There is an issue reported on platforms with VCN 4.0.5 that some jobs are timing out. This leads to a full GPU reset. That full GPU reset exposed a number of problems in DCN that have since been fixed, but still the GPU reset means a black screen for ~500ms while the GPU resets.
However the issue with VCN in this case doesn't require a full GPU reset to recover, just resetting VCN is enough. For a workaround while the VCN issue is still being root caused lessen the impact of VCN timeouts use the per-ring reset functionality to reset just VCN. v2: Also add support for per ring reset on other newer vcn4 and vcn5. Mario Limonciello (3): drm/amd: Add per-ring reset for vcn v4.0.5 use drm/amd: Add per-ring reset for vcn v4.0.0 use drm/amd: Add per-ring reset for vcn v5.0.0 use drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c | 17 ++++++++++++++++- drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c | 22 ++++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c | 17 ++++++++++++++++- 3 files changed, 54 insertions(+), 2 deletions(-) -- 2.49.0
