Ext3 users - Feb 2001 - Boom.. (ok so its 0.0.5b)

Big BadaBOOM

Not entirely sure what went wrong
BUT just after the reported load average on the box
started to go up at a steady rate of about 1.0 per 40 seconds

The box itself had been up for 2 weeks before this happened

Phil
=--

Feb 24 10:25:08 parcelfarce sshd[28545]: Received signal 15; terminating.
Feb 24 10:25:09 parcelfarce kernel: Unable to handle kernel NULL pointer derefer
ence at virtual address 00000000
Feb 24 10:25:09 parcelfarce kernel: current->tss.cr3 = 00101000, %%cr3 =
0010100
0
Feb 24 10:25:09 parcelfarce kernel: *pde = 00000000
Feb 24 10:25:09 parcelfarce kernel: Oops: 0000
Feb 24 10:25:09 parcelfarce kernel: CPU:    1
Feb 24 10:25:09 parcelfarce kernel: EIP:    0010:[ext3_get_inode_loc+49/340]
Feb 24 10:25:09 parcelfarce kernel: EFLAGS: 00010206
Feb 24 10:25:09 parcelfarce kernel: eax: 00000000   ebx: d7b7b000   ecx: d7b7b00
0   edx: c15c2e9c
Feb 24 10:25:09 parcelfarce kernel: esi: 39000003   edi: c1078aa0   ebp: c2a9de6
8   esp: c2a9de34
Feb 24 10:25:09 parcelfarce kernel: ds: 0018   es: 0018   ss: 0018
Feb 24 10:25:09 parcelfarce kernel: Process sshd (pid: 16284, process nr: 142, s
tackpage=c2a9d000)
Feb 24 10:25:09 parcelfarce kernel: Stack: c1078aa0 c2a9de68 0000a4bb d7b7b000 c
014d71a c15c2e9c c2a9de68 c025b4c0
Feb 24 10:25:09 parcelfarce kernel:        d73675e0 c1078aa0 00000001 d74b4c48 c
15c2fbc c1078aa0 c1078aa0 00000001
Feb 24 10:25:09 parcelfarce kernel:        c014d84e d73675e0 c1078aa0 c025b4c0 c
1078aa0 00000000 00000001 c1078b10
Feb 24 10:25:09 parcelfarce kernel: Call Trace: [ext3_orphan_del+370/544] [ext3_
delete_inode+134/328] [ext3_put_inode+10/16] [iput+155/588] [dput+166/328] [__fp
ut+62/72] [fput+23/72]
Feb 24 10:25:09 parcelfarce kernel:        [filp_close+82/92] [do_exit+361/744]
[do_signal+568/724] [sys_select+1014/1360] [sys_select+1346/1360] [signal_return
+20/24]
Feb 24 10:25:09 parcelfarce kernel: Code: 3b 30 76 1b 56 68 f0 65 21 c0 68 06 66
 21 c0 53 e8 be 35 00

Hi,

On Sat, Feb 24, 2001 at 03:25:59PM +0000, Bryce as root, 2600 Meridian Parkway
NC 27737 USA, +1 919 547 0012 x 217,+44 07802 448211
wrote:
> 
> Not entirely sure what went wrong
> BUT just after the reported load average on the box
> started to go up at a steady rate of about 1.0 per 40 seconds
Looks like you hit one of the two big ext3-0.0.6b-fixed bugs: the
original orphan code was reading the inode's orphan list pointers
before taking the superblock lock, so by the time it came to update
the pointers, the original list might have changed.

Bad boy, you shouldn't be running such an old version!  Upgrade
immediately!  This should be fixed in 6b.

Cheers,
 Stephen