Commit 648a0dc0 authored by Jesse.Zhang's avatar Jesse.Zhang Committed by Alex Deucher
Browse files

drm/amdgpu: Fix user queue deadlock by reordering mutex locking 



This resolves a deadlock between user queue management and GPU reset
paths by enforcing consistent lock ordering.

The deadlock occurred when:

1. Process exit path (amdgpu_userq_mgr_fini) would:
   - Take uqm->userq_mutex
   - Then try to take adev->userq_mutex for list operations

2. GPU reset path (amdgpu_userq_pre_reset) would:
   - Take adev->userq_mutex first (for list traversal)
   - Then take uqm->userq_mutex

The solution establishes a strict top-down locking order:
1. Always take adev->userq_mutex before any uqm->userq_mutex
2. Maintain this order consistently across all code paths

Changes made:
- Reordered locking in amdgpu_userq_mgr_fini() to take device lock first
- Kept existing proper order in amdgpu_userq_pre_reset()
- Simplified the fini flow by removing redundant operations

This prevents circular dependencies while maintaining thread safety
during both normal operation and GPU reset scenarios.

Fixes: 4ce60dba ("drm/amdgpu: store userq_managers in a list in adev")
Reviewed-by: default avatarChristian König <christian.koenig@amd.com>
Reviewed-by: default avatarArvind Yadav <Arvind.Yadav@amd.com>
Signed-off-by: default avatarJesse Zhang <Jesse.Zhang@amd.com>
Signed-off-by: default avatarAlex Deucher <alexander.deucher@amd.com>
parent f71509fd

drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c

+3 −2
Original line number Diff line number Diff line
@@ -779,22 +779,23 @@ void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr) 


	cancel_delayed_work_sync(&userq_mgr->resume_work);



	mutex_lock(&adev->userq_mutex);

	mutex_lock(&userq_mgr->userq_mutex);

	idr_for_each_entry(&userq_mgr->userq_idr, queue, queue_id) {

		amdgpu_userq_wait_for_last_fence(userq_mgr, queue);

		amdgpu_userq_unmap_helper(userq_mgr, queue);

		amdgpu_userq_cleanup(userq_mgr, queue, queue_id);

	}

	mutex_lock(&adev->userq_mutex);



	list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) {

		if (uqm == userq_mgr) {

			list_del(&uqm->list);

			break;

		}

	}

	mutex_unlock(&adev->userq_mutex);

	idr_destroy(&userq_mgr->userq_idr);

	mutex_unlock(&userq_mgr->userq_mutex);

	mutex_unlock(&adev->userq_mutex);

	mutex_destroy(&userq_mgr->userq_mutex);

}