drm/i915/gvt: Kick scheduler when new workload queued

The current schedule policy rely on a 1ms timer to execute workload. This
can introduce maximum 1ms unnecessary latency. This is especially bad for
small media workloads.

And I don't think we need this timer for QoS, but the change is not simply
remove the code. So I made a new API intel_gvt_kick_schedule() for future
change.

Signed-off-by: Changbin Du <changbin.du@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>

drivers/gpu/drm/i915/gvt/sched_policy.c

@@ -372,6 +372,11 @@ void intel_vgpu_start_schedule(struct intel_vgpu *vgpu)
 	vgpu->gvt->scheduler.sched_ops->start_schedule(vgpu);
 }
 
+void intel_gvt_kick_schedule(struct intel_gvt *gvt)
+{
+	intel_gvt_request_service(gvt, INTEL_GVT_REQUEST_EVENT_SCHED);
+}
+
 void intel_vgpu_stop_schedule(struct intel_vgpu *vgpu)
 {
 	struct intel_gvt_workload_scheduler *scheduler =

drivers/gpu/drm/i915/gvt/sched_policy.h

@@ -57,4 +57,6 @@ void intel_vgpu_start_schedule(struct intel_vgpu *vgpu);
 
 void intel_vgpu_stop_schedule(struct intel_vgpu *vgpu);
 
+void intel_gvt_kick_schedule(struct intel_gvt *gvt);
+
 #endif

drivers/gpu/drm/i915/gvt/scheduler.c

@@ -1337,5 +1337,6 @@ void intel_vgpu_queue_workload(struct intel_vgpu_workload *workload)
 {
 	list_add_tail(&workload->list,
 		workload_q_head(workload->vgpu, workload->ring_id));
+	intel_gvt_kick_schedule(workload->vgpu->gvt);
 	wake_up(&workload->vgpu->gvt->scheduler.waitq[workload->ring_id]);
 }