2002-02-12 07:57:36

by Raphael Manfredi

Subject: 2.4.18-pre7 - known issue with ext3?

I've got the following panic trace on the console this morning:

invalid operand: 0000
CPU: 0
EIP: 0010:[__insert_into_lru_list+28/92] Not tainted
EFLAGS: 00010286
eax: 00000000 ebx: 00000002 ecx: cce5ce40 edx: c02b5e94
esi: cce5ce40 edi: cce5ce40 ebp: cfa11640 esp: cefb7ea0
ds: 0018 es: 0018 ss: 0018
Process perl (pid: 14685, stackpage=cefb7000)
Stack: 00000002 c012b4c5 cce5ce40 00000002 cce5ce40 00000400 c012b4d5 cce5ce40
c012be08 cce5ce40 00001000 00027000 00000000 cfa11640 00000400 00000000
c012c357 cfa11640 c1225a80 00000000 00001000 c1225a80 40018000 cfa11640
Call Trace: [__refile_buffer+73/80] [refile_buffer+9/16]
[__block_commit_write+124/192] [generic_commit_write+51/92]
[ext3_commit_write+290/436] [generic_file_write+1184/1720]
[ext3_file_write+67/76] [sys_write+142/196] [system_call+51/64]

Are there known problems with the implementation of ext3 on this kernel, or
is the root cause elsewhere, and it's only an accident that it was in ext3
code at the time?

Raphael

2002-02-12 08:11:48

by Andrew Morton

Subject: Re: 2.4.18-pre7 - known issue with ext3?

Raphael Manfredi wrote:
>
> I've got the following panic trace on the console this morning:
>
> invalid operand: 0000
> CPU: 0
> EIP: 0010:[__insert_into_lru_list+28/92] Not tainted
> EFLAGS: 00010286
> eax: 00000000 ebx: 00000002 ecx: cce5ce40 edx: c02b5e94
> esi: cce5ce40 edi: cce5ce40 ebp: cfa11640 esp: cefb7ea0
> ds: 0018 es: 0018 ss: 0018
> Process perl (pid: 14685, stackpage=cefb7000)
> Stack: 00000002 c012b4c5 cce5ce40 00000002 cce5ce40 00000400 c012b4d5 cce5ce40
> c012be08 cce5ce40 00001000 00027000 00000000 cfa11640 00000400 00000000
> c012c357 cfa11640 c1225a80 00000000 00001000 c1225a80 40018000 cfa11640
> Call Trace: [__refile_buffer+73/80] [refile_buffer+9/16]
> [__block_commit_write+124/192] [generic_commit_write+51/92]
> [ext3_commit_write+290/436] [generic_file_write+1184/1720]
> [ext3_file_write+67/76] [sys_write+142/196] [system_call+51/64]
>
> Are there known problems with the implementation of ext3 on this kernel, or
> is the root cause elsewhere, and it's only an accident that it was in ext3
> code at the time?
>

Something corrupted a buffer-head from your freelist.

Please send me your .config and a description of the hardware.
Also a description of what sorts of things the machine is
being used for. Also please run memtest86 for 12 hours or
more and let me know the results.

Thanks.

-