Skip to contentSkip to conversation

sched/dl: Fix preemption checks

RFC In review Kirill Tkhai 3 participants 3 messages Oct 21, 2014 – Oct 28, 2014 Scheduler

lore
Share & tools
lore
#0 lore raw

RFC 1/1 sched/dl: Fix preemption checks

1)switched_to_dl() check is wrong. We reschedule only if rq->curr is deadline task, and we do not reschedule if it's a lower priority task. But we must always preempt a task of other classes.

2)dl_task_timer():
Policy does not change in case of priority inheritance. rt_mutex_setprio() changes prio, while policy remains old.

So we lose some balancing logic in dl_task_timer() and switched_to_dl() when we check policy instead of priority. Boosted task may be rq->curr.

(I didn't change switched_from_dl() because no check is necessary there at all).

I've looked at this place(switched_to_dl) several times and even fixed this function, but found just now... I suppose some performance tests may work better after this.

Completely untested, just RFC.

Juri how about this?

---
 kernel/sched/deadline.c |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
kernel/sched/deadline.c +7 -3
index 256e577..937a875 100644
@@ -532,7 +532,7 @@ static enum hrtimer_restart dl_task_timer(struct hrtimer *timer)
 	dl_se->dl_yielded = 0;
 	if (task_on_rq_queued(p)) {
 		enqueue_task_dl(rq, p, ENQUEUE_REPLENISH);
-		if (task_has_dl_policy(rq->curr))
+		if (dl_task(rq->curr))
 			check_preempt_curr_dl(rq, p, 0);
 		else
 			resched_curr(rq);
@@ -1607,8 +1607,12 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p)
 			/* Only reschedule if pushing failed */
 			check_resched = 0;
 #endif /* CONFIG_SMP */
-		if (check_resched && task_has_dl_policy(rq->curr))
-			check_preempt_curr_dl(rq, p, 0);
+		if (check_resched) {
+			if (dl_task(rq->curr))
+				check_preempt_curr_dl(rq, p, 0);
+			else
+				resched_curr(rq);
+		}
 	}
 }
 
#1 lore raw

Hi Kirill,

Quoted 17 lines · Kirill Tkhai

1)switched_to_dl() check is wrong. We reschedule only if rq->curr is deadline task, and we do not reschedule if it's a lower priority task. But we must always preempt a task of other classes.

2)dl_task_timer():
Policy does not change in case of priority inheritance. rt_mutex_setprio() changes prio, while policy remains old.

So we lose some balancing logic in dl_task_timer() and switched_to_dl() when we check policy instead of priority. Boosted task may be rq->curr.

(I didn't change switched_from_dl() because no check is necessary there at all).

Quoted 8 lines

I've looked at this place(switched_to_dl) several times and even fixed this function, but found just now... I suppose some performance tests may work better after this.

Completely untested, just RFC.

Juri how about this?

I'll have to give it a spin, but it looks good.

Thanks,

- Juri

Quoted 38 lines
---
 kernel/sched/deadline.c |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
kernel/sched/deadline.c +7 -3
index 256e577..937a875 100644
@@ -532,7 +532,7 @@ static enum hrtimer_restart dl_task_timer(struct hrtimer *timer)
 	dl_se->dl_yielded = 0;
 	if (task_on_rq_queued(p)) {
 		enqueue_task_dl(rq, p, ENQUEUE_REPLENISH);
-		if (task_has_dl_policy(rq->curr))
+		if (dl_task(rq->curr))
 			check_preempt_curr_dl(rq, p, 0);
 		else
 			resched_curr(rq);
@@ -1607,8 +1607,12 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p)
 			/* Only reschedule if pushing failed */
 			check_resched = 0;
 #endif /* CONFIG_SMP */
-		if (check_resched && task_has_dl_policy(rq->curr))
-			check_preempt_curr_dl(rq, p, 0);
+		if (check_resched) {
+			if (dl_task(rq->curr))
+				check_preempt_curr_dl(rq, p, 0);
+			else
+				resched_curr(rq);
+		}
 	}
 }
 
#2 lore raw
In reply to Kirill Tkhai
sched/dl: Fix preemption checks
Commit-ID:  f3a7e1a9c464a32ee186ab91388313c82e7ce018
Gitweb:     http://git.kernel.org/tip/f3a7e1a9c464a32ee186ab91388313c82e7ce018
Author:     Kirill Tkhai <[email protected]>
AuthorDate: Tue, 21 Oct 2014 20:35:56 +0400
Committer:  Ingo Molnar <[email protected]>
CommitDate: Tue, 28 Oct 2014 10:46:10 +0100

sched/dl: Fix preemption checks

1) switched_to_dl() check is wrong. We reschedule only if rq->curr is deadline task, and we do not reschedule if it's a lower priority task. But we must always preempt a task of other classes.

2) dl_task_timer():
Policy does not change in case of priority inheritance. rt_mutex_setprio() changes prio, while policy remains old.

So we lose some balancing logic in dl_task_timer() and switched_to_dl() when we check policy instead of priority. Boosted task may be rq->curr.

(I didn't change switched_from_dl() because no check is necessary there at all).

I've looked at this place(switched_to_dl) several times and even fixed this function, but found just now... I suppose some performance tests may work better after this.

---
 kernel/sched/deadline.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
kernel/sched/deadline.c +7 -3
index 4616789..5285332 100644
@@ -540,7 +540,7 @@ again:
 	dl_se->dl_yielded = 0;
 	if (task_on_rq_queued(p)) {
 		enqueue_task_dl(rq, p, ENQUEUE_REPLENISH);
-		if (task_has_dl_policy(rq->curr))
+		if (dl_task(rq->curr))
 			check_preempt_curr_dl(rq, p, 0);
 		else
 			resched_curr(rq);
@@ -1626,8 +1626,12 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p)
 			/* Only reschedule if pushing failed */
 			check_resched = 0;
 #endif /* CONFIG_SMP */
-		if (check_resched && task_has_dl_policy(rq->curr))
-			check_preempt_curr_dl(rq, p, 0);
+		if (check_resched) {
+			if (dl_task(rq->curr))
+				check_preempt_curr_dl(rq, p, 0);
+			else
+				resched_curr(rq);
+		}
 	}
 }