Subject: Re: Tap(4) interface in NetBSD?
To: Andrew Brown <atatat@atatdot.net>
From: Manuel Bouyer <bouyer@antioche.eu.org>
List: current-users
Date: 04/28/2003 21:19:29
On Sun, Apr 27, 2003 at 03:45:14PM -0400, Andrew Brown wrote:
> >> i guess i must be wrong, but i presume that if i read from it, i get
> >> copies of all ethernet frames received by the host (which is somewhat
> >> like reading from an AF_INET, SOCK_RAW socket), but that doesn't seem
> >> of much utility to me.
> >
> >Think a simulator like SIMH-VAX. 
> 
> right, i've used that, but that uses bpf.  because it has to.
> 
> it also asserts its own ethernet address, so if the kernel didn't
> rewrite the source ethernet address when transmitting the packet
> (never mind deciding which interface on which to transmit it), and
> then rewrite it *back* when replies were received, it wouldn't help
> much.

That would help a lot. You'd have a virtual ethernet network between
simh and the tap device. Then you can decide what to do with it:
route between the tap device and the real ethernet, possibly with nat,
or bridge them, or make it a completely private network.
Much more flexible than using bpf.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--