PCMCIA 100MBit Eth. Card doesn't work after update (Suse 8.2 -> 9.2)

Hi all,

I've just updated my notebook (an old PIII 600MHz) to Suse 9.2.
Unfortunately quite a few things that worked with Suse 8.2 don't work any more.
Most importantly: Some USB sticks and my PCMCIA 100 MBit Ethernet card
(it's an ALLNET 0142+ Card) :-(

The network problem is very strange:

I have configured a local network (NFS, ssh, ...)
and routing to the internet via another machine in the local net.
The network works. BUT: if I want to transfer a large
amount of data within the local network via the PCMCIA card then
the network connection hangs up within a few seconds.
I can restart the network (/etc/init/network restart) and the
network works again for a few seconds and then hangs up again.

The diagnostic output doesn't change much when the network dies
(lsmod still shows the loaded module; ifconfig still shows
eth0 with a similar output), but no data is transfered anymore.
When I start a data transfer and watch the ping output then ping
shows after a few seconds that the net becomes unavailable.

I would very much appreciate ANY hint which might help me solve the
problem!

Below I attached a lot of diagnostic output. Please tell me if there
is something else that would help you to give me some advice to analyze
the problem.

Thanks for your help in advance,

Jan


PS:
the diagnostic stuff:
1. after restarting the network I see in /var/log/messages:
Jan 15 20:10:15 gin kernel: eth0: found link beat
Jan 15 20:10:15 gin kernel: eth0: autonegotiation complete: 100baseT-FD
selected
Jan 15 20:10:24 gin kernel: eth0: no IPv6 routers present

2. lsmod shows:
...
pcmcia_core 66100 3 axnet_cs,yenta_socket,ds
...

3. dmesg shows:

...
eth0: Asix AX88190: io 0x300, irq 3, hw_addr 00:48:54:3A:82:30
...
Yenta: CardBus bridge found at 0000:00:0a.0 [1043:1386]
Yenta: ISA IRQ mask 0x0498, PCI irq 11
Socket status: 30000410
PCI: Found IRQ 9 for device 0000:00:0a.1
PCI: Sharing IRQ 9 with 0000:00:09.0
Yenta: CardBus bridge found at 0000:00:0a.1 [1043:1386]
Yenta: ISA IRQ mask 0x0098, PCI irq 9
Socket status: 30000820
...

4. /proc/bus/pccard/drivers shows:

axnet_cs 1 1

5. the card is a Allnet ALL 0142+ card

6. the card seems to identify itself as

"NE2000 Compatible Fast Ethernet"

because in /etc/pcmcia/config this string was bound to
the module "pcnet_cs" (which produced the errors in
/var/log/messages:

Dec 29 23:28:25 linux cardmgr[3160]: socket 0: NE2000 Compatible Fast
Ethernet
Dec 29 23:28:25 linux kernel: cs: memory probe 0xa0000000-0xa0ffffff:
clean.
Dec 29 23:28:26 linux kernel: pcnet_cs: this is an AX88190 card!
Dec 29 23:28:26 linux kernel: pcnet_cs: use axnet_cs instead.
Dec 29 23:28:26 linux kernel: pcnet_cs: unable to read hardware net
address for io base 0x300
Dec 29 23:28:26 linux cardmgr[3160]: get dev info on socket 0 failed:
No such device

Therefore I changed the 'bind "pcnet_cs"' to 'bind "axnet_cs"'
which workds (except, of course, that the network dies under
heavy load):

Jan 14 08:16:26 gin /etc/dev.d/net/50-ifup.dev[6193]: eth0
device: NE2000 Compatible Fast Ethernet
Jan 14 08:16:26 gin /etc/dev.d/net/50-ifup.dev[6193]: eth0
configuration: eth-bus-pcmcia

7. ifconfig shows:

eth0 Protokoll:Ethernet Hardware Adresse 00:48:54:3A:82:30
inet Adresse:192.168.0.2 Bcast:192.168.0.255
Maske:255.255.255.0
inet6 Adresse: fe80::248:54ff:fe3a:8230/64
Gültigkeitsbereich:Verbindung
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:22175 errors:0 dropped:0 overruns:0 frame:4
TX packets:12960 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 Sendewarteschlangenlänge:1000
RX bytes:3287020 (3.1 Mb) TX bytes:7015190 (6.6 Mb)
Interrupt:3 Basisadresse:0x300