Message-ID: <3C657C70.9050300@sgi.com>
Date: Sat, 09 Feb 2002 13:45:52 -0600
From: Stephen Lord <lord@sgi.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
MIME-Version: 1.0
To: Thomas Duffy <tduffy@afara.com>
CC: Linux XFS Mailing List <linux-xfs@oss.sgi.com>
Subject: Re: kernel panic (corruption of task struct)
References: <1013222059.2794.35.camel@tduffy-lnx.afara.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-linux-xfs@oss.sgi.com
Precedence: bulk

Thomas Duffy wrote:

>I am running the 2.4.9-13SGI_XFS_1.0.2smp xfs kernel on a 2P athlon box.
>
>I have been seeing random lockups every now and again for a few months
>now (ever since upgrading to redhat 7.2)
>
>I enabled kdb to see what happened.
>
>Basically, cc1 is always the offending process.  I have seen this thrice
>now.  It is causing the task_struct to be munged.  At offset 0x1c into
>the struct, 4 copies of the uid and 4 copies of gid of the user who is
>running the cc1 are being slammed in...causing the machine to panic.
>
>I don't know if this is XFS related or not, but I thought I would
>inquire on this list before taking it to a larger (lkml) audience.
>
>Any help would be great.
>
>Thanks!
>
>-tduffy
>
Stack overflow looks likely here. I don't know of any in xfs, but xfs in 
combination
with other things might cause problems. We had one case in LVM 
snapshotting where
LVM was putting large chunks of stuff on the stack. That code should not 
be in your
kernel though, unless you added it.

Steve