Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

STRIP (was Re: richochet modems)

0 views
Skip to first unread message

Matt Braithwaite

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
On Wed, 24 Nov 1999 12:09:08 -0800, Randy Bush <ra...@psg.com> said:
>
> i just run userland ppp and it cooks

Not that it's especially relevant, but: is anybody interested in
running STRIP under FreeBSD? I ported the NetBSD driver and it's been
working fine for me.

(STRIP is Starmode Radio IP---you use the Ricochet network to talk
directly to other radios, one of which may be connected to a router
that's attached to the Internet somewhere. It's nice because (a) you
don't have to use Metricom's Internet connectivity, and (b) you don't
have to pay all the penalties of PPP mode, like lengthy connection
setup and a changed IP address when it drops. The original work was
done by Stuart Cheshire at Stanford; the NetBSD port is due to
Jonathan Stone, also at Stanford.)

--
Matt Braithwaite Here in my car, I can only receive.


To Unsubscribe: send mail to majo...@FreeBSD.org
with "unsubscribe freebsd-mobile" in the body of the message

Brad Karp

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
I wrote my own StarMode driver for FreeBSD in the Fall of 1997. It's
been available at:

http://www.eecs.harvard.edu/~karp/humr.tar.gz

since then.

My StarMode code, HUMR, is superior to STRIP in that it requires *no*
hardwired database of IP-to-MAC address mappings--that is, it allows users
who don't know each other's MAC addresses to power up near one another and
*dynamically* discover the required ARP entries. In essence, HUMR implements
ARP over the non-broadcast Metricom medium.

John-Mark Gurney has my code, and was looking at committing it to FreeBSD a
while back, but I never heard from him in the end.

John-Mark: what's the status of this? If he's swamped, anyone else want to
take up reviewing my code for commit??

-Brad, ka...@eecs.harvard.edu

Dirk-Willem van Gulik

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to

On 24 Nov 1999, Matt Braithwaite wrote:

> On Wed, 24 Nov 1999 12:09:08 -0800, Randy Bush <ra...@psg.com> said:
> >
> > i just run userland ppp and it cooks
>
> Not that it's especially relevant, but: is anybody interested in
> running STRIP under FreeBSD? I ported the NetBSD driver and it's been
> working fine for me.

Actually this is kind of neat. I am in italy; and have just one of those
beasts... for use in the US. But this would be a hack to do something
usefull with them here as well when I get a second one.

Any specific pointer's for the freebsd code ; I just googled; but got
mainly linux/netbsd specific ptr's.

Dw

Randy Bush

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
hmmmmmm. sounds like a build-your-own wap, yes?

i do have a metricom for use when traveling, but no waps within range of
my home base. and i do have metricom-loaded guests. and i do have a t1
to the net. so this sounds as if it may be somewhat useful, though we have
been using wavelan and successor cards for this up to now.

randy

Matt Braithwaite

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
On Wed, 24 Nov 1999 15:34:48 -0500 (EST), Brad Karp <ka...@eecs.harvard.edu> said:
>
> My StarMode code, HUMR, is superior to STRIP in that it requires
> *no* hardwired database of IP-to-MAC address mappings--that is, it
> allows users who don't know each other's MAC addresses to power up
> near one another and *dynamically* discover the required ARP
> entries. In essence, HUMR implements ARP over the non-broadcast
> Metricom medium.

That *is* pretty cool. On the other hand, STRIP is an adequate
substitute for PPP mode (i.e., mobile node to Internet gateway, rather
than mobile node to mobile node), and it runs on Linux as well, for
people who are into that sort of thing.

--
Matt Braithwaite Here in my car, I can only receive.

Matt Braithwaite

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
On Wed, 24 Nov 1999 21:39:24 +0100 (CET), Dirk-Willem van Gulik <di...@webweaving.org> said:
>
> On 24 Nov 1999, Matt Braithwaite wrote:
>
>> Not that it's especially relevant, but: is anybody interested in
>> running STRIP under FreeBSD? I ported the NetBSD driver and it's been
>> working fine for me.
>
> Actually this is kind of neat. I am in italy; and have just one of those
> beasts... for use in the US. But this would be a hack to do something
> usefull with them here as well when I get a second one.

Sure---this *does* work; you do not need the Ricochet network to talk
modem-to-modem. It's also faster. When I'm in the same room as my
gateway, I can get over 6 KB/sec, as opposed to a few KB/sec over the
Ricochet network.

I believe the range for direct radio to radio communication is about
1/4 mile with typical urban obstacles; can anybody confirm?

> Any specific pointer's for the freebsd code ; I just googled; but
> got mainly linux/netbsd specific ptr's.

I haven't put the code out in public anywhere because it's so
embarassing. I know nothing about kernel programming, so what I did
was this: I diffed NetBSD's if_ppp.c and if_strip.c, and I diffed
FreeBSD's if_ppp.c against NetBSD's if_ppp.c, and I used the results
to turn a copy of FreeBSD's if_ppp.c into a STRIP driver. Patches to
arp? and slattach? are also necessary, but trivial, IIRC.

