2003-01-07 17:27:31

by Matt Bernstein

[permalink] [raw]
Subject: 2.5.54-mm2 oops under io load

I was trying to demonstrate that xmms won't skip in 2.5 no matter what
load you throw at the computer (by doing some stuff such as 'find /usr
-type f -exec cat "{}" \; >/dev/null &' a few times each after hopefully
long enough to defeat the caches), when:

kernel BUG at mm/slab.c:1516!
invalid operand: 0000
CPU: 0
EIP: 0060:[<c013405a>] Not tainted
EFLAGS: 00210002
EIP is at cache_alloc_refill+0x24a/0x2a0
eax: 00000003 ebx: c3318040 ecx: 00000014 edx: 00000013
esi: c3318058 edi: 00000013 ebp: 00000034 esp: ce1b5e48
ds: 007b es: 007b ss: 0068
Process pidof (pid: 22002, threadinfo=ce1b4000 task=c4c66640)
Stack: d0c34344 00000000 d7d2b010 d7fed9b0 d7fed9b8 d7d2b000 d7fed9a4 00200246
000001d0 c092814c c0134228 d7fed9a4 000001d0 00000007 c14bcddc fffffff4
c14bce44 c015fb9c d7fed9a4 000001d0 00000000 00000000 cc0aa00b 00b118fa
Call Trace:
[<c0134228>] kmem_cache_alloc+0x78/0x80
[<c015fb9c>] d_alloc+0x1c/0x1c0
[<c01562dd>] real_lookup+0xad/0x100
[<c015653f>] do_lookup+0x9f/0xb0
[<c015687e>] link_path_walk+0x32e/0x5f0
[<c0134fbb>] cache_free_debugcheck+0x15b/0x1f0
[<c01574ae>] open_namei+0x7e/0x390
[<c01481b3>] filp_open+0x43/0x70
[<c014860b>] sys_open+0x5b/0x90
[<c0109387>] syscall_call+0x7/0xb

Code: 0f 0b ec 05 89 d7 26 c0 e9 48 fe ff ff 8b 7c 24 2c 8b 57 38
<6>note: pidof[22002] exited with preempt_count 1

This is under a Redhat Phoebe set-up (only altered with Rusty's modutils,
and a viciously hacked rc.sysinit) using its (default) ext3-with-indexes
stuff:

Linux r2-pc 2.5.54 #1 Thu Jan 2 23:09:11 GMT 2003 i686 athlon i386 GNU/Linux

Gnu C 3.2.1
Gnu make 3.79.1
util-linux 2.11w
mount 2.11w
module-init-tools 0.9.7
e2fsprogs 1.32
jfsutils error:
reiserfsprogs 3.6.4
pcmcia-cs 3.1.31
PPP 2.4.1
isdn4k-utils 3.1pre4
Linux C Library 2.3.1
Dynamic linker (ldd) 2.3.1
Procps 2.0.10
Net-tools 1.60
Kbd 1.08
Sh-utils 4.5.3
Modules Loaded mousedev hid usbmouse uhci_hcd usbcore ohci1394 ieee1394
parport_pc parport es1371 soundcore gameport ac97_codec sd_mod aic7xxx scsi_mod
nfs lockd sunrpc autofs4 mga via_agp agpgart atkbd i8042 serio e100
af_packet unix

PS modules will be in a different order as it takes a fight to get my USB
mouse recognised..

PPS xmms was still playing :-) and. until the stress test, this kernel was
stable for 4 days