Hi there,
I've been having a problem with an NFS server on the same box as my
bridge/firewall. Everything else works perfectly (most useful patch I've
found in ages - nice one), but as soon as I create a bit of load on NFS,
I get a kernel panic. I have the complete ksymoops output if it would
help; I'll only include short spam in my first message to the list:
>>EIP; c01f663e <br_nf_post_routing_finish+50/132> <=====
Trace; c01c4f2c <ip_fragment+716/868>
Trace; c01f660c <br_nf_post_routing_finish+0/132>
Trace; c01e54a0 <ip_refrag+68/96>
Trace; c01f660c <br_nf_post_routing_finish+0/132>
Trace; c01f660c <br_nf_post_routing_finish+0/132>
Trace; c01bb310 <nf_iterate+56/136>
...
I am running 2.4.18 patched with
bridge-nf-0.0.6-against-2.4.18.diff
00_brnf_post_routing_paranoia.diff
01_brnf_clean_up_protocol_checks.diff
I added the extra patches in a desperate attempt to make it work
properly. The machine is a PIII450 with 1 EtherExpress Pro/100 and one
NE2000 clone. The interior interface to which NFS speaks is the Pro/100.
The same problem occurs with either the kernelspace NFS server or the
userspace NFS server.
My assumption is that this has something to do with the handling of
fragments. I have picked up bits and pieces from the list archive to
affirm this, but I don't really have the background knowledge to start
digging through the guts of netfilter myself. Any comments, even really
nasty sarcastic ones, would be appreciated at this point.
Oh, and "hiya" as well ;)
Cheers,
James
--
James McDonald <[EMAIL PROTECTED]> - http://www.xanth.co.uk/james/
_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge
