linux-kernel - Re: jbd2: don't wake kjournald unnecessarily

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <CA+icZUWOvk5i+XY22xdZYuxu=Vy3fG2QxJStEgNFQVS4u55y-w@mail.gmail.com>
Date:	Sun, 20 Jan 2013 00:44:31 +0100
From:	Sedat Dilek <sedat.dilek@...il.com>
To:	Eric Sandeen <sandeen@...hat.com>
Cc:	Jan Kara <jack@...e.cz>,
	linux-fsdevel <linux-fsdevel@...r.kernel.org>,
	Ext4 Developers List <linux-ext4@...r.kernel.org>,
	"Theodore Ts'o" <tytso@....edu>,
	LKML <linux-kernel@...r.kernel.org>,
	linux-next <linux-next@...r.kernel.org>
Subject: Re: jbd2: don't wake kjournald unnecessarily

Hi,

I and some others hit a similiar problem in Linux-Next
(next-20130118), please see [1] and [2].

[3] has a interim analyze of my problems.

After suspecting the problem was caused by TTY-NEXT, it turned out to
be a JBD2 problem finally.
The freezer/pm_test was helpful to hit the issue (Thanks Rafael for the hint!).

So, the issue has two faces: TTY and JBD2.
[4] gives a list and URLs of the patches I had to apply to have a
clean system again.

After applying the two TTY patches (without Eric's JBD2-fix!) the
call-trace after freezer/pm_test looked like this;

[  433.527986] PM: Syncing filesystems ... done.
[  433.843761] PM: Preparing system for mem sleep
[  436.306002] Freezing user space processes ...
[  456.304956] Freezing of tasks failed after 20.01 seconds (1 tasks
refusing to freeze, wq_busy=0):
[  456.305060] Cache I/O       D ffffffff8180d780     0  2132      1 0x00000004
[  456.305065]  ffff88007b9dfe18 0000000000000046 ffff88007b9dfdc8
0000000300000001
[  456.305069]  ffff880097e21720 ffff88007b9dffd8 ffff88007b9dffd8
ffff88007b9dffd8
[  456.305072]  ffff880119b32e40 ffff880097e21720 ffff88007b9dfe28
ffff880118077800
[  456.305076] Call Trace:
[  456.305085]  [<ffffffff816b7469>] schedule+0x29/0x70
[  456.305089]  [<ffffffff8129067d>] jbd2_log_wait_commit+0xcd/0x1a0
[  456.305094]  [<ffffffff8107eac0>] ? add_wait_queue+0x60/0x60
[  456.305098]  [<ffffffff812307a5>] ext4_sync_file+0x205/0x380
[  456.305103]  [<ffffffff811c43ad>] do_fsync+0x5d/0x90
[  456.305108]  [<ffffffff8119442b>] ? sys_write+0x6b/0xa0
[  456.305111]  [<ffffffff811c4790>] sys_fsync+0x10/0x20
[  456.305114]  [<ffffffff816c0edd>] system_call_fastpath+0x1a/0x1f
[  456.305122]
[  456.305124] Restarting tasks ... done.
[  456.315056] video LNXVIDEO:00: Restoring backlight state

After applying Eric's patch [5], I could not hit the call-trace again.
NOTE: The patch is from December 2012 and is not shipped in latest Linux-Next.

The attached testcase script was helpful to force the call-trace.
I have run 50 loops of it w/o an issue!

Feel free to add a Reported-by/Tested-by.
( The issue kept me busy for the last days. )

Regards,
- Sedat -

[1] http://marc.info/?t=135283664600002&r=1&w=2
[2] http://marc.info/?t=135862023700004&r=1&w=2
[3] http://marc.info/?l=linux-kernel&m=135862010419101&w=2
[4] http://marc.info/?l=linux-next&m=135863550923093&w=2
[5] http://patchwork.ozlabs.org/patch/207237/

Download attachment "run_pm-test_v2.sh" of type "application/x-sh" (875 bytes)