home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dcom.lans.misc
- Path: sparky!uunet!munnari.oz.au!mel.dit.csiro.au!mineng.dmpe.CSIRO.AU!dmssyd.syd.dms.CSIRO.AU!megadata!andrew
- From: andrew@megadata.mega.oz.au (Andrew McRae)
- Subject: Re: Arcnet for an Embedded system
- Message-ID: <1992Sep4.013425.26436@megadata.mega.oz.au>
- Originator: andrew@noah
- Sender: news@megadata.mega.oz.au
- Organization: Megadata P/L, North Ryde, Sydney, Aust.
- References: <4403@equinox.unr.edu>
- Date: Fri, 4 Sep 1992 01:34:25 GMT
- Lines: 56
-
- From article <4403@equinox.unr.edu>, by essary@pyramid.unr.edu (Bill Essary):
- > We have just settled on Arcnet for our system (386, 8050 and Siemens 166
- > processor mix. Has anyone seen an OSI type (well defined physical, network
- > and transport layer) protocol stack written to a small realtime kernel
- > interface? We have an in-house kernel and would love to port rather
- > than write the network. Actually, writing the network would be fun,
- > but... time.
- >
- > Bill Essary
-
- We currently have TCP/IP running over 802.4 Token Bus, with the
- physical media being in one case 5 Mbit carrier band, and in another
- case 2 Mbit RS-485 Twisted Pair (I'm not sure if the TP 802.4 has made it
- all the way to a final standard yet). The TP is dual media (very nice
- implementation). We fit this into a 68000 box where the entire
- application code plus kernel, networking, drivers etc. fits in under
- 128Kb ROM. A future plan is to run the BSD 4.4 OSI stack over this
- media and run MAP, though we would estimate that we would need our
- bigger 68020 processor with 1 Mbyte RAM.
-
- If you write your kernel interface to the BSD networking code, then
- the BSD 4.4 OSI code should drop in easily. I took the BSD 4.3 Tahoe
- code and interfaced it to our embedded kernel. One of the design goals
- was to keep the BSD code unchanged so that I could later drop the
- 4.4 code in without changes; as it is, some change would be needed
- as the socket address structure changed slightly and the mbuf stuff
- has changed. But the changes would be easily incorporated.
-
- This has been an interesting thread about what networking to
- use in embedded systems; it is indicative of the state of the
- industry that no one standard (de facto or de jure) has emerged as the
- prime candidate, not like ethernet in the office environment. When
- we went through the exercise of choosing a network (media and protocol)
- we had a set of design criteria, and no one network met
- all our requirements. A major problem with a lack of standards is that
- you do not get the chip makers designing and releasing integrated
- controllers that do most of the hard work in running a medium speed
- network. Some people have mentioned Arcnet controllers, and I
- wish I had known about them (Down Under here we unfortunately do not
- get informed about the entire range of available products from vendors).
-
- The Token Bus stuff seems very good, albeit a little more expensive,
- probably because it runs faster than most other PLC RS-485 based
- networks. What swayed us in the end is that we cannot afford not
- to go with a standard, even though that standard seems to be poorly
- supported. Catch-22, really.
-
- For more info about the porting of the BSD code, see the paper
- "Porting TCP/IP to 802.4 Token Bus" in the Australian Unix Users
- Group winter 1991 conference proceedings (I can mail the paper to
- anyone who is interested).
-
- Andrew McRae inet: andrew@megadata.mega.oz.au
- Megadata Pty Ltd, uucp: ..!uunet!megadata.mega.oz.au!andrew
- North Ryde 2113 Phone: +61 2 805 0899
- NSW AUSTRALIA Fax: +61 2 887 4847
-