Commit 4fa94425 authored by Pierre-Eric Pelloux-Prayer's avatar Pierre-Eric Pelloux-Prayer Committed by Alex Deucher
Browse files

drm/amdgpu: add missing lock to amdgpu_ttm_access_memory_sdma 



Users of ttm entities need to hold the gtt_window_lock before using them
to guarantee proper ordering of jobs.

Cc: stable@vger.kernel.org
Fixes: cb5cc4f5 ("drm/amdgpu: improve debug VRAM access performance using sdma")
Signed-off-by: default avatarPierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: default avatarChristian König <christian.koenig@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent 5427e32f

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c

+2 −0
Original line number Diff line number Diff line
@@ -1486,6 +1486,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, 
	if (r)

		goto out;



	mutex_lock(&adev->mman.gtt_window_lock);

	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);

	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +

		src_mm.start;
@@ -1500,6 +1501,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, 
	WARN_ON(job->ibs[0].length_dw > num_dw);



	fence = amdgpu_job_submit(job);

	mutex_unlock(&adev->mman.gtt_window_lock);



	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))

		r = -ETIMEDOUT;