One machine has a 3c590 that is about 9 months old; unlike many on this
list, I have had no problems whatsoever with this card. I'm running 0.30
with the one line memory leak patch posted to linux-kernel a few days ago.
This machine suddenly got stuck in a slow scrolling message below. The
machine was catatonic on the network; a module unload / reload fixed it.
eth0: Transmitter access conflict
eth0: Transmitter access conflict
eth0: Transmitter access conflict
About that same time another machine running an SMC Ultra started
occasionally displaying messages like those below; these are from 8390.c.
eth0: bogus packet size: 1532, status=0x1, nxpg=0xe.
eth0: bogus packet size: 1532, status=0x1, nxpg=0x3f.
eth0: bogus packet size: 1532, status=0x1, nxpg=0x1d.
eth0: bogus packet size: 1532, status=0x1, nxpg=0x11.
The messages have since stopped, so I wonder if something bogus was coming
in from the Internet. I did capture a tcpdump -w binary dump during that
time just in case it would be useful; it is about 7 MB ungzipped. If
someone could suggest what to look for in it I would be glad to scan the
log for what might have been going on.
Anyway, it seemed curious to me to wonder what might have tripped the
3c59x driver into locking up that also caused the Ultra driver to report
problems. It might not, but I thought this would perhaps give more
insight into the 3c59x driver problems.
-- Michael