Intel EtherExpress Pro Transmitter crash with 2.2.9 (Ipv6 related?)
Jason Gunthorpe
jgg@ualberta.ca
Mon May 24 19:17:59 1999
[Please CC, I am not on all the lists :>]
Hi all,
I just upgraded a machine from 2.2.5 to 2.2.9 and had the ethernet driver
fail completely within about 4 hours. I suspect this may have something to
do with the fact that I enabled IPv6 support and connected a tunnel to the
machine (it was live on the 6bone at the time) These are the messages I
got:
May 24 05:33:10 pandora kernel: eth0: Trying to restart the transmitter...
May 24 05:33:20 pandora kernel: eth0: Transmit timed out: status 0050
0000 at
36558/36574 command 000c0000.
May 24 05:33:20 pandora kernel: eth0: Trying to restart the transmitter...
May 24 05:34:35 pandora kernel: eth0: Transmit timed out: status 0050
0000 at
37451/37466 command 00000000.
May 24 05:34:35 pandora kernel: eth0: Trying to restart the transmitter...
May 24 05:34:55 pandora kernel: eth0: Transmit timed out: status 0090
0000 at
37466/37482 command 000c0000.
May 24 05:35:30 pandora last message repeated 7 times
May 24 05:36:30 pandora last message repeated 12 times
May 24 05:37:30 pandora last message repeated 12 times
May 24 05:38:30 pandora last message repeated 12 times
It repeats the last message one once a minute till it was rebooted. The
first messages started almost instantly after booting but the card
appeared fine so I let it go.
The driver prints this out when it boots:
May 24 03:57:33 pandora kernel: eth0: Intel EtherExpress Pro 10/100 at
0x1420, 00:90:27:2D:97:A7, IRQ 10.
May 24 03:57:33 pandora kernel: Board assembly 689661-004, Physical
connectors present: RJ45
May 24 03:57:33 pandora kernel: Primary interface chip i82555 PHY #1.
May 24 03:57:33 pandora kernel: General self-test: passed.
May 24 03:57:33 pandora kernel: Serial sub-system self-test: passed.
May 24 03:57:33 pandora kernel: Internal registers self-test: passed.
May 24 03:57:33 pandora kernel: ROM checksum self-test: passed
(0x24c9f043).
May 24 03:57:33 pandora kernel: Receiver lock-up workaround activated.
If necessary I can probably do some testing, it comes up rather quickly.
Thanks,
Jason