XFS & cryptoloop problems

[Udo Wolter <uwp@xxxxxxxxxxxxxxxx>]

Hi!

I'm using XFS for more than 2 years now and normally I've been very
happy with it. Since some months I'm using cryptoloop to crypt my disks
and put XFS systems onto that loop, e.g.:

losetup -e twofish /dev/loop0 /dev/sda5
mkfs.xfs /dev/loop0

Now it's perfectly mountable and usable. I also using IDE-disks as USB-disks
for having the convenience to swap them, switch them on and off etc. So,
even though it's device sda, it's still an IDE disk behind it.

Yesterday one of my disks had problems. During a filetransfer to the
disk it just got stuck and I had to reboot. After reboot I wanted to
mount the disk again:

mount -o encryption=twofish /dev/sda5 /mnt

I got this:

XFS mounting filesystem loop0
Starting XFS recovery on filesystem: loop0 (dev: loop0)
Access to block zero: fs: <loop0> inode: 61277749 start_block : 0 start_off : 0 blkcnt : 0 extent-state : 0 
------------[ cut here ]------------
kernel BUG at fs/xfs/support/debug.c:106!
invalid operand: 0000 [#1]
PREEMPT SMP 
Modules linked in: ipt_MASQUERADE iptable_nat ip_conntrack_irc bridge tun ipv6 lp binfmt_misc iptable_filter ip_tables ide_cd cdrom parport_pc parport 8250_pnp 8250 serial_core pcspkr rtc 8139cp i2c_i801 sd_mod usblp uhci_hcd eth1394 intel_mch_agp intel_agp ohci1394 ieee1394 usb_storage ehci_hcd usbcore i2c_isa it87 i2c_sensor evdev i2c_core cryptoloop loop blowfish psmouse snd_seq_oss snd_seq_midi snd_seq_midi_event snd_seq snd_ice1724 snd_ice17xx_ak4xxx snd_ac97_codec snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd_page_alloc snd_ak4xxx_adda snd_mpu401_uart snd_rawmidi snd_seq_device snd soundcore ip_conntrack_ftp ip_conntrack pppoe pppox ppp_generic slhc r128 drm agpgart 8139too mii crc32 e1000 af_packet
CPU:    0
EIP:    0060:[<c029db91>]    Not tainted VLI
EFLAGS: 00010246   (2.6.11.6-grsec) 
EIP is at cmn_err+0xa1/0xc0
eax: 00000000   ebx: c03e2b20   ecx: 00000000   edx: 00000001
esi: 00000000   edi: c04bcb40   ebp: 00000000   esp: f56399c4
ds: 007b   es: 007b   ss: 0068
Process mount (pid: 8341, threadinfo=f5639000 task=f5b92000)
Stack: c03d4128 c03d9fc8 c04bcb40 00000286 c03e2b20 00000000 00000001 f5639ad8
       c0242d87 00000000 c03e2b20 f5a33c40 03a70635 00000000 00000000 00000000
       00000000 00000000 00000000 00000000 00000000 00000000 00000000 f546d62c 
Call Trace:
 [<c0242d87>] xfs_bmap_search_extents+0x107/0x130
 [<c0245ba3>] xfs_bunmapi+0x1b3/0x10c0
 [<c0186ba3>] __alloc_pages+0x2e3/0x420
 [<c03b6e86>] _spin_lock+0x16/0x90
 [<c03b728d>] _spin_unlock+0xd/0x30
 [<c018aa02>] cache_alloc_refill+0x142/0x230
 [<c028658d>] xfs_trans_log_inode+0x2d/0x60
 [<c026d70a>] xfs_itruncate_finish+0x1fa/0x460
 [<c028c62d>] xfs_inactive+0x50d/0x560
 [<c02956d3>] pagebuf_offset+0x33/0x40
 [<c026b334>] xfs_itobp+0x114/0x260
 [<c029da31>] vn_rele+0xd1/0xe0
 [<c029bfc8>] linvfs_clear_inode+0x18/0x30
 [<c01bd4d6>] clear_inode+0xe6/0x130
 [<c01be478>] generic_delete_inode+0x128/0x150
 [<c02ba331>] _atomic_dec_and_lock+0x31/0x50
 [<c01be673>] iput+0x63/0x90
 [<c027d65e>] xlog_recover_process_iunlinks+0x33e/0x3e0
 [<c027e9e8>] xlog_recover_finish+0xa8/0xe0
 [<c02754cc>] xfs_log_mount_finish+0x2c/0x30
 [<c0280176>] xfs_mountfs+0x826/0xef0
 [<c03b728d>] _spin_unlock+0xd/0x30
 [<c027f668>] xfs_readsb+0x198/0x200
 [<c0270baf>] xfs_ioinit+0x1f/0x40
 [<c0287ede>] xfs_mount+0x2ae/0x4c0
 [<c029cbf3>] vfs_mount+0x43/0x50
 [<c029ca0e>] linvfs_fill_super+0x9e/0x200
 [<c02b9987>] snprintf+0x27/0x30
 [<c01dc0a4>] disk_name+0xb4/0xc0
 [<c01aa19e>] sb_set_blocksize+0x2e/0x60
 [<c01a9b2a>] get_sb_bdev+0x10a/0x150
 [<c01bff70>] alloc_vfsmnt+0x90/0xd0
 [<c029cba0>] linvfs_get_sb+0x30/0x40
 [<c029c970>] linvfs_fill_super+0x0/0x200
 [<c01a9d87>] do_kern_mount+0x57/0xe0
 [<c01c11cb>] do_new_mount+0xbb/0x100
 [<c01c18b1>] do_mount+0x181/0x1d0
 [<c01c16d3>] copy_mount_options+0x63/0xc0
 [<c01c1ccf>] sys_mount+0x9f/0xe0
 [<c014a8f7>] syscall_call+0x7/0xb
Code: b8 40 cb 4b c0 89 44 24 08 8b 04 ad 80 27 42 c0 89 44 24 04 e8 41 59 ec ff 8b 54 24 0c b8 64 27 42 c0 e8 83 97 11 00 85 ed 75 08 <0f> 0b 6a 00 d8 9f 3d c0 83 c4 10 5b 5e 5f 5d c3 eb 0d 90 90 90

