From: Jan Kara <jack@suse.cz>
To: Witold Baryluk <baryluk@smp.if.uj.edu.pl>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Kernel Testers List <kernel-testers@vger.kernel.org>,
Maciej Rutecki <maciej.rutecki@gmail.com>,
Florian Mickler <florian@mickler.org>,
linux-ext4@vger.kernel.org,
Andreas Dilger <adilger.kernel@dilger.ca>,
Theodore Ts'o <tytso@mit.edu>
Subject: Re: [Bug #39732] JBD: Spotted dirty metadata buffer (dev = dm-2, blocknr = 2512725). There's a risk of filesystem corruption in case of system crash.
Date: Tue, 13 Dec 2011 18:35:42 +0100 [thread overview]
Message-ID: <20111213173542.GH11747@quack.suse.cz> (raw)
In-Reply-To: <20111213041144.GF28734@smp.if.uj.edu.pl>
Hi,
On Tue 13-12-11 05:11:44, Witold Baryluk wrote:
> On 08-28 21:01, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a report
> > of regressions introduced between 2.6.39 and 3.0.
> >
> > The following bug entry is on the current list of known regressions
> > introduced between 2.6.39 and 3.0. Please verify if it still should
> > be listed and let the tracking team know (either way).
> >
> >
> > Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=39732
> > Subject : JBD: Spotted dirty metadata buffer (dev = dm-2, blocknr = 2512725). There's a risk of filesystem corruption in case of system crash.
> > Submitter : Witold Baryluk <baryluk@smp.if.uj.edu.pl>
> > Date : 2011-07-22 04:22 (38 days old)
>
> This bug still exists in 3.1.0 and 3.2.0-rc5
>
> I spoted it again few days ago, because I moved from me laptop with 2GB
> to temporary desktop with 256MB of RAM, and not it
> triggers like a mad.
>
> So my initial suspicious that it only shows up on big memory
> presure was correct.
>
>
> It is just like this
> # dmesg | grep dirty
> [122647.953872] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1019852). There's a risk of filesystem corruption in case of system crash.
> [122647.954598] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1043195). There's a risk of filesystem corruption in case of system crash.
> [122647.955213] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1043197). There's a risk of filesystem corruption in case of system crash.
> [122689.978864] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 498559). There's a risk of filesystem corruption in case of system crash.
> [122705.485461] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 68282). There's a risk of filesystem corruption in case of system crash.
> [122705.486046] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 68284). There's a risk of filesystem corruption in case of system crash.
> [122705.496916] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 68365). There's a risk of filesystem corruption in case of system crash.
> [122705.612423] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 69218). There's a risk of filesystem corruption in case of system crash.
> [122706.531297] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 833896). There's a risk of filesystem corruption in case of system crash.
> [122720.863807] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 914979). There's a risk of filesystem corruption in case of system crash.
> [122721.046789] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 890090). There's a risk of filesystem corruption in case of system crash.
> [122721.088280] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 916474). There's a risk of filesystem corruption in case of system crash.
> [122721.176979] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 928182). There's a risk of filesystem corruption in case of system crash.
> [122721.177593] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 927783). There's a risk of filesystem corruption in case of system crash.
> [122721.371654] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 920655). There's a risk of filesystem corruption in case of system crash.
> [122721.543676] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 967035). There's a risk of filesystem corruption in case of system crash.
> [122728.396687] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1043474). There's a risk of filesystem corruption in case of system crash.
> [122736.283876] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 713478). There's a risk of filesystem corruption in case of system crash.
> [122743.576350] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 745551). There's a risk of filesystem corruption in case of system crash.
> [122743.889259] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 828117). There's a risk of filesystem corruption in case of system crash.
> [122752.077022] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 905082). There's a risk of filesystem corruption in case of system crash.
> [122752.125005] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 905294). There's a risk of filesystem corruption in case of system crash.
> [122752.784390] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1030306). There's a risk of filesystem corruption in case of system crash.
> [122752.785053] JBD: Spotted dirty metadata buffer (dev = dm-3, blocknr = 1031593). There's a risk of filesystem corruption in case of system crash.
> #
>
> dm-4 is ext4 partition mounted at /var
> dm-4 is on LVM on luksCrypt on ATA drive.
>
> I'm using currently stock Debian testing kernel:
> $ uname -a
> Linux sredniczarny 3.1.0-1-686-pae #1 SMP Tue Nov 29 19:01:56 UTC 2011 i686 GNU/Linux
> $ cat /proc/version
> Linux version 3.1.0-1-686-pae (Debian 3.1.4-1) (waldi@debian.org) (gcc version 4.6.2 (Debian 4.6.2-4) ) #1 SMP Tue Nov 29 19:01:56 UTC 2011
> $
>
> 32-bit intel compatible cpu
Thanks for report. Actually I was able to reproduce these messages with
data=journal mode as well when running bash_shared_mapping test. The
culprit of the problem is that block_mkwrite() as used in
ext4_page_mkwrite() marks buffers dirty and JBD2 notices this before
we clear dirty bits later in ext4_page_mkwrite() (since JBD2 doesn't use
page lock for protection when inspecting buffers). So for data=journal mode
we'll probably have to reimplement what block_page_mkwrite() does in a
slightly different manner tweaked for the needs of data=journal mode.
Honza
--
Jan Kara <jack@suse.cz>
SUSE Labs, CR
next prev parent reply other threads:[~2011-12-13 17:35 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <Sm7qbRR0zn.A.yAG.LepWOB@chimera>
[not found] ` <gUVi6eNP8RE.A.z7G.OfpWOB@chimera>
2011-12-13 4:11 ` [Bug #39732] JBD: Spotted dirty metadata buffer (dev = dm-2, blocknr = 2512725). There's a risk of filesystem corruption in case of system crash Witold Baryluk
2011-12-13 17:35 ` Jan Kara [this message]
[not found] ` <20111213173542.GH11747-+0h/O2h83AeN3ZZ/Hiejyg@public.gmane.org>
2011-12-13 23:43 ` Witold Baryluk
2011-07-22 4:22 [Bug 39732] New: " bugzilla-daemon
2011-07-22 22:13 ` [Bug 39732] " bugzilla-daemon
2011-07-25 3:52 ` bugzilla-daemon
2012-01-24 23:17 ` bugzilla-daemon
2012-07-02 13:53 ` bugzilla-daemon
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=20111213173542.GH11747@quack.suse.cz \
--to=jack@suse.cz \
--cc=adilger.kernel@dilger.ca \
--cc=baryluk@smp.if.uj.edu.pl \
--cc=florian@mickler.org \
--cc=kernel-testers@vger.kernel.org \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=maciej.rutecki@gmail.com \
--cc=rjw@sisk.pl \
--cc=tytso@mit.edu \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).