On 2019-05-10 02:01, Paul Koning wrote:
On May 9, 2019, at 7:55 PM, Johnny Billquist <b...@softjar.se> wrote:
On 2019-05-10 01:46, Paul Koning wrote:
On May 9, 2019, at 3:20 PM, Hittner, David T [US] (MS) <david.hitt...@ngc.com>
wrote:
(It's been a long time since I've played with SMAC on wireless. I did get it to
work, but it wasn't worth the pain to me, so I upgraded to DECNET/OSI
non-compatibility mode.)
IIRC, when you start DECNET IV, it sends a broadcast packet to see if there is
an address collision with the hard-coded DECNET IV address before it changes
the MAC to the DECNET IV MAC.
That's not in any DECnet standard. It may be someone did that, and it wouldn't
be a bad idea to do so.
BTW, some NICs allow enabling multiple individual addresses and choosing which one you
want. DEC made that standard fairly early on, once it became clear that combining LAT
and DECnet on a single interface was a pain. DEUNA doesn't do this, DEQNA does, and all
the DEC single-chip Ethernet interfaces support it. On such interfaces you'd use the
aa-04 address for DECnet and the "hardware address" for other things.
Hum? My understanding is that the DEUNA and DELUA works exactly the same in
this aspect. And combining DECnet and LAT is not a problem on of those
interfaces, so now I'm curious what you are thinking of?
Also, at least under RSX, all software is definitely using the same MAC address
for any and all network protocols you might be running, which includes DECnet,
LAT, MOP, IP, ARP, and anything else you might want to throw at it.
That's certainly allowed, but as I said, most DEC NICs allow you to avoid the
DECnet address for other protocols.
The difficulty was with VMS, which would start LAT before DECnet and use the
hardware address for LAT if the individual address hadn't been overridden. So
your sessions would fail because the address would change. The solution, with
the DEUNA, was either to start LAT after DECnet, or to teach VMS the DECnet
address via some other system parameter so it would be set at boot time, not
wait until DECnet startup. The multiple individual address feature was made a
standard part of the DEC Ethernet architecture so you wouldn't need to do any
of this; if a protocol needed a specific format address it could just use it,
without bothering other protocols on the same machine.
Ok. Yes, changing the MAC address is always a potential problem.
But I wonder if you are mixing controllers up. I just re-checked the
DELUA manual, and it only allows one physical address. You can change
it, but there can be only one, and if you want to receive any others,
you'll have to go into promiscuous mode. It allows up to 10 multicast
addresses.
And this is just the same as for the DEUNA.
Maybe later DEC controllers allowed more, but as far as I know, the
DELUA do not.
Also, the DEUNA and DELUA does not allow you to set the source MAC
address on transmitted packets. You need to have the room for it, but at
transmission, the controller will on its own insert the actual physical
source address in the transmitted message. So it's probably totally
impractical/impossible to use several MAC addresses, even if you were to
enable promiscuous mode. Promiscuous mode can be used to get packets on
the ethernet, but you still can't act as if you have multiple MAC
addresses. And changing the MAC address is a very disruptive process
that you cannot really be doing during normal operation.
I should probably go and dig some in VMS, but my recollection is that
the address changes halfway through the booting, and for that reason the
network goes down and up again during booting, and other protocols like
LAT handles this. But I might be misremembering things in VMS. I use it
too seldom these days...
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: b...@softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
Simh mailing list
Simh@trailing-edge.com
http://mailman.trailing-edge.com/mailman/listinfo/simh