drm/amdgpu: move sched job process from isr to fence callback

This way can avoid interrupt lost, and can process sched job exactly.

Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Jammy Zhou <Jammy.Zhou@amd.com>

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -404,7 +404,7 @@ struct amdgpu_fence_driver {
 
 struct amdgpu_fence {
 	struct fence base;
-
+	struct fence_cb cb;
 	/* RB, DMA, etc. */
 	struct amdgpu_ring		*ring;
 	uint64_t			seq;

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

@@ -350,25 +350,8 @@ void amdgpu_fence_process(struct amdgpu_ring *ring)
 		}
 	} while (atomic64_xchg(&ring->fence_drv.last_seq, seq) > seq);
 
-	if (wake) {
-		if (amdgpu_enable_scheduler) {
-			uint64_t handled_seq =
-				amd_sched_get_handled_seq(ring->scheduler);
-			uint64_t latest_seq =
-				atomic64_read(&ring->fence_drv.last_seq);
-			if (handled_seq == latest_seq) {
-				DRM_ERROR("ring %d, EOP without seq update (lastest_seq=%llu)\n",
-					  ring->idx, latest_seq);
-				goto exit;
-			}
-			do {
-				amd_sched_isr(ring->scheduler);
-			} while (amd_sched_get_handled_seq(ring->scheduler) < latest_seq);
-		}
-
+	if (wake)
 		wake_up_all(&ring->fence_drv.fence_queue);
-	}
-exit:
 	spin_unlock_irqrestore(&ring->fence_lock, irqflags);
 }
 

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

@@ -43,12 +43,20 @@ static int amdgpu_sched_prepare_job(struct amd_gpu_scheduler *sched,
 	return r;
 }
 
+static void amdgpu_fence_sched_cb(struct fence *f, struct fence_cb *cb)
+{
+	struct amdgpu_fence *fence =
+		container_of(cb, struct amdgpu_fence, cb);
+	amd_sched_isr(fence->ring->scheduler);
+}
+
 static void amdgpu_sched_run_job(struct amd_gpu_scheduler *sched,
 				 struct amd_context_entity *c_entity,
 				 void *job)
 {
 	int r = 0;
 	struct amdgpu_cs_parser *sched_job = (struct amdgpu_cs_parser *)job;
+	struct amdgpu_fence *fence;
 
 	mutex_lock(&sched_job->job_lock);
 	r = amdgpu_ib_schedule(sched_job->adev,
@@ -57,6 +65,11 @@ static void amdgpu_sched_run_job(struct amd_gpu_scheduler *sched,
 			       sched_job->filp);
 	if (r)
 		goto err;
+	fence = sched_job->ibs[sched_job->num_ibs - 1].fence;
+	if (fence_add_callback(&fence->base,
+			       &fence->cb, amdgpu_fence_sched_cb))
+		goto err;
+
 	if (sched_job->run_job) {
 		r = sched_job->run_job(sched_job);
 		if (r)