diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2026-11-20 20:43:06 -0800 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2026-11-20 20:43:06 -0800 |
| commit | c44104f3642da720164042623e4832836aefdd8e (patch) | |
| tree | 76fc4494b40b186831ee690b8a335a0cb6b87480 /drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c | |
| parent | aa31d5f8b151b786b748c6bfc9a8c4f1855c8adc (diff) | |
| parent | ce29b8f9fb606f87014c2d05483a3fef05a11f38 (diff) | |
Pull drm fixes from Dave Airlie:
"Weekly fixes, around the usual size, though it looks like vmwgfx got
hit with a lot of analysis and has a bunch of validation fixes.
nouveau fixes for a 7.3 regression.
I know of one open i915 7.3 regressions that hopefully a fix will land
next week. Otherwise amdgpu and xe have the most with a scattering for
others.
core:
- syncobj: Limit allocations
- vblank: Resolve deadlock in timer
amdgpu:
- GPUVM fixes
- SMU 8 fix
- PCO quirk
- ATPX lifetime fix
- Unwinding fix for module loading failure
- UserQ fix
- HDMI fixes
- DML2 fix
- DC DRAM config fixes
- DCN 4.2 fixes
- DC pre-OS clock handling fixes
- DPMS fix
- Cleaner shader fixes for multi-XCC parts
- Vblank fix for GPU resets
- Properly handle firmware with different trust athorities
- Freesync/VRR fixes
- Backlight fix
- Display warning fixes
- PSR fixes
amdkfd:
- SVM fixes
radeon:
- Bounds checks in vbios parsing
- ATPX lifetime fix
xe:
- hwmon error handling fixes
- Treat CT-disabled TLB invalidation as cancelled
- Validate BOs on export to charge the exporter
- Always free requested IRQs on uninstall
- Reset TLB invalidation state in submit fini
- Use ordered clock reads for engine cycles cpu_delta
- Treat non-fatal PCIe errors as recovered
i915:
- Regression fix for GitLab #16595 and #17423: 4K displays
corrupt/black on CML/KBL
- DSB programming fix for GOSUB
imagination:
- Use correct function for polling non-iomem
ivpu:
- Fix race condition in recovery
nouveau:
- Fix instmem io_mapping faults and slowdowns
omap:
- dss: Fix handling deferrals
qaic:
- Harden ioctl checks
rocket:
- iommu fixes
- lifetime and validation fixes
- clock request fix
- devicetre core count fix
tilcdc:
- Fix drvdata in suspend/resume
vc4:
- async flip fixes
- off by one shader record
- object lifetime fixes
- two dsi fixes
vmwgfx:
- bounds checking fixes
- validation fixes"
* tag 'drm-fixes-2026-11-21' of https://gitlab.freedesktop.org/drm/kernel: (111 commits)
drm/vc4: Clear the slice caches before resetting a stalled GPU
drm/vc4: Start shader code at varying offsets in its BO
drm/vc4: Check for a validated shader before reading its thread mode
drm/vc4: Fix off-by-one in the shader record BO handle check
MAINTAINERS: Add Dmitry Baryshkov as bridge maintainer
MAINTAINERS: Remove myself for VC4 DRM driver
MAINTAINERS: Remove Andrzej Hajda from bridge maintainers
MAINTAINERS: Remove Jonas Karlman from bridge reviewers
MAINTAINERS: Move Tegra DRM under drm-misc
MAINTAINERS: Move Keem Bay DRM under drm-misc
MAINTAINERS: Move armada DRM under drm-misc
drm/xe/hwmon: Propagate all hardware read errors
drm/xe/hwmon: Report errors using structured xe_log API
drm/xe/log: Add VRAM, PAT, and HWMON DRIVER_HARDWARE components
drm/xe/guc: Treat CT-disabled TLB invalidation as cancelled
drm/xe/dma-buf: Validate BOs on export to charge the exporter
drm/xe/irq: Always free requested IRQs on uninstall
drm/xe/guc: Reset TLB invalidation state in submit fini
drm/xe/query: Use ordered clock reads for engine cycles cpu_delta
drm/xe/ras: Treat non-fatal PCIe errors as recovered
...
Diffstat (limited to 'drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c')
| -rw-r--r-- | drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c | 19 |
1 files changed, 17 insertions, 2 deletions
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c index 0be4506e9ae5..f2539bcf824f 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_freesync.c @@ -251,6 +251,13 @@ void amdgpu_dm_update_freesync_state_on_stream( mod_build_adaptive_sync_infopacket(new_stream, aconn->as_type, NULL, &new_stream->adaptive_sync_infopacket); + } else if (aconn && aconn->vsdb_info.amd_vsdb_version == 3 && + aconn->max_vfreq > 255) { + /* + * v1 saturates at 255 Hz. Keyed on the sink's max, not the mode's + * rate, so the packet version stays fixed per sink. + */ + packet_type = PACKET_TYPE_FS_V3; } app_tf = dm_dp_in_hdr_mode(new_stream) ? @@ -400,8 +407,16 @@ void amdgpu_dm_handle_vrr_transition(struct amdgpu_display_manager *dm, struct dm_crtc_state *new_state) { struct amdgpu_device *adev = dm->adev; - bool old_vrr_active = amdgpu_dm_crtc_vrr_active(old_state); - bool new_vrr_active = amdgpu_dm_crtc_vrr_active(new_state); + /* + * A disabled CRTC keeps its freesync state, but drm_crtc_vblank_get() + * fails there because vblank is off. Hold the reference only while + * the CRTC is active as well, which also covers resume, where the + * reset state lost track of the reference taken before suspend + */ + bool old_vrr_active = old_state->base.active && + amdgpu_dm_crtc_vrr_active(old_state); + bool new_vrr_active = new_state->base.active && + amdgpu_dm_crtc_vrr_active(new_state); /* Only DCE gates vupdate on VRR, keep it enabled for DCN */ bool vrr_gates_vupdate = amdgpu_ip_version(adev, DCE_HWIP, 0) == 0; |
