The patch solves the UVD issue.
By the way, I had to change one of the "->"'s to a "." to compile:
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c: In function ‘amdgpu_job_free_resources’:
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c:159:61: error: invalid type argument of ‘->’ (have ‘struct dma_fence’)
159 | if (job->base.s_fence && job->base.s_fence->finished->ops)
| ^~
Thanks,
Thong Thai
On 2023-01-12 09:05, Alex Deucher
wrote:
On Thu, Jan 12, 2023 at 8:48 AM Christian König <ckoenig.leichtzumerken@xxxxxxxxx> wrote:It can be that neither fence were initialized when we run out of UVD streams for example. Signed-off-by: Christian König <christian.koenig@xxxxxxx>Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/2324 Reviewed-by: Alex Deucher <alexander.deucher@xxxxxxx>--- drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 9e549923622b..28929c6219a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -161,8 +161,14 @@ void amdgpu_job_free_resources(struct amdgpu_job *job) struct dma_fence *f; unsigned i; - /* use sched fence if available */ - f = job->base.s_fence ? &job->base.s_fence->finished : &job->hw_fence; + /* Check if any fences where initialized */ + if (job->base.s_fence && job->base.s_fence->finished->ops) + f = &job->base.s_fence->finished; + else if (job->hw_fence.ops) + f = &job->hw_fence; + else + f = NULL; + for (i = 0; i < job->num_ibs; ++i) amdgpu_ib_free(ring->adev, &job->ibs[i], f); } -- 2.25.1
