[OpenAFS] recurring 1.3.85 server Oops

Kevin openafs@gnosys.biz
Tue, 26 Jul 2005 15:07:52 -0400

Chaskiel M Grundman wrote:

> This is not an openafs problem. However, if you do follow it up with
> anyone else, first disable the afs kernel module and re-reproduce the
> problem, so that afs is not blindly blamed for the problem

Thanks for your reply.  After much pulling of hair and wringing of hands and
trial after trial after trial to find corrupted filesystems, I've found that 
that the root cause is almost certainly bad memory.

This baffles me because this memory is a replacement for another set of bad
memory and I tested it less than a year ago immediately after installing this
set and it passed all tests more than 10 times without even a single error.

Now, the memory tests fail within 10 seconds of starting memtest86+.

Sorry for the unintentional red herring.  Bizarre coincidence that upgrading
to 1.3.85 and a failure of RAM happened simultaneously.

Does this mean anything noteworthy to the openafs developers?  I'm not
sure if this file (mm/rmap.c) is a part of openafs or not.  Since I've confirmed
bad RAM, it may be meaningless, but thought I'd bring it up since it's so
clear ("kernel BUG at..."):

Jul 22 08:51:32 aphrodite ------------[ cut here ]------------
Jul 22 08:51:32 aphrodite kernel BUG at mm/rmap.c:482!
Jul 22 08:51:32 aphrodite invalid operand: 0000 [#4]
Jul 22 08:51:32 aphrodite Modules linked in: libafs parport_pc lp
parport snd_pcm_oss snd_mixer_oss snd_seq_oss snd_seq_midi_event snd_seq
snd_seq_device snd_intel8x0 snd_ac97_codec snd_pcm snd_timer snd
soundcore snd_page_alloc