[comp.os.vms] Vaxstation-2000 broadcast problem

KVC@ENGVAX.SCG.HAC.COM (Kevin Carosso) (07/17/87)

If anyone's still interested, here's the outcome of the problem I reported
to the list a while ago with receiving ethernet broadcasts on
a VAXstation-2000.

It turns out that DEC filters out broadcasts on "newer ethernet interfaces".
I assume this is a change to the device driver somewhere.  If you want to
receive broadcasts and multicasts, you must enable them for your protocol
type.  You can use the NMA$C_PCLI_MCA or NMA$C_PCLI_MLT to do so.  The former
is what DEC recommends.  It works fine for multicast addresses.  I gave
it a try for the ethernet broadcast address (FF-FF-FF-FF-FF-FF) but it was
unhappy.  So, for broadcasts you have to use MLT.  It means putting something
like:
                .word   NMA$C_PCLI_MLT
                .long   NMA$C_STATE_ON

in the nasty structure you send to the ethernet device when you start up
your protocol.  Both are documented in the "I/O User's Guide", so look it
up if you need it.  (though I never found documention indicating that
the driver REQUIRED you to do so on the -2000)

In my case (fixing CMU/Tek TCP/IP for the VS-2000) I enabled MLT for the
ARP protocol, but disabled it on IP.  If there's a need to accept ethernet
broadcasts on IP, could someone let me know?  It works just fine on the
VS-2000 now.

Do the BI ethernet controllers (DEBNT clunker and DEBNA (?) new widget)
do the same thing as the DESVA on the VS-2000?  If so, you'll need my fixes.

To the people who commented that the CMU stuff worked on their VS-2000,
maybe CMU had the fix in a later release?  Alternately, if you were always
going from the VMS VS-2000 *to* a UNIX 4.2 or 4.3 machine, that will work
anyway because UNIX doesn't need to ARP since it saves the ARP address when
it gets the ARP from the VS-2000 who started the connection.  The CMU stuff
should probably do that too, it saves on the number of broadcasts on the
net.  I will see about adding that when I have some time.

Meanwhile, those of you with the CMU TCP/IP, I'm sending out a collection
of my fixes to the TEKTCP mailing list.  Tune in there (if it still works)
for the goods.

        /Kevin Carosso                     kvc@engvax.scg.hac.com
         Hughes Aircraft Co.               kvc%engvax@oberon.usc.edu

Ps.  Thanks to the people who relayed (second or third hand!) the clue
     I needed to track this one down.