kernel oopses

[Martin Siegert]

On Tue, Jan 29, 2002 at 07:40:21PM +0100, Joachim Worringen wrote:
> 
> When does the crash happen? When loading the driver or after a zillion
> of packets have been transmitted? Anyway, check if you can turn on some
> debug output (use the source, Luke) and see what the driver just did
> before it crashed (the debug output does go into the system log).

It crashes after quite a few (probably not a zillion :-) packets have been
sent:

# /NPtcp -t -h b002 -o sk9dlin.out -P
Latency: 0.000115
Now starting main loop
  0:         1 bytes 5912 times -->    0.17 Mbps in 0.000115 sec
  1:         2 bytes 5699 times -->    0.35 Mbps in 0.000115 sec
  2:         3 bytes 5709 times -->    0.52 Mbps in 0.000115 sec
  3:         4 bytes 3775 times -->    0.68 Mbps in 0.000123 sec
  4:         6 bytes 4205 times -->    1.03 Mbps in 0.000122 sec
  5:         8 bytes 2805 times -->    1.37 Mbps in 0.000114 sec
  6:        12 bytes 3495 times -->

... that's where it crashes.

> And what does Syskonnect say? If it's such a harsh problem, they should
> have noticed by now - if not, they'll be happy to hear about it!

that's basically what started this thread: I had the two cards for evaluation
and thought that I send them the oops output.

OTOH - no I'm not going to spend more time on debugging their driver.
I took the opportunity to learn how to capture the oops output
(thanks for the responses!) because that is - as Rob Latham pointed
out - something I may need in the future: my 96 node dual athlon cluster
goes into production within the next few weeks (without a sk9d21 card, but
with a sk9843 which works fine).

Cheers,
Martin