All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jens Axboe <jens.axboe@oracle.com>
To: Alexander Beregalov <a.beregalov@gmail.com>
Cc: "linux-next@vger.kernel.org" <linux-next@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: next-20081215: kernel BUG at block/cfq-iosched.c:1243
Date: Mon, 15 Dec 2008 19:44:41 +0100	[thread overview]
Message-ID: <20081215184440.GI32491@kernel.dk> (raw)
In-Reply-To: <a4423d670812151038p4bab739fo3a6f9afd1ad3e358@mail.gmail.com>

On Mon, Dec 15 2008, Alexander Beregalov wrote:
> 2008/12/15 Jens Axboe <jens.axboe@oracle.com>:
> >> [   23.080733] kernel BUG at block/cfq-iosched.c:1243!
> >> [   23.080733] invalid opcode: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
> 
> > Odd, is it reproducible? What is your .config?
> Yes, it is
> 
> Config is in attach

It's weird, because the only change for CFQ in that area is the one
below. Does the oops go away if you back this out?


>From 030a3f02bb42fad91ae84c99924fe3ee162a634d Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Fri, 12 Dec 2008 15:56:31 +0100
Subject: [PATCH] cfq-iosched: fix race between exiting queue and exiting task

Original patch from Nikanth Karthikesan <knikanth@suse.de>

When a queue exits the queue lock is taken and cfq_exit_queue() would free all
the cic's associated with the queue.

But when a task exits, cfq_exit_io_context() gets cic one by one and then
locks the associated queue to call __cfq_exit_single_io_context. It looks like
between getting a cic from the ioc and locking the queue, the queue might have
exited on another cpu.

Fix this by rechecking the cfq_io_context queue key inside the queue lock
again, and not calling into __cfq_exit_single_io_context() if somebody
beat us to it.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
---
 block/cfq-iosched.c |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index ee8a90c..8a2485d 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1314,7 +1314,16 @@ static void cfq_exit_single_io_context(struct io_context *ioc,
 		unsigned long flags;
 
 		spin_lock_irqsave(q->queue_lock, flags);
+
+		/*
+		 * Ensure we get a fresh copy of the ->key to prevent
+		 * race between exiting task and queue
+		 */
+		smp_read_barrier_depends();
+		if (cic->key)
+			__cfq_exit_single_io_context(cfqd, cic);
 		__cfq_exit_single_io_context(cfqd, cic);
+
 		spin_unlock_irqrestore(q->queue_lock, flags);
 	}
 }
-- 
1.6.0.4.755.g6e139


-- 
Jens Axboe

  reply	other threads:[~2008-12-15 18:45 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-12-15 18:29 next-20081215: kernel BUG at block/cfq-iosched.c:1243 Alexander Beregalov
2008-12-15 18:32 ` Jens Axboe
2008-12-15 18:38   ` Alexander Beregalov
2008-12-15 18:44     ` Jens Axboe [this message]
2008-12-15 20:09       ` Alexander Beregalov
2008-12-15 20:18         ` Jens Axboe
2008-12-15 20:29           ` Alexander Beregalov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20081215184440.GI32491@kernel.dk \
    --to=jens.axboe@oracle.com \
    --cc=a.beregalov@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-next@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.