sched/deadline: Do not try to push tasks if pinned task switches to dl
authorWanpeng Li <wanpeng.li@linux.intel.com>
Wed, 22 Oct 2014 00:36:43 +0000 (08:36 +0800)
committerIngo Molnar <mingo@kernel.org>
Tue, 28 Oct 2014 09:47:57 +0000 (10:47 +0100)
As Kirill mentioned (https://lkml.org/lkml/2013/1/29/118):

 | If rq has already had 2 or more pushable tasks and we try to add a
 | pinned task then call of push_rt_task will just waste a time.

Just switched pinned task is not able to be pushed. If the rq has had
several dl tasks before they have already been considered as candidates
to be pushed (or pulled). This patch implements the same behavior as rt
class which introduced by commit 10447917551e ("sched/rt: Do not try to
push tasks if pinned task switches to RT").

Suggested-by: Kirill V Tkhai <tkhai@yandex.ru>
Acked-by: Juri Lelli <juri.lelli@arm.com>
Signed-off-by: Wanpeng Li <wanpeng.li@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1413938203-224610-1-git-send-email-wanpeng.li@linux.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/deadline.c

index 5285332..9d1e76a 100644 (file)
@@ -1622,7 +1622,8 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p)
 
        if (task_on_rq_queued(p) && rq->curr != p) {
 #ifdef CONFIG_SMP
-               if (rq->dl.overloaded && push_dl_task(rq) && rq != task_rq(p))
+               if (p->nr_cpus_allowed > 1 && rq->dl.overloaded &&
+                       push_dl_task(rq) && rq != task_rq(p))
                        /* Only reschedule if pushing failed */
                        check_resched = 0;
 #endif /* CONFIG_SMP */