasid, &exec); + drm_exec_retry_on_contention(&exec); + if (!vm) + break; + } + if (!vm) { + drm_exec_fini(&exec); return false; + } is_compute_context = vm->is_compute_context; if (is_compute_context) { - /* Unreserve root since svm_range_restore_pages might try to reserve it. */ - /* TODO: rework svm_range_restore_pages so that this isn't necessary. */ - amdgpu_bo_unreserve(root); + /* Release the root PD lock since svm_range_restore_pages + * might try to take it. + * TODO: rework svm_range_restore_pages so that this isn't + * necessary. + */ + drm_exec_fini(&exec); if (!svm_range_restore_pages(adev, pasid, vmid, - node_id, addr >> PAGE_SHIFT, ts, write_fault)) { - amdgpu_bo_unref(&root); + node_id, addr >> PAGE_SHIFT, ts, write_fault)) return true; - } - amdgpu_bo_unref(&root); /* Re-acquire the VM lock, could be that the VM was freed in between. */ - vm = amdgpu_vm_lock_by_pasid(adev, &root, pasid); - if (!vm) + drm_exec_init(&exec, 0, 1); + drm_exec_until_all_locked(&exec) { + vm = amdgpu_vm_lock_by_pasid(adev, pasid, &exec); + drm_exec_retry_on_contention(&exec); + if (!vm) + break; + } + if (!vm) { + drm_exec_fini(&exec); return false; + } } addr /= AMDGPU_GPU_PAGE_SIZE; @@ -3032,7 +3057,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, value = 0; } - r = dma_resv_reserve_fences(root->tbo.base.resv, 1); + r = dma_resv_reserve_fences(vm->root.bo->tbo.base.resv, 1); if (r) { pr_debug("failed %d to reserve fence slot\n", r); goto error_unlock; @@ -3046,12 +3071,10 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, r = amdgpu_vm_update_pdes(adev, vm, true); error_unlock: - amdgpu_bo_unreserve(root); + drm_exec_fini(&exec); if (r < 0) dev_err(adev->dev, "Can't handle page fault (%d)\n", r); - amdgpu_bo_unref(&root); - return false; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 3695299f1a03..b32f51a78cd8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -592,7 +592,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid, bool write_fault); struct amdgpu_vm *amdgpu_vm_lock_by_pasid(struct amdgpu_device *adev, - struct amdgpu_bo **root, u32 pasid); + u32 pasid, struct drm_exec *exec); void amdgpu_vm_set_task_info(struct amdgpu_vm *vm); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 47721d0c3781..81a759a98725 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -4071,6 +4071,41 @@ err_priv_inst: return r; } +static void gfx_v9_0_deactivate_kcq_hqd(struct amdgpu_device *adev) +{ + amdgpu_gfx_rlc_enter_safe_mode(adev, 0); + for (int i = 0; i < adev->gfx.num_compute_rings; i++) { + u32 tmp; + struct amdgpu_ring *ring = &adev->gfx.compute_ring[i]; + + mutex_lock(&adev->srbm_mutex); + soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0, 0); + tmp = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE); + /* disable the queue if it's active */ + if (tmp & CP_HQD_ACTIVE__ACTIVE_MASK) { + int j; + + WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 1); + for (j = 0; j < adev->usec_timeout; j++) { + tmp = RREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE); + if (!(tmp & CP_HQD_ACTIVE__ACTIVE_MASK)) + break; + udelay(1); + } + if (j == AMDGPU_MAX_USEC_TIMEOUT) { + DRM_DEBUG("comp_%u_%u_%u dequeue request failed.\n", + ring->me, ring->pipe, ring->queue); + /* Manual disable if dequeue request times out */ + WREG32_SOC15(GC, 0, mmCP_HQD_ACTIVE, 0); + } + WREG32_SOC15(GC, 0, mmCP_HQD_DEQUEUE_REQUEST, 0); + } + soc15_grbm_select(adev, 0, 0, 0, 0, 0); + mutex_unlock(&adev->srbm_mutex); + } + amdgpu_gfx_rlc_exit_safe_mode(adev, 0); +} + static int gfx_v9_0_hw_fini(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; @@ -4095,6 +4130,10 @@ static int gfx_v9_0_hw_fini(struct amdgpu_ip_block *ip_block) return 0; } + if ((adev->flags & AMD_IS_APU) && amdgpu_in_reset(adev) && + amdgpu_asic_reset_method(adev) == AMD_RESET_METHOD_MODE2) + gfx_v9_0_deactivate_kcq_hqd(adev); + /* Use deinitialize sequence from CAIL when unbinding device from driver, * otherwise KIQ is hanging when binding back */ diff --git a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c index 95b3f4e55ec3..699c274d357e 100644 --- a/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c +++ b/drivers/gpu/drm/amd/amdgpu/ih_v6_1.c @@ -790,7 +790,7 @@ static void ih_v6_1_set_interrupt_funcs(struct amdgpu_device *adev) const struct amdgpu_ip_block_version ih_v6_1_ip_block = { .type = AMD_IP_BLOCK_TYPE_IH, .major = 6, - .minor = 0, + .minor = 1, .rev = 0, .funcs = &ih_v6_1_ip_funcs, }; |