drm/sched: fix timeout handling v2

We need to make sure that we don't race between job completion and
timeout.

v2: put revert label after calling the handling manually

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/scheduler/sched_main.c

@@ -249,13 +249,41 @@ static void drm_sched_job_timedout(struct work_struct *work)
 {
 	struct drm_gpu_scheduler *sched;
 	struct drm_sched_job *job;
+	int r;
 
 	sched = container_of(work, struct drm_gpu_scheduler, work_tdr.work);
+
+	spin_lock(&sched->job_list_lock);
+	list_for_each_entry_reverse(job, &sched->ring_mirror_list, node) {
+		struct drm_sched_fence *fence = job->s_fence;
+
+		if (!dma_fence_remove_callback(fence->parent, &fence->cb))
+			goto already_signaled;
+	}
+
 	job = list_first_entry_or_null(&sched->ring_mirror_list,
 				       struct drm_sched_job, node);
+	spin_unlock(&sched->job_list_lock);
 
 	if (job)
-		job->sched->ops->timedout_job(job);
+		sched->ops->timedout_job(job);
+
+	spin_lock(&sched->job_list_lock);
+	list_for_each_entry(job, &sched->ring_mirror_list, node) {
+		struct drm_sched_fence *fence = job->s_fence;
+
+		if (!fence->parent || !list_empty(&fence->cb.node))
+			continue;
+
+		r = dma_fence_add_callback(fence->parent, &fence->cb,
+					   drm_sched_process_job);
+		if (r)
+			drm_sched_process_job(fence->parent, &fence->cb);
+
+already_signaled:
+		;
+	}
+	spin_unlock(&sched->job_list_lock);
 }
 
 /**