[PATCH 6.12.y 2/2] drm/amdkfd: use the queue's process info for the WPTR BO eviction fence

[Sasha Levin <sashal@kernel.org>]

From: Jesse Zhang <Jesse.Zhang@amd.com>

[ Upstream commit da2897ee8b811945fd40ecc4aaf03291c14ce61c ]

amdgpu_amdkfd_map_gtt_bo_to_gart() removes the eviction fence of
bo->vm_bo->vm->process_info from the WPTR BO of a new queue. bo->vm_bo
is the VM which opened the BO last, not necessarily the KFD process' VM,
and a render node VM has no process info.

On navi31, an unprivileged process which exports its WPTR BO with
AMDKFD_IOC_EXPORT_DMABUF and imports it on a second render node fd
oopses in AMDKFD_IOC_CREATE_QUEUE:

  BUG: kernel NULL pointer dereference, address: 0000000000000060
  RIP: 0010:amdgpu_amdkfd_map_gtt_bo_to_gart+0x86/0x1a0 [amdgpu]
   init_user_queue.part.0+0x180/0x250 [amdgpu]
   pqm_create_queue+0x32d/0x730 [amdgpu]
   kfd_ioctl_create_queue+0x583/0x9f0 [amdgpu]

Pass the process info of the queue instead. With the fix, the queue is
created and destroyed normally.

Fixes: 7f347e3f8217 ("drm/amdkfd: Fix NULL pointer error for GC 11.0.1 on mGPU")
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 6b4157498a176bd1c6083f6bb5aaf8a257a2027e)
Cc: stable@vger.kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h             | 3 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c       | 8 +++++---
 drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c | 3 ++-
 3 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index b921587d428cc..f5374b6f2501b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -324,7 +324,8 @@ int amdgpu_amdkfd_gpuvm_map_gtt_bo_to_kernel(struct kgd_mem *mem,
 					     void **kptr, uint64_t *size);
 void amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(struct kgd_mem *mem);
 
-int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart);
+int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart,
+				     struct amdkfd_process_info *process_info);
 
 int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
 					    struct dma_fence __rcu **ef);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 55834719e9848..318d731bf9ecd 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -2248,11 +2248,13 @@ int amdgpu_amdkfd_gpuvm_sync_memory(
  * amdgpu_amdkfd_map_gtt_bo_to_gart - Map BO to GART and increment reference count
  * @bo: Buffer object to be mapped
  * @bo_gart: Return bo reference
+ * @process_info: KFD process info of the queue the BO is for
  *
  * Before return, bo reference count is incremented. To release the reference and unpin/
  * unmap the BO, call amdgpu_amdkfd_free_gtt_mem.
  */
-int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart)
+int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo_gart,
+				     struct amdkfd_process_info *process_info)
 {
 	int ret;
 
@@ -2274,8 +2276,8 @@ int amdgpu_amdkfd_map_gtt_bo_to_gart(struct amdgpu_bo *bo, struct amdgpu_bo **bo
 		goto err_map_bo_gart_failed;
 	}
 
-	amdgpu_amdkfd_remove_eviction_fence(
-		bo, bo->vm_bo->vm->process_info->eviction_fence);
+	/* bo->vm_bo can be the VM of any file which opened the BO */
+	amdgpu_amdkfd_remove_eviction_fence(bo, process_info->eviction_fence);
 
 	amdgpu_bo_unreserve(bo);
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
index 4a3ed391b3c7b..819c7f5fb16f8 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
@@ -285,7 +285,8 @@ static int init_user_queue(struct process_queue_manager *pqm,
 		}
 
 		retval = amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo,
-							  &(*q)->wptr_bo_gart);
+							  &(*q)->wptr_bo_gart,
+							  pqm->process->kgd_process_info);
 		if (retval) {
 			pr_err("Failed to map wptr bo to GART\n");
 			goto free_gang_ctx_bo;
-- 
2.53.0