2004-01-22 23:08:18

by Adam Kropelin

Subject: 2.6.1 oops in prune_dcache()

At 4 AM this morning (during cron run, I suppose) a box running 2.6.1
hit the oops below. It locked solid, had to hit the reset button to
reboot it. The machine had been running 2.6.1 for about a week prior
with no problems.

Hardware is single Pentium Pro 200, 128 MB RAM (extensively
memtest86'ed). Kernel is no-SMP, no-preempt, and (obviously) no-highmem.

Feel free to ask for more details if I can help. This is the first oops
I've seen since mid 2.5.x on this machine.

--Adam


Jan 21 04:06:32 print kernel: Unable to handle kernel paging request at virtual address 00008014
Jan 21 04:06:32 print kernel: printing eip:
Jan 21 04:06:32 print kernel: c01570e5
Jan 21 04:06:32 print kernel: *pde = 00000000
Jan 21 04:06:32 print kernel: Oops: 0000 [#1]
Jan 21 04:06:32 print kernel: CPU: 0
Jan 21 04:06:32 print kernel: EIP: 0060:[<c01570e5>] Not tainted
Jan 21 04:06:32 print kernel: EFLAGS: 00010206
Jan 21 04:06:32 print kernel: EIP is at prune_dcache+0xe5/0x130
Jan 21 04:06:32 print kernel: eax: 00008000 ebx: c1dc83e0 ecx: c577da74 edx: c577da74
Jan 21 04:06:32 print kernel: esi: c577da64 edi: c1dc83e0 ebp: 00000072 esp: c11d7e70
Jan 21 04:06:32 print kernel: ds: 007b es: 007b ss: 0068
Jan 21 04:06:32 print kernel: Process kswapd0 (pid: 7, threadinfo=c11d6000 task=c11db2f0)
Jan 21 04:06:37 print kernel: Stack: c11d6000 c11d7eac c0114641 00000202 c11db2f0 c11db310 000658fc 9b6c7589
Jan 21 04:06:37 print kernel: 00000080 0000031b c11d6000 00000080 c0157445 00000080 c0134689 00000080
Jan 21 04:06:37 print kernel: 000000d0 000031d4 000000a8 01620ff0 00000000 00000001 c7ffeb80 000000a8
Jan 21 04:06:37 print kernel: Call Trace:
Jan 21 04:06:37 print kernel: [<c0114641>] schedule+0x41/0x4c0
Jan 21 04:06:37 print kernel: [<c0157445>] shrink_dcache_memory+0x15/0x20
Jan 21 04:06:37 print kernel: [<c0134689>] shrink_slab+0x109/0x160
Jan 21 04:06:37 print kernel: [<c01356cf>] balance_pgdat+0x13f/0x1f0
Jan 21 04:06:37 print kernel: [<c013588b>] kswapd+0x10b/0x110
Jan 21 04:06:37 print kernel: [<c0115d00>] autoremove_wake_function+0x0/0x40
Jan 21 04:06:37 print kernel: [<c0115d00>] autoremove_wake_function+0x0/0x40
Jan 21 04:06:37 print kernel: [<c0135780>] kswapd+0x0/0x110
Jan 21 04:06:37 print kernel: [<c0106fb5>] kernel_thread_helper+0x5/0x10
Jan 21 04:06:37 print kernel:
Jan 21 04:06:37 print kernel: Code: 8b 40 14 85 c0 74 08 56 53 ff d0 5a 59 eb 07 56 e8 66 20 00

2004-01-23 00:22:31

by Andrew Morton

Subject: Re: 2.6.1 oops in prune_dcache()

Adam Kropelin <[email protected]> wrote:
>
> At 4 AM this morning (during cron run, I suppose) a box running 2.6.1
> hit the oops below. It locked solid, had to hit the reset button to
> reboot it. The machine had been running 2.6.1 for about a week prior
> with no problems.
>
> Hardware is single Pentium Pro 200, 128 MB RAM (extensively
> memtest86'ed).

Yes, but it is old.

> Kernel is no-SMP, no-preempt, and (obviously) no-highmem.
>
> Feel free to ask for more details if I can help. This is the first oops
> I've seen since mid 2.5.x on this machine.
>
> --Adam
>
>
> Jan 21 04:06:32 print kernel: Unable to handle kernel paging request at virtual address 00008014
> Jan 21 04:06:32 print kernel: printing eip:
> Jan 21 04:06:32 print kernel: c01570e5
> Jan 21 04:06:32 print kernel: *pde = 00000000
> Jan 21 04:06:32 print kernel: Oops: 0000 [#1]
> Jan 21 04:06:32 print kernel: CPU: 0
> Jan 21 04:06:32 print kernel: EIP: 0060:[<c01570e5>] Not tainted
> Jan 21 04:06:32 print kernel: EFLAGS: 00010206
> Jan 21 04:06:32 print kernel: EIP is at prune_dcache+0xe5/0x130
> Jan 21 04:06:32 print kernel: eax: 00008000 ebx: c1dc83e0 ecx: c577da74 edx: c577da74

Bit 15 of %eax got flipped. The kernel indexed off it and oopsed.

This is most likely a hardware failure.

2004-01-23 00:31:47

by Adam Kropelin

Subject: Re: 2.6.1 oops in prune_dcache()

Andrew Morton wrote:
> Adam Kropelin <[email protected]> wrote:
>>
>> At 4 AM this morning (during cron run, I suppose) a box running 2.6.1
>> hit the oops below. It locked solid, had to hit the reset button to
>> reboot it. The machine had been running 2.6.1 for about a week prior
>> with no problems.

<snip>

>> Jan 21 04:06:32 print kernel: eax: 00008000 ebx: c1dc83e0 ecx:
>> c577da74 edx: c577da74
>
> Bit 15 of %eax got flipped. The kernel indexed off it and oopsed.
>
> This is most likely a hardware failure.

Yikes...I should have noticed that bit. You're certainly right. Guess I'm in
the market for some RAM. Sorry for the noise.

--Adam