From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932897Ab2JWJ6a (ORCPT ); Tue, 23 Oct 2012 05:58:30 -0400 Received: from cantor2.suse.de ([195.135.220.15]:48934 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932503Ab2JWJ63 (ORCPT ); Tue, 23 Oct 2012 05:58:29 -0400 Date: Tue, 23 Oct 2012 10:41:58 +0200 From: Jan Kara To: Fabio Coatti Cc: Jan Kara , NeilBrown , "Myklebust, Trond" , Paul Bolle , linux-kernel@vger.kernel.org, Jeff Layton Subject: Re: ext3 issue on 3.6.1 Message-ID: <20121023084158.GA3064@quack.suse.cz> References: <20121018220809.GD17646@quack.suse.cz> <20121019124454.316ecfac@notabene.brown> <20121022134541.GA9438@quack.suse.cz> <20121022154258.GB9438@quack.suse.cz> <20121022161040.GC9438@quack.suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon 22-10-12 18:18:10, Fabio Coatti wrote: > 2012/10/22 Jan Kara : > > On Mon 22-10-12 18:01:56, Fabio Coatti wrote: > >> 2012/10/22 Jan Kara : > >> >> Uhu, my fault, I was sure to have answered to your request, sorry. > >> >> So yes, we have an ext3 R/O partition mounted on that machine. The > >> >> 3.6.3, with patch applied, dmesg is here: > >> > Interesting. You seem to be using data=journal on R/O partition, don't > >> > you? Can you also attach /proc/mounts? > >> > > >> > Also please try running a kernel with the attached patch and report the > >> > warning. Thanks! > >> > >> Do you want to try the kernel without touching the mount > >> configuration, right? (I mean, without (say) mounting ext2 the ext3 ro > >> partition, right?) > > Yes, leave things as they are now... > > > >> right now I can't access the machine, but I recall that the partition > >> is mounted rw, something is wrote onto it and then changed to r/o and > >> among the options there is data=ordered. > > /proc/mounts should show us whether data=ordered is in use. Actually the > > warning is in function which is called only in data=journal mode so that's > > why I think you are using it... > > Ok, got the options: > > ro,noatime,errors=continue,barrier=1,data=ordered 0 0 Can you please post *full* /proc/mounts content? Thanks. It is hard to believe ext3_journalled_writepage() gets called on data=ordered filesystem. Honza > and as expected, the usual warning triggered, but not the message from > your patch: > > [ 173.408622] ------------[ cut here ]------------ > [ 173.408725] WARNING: at fs/ext3/inode.c:1754 > ext3_journalled_writepage+0x55/0x1a7() > [ 173.408850] Hardware name: ProLiant BL465c G7 > [ 173.408940] Pid: 2142, comm: flush-8:0 Not tainted 3.6.3-p #2 > [ 173.409034] Call Trace: > [ 173.409123] [] ? warn_slowpath_common+0x73/0x87 > [ 173.409219] [] ? ext3_journalled_writepage+0x55/0x1a7 > [ 173.409320] [] ? __writepage+0xa/0x21 > [ 173.409414] [] ? write_cache_pages+0x206/0x2f8 > [ 173.409509] [] ? set_page_dirty+0x5e/0x5e > [ 173.409604] [] ? queue_unplugged+0x28/0x34 > [ 173.409698] [] ? generic_writepages+0x3e/0x55 > [ 173.409793] [] ? __writeback_single_inode+0x39/0xd1 > [ 173.409889] [] ? writeback_sb_inodes+0x206/0x392 > [ 173.409983] [] ? __writeback_inodes_wb+0x67/0xa2 > [ 173.410079] [] ? wb_writeback+0xfd/0x18b > [ 173.410172] [] ? bdi_dirty_limit+0x27/0x81 > [ 173.410348] [] ? wb_do_writeback+0x13d/0x1a2 > [ 173.410461] [] ? add_timer_on+0x61/0x61 > [ 173.410584] [] ? bdi_writeback_thread+0x7f/0x13e > [ 173.410700] [] ? queue_work_on+0x16/0x1e > [ 173.410809] [] ? wb_do_writeback+0x1a2/0x1a2 > [ 173.410902] [] ? wb_do_writeback+0x1a2/0x1a2 > [ 173.410996] [] ? kthread+0x81/0x89 > [ 173.411089] [] ? kernel_thread_helper+0x4/0x10 > [ 173.411184] [] ? kthread_worker_fn+0xe0/0xe0 > [ 173.411277] [] ? gs_change+0xb/0xb > [ 173.411367] ---[ end trace 8243403b245b6cfc ]--- > > -- > Fabio -- Jan Kara SUSE Labs, CR