[HECnet] DECnet User Mode Router - Encapsulation Formats

[Paul_Koning at Dell.com]

On Jun 22, 2012, at 6:11 PM, Rob Jarratt wrote:

-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE]
On Behalf Of Paul_Koning at Dell.com
Sent: 22 June 2012 22:25
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] DECnet User Mode Router - Encapsulation Formats


On Jun 22, 2012, at 4:07 PM, Rob Jarratt wrote:


DDCMP over some kind of serial line? Would need appropriate hardware
and driver code to work against that.

DDCMP over sync, yes.   But DDCMP also works on an async line, and that
just
requires protocol code that can talk to a raw mode TTY device.   That would
work for what I need -- I have async DDCMP on my PRO-380 running RSTS,
and in fact it's a standard feature of some of the PDP11 DECnet products.
(I'm not sure if it was ever official in DECnet/E, but the machinery is
there in
the latest versions of the kernel.)

	paul

Hmmm, so if we can use async serial then we have a chance to add that to the
router once I have had a chance to do the main part. It would be good to do
something like that. I don't know enough about DDCMP yet to know what the
implications are (can it work with a Version 2 router for example?).

Regards

Rob

Version 2?   Do you mean Level 2?   DDCMP works in DECnet phase 1 through 4, I should remember about 5 (a.k.a., DECnet/OSI) but I don't.   It's another datalink type, different from Ethernet, partly because it doesn't have addresses (it's point to point) and partly because the headers are different ("short" rather than "long" headers).   But much of it is the same: routing exchanges work the same, for example.   It's all spelled out in detail in the routing layer spec.   The DDCMP protocol itself is in the DDCMP spec.   My first exposure to its details was when I read the spec to implement the RSTS driver, and the spec is good enough that I could simply do what it says and end up with code that works.   (Most of the DECnet specs are that good, actually.)   It's not exactly a hard protocol; maybe a few thousand lines of assembly language code, so probably a few hundred if done in C.   The point to point specific aspects of the routing layer would actually be the larger part, though not all that much larger, either.

	paul