[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20070605173949.GA27618@ms2.inr.ac.ru>
Date: Tue, 5 Jun 2007 21:39:49 +0400
From: Alexey Kuznetsov <kuznet@....inr.ac.ru>
To: Thomas Gleixner <tglx@...utronix.de>
Cc: Ingo Molnar <mingo@...e.hu>, linux-kernel@...r.kernel.org,
Andrew Morton <akpm@...ux-foundation.org>,
Ulrich Drepper <drepper@...hat.com>
Subject: Re: [RFC][PATCH] muptiple bugs in PI futexes
Hello!
> Hmm, what means not expected ? -ESRCH is returned, when the owner task
> is not found.
This is not supposed to happen with robust futexes.
glibs aborts (which is correct), or for build with disabled debugging
enters simulated deadlock (which is confusing).
> lock. Also using uval is wrong.
Yup. You are right.
This means those RETRY messages could be spurious. I must rerun the test.
> This does not really explain, why you do prevent the -ESRCH return value
> in the next cycle,
Because right curval is refetched, it already has FUTEX_OWNER_DIED bit set
and we succesfully take the lock.
> No, -EDEADLK is returned from here:
>
> ret = rt_mutex_timed_lock(&q.pi_state->pi_mutex, to, 1);
Of course. It is the only place where ret is set. :-)
>
> The rtmutex code only returns -EDEADLK, when the lock is already held by
> the task
This case.
> The fix is to remove it and to find the real cause of the problem.
>
> I'm running the glibc tests since hours w/o tripping into it.
OK.
You need run only tst-robustpi8 in loop. It should be triggered quickly,
a few of minutes on 8-way smp here.
If you want, I can insert some debugging printks, which you need,
and run the test here.
Alexey
-
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