> - what does the m0n0 route table look like?
$ netstat -r
Routing tables

Internet:
Destination        Gateway            Flags    Refs      Use  Netif Expire
default            zur01a06.sunrise.c UGS         0   302484    ng0
localhost          localhost          UH          0      468    lo0
192.168.0          link#2             UC          0        0    vr1
192.168.10         link#1             UC          0        0    vr0
192.168.10.20      00:90:4b:b1:c8:73  UHLW        1    24978    vr0   1069
192.168.100        link#3             UC          0        0    vr2
zur01a06.sunrise.c adsl-89-217-1-4.ad UH          1        0    ng0

As you can see I'm using PPPOE on the WAN interface. 192.168.10.20 is the XP
box that I'm just typing this very message ;-).
 
> - have you tried to reboot the m0n0 box after switching from interface 1
to interface 2 with the XP box?
>
> The reason I ask the second question is because many times (with Cisco and
other branded routers, and some switches for that matter), if I move a
device from one physical > interface to another, the ARP cache requires a
flush in order to realize that the device has moved.

In fact I did, with no change in my inability to connect. Also I did not
have a switch between XP box and m0n0 for these tests, just plain cat5e
cables (even tried if a cross-over cable would make a difference - it did
not which was what I expected). 
What baffles me is the fact that I can connect using vr0 without any
problems, while connecting to vr2 does not result in any data between the
connected hosts (using even the same cable) while both hosts pretend to have
a valid link (and agree on using 100MB/s). It seems smehow that there is a
barrier between the physical port and the Ethernet stack within m0n0 or the
Alix board.

Thanks for bearing with me

Stefan