>> 1) hinv reports 1gb, netbsd kernel only 256mb. Found this:
>> http://mail-index.netbsd.org/port-sgimips/2004/03/24/0000.html anyone
knows about changes on that point until now?
>
> No changes, until someone implements real mips64 support.
> (maybe we should take a look openbsd/mips64)

I've tried openbsd and it has the same problem, although I didn't try to
force it with a kernel parameter. I might try it again later.

>> Actually I'd really like to know what this crime
>> device is :)
>
> See crime(4) man page.

You've got me there :)

> Hmm, what does "ifconfig mec0" say?

When on the switch:

mec0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        address: 08:00:69:0e:bd:18
        media: Ethernet autoselect (100baseTX full-duplex)
        status: active
        inet 192.168.48.139 netmask 0xffffff00 broadcast 192.168.48.255
inet6 fe80::a00:69ff:fe0e:bd18%mec0 prefixlen 64 scopeid 0x1

The box is on another location, so I can't put it on the hub but in a few
days...

>
>> 4) shall I be punished for ignoring 'arplookup: unable to enter address
for xxx.xxx.xxx.xxx@00:12:17:51:e0:f1 on mec0 (host is not on local
network)'. The address mentioned, appears to be the public address of
the
>> machine that nats for the NetBSD box...
>
> I'm not sure if it's mec specific. Does it occur on tl(4) too?

Haven't tried tl(4) yet... But it forgot to mention that the natting
machine only has one nic, so it has local and public address are on the
same card. I guess NetBSD complains when it sees multiple subnets on one
card...

>
>> 5) everybody ignores 'pci_addr_fixup' right?
>
> It's just info, not a warning. Maybe we could wrap
> those printf messages with #ifdef DEBUG, though.
> ---
> Izumi Tsutsui
>

Thanks for the info,
Wouter