On Sat, 22 Apr 2000, der Mouse wrote:

> >> The hardware/firmware in normal Ethernet interfaces receives packets
> >> sent to all-ones and packets sent to its own ("unicast") address.
> > Not neccesarily.  The DELUA for instance don't think
> > ff:ff:ff:ff:ff:ff is any different from 01:00:00:00:00.
> 
> Weird.  That's certainly a lovely little gotcha waiting to bite someone
> someday.

Well, it bit me when I was writing the ethernet stuff for my IP stack
under RSX. Didn't get a thing until I said that I wanted the broadcast
address.
(Hint, ARP failed pretty bad too... :-)

> >> [...]
> > Yup. But like I said, some of them don't treat all ones any special.
> > There is really no good reason to do so either.
> 
> I'm not sure that last sentence is true - doesn't the Ethernet spec
> specify all-ones as the broadcast address, in which case not receiving
> it is *very* unusual behavior and hence it would be reasonable for
> hardware to do so, either always or by default?

I haven't check the real ethernet spec, but atleast a DEQNA manual I have
here separates all-ones from other multicast addresses, so you might have
a point. Unfortunately I haven't got a DELUA manual close by...

> Whether or not the Ethernet spec says so, that's definitely how current
> Ethernet practice works. :-)

That's definitely true. But that don't prevent the hardware from expecting
you to list the all-ones address.

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol