[tulip] link freezing up, switching to 100M?

Jeremy M. Dolan jmd@foozle.turbogeek.org
Sat, 11 Nov 2000 01:53:41 -0600


I have a strange problem, perhaps someone here can help me diagnose.

I have a linux box (linux 2.2.16) with two tulip cards in it, Netgear
FA-310TX's, or, as tulip.c call's them:

Found Lite-On 82c168 PNIC at PCI I/O address 0x9100.
tulip.c:v0.91g-ppc 7/16/99 becker@cesdis.gsfc.nasa.gov
eth0: Lite-On 82c168 PNIC rev 32 at 0x9100, 00:A0:CC:5C:07:8F, IRQ 10.
eth0:  MII transceiver #1 config 3000 status 7809 advertising 01e1.
Found Lite-On 82c168 PNIC at PCI I/O address 0x9200.
eth1: Lite-On 82c168 PNIC rev 32 at 0x9200, 00:A0:CC:58:2F:5C, IRQ 7.
eth1:  MII transceiver #1 config 3000 status 7809 advertising 01e1.

These were both working fine at my old apartment, acting as a
router/NAT machine. Recently I moved into a new apartment and started
having trouble. The card will work fine for a little while, but
eventually (depending on the load) it will die. By die, I mean, the
link basically drops. On the back of the card, the link light is still
on, and the activity light is blinking, but no packets will send or
recieve. As soon as i pull the patch cable out of the card, the 100M
light goes ON (the hub is 10M). That card won't send until I reboot,
but i can take the interface on it down with ifconfig and run of the
other one until it locks up in the same way.

I've switched patch cables, and ports on the hub, but this is the same
hub that everything used to work fine on, a netgear 4 port 10BaseT.

I tried to turn off full duplex and set the options= to a few
differant things, even tried the latest 'test' drivers off the tulip
page.

wierd thing is, now that i've set debug to level 7, it wont do it (of
course). It would do it every time when I went to download this
certain 9 meg file (probably 10 for 10, as i was testing differant
drivers/settings/cables, and usually in the first 100-200k), but with
debug 7, I can download the 9M file fine. Did it twice, no problems.

Since that wouldn't work, I used tulip-diag.c to get whatever info I
could. The tarball attached expands to three directories, 'before',
'after', '100'. Before is when everything is fine. After is when the
lockup occurs. 100 is when i remove the patch cable and the 100M light
goes on on the card. I'm not too familiar with all of the things it
repots (MII, etc), but the differance I noticed was normally, the port
is 'MII, half-duplex', when the 100M light goes on, tulip-diag claims
the port is '10mpbs-serial'.

Anyone have any clues? I'm about ready to chuck all these cables, the
hub, and both NICs, and just grab some new stuff.

PLEASE CC ME ON REPLIES. Thanks for any help!

-- 
Jeremy M. Dolan <jmd@turbogeek.org>
OpenPGP key = http://turbogeek.org/openpgp-key
OpenPGP fingerprint = 494C 7A6E 19FB 026A 1F52  E0D5 5C5D 6228 DC43 3DEE