drm/amdgpu: fence wait old rcu slot

since the rcu slot was initialized to be num_hw_submission,
if command submission doesn't use scheduler, this limitation
will be invalid like uvd test.

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>
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
index 4303b44..d81f1f4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@@ -121,7 +121,7 @@
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_fence *fence;
-	struct fence **ptr;
+	struct fence *old, **ptr;
 	uint32_t seq;
 
 	fence = kmem_cache_alloc(amdgpu_fence_slab, GFP_KERNEL);
@@ -141,7 +141,11 @@
 	/* This function can't be called concurrently anyway, otherwise
 	 * emitting the fence would mess up the hardware ring buffer.
 	 */
-	BUG_ON(rcu_dereference_protected(*ptr, 1));
+	old = rcu_dereference_protected(*ptr, 1);
+	if (old && !fence_is_signaled(old)) {
+		DRM_INFO("rcu slot is busy\n");
+		fence_wait(old, false);
+	}
 
 	rcu_assign_pointer(*ptr, fence_get(&fence->base));