It's a 2.6.11.6 kernel with grsecurity enabled. Ok, I thought, maybe
it's something old or weird. After this I took the disk and hung it to
another PC, similar problem:

May 30 10:25:43 roeddirs kernel: XFS mounting filesystem loop0
May 30 10:25:43 roeddirs kernel: Starting XFS recovery on filesystem: loop0 (dev: loop0)
May 30 10:25:43 roeddirs kernel: ------------[ cut here ]------------
May 30 10:25:43 roeddirs kernel: PREEMPT SMP 
May 30 10:25:43 roeddirs kernel: Modules linked in: lp ide_cd cdrom ns558 gameport parport_pc parport 8250_pnp sd_mod dvb_ttpci dvb_core saa7146_vv video_buf saa7146 v4l1_compat v4l2_common videodev ves1820 stv0299 tda8083 stv0297 sp8870 firmware_class ves1x93 ttpci_eeprom i2c_i801 i2c_core ehci_hcd usb_storage uhci_hcd usbcore intel_mch_agp intel_agp ipv6 binfmt_misc tun crc32 eth1394 8250 serial_core cryptoloop loop blowfish video1394 evdev ohci1394 ieee1394 psmouse snd_intel8x0 snd_ac97_codec snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd soundcore snd_page_alloc hpt366 autofs4 af_packet nls_iso8859_1 ntfs reiserfs agpgart e1000 rtc
May 30 10:25:43 roeddirs kernel: CPU:    0
May 30 10:25:43 roeddirs kernel: EIP:    0060:[<c02516b1>]    Not tainted VLI
May 30 10:25:43 roeddirs kernel: EFLAGS: 00010246   (2.6.11.10) 
May 30 10:25:43 roeddirs kernel: EIP is at cmn_err+0xa1/0xc0
May 30 10:25:43 roeddirs kernel: eax: 00000000   ebx: c03835e0   ecx: f6288020   edx: 00000000
May 30 10:25:43 roeddirs kernel: esi: 00000000   edi: c049fb40   ebp: 00000000   esp: f56a59cc
May 30 10:25:43 roeddirs kernel: ds: 007b   es: 007b   ss: 0068
May 30 10:25:43 roeddirs kernel: Process mount (pid: 4613, threadinfo=f56a4000 task=f6288550)
May 30 10:25:43 roeddirs kernel: Stack: c0374bf8 c037aa98 c049fb40 00000202 c03835e0 00000000 00000001 f56a5ae0 
May 30 10:25:43 roeddirs kernel:        c01f6c07 00000000 c03835e0 f7cb98c0 03a70635 00000000 00000000 00000000 
May 30 10:25:43 roeddirs kernel:        00000000 00000000 00000000 00000000 00000000 00000000 00000000 f5f5692c 
May 30 10:25:43 roeddirs kernel: Call Trace:
May 30 10:25:43 roeddirs kernel:  [<c01f6c07>] xfs_bmap_search_extents+0x107/0x130
May 30 10:25:43 roeddirs kernel:  [<c01f9a23>] xfs_bunmapi+0x1b3/0x10c0
May 30 10:25:43 roeddirs kernel:  [<c0115113>] scheduler_tick+0x63/0x320
May 30 10:25:43 roeddirs kernel:  [<c013cb23>] __alloc_pages+0x2e3/0x420
May 30 10:25:43 roeddirs kernel:  [<c0140940>] cache_alloc_refill+0x140/0x230
May 30 10:25:43 roeddirs kernel:  [<c023a3dd>] xfs_trans_log_inode+0x2d/0x60
May 30 10:25:43 roeddirs kernel:  [<c022158a>] xfs_itruncate_finish+0x1fa/0x460
May 30 10:25:43 roeddirs kernel:  [<c024038d>] xfs_inactive+0x50d/0x560
May 30 10:25:43 roeddirs kernel:  [<c0249403>] pagebuf_offset+0x33/0x40
May 30 10:25:43 roeddirs kernel:  [<c021f1b4>] xfs_itobp+0x114/0x260
May 30 10:25:43 roeddirs kernel:  [<c0251551>] vn_rele+0xd1/0xe0
