drm/amdkfd: Use ref count to prevent kfd_process destruction

Use a reference counter instead of a lock to prevent process
destruction while functions running out of process context are using
the kfd_process structure. In many cases these functions don't need
the structure to be locked. In the few cases that really do need the
process lock, take it explicitly.

This helps simplify lock dependencies between the process lock and
other locks, particularly amdgpu and mm_struct locks. This will be
important when amdgpu calls back to amdkfd for memory evictions.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

1 files changed, 7 insertions, 7 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
index cb92d4b72400..93aae5c1e78b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@@ -441,7 +441,7 @@ void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
 	/*
 	 * Because we are called from arbitrary context (workqueue) as opposed
 	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
+	 * running so the lookup function increments the process ref count.
 	 */
 	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
 
@@ -493,7 +493,7 @@ void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
 	}
 
 	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+	kfd_unref_process(p);
 }
 
 static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events)
@@ -847,7 +847,7 @@ void kfd_signal_iommu_event(struct kfd_dev *dev, unsigned int pasid,
 	/*
 	 * Because we are called from arbitrary context (workqueue) as opposed
 	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
+	 * running so the lookup function increments the process ref count.
 	 */
 	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
 	struct mm_struct *mm;
@@ -860,7 +860,7 @@ void kfd_signal_iommu_event(struct kfd_dev *dev, unsigned int pasid,
 	 */
 	mm = get_task_mm(p->lead_thread);
 	if (!mm) {
-		mutex_unlock(&p->mutex);
+		kfd_unref_process(p);
 		return; /* Process is exiting */
 	}
 
@@ -903,7 +903,7 @@ void kfd_signal_iommu_event(struct kfd_dev *dev, unsigned int pasid,
 			&memory_exception_data);
 
 	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+	kfd_unref_process(p);
 }
 
 void kfd_signal_hw_exception_event(unsigned int pasid)
@@ -911,7 +911,7 @@ void kfd_signal_hw_exception_event(unsigned int pasid)
 	/*
 	 * Because we are called from arbitrary context (workqueue) as opposed
 	 * to process context, kfd_process could attempt to exit while we are
-	 * running so the lookup function returns a locked process.
+	 * running so the lookup function increments the process ref count.
 	 */
 	struct kfd_process *p = kfd_lookup_process_by_pasid(pasid);
 
@@ -924,5 +924,5 @@ void kfd_signal_hw_exception_event(unsigned int pasid)
 	lookup_events_by_type_and_signal(p, KFD_EVENT_TYPE_HW_EXCEPTION, NULL);
 
 	mutex_unlock(&p->event_mutex);
-	mutex_unlock(&p->mutex);
+	kfd_unref_process(p);
 }