In other words, my idea of a port was very nearly a purely textual
merge. Yeek. Still, it works fine, and interoperates with the Linux
implementation, although it doesn't have ARP support.

Still, if it'd be useful to you, I'll swallow my pride and send it
off. :-)

Brad Karp

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
Matt Braithwaite wrote:

> That *is* pretty cool. On the other hand, STRIP is an adequate
> substitute for PPP mode (i.e., mobile node to Internet gateway, rather
> than mobile node to mobile node), and it runs on Linux as well, for
> people who are into that sort of thing.

HUMR, of course, works mobile-node-to-Internet-gateway *and* mobile-node-to-
mobile-node.

STRIP isn't *quite* a PPP substitute, even when used as a single hop from a
mobile node to an Internet gateway. PPP can assign "pseudo" link level
addresses to either side of a link, without either side knowing any sort
of "MAC address" for the other. STRIP requires configuration of MAC addresses
for the two sides of the link into an ARP table.

You don't need STRIP to do PPP--you can "dial" the MAC address of the other
modem, and get a reliable byte stream and *literally* run PPP over it. So if
you know the MAC addresses, you don't need the STRIP code at all. StarMode's
only added value is multi-access (and multi-hop without Metricom poletops
in your area, though that's no longer a PPP substitute).

As for Linux, HUMR is fully user-level (runs on tun), so it's more portable
than STRIP, which is a Linux kernel driver.

-Brad, ka...@eecs.harvard.edu

Matt Braithwaite

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
On Wed, 24 Nov 1999 19:53:07 -0500 (EST), Brad Karp <ka...@eecs.harvard.edu> said:
>
> HUMR, of course, works mobile-node-to-Internet-gateway *and* mobile-node-to-
> mobile-node.

Right. STRIP does this too, but it's a kludge: a central node with a
known hardware address has to be the ``ARP server'' for everybody
else.

By the way, have you noticed that the new radios use a different
address format? They're ##-####-#### rather than ####-####. The
Linux STRIP driver doesn't deal well with this, and I haven't updated
mine either.

> STRIP isn't *quite* a PPP substitute, even when used as a single hop from a
> mobile node to an Internet gateway. PPP can assign "pseudo" link level
> addresses to either side of a link, without either side knowing any sort
> of "MAC address" for the other. STRIP requires configuration of MAC addresses
> for the two sides of the link into an ARP table.

Oh, yeah, I meant something more approximate, namely that STRIP works
pretty well if all you want to do with your radio is talk to the
Internet.

> You don't need STRIP to do PPP--you can "dial" the MAC address of
> the other modem, and get a reliable byte stream and *literally* run
> PPP over it. So if you know the MAC addresses, you don't need the
> STRIP code at all.

Sorry, can you explain that a little more? Are you saying that given
any two radios, you can set up a reliable byte stream between them
using AT commands?

Maybe I'm just not understanding what you're saying here.


--
Matt Braithwaite Here in my car, I can only receive.

Brad Karp

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
[forgot to cc -mobile on my reply to Matt]

To: ma...@braithwaite.net
Subject: Re: STRIP (was Re: richochet modems)

> Right. STRIP does this too, but it's a kludge: a central node with a
> known hardware address has to be the ``ARP server'' for everybody
> else.

Yes, my point exactly: HUMR requires no central ARP server.

> Sorry, can you explain that a little more? Are you saying that given
> any two radios, you can set up a reliable byte stream between them
> using AT commands?

Say you have two hosts, A and B, each with a Ricochet, MAC addresses
0000-0001, and 0000-0002, respectively.

B can tell its radio:

ATS0=1

A can tell its radio:

ATDT0000-0002

This will *literally* emulate a plain-old-modem-style byte stream
connection from host A to host B. You can just use standard client and
server PPP on the hosts on either side, then. No STRIP required.

So my overall point is: for a single hop of PPP over Metricom, there's
no need to use STRIP at all. And if you want multi-access, or multi-hop, the
central ARP server STRIP requires makes HUMR more convenient.

About the newest radios' MAC addresses: yes, the string parsing routines need
to be changed accordingly. I've not done this for HUMR yet, either.

-Brad, ka...@eecs.harvard.edu

Matt Braithwaite

unread,
Nov 24, 1999, 3:00:00 AM11/24/99
to
On Wed, 24 Nov 1999 20:27:26 -0500 (EST), Brad Karp <ka...@eecs.harvard.edu> said:
>
>> Sorry, can you explain that a little more? Are you saying that given
>> any two radios, you can set up a reliable byte stream between them
>> using AT commands?
>
> Say you have two hosts, A and B, each with a Ricochet, MAC addresses
> 0000-0001, and 0000-0002, respectively.
>
> B can tell its radio:
>
> ATS0=1
>
> A can tell its radio:
>
> ATDT0000-0002

Wow, I had no idea they carried modem emulation to such extremes!
ATS0=1, if anybody's out there who grew up with DSL, means ``answer''
the ``phone'' after one ``ring''. :-)

> So my overall point is: for a single hop of PPP over Metricom,
> there's no need to use STRIP at all.

