network problems (was: Re: booting with floppy)

[Klaus Heinz <heinz%netbsd.org@localhost>]

I wrote:

> I'm getting a lot of device timeouts from ne0. In fact, after the shell prompt
> above, the system is not really usable due to those timeouts.
> 
> The network interface card is a simple ne2000 clone:
> 
>   ne0 at isa0 port 0x300-0x31f irq 5
>   ne0: NE2000 (RTL8019) Ethernet
>   ne0: 10base2, 10baseT, 10baseT-FDX, auto, default [0x00 0x30] auto
> 
> and I had no problems with it in BeOS. I guess, I will try a different
> ISA network card (3com Etherlink III) and a new config file and kernel.

I tried 3 different cards, apart from the NE2000 clone mentioned above:

  SMC 8416
    I'm not sure, NetBSD has a driver for this one. I tried the we driver
    and it did not detect the card.

  3com Etherlink III (3c509) 
    The ep driver did not detect the card. On NetBSD/i386 it works for me.

  WD8013EPC
    The we driver detected the card but I cannot see packets with
    tcpdump (NetBSD) or snoop (Solaris) although the hub's
    activity LED is blinking when the BeBox is sending BOOTP or RARP packets.
    The only explanation I have for this behaviour is the card sending
    invalid Ethernet packets which the other NetBSD or Solaris box do not
    bother to process :-/

The SMC8416 and 3com cards were set to 0x300, irq 5, no plug-and-play,
the same as the (badly) working NE2000 clone.
(btw. does the BeBox hardware support PnP at all? With BeOS I never cared
about that).
The detected WD8013 uses 0x300 and irq 10, iomem at 0xcc000.

Is there some extended debug information for card detection in NetBSD?
I thought the 3com card would work best (the other cards are older,
crappier or maybe not supported), but the kernel never found this one.

I will probably try a different hub and different cables but I don't think
one of those is the cause for my problems.

ciao
     Klaus