networking problems on PC164

[Tobias Stefan Richter <tsr@achos.com>]

I have problems using the network on my PC164 alpha.
After modprobe'ing the NIC's module and ifconfig'ing
eth0 the system completely locks up after a random 
time of usage (immediately to some minutes up to about
an hour). But no network (usage) - no lock.

The problem should not be located in the NIC's driver
as I tried an ISA NE2000 clone and PCI cards from
Kingston (tulip driver), 3Com and Realtek (8139too).

The time it needs to lock up depends a bit on the 
kernel used, but none of 2.2.18, 2.4.2, 2.4.3 and 
2.4.4pre6 works correctly.

With 2.4.4pre6 and the 8139too card I also tried 
the 'PIO instead of MMIO' driver option. It said:

  8139too Fast Ethernet driver 0.9.16
  Assertion failed! ioaddr != NULL, 8139too.c,rtl_init_one,line=927
  eth0 Realtek RTL8139 Fast Ethernet at 0x8000, 01:00:aa:aa:00:40 IRQ 18

And then locked up the machine right after 'ifconfig eth0 <IPADDR>'

The PIO style again yields no such error messages but brings
these messages when pinging another host:

PING 10.17.26.2: 56 data bytes
64 bytes from 10.17.26.2: icmp_seq=0 ttl=255 time=0.4 ms
wrong data byte #8 should be 0x8 but was 0x6b
        c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 27 28 29 2a 2b 
        2c 2d 2e 2f db db db db db db db db db db db db 
64 bytes from 10.17.26.2: icmp_seq=1 ttl=255 time=0.4 ms
wrong data byte #8 should be 0x8 but was 0xf1
        c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 27 28 29 2a 2b 
        2c 2d 2e 2f db db db db db db db db db db db db 
64 bytes from 10.17.26.2: icmp_seq=2 ttl=255 time=0.4 ms
wrong data byte #8 should be 0x8 but was 0xc8
        c d e f 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f 20 21 22 23 24 25 26 27 28 29 2a 2b 
        2c 2d 2e 2f db db db db db db db db db db db db 

on the console and also

Apr 24 22:43:23 dafoe kernel: ping(1977): unaligned trap at 00000001200030e4: 0000000120027444 29 1
Apr 24 22:43:23 dafoe kernel: ping(1977): unaligned trap at 0000000120003110: 000000012002743c 29 2

in the syslog.

Trying other NICs and kernel versions gives similar output.

The should have worked alright before with Windows NT and Red Hat,
so I hope the cause to trouble is not its hardware, although I
already suspect some hardware failure (motherboard maybe).

Any ideas?

tobias

P.S. I tried my best to send helpful information, if there's anything
missing blame it on me being new to kernel issues like this.