[epic-bug] EPIC100 problems SMSC EPIC/C 83c175 based PCMCIA card.

[mr matsui]

I'm trying to get my network up and running under the 2.4 test kernel
series. I've installed the latest version 2.4.0-test11-ac2 as I've noted
that the changelog says there are Epic100 updates.
My network consists of a laptop with a Psion Gold Card Netglobal
56K+10/100Mb CardBus card and a duron based server with a D-Link DFE 530TX
NIC connected using a crossover cable.

I've had great problems getting a stable network connection. There are a
number of tests  I've used to test the network,
1- copy of a file from one machine to another, generally the gzipped kernel
(command scp)
2- start X and log into the server, run gnomehack on the server (command
X -query server)
3 - ping the server/laptop (ping -f server/laptop).
4 - share the modem connection with the server( using ipchains).

Part of bootup messages,
Nov 23 09:12:39 Elric kernel: cs: cb_alloc(bus 6): vendor 0x10b8, device
0x0006
Nov 23 09:12:39 Elric kernel: PCI: Enabling device 06:00.0 (0000 -> 0003)
Nov 23 09:12:39 Elric kernel: PCI: Found IRQ 10 for device 06:00.0
Nov 23 09:12:39 Elric kernel: PCI: The same IRQ used for device 00:04.0
Nov 23 09:12:39 Elric kernel: PCI: The same IRQ used for device 00:04.1
Nov 23 09:12:39 Elric kernel: epic100.c:v1.09 5/29/2000 Written by Donald
Becker <becker@scyld.com>
Nov 23 09:12:39 Elric kernel:   http://www.scyld.com/network/epic100.html
Nov 23 09:12:39 Elric kernel:  (unofficial 2.4.x kernel port, version 1.1.5,
September 7, 2000)
Nov 23 09:12:39 Elric kernel: PCI: Setting latency timer of device 06:00.0
to 64
Nov 23 09:12:39 Elric kernel: eth0: SMSC EPIC/C 83c175 at 0xc8810000, IRQ
10, 00:60:63:06:1b:0e.
Nov 23 09:12:39 Elric kernel: eth0: MII transceiver #1 control 3100 status
7809.
Nov 23 09:12:39 Elric kernel: eth0: Autonegotiation advertising 01e1 link
partner 0000.
Nov 23 09:12:39 Elric kernel: call_usermodehelper[/sbin/hotplug]: no root fs
Nov 23 09:12:39 Elric kernel: PCI: Enabling device 06:00.1 (0000 -> 0003)
Nov 23 09:12:39 Elric kernel: PCI: Found IRQ 10 for device 06:00.1
Nov 23 09:12:39 Elric kernel: PCI: The same IRQ used for device 00:04.0
Nov 23 09:12:39 Elric kernel: PCI: The same IRQ used for device 00:04.1
Nov 23 09:12:39 Elric kernel: PCI: The same IRQ used for device 06:00.0
Nov 23 09:12:39 Elric kernel: call_usermodehelper[/sbin/hotplug]: no root fs

Messages after starting X and logging into my server (X -query corum).
Nov 23 09:12:40 Elric kernel: eth0: Setting full-duplex based on MII #1 link
partner capability of 41e1.
Nov 23 09:12:41 Elric /usr/sbin/cron[171]: (CRON) INFO (pidfile fd = 3)
Nov 23 09:12:41 Elric /usr/sbin/cron[172]: (CRON) STARTUP (fork ok)
Nov 23 09:14:39 Elric kernel: NETDEV WATCHDOG: eth0: transmit timed out
Nov 23 09:14:39 Elric kernel: eth0: Transmit timeout using MII device, Tx
status 4003.
Nov 23 09:14:39 Elric kernel: eth0: Restarting the EPIC chip, Rx 1149/1149
Tx 1055/1065.
Nov 23 09:14:39 Elric kernel: eth0: epic_restart() done, cmd status 000a,
ctl 0512 interrupt 240000.
Nov 23 09:15:09 Elric kernel: NETDEV WATCHDOG: eth0: transmit timed out
Nov 23 09:15:09 Elric kernel: eth0: Transmit timeout using MII device, Tx
status 4003.
Nov 23 09:15:09 Elric kernel: eth0: Restarting the EPIC chip, Rx 1149/1149
Tx 1065/1075.
Nov 23 09:15:09 Elric kernel: eth0: epic_restart() done, cmd status 000a,
ctl 0512 interrupt 240000.
Nov 23 09:15:49 Elric kernel: eth0: Setting full-duplex based on MII #1 link
partner capability of 41e1.

Messages copying 2.4.0test11 kernel between both machines
Nov 23 09:17:49 Elric kernel: spurious 8259A interrupt: IRQ7.
Nov 23 09:19:10 Elric kernel: eth0: Setting full-duplex based on MII #1 link
partner capability of 41e1.

Tests 1 and 2 fail. The network can be recovered by taking the network down
(ifdown eth0) and restarting (ifup eth0).

Is there anything I can do to correct these errors, is there any other
information that I can provide that might help identify/fix the errors ?

Diarmuid Drew.