Subject: RE: we0 driver broken?
To: tech-kern@netbsd. org <>
From: Martin Husemann <>
List: tech-kern
Date: 04/23/1999 08:21:29
Sorry to get on your nerves again, but this realy is a killer for me:

with -current sources the ethernet card on my file server stoped working.
With a kernel from 6 or 7 of April it works fine. The -current sources work
for me on other machines with different network cards. It's not a hardware
problem, as the old kernel still works fine on the problematic machine. What
happens for each and every packet received is this:

[/tmp] root@rumolt > tcpdump -vv -i we0 -s1500 -x
tcpdump: listening on we0
08:03:52.292001 > icmp: echo request (ttl
32, id 26381)
                         4500 003c 670d 0000 2001 f4fd d408 cb53
                         d408 cb51 0800 d25b 0100 7a00 6162 6364
                         6566 6768 696a 6b6c 6d6e 6f70 7172 7374
                         7576 7761 6263 6465 6667 6869
08:03:52.300101 > icmp: time exceeded
in-transit (ttl 255, id 215)
                         4500 0038 00d7 0000 ff01 7c37 d408 cb51
                         d408 cb53 0b00 9fa3 0000 0000 4500 003c
                         670d 0000 0101 13fe d408 cb53 d408 cb51
                         0800 d25b 0100 7a00

Sending seems to be OK, at least hwart reports it received a icmp time
exceeded in-transit message.

Can someone with the necessary background knowledge please check if this is
a driver bug, my fault, whatever? (I compiled that kernel without
optimization, so if its a compiler bug, it's not optimization.)

Can someone with CVS access please check if something changed in
sys/dev/isa/if_we.c or sys/dev/ic/dp8390.c and maybe mail me the old version
(I don't have a backup of the old working kernel sources)?

I can't debug that machine with kgdb easily, as it's the file server and all
my other machines are currently diskless :-(

Thanks very much in advance,