I am running into a problem that I suspect has been discussed earlier, but
I cannot find any information on it despite aggressive searching.

I boot into NetBSD on my IIci in my office and it comes up OK. I configured
networking and that starts OK. I can then telnet in or out. Everything is
wonderful.

This lasts for 2-5 minutes and then the system freezes and isn't so
wonderful any more :-(  I am certain that the lockup is related to ae0, as
if I leave the interface unplugged into the net, the system stays up
indefinitely. This is how I did all the configuring, in fact. I want to run
httpd on it for the office intranet though, so I am looking for ideas.

When it hangs, I can hit the panic button and raom around in the debugger.
Here's the traceback of the stack:
_nmihand (5c90000,7f,20,22,100) +2c
_lev7intr (7cac00, c) +12
_lbv_nubus_intr (1) +56
_rbv_intr(ae1dac) +4a
_lev2int (?)
_mi_switch(2004,4,0,ae1f08) +14
_sys_select (7f3a00,ae1f80,ae1f80) +218
_syscall (5d) +133
_trap() + c

So I can step a few instructions and continue. Sometimes (not always) this
activity, just hitting the panic button, I suspect, results in a miraculous
recovery. The system will print
blah-logging-blah ae0: warning - receive ring buffer overrun
and then we're back up, for a minute or two, anyway.

Now there is no guarantee that these receive overruns had anything to do
with anything. They are probably just a symptom of having the system
stopped while the network filled the buffer.

I can repeat the failure mode in about 5 minutes max at any time, and I'm
willing to play around in the debugger some, so as to provide info if
needed. The office Lan has all different kinds of traffic - IP, IPX, Apple,
DEC, etc.

Info about my setup:

Mac IIci 8MB internal video, 3COM EtherLink NB network interface.
NetBSD 1.1 release (obtained from ftp.netbsd.org).

Originally I ran the full NetBSD 1.1 release, but the kernel panicked (jump
to 0) upon boot with that one, so at his suggestion I am now running a
slightly newer kernel compiled by jweiss@mit.edu (yes, a relation). No more
panics. It's a bit newer, but probably not much. Is there any reason to
believe this has been fixed in an even newer kernel?

My interface card is a 3Com EtherLink NB card.
I thought that this was compatible with the apple cards and thus was
supported so I went ahead with this, but I can't find it listed explicitly
anywhere. I doubt it is the cause of the problem, as I suspect if it
weren't supported I would never have gotten as far as I did. (Please check
my logic here.)

Sorry, it's a long post - I tried to cram everything into it.

-srw
_________________________________________________________________________
 Steven R. Weiss   |    srw@hvcn.org      | If you think education is
Computer Scientist | (313) 995-8250 x5632 |  expensive, try ignorance.