[eepro100] eepro100.c v1.13, i82555? and "Command unit failed to mark command 00000000 as complete at 70053"

[Jan Just Keijser]

hi all,

we've got a bunch of eepro100 cards that show some weird behaviour even with
the eepro100 v1.13 driver:

from /var/log/messages:

May  8 01:05:35 localhost kernel: eepro100.c:v1.13 1/9/2001 Donald Becker
<becker@scyld.com>
May  8 01:05:35 localhost kernel:   http://www.scyld.com/network/eepro100.html
May  8 01:05:35 localhost kernel: eth0: Intel PCI EtherExpress Pro100 at
0xd0846000, 00:02:B9:45:BA:17, IRQ 9.
May  8 01:05:35 localhost kernel:   Board assembly 668081-002, Physical
connectors present: RJ45
May  8 01:05:35 localhost kernel:   Primary interface chip i82555 PHY #1.
May  8 01:05:35 localhost kernel:   General self-test: passed.
May  8 01:05:35 localhost kernel:   Serial sub-system self-test: passed.
May  8 01:05:35 localhost kernel:   Internal registers self-test: passed.
May  8 01:05:35 localhost kernel:   ROM checksum self-test: passed
(0x04f4518b).
May  8 01:05:35 localhost kernel:   Receiver lock-up workaround activated.
May  8 01:05:35 localhost kernel: eth1: Intel PCI EtherExpress Pro100 at
0xd0848000, 00:02:B9:45:BA:18, IRQ 10.
May  8 01:05:35 localhost kernel:   Board assembly 668081-002, Physical
connectors present: RJ45
May  8 01:05:35 localhost kernel:   Primary interface chip i82555 PHY #1.
May  8 01:05:35 localhost kernel:   General self-test: passed.
May  8 01:05:35 localhost kernel:   Serial sub-system self-test: passed.
May  8 01:05:35 localhost kernel:   Internal registers self-test: passed.
May  8 01:05:35 localhost kernel:   ROM checksum self-test: passed
(0x04f4518b).
May  8 01:05:35 localhost kernel:   Receiver lock-up workaround activated.
...
May  8 17:46:04 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 506058.
May  8 17:46:26 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 506093.
May  8 17:46:48 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 513126.
May  8 17:46:54 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 513132.
May  8 17:49:04 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 838966.
May  8 17:52:08 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 839163.
May  8 17:55:08 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 909378.
May  8 17:56:08 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 952172.
May  8 17:56:30 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 962960.
May  8 17:59:02 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 1212885.
May  8 17:59:14 localhost kernel: eth0: Command unit failed to mark command
00000000 as complete at 1212903.


These error/warning messages occur esp when the box is hit with unicast and
multicast traffic.
Fortunately, the network never goes down though , but performance does take a
hit.

Is this a known bug/anomaly? How can we get rid of this?

TIA,

JJK / Jan Just Keijser
Cisco Systems International BV