Unless, of course, you want multiple machines to be able to use the
same gateway (PPP over a reliable byte stream only allows one machine
at a time.). Also, STRIP probably uses precious bandwidth a little
more efficiently than PPP does.

(These are really just quibbles, of course. I'm not trying to defend
STRIP against HUMR; just pointing out that STRIP is useful for a
certain set of problems, though a smaller set.)

> And if you want multi-access, or multi-hop, the central ARP server
> STRIP requires makes HUMR more convenient.

Yeah. I'm curious which has more users. I've never run into any
Ricochet users who aren't using straight PPP.

--
Matt Braithwaite Here in my car, I can only receive.

Dirk-Willem van Gulik

unread,
Nov 25, 1999, 3:00:00 AM11/25/99
to

On 24 Nov 1999, Matt Braithwaite wrote:

> I believe the range for direct radio to radio communication is about
> 1/4 mile with typical urban obstacles; can anybody confirm?

Any guess of the frequency ?



> Still, if it'd be useful to you, I'll swallow my pride and send it
> off. :-)

That'd be cool !

Dw

Doug White

unread,
Nov 25, 1999, 3:00:00 AM11/25/99
to
On Thu, 25 Nov 1999, Dirk-Willem van Gulik wrote:

>
>
> On 24 Nov 1999, Matt Braithwaite wrote:
>
> > I believe the range for direct radio to radio communication is about
> > 1/4 mile with typical urban obstacles; can anybody confirm?
>
> Any guess of the frequency ?

Frequency hopping in the 900MHz range.

I believe you can go modem-modem even through the repeater network. As
long as the modem is set to auto-answer and you know the number, you can
call each other.

You have to use starmode if you want to handle multiple connections, but
for single connections modem-modem is fine. Metricom also sells
(sold?) gateway units that you could use to drop onto the POTS network.

Doug White | FreeBSD: The Power to Serve
dwh...@resnet.uoregon.edu | www.FreeBSD.org

Doug White

unread,
Nov 25, 1999, 3:00:00 AM11/25/99
to
On 24 Nov 1999, Matt Braithwaite wrote:

> By the way, have you noticed that the new radios use a different
> address format? They're ##-####-#### rather than ####-####. The
> Linux STRIP driver doesn't deal well with this, and I haven't updated
> mine either.

"new radios"? Did you get access to 128kb gear somehow?

Matt Braithwaite

unread,
Nov 26, 1999, 3:00:00 AM11/26/99
to
On Thu, 25 Nov 1999 12:49:28 -0800 (PST), Doug White <dwh...@resnet.uoregon.edu> said:
>
> On 24 Nov 1999, Matt Braithwaite wrote:
>> By the way, have you noticed that the new radios use a different
>> address format? They're ##-####-#### rather than ####-####. The
>> Linux STRIP driver doesn't deal well with this, and I haven't updated
>> mine either.
>
> "new radios"? Did you get access to 128kb gear somehow?

I only wish. No, I meant the SX modems (the flat ones).

--
Matt Braithwaite Here in my car, I can only receive.

Guido van Rooij

unread,
Nov 29, 1999, 3:00:00 AM11/29/99
to
On Thu, Nov 25, 1999 at 12:52:22PM -0800, Doug White wrote:
> On Thu, 25 Nov 1999, Dirk-Willem van Gulik wrote:
>
> >
> >
> > On 24 Nov 1999, Matt Braithwaite wrote:
> >
> > > I believe the range for direct radio to radio communication is about
> > > 1/4 mile with typical urban obstacles; can anybody confirm?
> >
> > Any guess of the frequency ?
>
> Frequency hopping in the 900MHz range.
>
> I believe you can go modem-modem even through the repeater network. As
> long as the modem is set to auto-answer and you know the number, you can
> call each other.
>
> You have to use starmode if you want to handle multiple connections, but
> for single connections modem-modem is fine. Metricom also sells
> (sold?) gateway units that you could use to drop onto the POTS network.

Anyone know the price of such a thing? Mighht be worthwhile here at home
(though I guess the frequency rabnge is already taken in Europe...
does anyone know (sounds like GSM))

-Guido

Poul-Henning Kamp

unread,
Nov 29, 1999, 3:00:00 AM11/29/99
to

>Anyone know the price of such a thing? Mighht be worthwhile here at home
>(though I guess the frequency rabnge is already taken in Europe...
>does anyone know (sounds like GSM))

You want stuff in the 2.4GHz range and it is too expensive and may
require license permit and approval.

--
Poul-Henning Kamp FreeBSD coreteam member
p...@FreeBSD.ORG "Real hackers run -current on their laptop."
FreeBSD -- It will take a long time before progress goes too far!

Guido van Rooij

unread,
Nov 29, 1999, 3:00:00 AM11/29/99
to
On Mon, Nov 29, 1999 at 11:16:17PM +0100, Poul-Henning Kamp wrote:
>
> >Anyone know the price of such a thing? Mighht be worthwhile here at home
> >(though I guess the frequency rabnge is already taken in Europe...
> >does anyone know (sounds like GSM))
>
> You want stuff in the 2.4GHz range and it is too expensive and may
> require license permit and approval.
>

The end of a nice idea :-(

-Guido

0 new messages