Ext3 users - May 2003 - kjournald oops on 2.4.20

I have a NFS server with stock 2.4.20 patched with the 5 data=journal
patches, plus the 2.4 LSM and NSA's selinux.

The partition is mounted with the options: 
rw,noatime,nosuid,nodev,data=journal,errors=remount-ro

The oops seems to happen when the fs is under load.  I don't think I
encountered this behavior with a non LSM/selinux kernel, so it may have
something to do with those, but I figured that I'd check here first.


Oops: 0000
CPU:    0
EIP:    0010:[<c017aed2>]    Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010287
eax: 00000000   ebx: 00000000   ecx: c20984f0   edx: c2098390
esi: 00000000   edi: c5e5bfa0   ebp: c5b6c0c0   esp: c9849e4c
ds: 0018   es: 0018   ss: 0018
Process kjournald (pid: 277, stackpage=c9849000)
Stack: c2098390 00000000 c63510e0 c63510e0 00000000 c1360000 c1360108
00000000
       c0178a30 c1360000 c63510e0 00000004 001be827 00000296 c1360094
00000000
       00000000 00000000 00000000 c31576a0 c5e5b5b0 001be827 c1d20780
c9ae10e0
Call Trace:    [<c0178a30>] [<c017c0aa>] [<c017bf10>]
[<c010591e>]
[<c017bf30>]
Code: 8b 76 28 89 1c 24 e8 e3 f8 ff ff 01 44 24 04 39 fb 75 eb 3b
 
 
>>EIP; c017aed2 <__journal_clean_checkpoint_list+52/80>   <==== 
>>ecx; c20984f0 <_end+1cee46c/c45dfdc>
>>edx; c2098390 <_end+1cee30c/c45dfdc>
>>edi; c5e5bfa0 <_end+5ab1f1c/c45dfdc>
>>ebp; c5b6c0c0 <_end+57c203c/c45dfdc>
>>esp; c9849e4c <_end+949fdc8/c45dfdc>
 
Trace; c0178a30 <journal_commit_transaction+150/12b0>
Trace; c017c0aa <kjournald+17a/230>
Trace; c017bf10 <commit_timeout+0/10>
Trace; c010591e <arch_kernel_thread+2e/40>
Trace; c017bf30 <kjournald+0/230>
 
Code;  c017aed2 <__journal_clean_checkpoint_list+52/80>
00000000 <_EIP>:
Code;  c017aed2 <__journal_clean_checkpoint_list+52/80>   <====   0:  
8b 76 28                  mov    0x28(%esi),%esi   <====Code;  c017aed5
<__journal_clean_checkpoint_list+55/80>
   3:   89 1c 24                  mov    %ebx,(%esp,1)
Code;  c017aed8 <__journal_clean_checkpoint_list+58/80>
   6:   e8 e3 f8 ff ff            call   fffff8ee <_EIP+0xfffff8ee>
Code;  c017aedd <__journal_clean_checkpoint_list+5d/80>
   b:   01 44 24 04               add    %eax,0x4(%esp,1)
Code;  c017aee1 <__journal_clean_checkpoint_list+61/80>
   f:   39 fb                     cmp    %edi,%ebx
Code;  c017aee3 <__journal_clean_checkpoint_list+63/80>
  11:   75 eb                     jne    fffffffe <_EIP+0xfffffffe>
Code;  c017aee5 <__journal_clean_checkpoint_list+65/80>
  13:   3b 00                     cmp    (%eax),%eax
 

-- 
Chris PeBenito
<pebenito@ieee.org>
AIM: PeBenito78
ICQ#: 10434387

"Engineering does not require science. Science helps
a lot, but people built perfectly good brick walls 
long before they knew why cement works."-Alan Cox

Hi,

On Mon, 2003-05-05 at 05:57, Christopher J. PeBenito
wrote:
> I have a NFS server with stock 2.4.20 patched with the 5 data=journal
> patches, plus the 2.4 LSM and NSA's selinux.
> 
> The partition is mounted with the options: 
> rw,noatime,nosuid,nodev,data=journal,errors=remount-ro
> 
> The oops seems to happen when the fs is under load.  
You mean it's reproducible, and always shows the same footprint?
> I don't think I
> encountered this behavior with a non LSM/selinux kernel, 
Me neither.  If you can reproduce this on such an unpatched kernel, that
would be really useful info.

Cheers,
 Stephen