drm/amdgpu: block scheduler when gpu reset

Signed-off-by: Chunming Zhou <David1.Zhou@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -25,6 +25,7 @@
  *          Alex Deucher
  *          Jerome Glisse
  */
+#include <linux/kthread.h>
 #include <linux/console.h>
 #include <linux/slab.h>
 #include <linux/debugfs.h>
@@ -1895,6 +1896,14 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 
 	atomic_inc(&adev->gpu_reset_counter);
 
+	/* block scheduler */
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring)
+			continue;
+		kthread_park(ring->sched.thread);
+	}
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
 
@@ -1928,7 +1937,7 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 			struct amdgpu_ring *ring = adev->rings[i];
 			if (!ring)
 				continue;
-
+			kthread_unpark(ring->sched.thread);
 			amdgpu_ring_restore(ring, ring_sizes[i], ring_data[i]);
 			ring_sizes[i] = 0;
 			ring_data[i] = NULL;
@@ -1946,8 +1955,10 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 	} else {
 		amdgpu_fence_driver_force_completion(adev);
 		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
-			if (adev->rings[i])
+			if (adev->rings[i]) {
+				kthread_unpark(adev->rings[i]->sched.thread);
 				kfree(ring_data[i]);
+			}
 		}
 	}
 

drivers/gpu/drm/amd/scheduler/gpu_scheduler.c

@@ -476,6 +476,16 @@ static void amd_sched_process_job(struct fence *f, struct fence_cb *cb)
 	wake_up_interruptible(&sched->wake_up_worker);
 }
 
+static bool amd_sched_blocked(struct amd_gpu_scheduler *sched)
+{
+	if (kthread_should_park()) {
+		kthread_parkme();
+		return true;
+	}
+
+	return false;
+}
+
 static int amd_sched_main(void *param)
 {
 	struct sched_param sparam = {.sched_priority = 1};
@@ -485,14 +495,15 @@ static int amd_sched_main(void *param)
 	sched_setscheduler(current, SCHED_FIFO, &sparam);
 
 	while (!kthread_should_stop()) {
-		struct amd_sched_entity *entity;
+		struct amd_sched_entity *entity = NULL;
 		struct amd_sched_fence *s_fence;
 		struct amd_sched_job *sched_job;
 		struct fence *fence;
 
 		wait_event_interruptible(sched->wake_up_worker,
-			(entity = amd_sched_select_entity(sched)) ||
-			kthread_should_stop());
+					 (!amd_sched_blocked(sched) &&
+					  (entity = amd_sched_select_entity(sched))) ||
+					 kthread_should_stop());
 
 		if (!entity)
 			continue;