[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#550733: xfsprogs: kernel oops on log replay during XFS mount when number of regions is corrupt



On Mon, 2009-10-12 at 17:47 +0200, Mateusz Kijowski wrote:
> Package: xfsprogs
> Version: 2.9.8-1lenny1
> Severity: important

This is really a kernel bug, and has been reassigned accordingly.

> Recently I got a corrupt filesystem in qemu image. xfs_repair insists that the filesystem must be
> unmounted cleanly before running. When trying to mount I get:
> 
> [141946.555463] Filesystem "loop0": Disabling barriers, not supported by the underlying device                                                               
> [141946.555463] XFS: correcting sb_features alignment problem                                                                                                
> [141946.555463] XFS mounting filesystem loop0                                                                                                                
> [141946.587706] Starting XFS recovery on filesystem: loop0 (logdev: internal)                                                                                
> [141946.587926] BUG: unable to handle kernel NULL pointer dereference at 0000000000000010                                                                    
> [141946.587926] IP: [<ffffffffa041ad5a>] :xfs:xlog_recover_add_to_trans+0xc1/0xe7                                                                            
> [141946.587926] PGD 19242b067 PUD 24f3f6067 PMD 0                                                                                                            
> [141946.587926] Oops: 0002 [1] SMP                                                                                                                           
> [141946.587926] CPU 1                                                                                                                                        
> [141946.587926] Modules linked in: xfs tun ipt_REJECT xt_tcpudp iptable_filter ip_tables x_tables iscsi_trgt crc32c libcrc32c ipv6 nfsd auth_rpcgss 
> exportfs nfs lockd nfs_acl sunrpc bridge ext4dev jbd2 crc16 kvm_intel kvm loop serio_raw snd_pcm snd_timer snd psmouse soundcore i2c_i801 rng_core 
> snd_page_alloc pcspkr i2c_core i5000_edac edac_core button shpchp pci_hotplug evdev ext3 jbd mbcache dm_mirror dm_log dm_snapshot dm_mod raid1 md_mod 
> ide_cd_mod cdrom ata_generic piix ide_pci_generic ide_core ehci_hcd uhci_hcd e1000e 3w_9xxx sd_mod thermal processor fan thermal_sys ahci libata 
> scsi_mod dock                         
> [141946.587926] Pid: 13307, comm: mount Not tainted 2.6.26-1-amd64 #1
[...]

This was an old kernel version even when you reported this bug.  You
should install the meta-package linux-image-2.6-amd64 which currently
(in Debian 5.0) depends on linux-image-2.6.26-2-amd64.  I don't see any
bug fixes to XFS, but there have been many security fixes.

Anyway, I had a look through the changes since Linux 2.6.26, and I found
that this bug has been fixed.  I intend to include that fix in a stable
update.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: