2.5.75 slab corruption

Hello,

For me 2.5.75 is worse than anything since I am testing 2.5.x (2.5.66), how can I debug this one ?,
the call stack seems very small,
where can I begin ?
I never had a stable 2.5 on this computer,
thinking about hard problem ?

Nicolas.

Jul 12 19:39:57 hal9003 kernel:  printing eip:
Jul 12 19:39:57 hal9003 kernel: 78116ef1
Jul 12 19:39:57 hal9003 kernel: Oops: 0000 [#1]
Jul 12 19:39:57 hal9003 kernel: CPU:    0
Jul 12 19:39:57 hal9003 kernel: EIP:    0060:[<78116ef1>]    Not tainted
Jul 12 19:39:57 hal9003 kernel: EFLAGS: 00010047
Jul 12 19:39:57 hal9003 kernel: EIP is at 0x78116ef1
Jul 12 19:39:57 hal9003 kernel: eax: 00000088   ebx: e77ca73c   ecx: c02e00c0   
edx: 00000027
Jul 12 19:39:57 hal9003 kernel: esi: f799e100   edi: c02de788   ebp: e5565fbc   
esp: e5565f9c
Jul 12 19:39:57 hal9003 kernel: ds: 007b   es: 007b   ss: 0068
Jul 12 19:39:57 hal9003 kernel: Process kpat (pid: 14040, threadinfo=e5564000 
task=e5fd1880)
Jul 12 19:39:57 hal9003 kernel: Stack: c0147abc e5a6ad98 080986f8 0000061c 
e5fd1880 00000003 080986f8 080981b8
Jul 12 19:39:57 hal9003 kernel:        e5564000 c010900a 00000003 080986f8 
0000061c 080986f8 080981b8 bfffeb08
Jul 12 19:39:57 hal9003 kernel:        0000061c c010007b 0000007b 00000004 
40f6d728 00000073 00000246 bfffeae8
Jul 12 19:39:57 hal9003 kernel: Call Trace:
Jul 12 19:39:57 hal9003 kernel:  [sys_write+63/93] sys_write+0x3f/0x5d
Jul 12 19:39:57 hal9003 kernel:  [<c0147abc>] sys_write+0x3f/0x5d
Jul 12 19:39:57 hal9003 kernel:  [work_resched+5/22] work_resched+0x5/0x16
Jul 12 19:39:57 hal9003 kernel:  [<c010900a>] work_resched+0x5/0x16
Jul 12 19:39:57 hal9003 kernel:
Jul 12 19:39:57 hal9003 kernel: Code:  Bad EIP value.