From: Joseph Fannin Subject: JBD2/ext4 error Date: Sun, 2 Nov 2008 19:48:27 -0500 Message-ID: <20081103004827.GA2766@nineveh.local> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii To: linux-ext4@vger.kernel.org Return-path: Received: from mail-gx0-f12.google.com ([209.85.217.12]:64398 "EHLO mail-gx0-f12.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754026AbYKCAsc (ORCPT ); Sun, 2 Nov 2008 19:48:32 -0500 Received: by gxk5 with SMTP id 5so715090gxk.13 for ; Sun, 02 Nov 2008 16:48:31 -0800 (PST) Content-Disposition: inline Sender: linux-ext4-owner@vger.kernel.org List-ID: Hi, I'm hitting what's probably a bug in ext4 on one of my boxes. It always happens on the boot partition, which is extentless, since it seems likely GRUB will choke on extents. On with the backtrace, already: [ 4586.330338] JBD: grub-probe wants too many credits (266 > 256) [ 4586.330384] ext4_da_writepages: jbd2_start: 18691 pages, ino 39; err -28 [ 4586.330398] Pid: 7365, comm: grub-probe Not tainted 2.6.27.4-1jhf-686 #4 [ 4586.330456] [] ext4_da_writepages+0x165/0x2c0 [ext4] [ 4586.330562] [] scsi_pool_alloc_command+0x30/0x4a [scsi_mod] [ 4586.330647] [] blk_rq_map_sg+0xf1/0x240 [ 4586.330671] [] kobject_get+0xf/0x13 [ 4586.330690] [] ext4_da_writepages+0x0/0x2c0 [ext4] [ 4586.330726] [] do_writepages+0x20/0x30 [ 4586.330745] [] __writeback_single_inode+0x15e/0x314 [ 4586.330767] [] _spin_unlock_irq+0x24/0x2e [ 4586.330779] [] mark_page_accessed+0x37/0x53 [ 4586.330793] [] read_cache_page_async+0xf9/0x101 [ 4586.330802] [] blkdev_readpage+0x0/0xc [ 4586.330823] [] generic_sync_sb_inodes+0x229/0x32d [ 4586.330835] [] sync_inodes_sb+0x7e/0x86 [ 4586.330845] [] __fsync_super+0xa/0x61 [ 4586.330861] [] fsync_super+0x8/0x14 [ 4586.330869] [] fsync_bdev+0x14/0x2b [ 4586.330879] [] blkdev_ioctl+0xac/0x70d [ 4586.330888] [] kobject_get+0xf/0x13 [ 4586.330897] [] exact_lock+0x7/0xd [ 4586.330907] [] kobj_lookup+0x112/0x140 [ 4586.330927] [] exact_match+0x0/0x7 [ 4586.330936] [] unlock_kernel+0x3b/0x45 [ 4586.330944] [] do_open+0x20c/0x288 [ 4586.330956] [] blkdev_open+0x0/0x4d [ 4586.330966] [] blkdev_open+0x25/0x4d [ 4586.330975] [] __dentry_open+0x15a/0x226 [ 4586.330985] [] nameidata_to_filp+0x1c/0x2c [ 4586.330994] [] do_filp_open+0x327/0x64c [ 4586.331006] [] block_ioctl+0x13/0x16 [ 4586.331016] [] block_ioctl+0x0/0x16 [ 4586.331024] [] vfs_ioctl+0x1c/0x5f [ 4586.331035] [] do_vfs_ioctl+0x3c7/0x3f9 [ 4586.331044] [] _spin_lock+0x10/0x12 [ 4586.331067] [] virt_to_head_page+0x1f/0x2a [ 4586.331086] [] do_sys_open+0xaa/0xb2 [ 4586.331094] [] sys_ioctl+0x41/0x58 [ 4586.331103] [] sysenter_do_call+0x12/0x2f This kernel is simply 2.6.27.4 with the 2.6.27-ext4-2 patch applied, though I've seen it also in a kernel built from the Debian 2.6.26 sources with some ext4 patch applied too -- I'm not sure which, I don't have the patch or the source around anymore. The backtrace varies a bit, but ext4_da_writepages is always on top. It's not always grub-probe that sets it off -- often it happens while generating a new initrd. The error message will continue to repeat until the fs is unmounted. It can then be mounted without a problem with ext3, though the files that were being modified when ext4 started complaining will be corrupted. Ext3 works fine; fsck thinks the fs is fine. The filesystem is mounted in writeback mode, and the journal size is 4100k, according to dumpe2fs. Another box here has a similar boot partition with a 4114k journal, and I don't think it's ever done this. Actually, I have two other boxes with the same setup. Their disks are ide, though -- a sata disk in one, and ide driven via the pata drivers in the other. The one I'm seeing a problem with is SCSI-only, using that aic7000-series driver. I don't know what's causing this. Is there something I can do to track this down? -- Joseph Fannin jfannin@gmail.com