[PATCH 1/1] amdkfd: Fix memory availability double accounting of kfd pinned objects

[Daniel Phillips]

Pinned objects that are not kfd objects reduce the total vram available
to kfd, so we subtract the total size of pinned objects from kdf vram
availability. However this double counts objects pinned by kfd itself
because they are counted both as used and pinned. So track the total
size of objects pinned by kfd and add it back to kfd availability to
remove the double accounting.

Signed-off-by: Daniel Phillips <daniel.phillips at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h       | 1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 7 +++++++
 2 files changed, 8 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index 01ba3589b60a..f53514c66973 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -99,6 +99,7 @@ struct amdgpu_kfd_dev {
 	struct kfd_dev *dev;
 	int64_t vram_used;
 	uint64_t vram_used_aligned;
+	atomic64_t vram_pinned;
 	bool init_complete;
 	struct work_struct reset_work;
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index e7403f8e4eba..97a4e3dba053 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -1414,6 +1414,9 @@ static int amdgpu_amdkfd_gpuvm_pin_bo(struct amdgpu_bo *bo, u32 domain)
 	amdgpu_bo_sync_wait(bo, AMDGPU_FENCE_OWNER_KFD, false);
 	amdgpu_bo_unreserve(bo);
 
+	atomic64_add(amdgpu_bo_size(bo),
+		&amdgpu_ttm_adev(bo->tbo.bdev)->kfd.vram_pinned);
+
 	return ret;
 }
 
@@ -1435,6 +1438,9 @@ static void amdgpu_amdkfd_gpuvm_unpin_bo(struct amdgpu_bo *bo)
 
 	amdgpu_bo_unpin(bo);
 	amdgpu_bo_unreserve(bo);
+
+	atomic64_sub(amdgpu_bo_size(bo),
+		&amdgpu_ttm_adev(bo->tbo.bdev)->kfd.vram_pinned);
 }
 
 int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev,
@@ -1588,6 +1594,7 @@ size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev)
 	available = adev->gmc.real_vram_size
 		- adev->kfd.vram_used_aligned
 		- atomic64_read(&adev->vram_pin_size)
+		+ atomic64_read(&adev->kfd.vram_pinned)
 		- reserved_for_pt;
 	spin_unlock(&kfd_mem_limit.mem_limit_lock);
 
-- 
2.39.0