random network lockups

Steven S. Dick (ssd@nevets.oau.org)
Mon, 8 Apr 96 01:05 EDT

I have four network cards I have used on my two machine home network.
(Obviously, two at a time.) Using these cards, I have been getting
random lockups from every kernel I've tried.

The main machine runs only Linux. I've tried both an NE200 card and a WD8013
card. The other machine can run either DOS, or Linux with nfs mounted /usr,
and I've only tried the ne2000 in that machine.

For kernels versions 1.2.x and 1.3.58 - 1.3.71, the main machine will lock up
hard with the ne2000 if I run X on the remote machine _and_ start up PPP
on the main machine. Sometimes the lockup occurs in the first 10 minutes,
sometimes within the first two hours. It never locked up under kernel 1.0,
and it did not lock up with the WD8013 card in place.

However, with the WD8013 card and kernels 1.3.69 - 1.3.81, telnet
sessions from the dos machine would occasionally freeze for
between 10 and 20 seconds. The dos machine is using an old version of
Clarkson telnet which I've not previously had a problem with.
During one of these lockups, I tried pinging the dos machine.
The first 10 pings went through, and then pings 11-101 did not,
and then the session unlocked and further pings did get through.

I've been running with the NE2000 card in the main machine for
a week now (1.3.81), and not had a problem with telnet locking up, so it
isn't a problem specific to the kernel version--apparently it's
a driver problem of some sort. I haven't had a chance to attempt
to see of the hard lockup problem still occurs.

I'm now running 1.3.84 with the ne2000 card. I'll see if I can
trigger the ne2000 lockup this week, and possibly switch to the wd8013
card sometime later to see if it still gives problems under .84.

Can anyone suggest ways to debug either problem?