[Intel-gfx] [PATCH] drm/i915: Don't set queue_priority_hint if we don't kick the submission

Chris Wilson chris at chris-wilson.co.uk
Fri Oct 18 16:48:08 UTC 2019


If we change the priority of the active context, then it has no impact
on the decision of whether to preempt the active context -- we don't
preempt the context with itself. In this situation, we elide the tasklet
rescheduling and should *not* be marking up the queue_priority_hint as
that may mask a later submission where we decide we don't have to kick
the tasklet as a higher priority submission is pending (spoiler alert,
it was not).

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
---
 drivers/gpu/drm/i915/i915_scheduler.c | 37 ++++++++++++++++-----------
 1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_scheduler.c b/drivers/gpu/drm/i915/i915_scheduler.c
index 0ca40f6bf08c..d2edb527dcb8 100644
--- a/drivers/gpu/drm/i915/i915_scheduler.c
+++ b/drivers/gpu/drm/i915/i915_scheduler.c
@@ -189,22 +189,34 @@ static inline bool need_preempt(int prio, int active)
 	return prio >= max(I915_PRIORITY_NORMAL, active);
 }
 
-static void kick_submission(struct intel_engine_cs *engine, int prio)
+static void kick_submission(struct intel_engine_cs *engine,
+			    const struct i915_request *rq,
+			    int prio)
 {
-	const struct i915_request *inflight =
-		execlists_active(&engine->execlists);
+	const struct i915_request *inflight;
+
+	/*
+	 * We only need to kick the tasklet once for the high priority
+	 * new context we add into the queue.
+	 */
+	if (prio <= engine->execlists.queue_priority_hint)
+		return;
+
+	/* Nothing currently active? We're overdue for a submission! */
+	inflight = execlists_active(&engine->execlists);
+	if (!inflight)
+		return;
 
 	/*
 	 * If we are already the currently executing context, don't
-	 * bother evaluating if we should preempt ourselves, or if
-	 * we expect nothing to change as a result of running the
-	 * tasklet, i.e. we have not change the priority queue
-	 * sufficiently to oust the running context.
+	 * bother evaluating if we should preempt ourselves.
 	 */
-	if (!inflight || !need_preempt(prio, rq_prio(inflight)))
+	if (inflight->hw_context == rq->hw_context)
 		return;
 
-	tasklet_hi_schedule(&engine->execlists.tasklet);
+	engine->execlists.queue_priority_hint = prio;
+	if (need_preempt(prio, rq_prio(inflight)))
+		tasklet_hi_schedule(&engine->execlists.tasklet);
 }
 
 static void __i915_schedule(struct i915_sched_node *node,
@@ -330,13 +342,8 @@ static void __i915_schedule(struct i915_sched_node *node,
 			list_move_tail(&node->link, cache.priolist);
 		}
 
-		if (prio <= engine->execlists.queue_priority_hint)
-			continue;
-
-		engine->execlists.queue_priority_hint = prio;
-
 		/* Defer (tasklet) submission until after all of our updates. */
-		kick_submission(engine, prio);
+		kick_submission(engine, node_to_request(node), prio);
 	}
 
 	spin_unlock(&engine->active.lock);
-- 
2.23.0



More information about the Intel-gfx mailing list