Re: Kernel panic issues

[Marcelo Tosatti <marcelo.tosatti@cyclades.com>]

Nathan,

On Sun, Sep 05, 2004 at 05:05:27PM -0700, Randy.Dunlap wrote:
> On Sun,  5 Sep 2004 12:12:24 -0700 Nathan wrote:
> 
> | Hi,  I have a server running debian 3.0r1 kernel 2.4.25 and I get these kernel 
> | panic about 5 times this week.  If anyone can tell me what it means it would be 
> | greatly appreciated.  Any additional instructions on how to read kernel panic 
> | dumps would also be appreciated.
> 
> Denis Vlasenko recently did a "howto find oops location" for 2.6.x,
> but it's probably the best reference for you to look at.
> It's here:
>   http://marc.theaimsgroup.com/?l=linux-kernel&m=109257016020612&w=2
> 
> 
> | asdasdkernel BUG as slab.c:1263!
> | Invalid operand: 0000
> | CPU:	0
> | EIP:	0010:[<c012609d>] Not tainted
> | EFLAGS: 00010012
> | eax: f31eafff	ebx: c19ad700	ecx: 00000001	edx: 00000001
> | esi: f31ea800	edi: f31eabd3	ebp: c02cfca8	esp: c02cfc8c
> | ds: 0018	es: 0018	ss: 0018
> | Process swapper (pid: 0, stackpage=c02cf000)
> | Stack:	f69657fc c03397e0 00000020 00000800 00012800 f31eabd3 00000246 c02cfcc4
> | 	c01f6b5e 0000065c 00000020 00000008 0000001c f74ec160 c02cfcf f887afe3
> | 	00000620 00000020 00000008 0000001c f74ec160 c01fa090 00000000 f6ebec
> | Call Trace:	[<c01f6b5e>] [<f887afe3>] [<c01fa090>] [<f887ae58>] [<f887ae58>]
> | 	[<c0107ee0>] [<c010806f>] [<c0125f2c>] [<c0231d11>] [<c02320c8>] 
> | [<c0207b60>]
> | 	[<f887b4ef>] [<c010806f>] [<c0207b60>] [<c02010b7>] [<c0207b60>] 
> | [<c02079f5>]
> | 	[<c0207b60>] [<c01fa40b>] [<c01fa4ad>] [<c01fa5bf>] [<c011552b>] 
> | [<c010809d>]
> | 	[<c0105260>] [<c0105260>] [<c0105260>] [<c0105260>] [<c0105286>] 
> | [<c01052f9>]
> | 	[<c0105000>] [<c010502a>]
> | 
> | Code: 0f 0b ef 04 60 33 26 c0 8b 7d f4 f7 c7 00 04 00 00 74 36 b8
> |  <0>Kernel panic: Aiee, Killing interrupt handler!
> | In interrupt handler - not syncing
> 
> The stack addresses are useless without associating some of (your)
> kernel symbols with them.  Please read REPORTING-BUGS in the top
> level of the kernel source tree for full bug-reporting info, and see
> Documentation/Changes on where to get 'ksymoops' if you don't
> already have it, then run this panic message text thru ksymoops.
> That should tell the function call chain to get to slab.c.

Yes please run ksymoops on the output as Randy suggests.


> kernel BUG as slab.c:1263!

Are you using SMP? 

The BUG happens because kmem_check_poison_obj finds a POISON_END 
byte not at the end the object .

        if (cachep->flags & SLAB_POISON)
                if (kmem_check_poison_obj(cachep, objp))
                        BUG();
Whats your config?

Have you been able to capture the same oops ie same or similar backtrace,
also "BUG at slab.c:1263!" more than once?

Maybe hardware bug, but potentially not (maybe some sort of kernel memory
overwrite).

Manfred, any clues?