]> git.karo-electronics.de Git - linux-beck.git/commit
sched/core: Fix endless loop in pick_next_task()
authorKirill Tkhai <ktkhai@parallels.com>
Thu, 6 Mar 2014 09:32:01 +0000 (13:32 +0400)
committerIngo Molnar <mingo@kernel.org>
Tue, 11 Mar 2014 11:05:39 +0000 (12:05 +0100)
commit4c6c4e38c4e9a454889298dcc498174968d14a09
treeffb6a22c1f3a58667b3c745fe055b6dab3065443
parente4aa358b6c23f98b2715594f6b1e9a4996a55f04
sched/core: Fix endless loop in pick_next_task()

1) Single cpu machine case.

When rq has only RT tasks, but no one of them can be picked
because of throttling, we enter in endless loop.

pick_next_task_{dl,rt} return NULL.

In pick_next_task_fair() we permanently go to retry

if (rq->nr_running != rq->cfs.h_nr_running)
return RETRY_TASK;

(rq->nr_running is not being decremented when rt_rq becomes
throttled).

No chances to unthrottle any rt_rq or to wake fair here,
because of rq is locked permanently and interrupts are
disabled.

2) In case of SMP this can cause a hang too. Although we unlock
   rq in idle_balance(), interrupts are still disabled.

The solution is to check for available tasks in DL and RT
classes instead of checking for sum.

Signed-off-by: Kirill Tkhai <ktkhai@parallels.com>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1394098321.19290.11.camel@tkhai
Signed-off-by: Ingo Molnar <mingo@kernel.org>
kernel/sched/fair.c
kernel/sched/rt.c
kernel/sched/sched.h