May 30 10:25:43 roeddirs kernel:  [<c024fae8>] linvfs_clear_inode+0x18/0x30
May 30 10:25:43 roeddirs kernel:  [<c0172096>] clear_inode+0xe6/0x130
May 30 10:25:43 roeddirs kernel:  [<c0173038>] generic_delete_inode+0x128/0x150
May 30 10:25:43 roeddirs kernel:  [<c025ed11>] _atomic_dec_and_lock+0x31/0x50
May 30 10:25:43 roeddirs kernel:  [<c0173233>] iput+0x63/0x90
May 30 10:25:43 roeddirs kernel:  [<c02314ae>] xlog_recover_process_iunlinks+0x33e/0x3e0
May 30 10:25:43 roeddirs kernel:  [<c0232838>] xlog_recover_finish+0xa8/0xe0
May 30 10:25:43 roeddirs kernel:  [<c022931c>] xfs_log_mount_finish+0x2c/0x30
May 30 10:25:43 roeddirs kernel:  [<c0233fc6>] xfs_mountfs+0x826/0xef0
May 30 10:25:43 roeddirs kernel:  [<c02334b8>] xfs_readsb+0x198/0x200
May 30 10:25:43 roeddirs kernel:  [<c02249ff>] xfs_ioinit+0x1f/0x40
May 30 10:25:43 roeddirs kernel:  [<c023bd2e>] xfs_mount+0x2ae/0x4c0
May 30 10:25:43 roeddirs kernel:  [<c0250713>] vfs_mount+0x43/0x50
May 30 10:25:43 roeddirs kernel:  [<c025052e>] linvfs_fill_super+0x9e/0x200
May 30 10:25:43 roeddirs kernel:  [<c025e367>] snprintf+0x27/0x30
May 30 10:25:43 roeddirs kernel:  [<c01906c4>] disk_name+0xb4/0xc0
May 30 10:25:43 roeddirs kernel:  [<c015f89e>] sb_set_blocksize+0x2e/0x60
May 30 10:25:43 roeddirs kernel:  [<c015f22a>] get_sb_bdev+0x10a/0x150
May 30 10:25:43 roeddirs kernel:  [<c0174a00>] alloc_vfsmnt+0x90/0xd0
May 30 10:25:43 roeddirs kernel:  [<c02506c0>] linvfs_get_sb+0x30/0x40
May 30 10:25:43 roeddirs kernel:  [<c0250490>] linvfs_fill_super+0x0/0x200
May 30 10:25:43 roeddirs kernel:  [<c015f487>] do_kern_mount+0x57/0xe0
May 30 10:25:43 roeddirs kernel:  [<c0175bbc>] do_new_mount+0x9c/0xe0
May 30 10:25:43 roeddirs kernel:  [<c0176277>] do_mount+0x157/0x1b0
May 30 10:25:43 roeddirs kernel:  [<c0170062>] dput+0x152/0x1d0
May 30 10:25:43 roeddirs kernel:  [<c01760c3>] copy_mount_options+0x63/0xc0
May 30 10:25:43 roeddirs kernel:  [<c017666f>] sys_mount+0x9f/0xe0
May 30 10:25:43 roeddirs kernel:  [<c01028af>] syscall_call+0x7/0xb
May 30 10:25:43 roeddirs kernel: Code: b8 40 fb 49 c0 89 44 24 08 8b 04 ad 80 dc 3b c0 89 44 24 04 e8 f1 85 ec ff 8b 54 24 0c b8 64 dc 3b c0 e8 f3 bd 10 00 85 ed 75 08 <0f> 0b 6a 00 a8 aa 37 c0 83 c4 10 5b 5e 5f 5d c3 eb 0d 90 90 90

This time it's a standard 2.6.11.10 kernel. Both kernels have SMP/SMT
enabled because they should do hyperthreading (intel machines). What I'll
try to do is to losetup the disk again and copy the looped disk with dd
to another disk. Maybe I'm getting my data back. BTW, the mount itself hung
up with segmentation fault.

Any hints for this problem? Thanx!

Mermgfurt,
                Udo
--
Udo Wolter                      |  /"\
email:    uwp@xxxxxxxxxxxxxxxx  |  \ / ASCII RIBBON CAMPAIGN
www:      www.dicke-aersche.de  |   X    AGAINST HTML MAIL
dark:    heaven@xxxxxxxxxxxxxx  |  / \