drm/amdgpu: clear freed mappings immediately when BO may be freed
authorNicolai Hähnle <nicolai.haehnle@amd.com>
Thu, 23 Mar 2017 18:34:11 +0000 (19:34 +0100)
committerAlex Deucher <alexander.deucher@amd.com>
Thu, 30 Mar 2017 03:55:32 +0000 (23:55 -0400)
commit23e0563e48f7e9e98003df5b43d6a48e162782c6
tree29ecc7cfb4a1725370da475d0960720887868dd3
parent2de6a7c52a412985446ee358d8e27b7f3de5e3f3
drm/amdgpu: clear freed mappings immediately when BO may be freed

Also, add the fence of the clear operations to the BO to ensure that
the underlying memory can only be re-used after all PTEs pointing to
it have been cleared.

This avoids the following sequence of events that could be triggered
by user space:

1. Submit a CS that accesses some BO _without_ adding that BO to the
   buffer list.
2. Free that BO.
3. Some other task re-uses the memory underlying the BO.
4. The CS is submitted to the hardware and accesses memory that is
   now already in use by somebody else.

By clearing the page tables immediately in step 2, a GPU VM fault will
be triggered in step 4 instead of wild memory accesses.

v2: use amdgpu_bo_fence directly

Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c