Am 12.07.22 um 00:44 schrieb sunpeng.li@xxxxxxx:
From: Leo Li <sunpeng.li@xxxxxxx>
When pinning a buffer, we should check to see if there are any
additional restrictions imposed by bo->preferred_domains. This will
prevent the BO from being moved to an invalid domain when pinning.
For example, this can happen if the user requests to create a BO in GTT
domain for display scanout. amdgpu_dm will allow pinning to either VRAM
or GTT domains, since DCN can scanout from either or. However, in
amdgpu_bo_pin_restricted(), pinning to VRAM is preferred if there is
adequate carveout. This can lead to pinning to VRAM despite the user
requesting GTT placement for the BO.
Signed-off-by: Leo Li <sunpeng.li@xxxxxxx>
---
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index 8a7b0f6162da..50400d70ef12 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -883,6 +883,9 @@ int amdgpu_bo_pin_restricted(struct amdgpu_bo *bo, u32 domain,
if (WARN_ON_ONCE(min_offset > max_offset))
return -EINVAL;
+ /* Check domain to be pinned to against preferred domains */
+ domain = bo->preferred_domains & domain
Please guard that with an "if (bo->preferred_domains & domain)" to
handle the case when the kernel needs to override the pin domain. That
can happen when we export a BO to a V4L camera (for example).
Apart from that looks like a rather nice bugfix.
Alex should we add a CC stable to this?
Regards,
Christian.
+
/* A shared bo cannot be migrated to VRAM */
if (bo->tbo.base.import_attach) {
if (domain & AMDGPU_GEM_DOMAIN_GTT)
