Am 15.06.2018 um 17:14 schrieb Michel Dänzer: > On 2018-06-15 05:11 PM, Christian König wrote: >> Am 15.06.2018 um 16:42 schrieb Michel Dänzer: >>> From: Michel Dänzer <michel.daenzer at amd.com> >>> >>> Even BOs with AMDGPU_GEM_CREATE_NO_CPU_ACCESS may end up at least >>> partially in CPU visible VRAM, in particular when all VRAM is visible. >>> >>> Cc: stable at vger.kernel.org >>> Signed-off-by: Michel Dänzer <michel.daenzer at amd.com> >>> --- >>>  drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 26 ++++++++++++++++++-- >>>  1 file changed, 24 insertions(+), 2 deletions(-) >>> >>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >>> index 491fd0a789f7..9b34df7017f0 100644 >>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c >>> @@ -106,10 +106,32 @@ static u64 amdgpu_vram_mgr_vis_size(struct >>> amdgpu_device *adev, >>>   */ >>>  u64 amdgpu_vram_mgr_bo_invisible_size(struct amdgpu_bo *bo) >>>  { >>> -   if (bo->flags & AMDGPU_GEM_CREATE_NO_CPU_ACCESS) >>> +   struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); >>> +   struct amdgpu_vram_mgr *mgr = adev->mman.bdev.man[TTM_PL_VRAM].priv; >>> +   struct ttm_mem_reg *mem = &bo->tbo.mem; >>> +   struct drm_mm_node *nodes = mem->mm_node; >>> +   unsigned pages = mem->num_pages; >>> +   u64 usage = 0; >>> + >>> +   if (adev->gmc.visible_vram_size == adev->gmc.real_vram_size) >>> +       return 0; >> Andrey just added a helper for that into amdgpu_gmc.h. > Yep, see patch 4 and the cover letter. :) Yeah, seen that after hitting send :) >>> + >>> +   if (mem->start >= adev->gmc.visible_vram_size >> PAGE_SHIFT) >>>          return amdgpu_bo_size(bo); >>>  -   return 0; >>> +   if (!nodes) >>> +       return 0; >>> + >>> +   spin_lock(&mgr->lock); >> I actually don't see any need to grab the lock here. > Ah, so amdgpu_vram_mgr_del only grabs it for drm_mm_remove_node? Yes, exactly. Christian.