sched, kvm: Fix race condition involving sched_in_preempt_notifers
authorTejun Heo <tj@kernel.org>
Fri, 13 Nov 2009 09:33:53 +0000 (18:33 +0900)
committerIngo Molnar <mingo@elte.hu>
Sun, 15 Nov 2009 08:59:54 +0000 (09:59 +0100)
commit498657a478c60be092208422fefa9c7b248729c2
treec04542330fbbfaaa863fbec0f565b79c89df7033
parent055a00865dcfc8e61f3cbefbb879c9577bd36ae5
sched, kvm: Fix race condition involving sched_in_preempt_notifers

In finish_task_switch(), fire_sched_in_preempt_notifiers() is
called after finish_lock_switch().

However, depending on architecture, preemption can be enabled after
finish_lock_switch() which breaks the semantics of preempt
notifiers.

So move it before finish_arch_switch(). This also makes the in-
notifiers symmetric to out- notifiers in terms of locking - now
both are called under rq lock.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Avi Kivity <avi@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <4AFD2801.7020900@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/sched.c