[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHGf_=pLJcoSuKZb+P0c_7uW16hO35cMkQfJQB1SUJt8uLkGYw@mail.gmail.com>
Date: Thu, 26 Jan 2012 16:21:30 -0500
From: KOSAKI Motohiro <kosaki.motohiro@...il.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Oleg Nesterov <oleg@...hat.com>,
Yasunori Goto <y-goto@...fujitsu.com>,
Ingo Molnar <mingo@...e.hu>,
Hiroyuki KAMEZAWA <kamezawa.hiroyu@...fujitsu.com>,
Linux Kernel ML <linux-kernel@...r.kernel.org>
Subject: Re: [BUG] TASK_DEAD task is able to be woken up in special condition
2012/1/25 Peter Zijlstra <peterz@...radead.org>:
> On Wed, 2011-12-28 at 16:07 -0500, KOSAKI Motohiro wrote:
>>
>> CPU0 CPU1
>> --------------------------------------------------------
>> deactivate_task()
>> task->state = TASK_UNINTERRUPTIBLE;
>> activate_task()
>> rq->nr_uninterruptible--;
>>
>> schedule()
>> deactivate_task()
>> rq->nr_uninterruptible++;
>>
>>
>
> Hmm, I think you're right, when CPU0 does __sched_setscheduler() on the
> task running on CPU1 and CPU1's @task is current.
>
> I think only __sched_setscheduler() is really a problem, the other
> activate/deactivate users not schedule or wakeup are __migrate_task()
> and normalize_task().
>
> __migrate_task() will only run on tasks that aren't actually running
> anywhere so the above scenario can't happen, normalize_task() is never
> used on normal systems (sysrq-n).
>
> So I guess the below cures things and cleans up a bit.. no?
Sorry for the delay. Yeah, I think this should work. (even though I
don't understand __migrate_task() detail)
Thank you!
Ack-by: KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists