kernel crash

[Gordon Messmer <yinyang@eburg.com>]

[-- Attachment #1: Type: text/plain, Size: 1041 bytes --]

We recently have had trouble with kernel crashes on a Linux 2.4.20 box 
used as a NAT gateway.  The machine is running Slackware 8.1 with a 
vanilla 2.4.20 kernel.  The kernel configuration is attached.  The 
message-log attachment contains the full dmesg output, as captured by a 
remote syslog server, as well as the last messages received before the 
box hung:

May 28 16:18:08 tog-wakko4 kernel: LIST_DELETE: ip_conntrack_core.c:303 
`&ct->tuplehash[IP_CT_DIR_REPLY]'(f687fbc4) not in &ip_conntrack_hash 
[hash_conntrack(&ct->tuplehash[IP_CT_DIR_REPLY].tuple)].
May 28 16:18:52 tog-wakko4 kernel: LIST_DELETE: ip_conntrack_core.c:303 
`&ct->tuplehash[IP_CT_DIR_REPLY]'(f687cd24) not in &ip_conntrack_hash 
[hash_conntrack(&ct->tuplehash[IP_CT_DIR_REPLY].tuple)].
May 28 16:18:52 tog-wakko4 kernel: LIST_DELETE: ip_conntrack_core.c:303 
`&ct->tuplehash[IP_CT_DIR_REPLY]'(f687ce64) not in &ip_conntrack_hash 
[hash_conntrack(&ct->tuplehash[IP_CT_DIR_REPLY].tuple)].

What further information would be useful in tracking down this problem?

[-- Attachment #2: message-log.txt.gz --]
[-- Type: application/x-gzip, Size: 5563 bytes --]

[-- Attachment #3: kernel-config.gz --]
[-- Type: application/x-gzip, Size: 3915 bytes --]