-----Original Message-----
From: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Sunday, October 24, 1999 3:46 PM
Subject: Re: dhcpd + sv2agw
>On 24 Oct, Cathryn Mataga wrote:
>> Would it be like a kernel compile time thing? Or like a setsockopt type of
>mechanism
>> so each broadcast socket would have a different callsign? What's the application or
>> bug fix you have in mind for this? (That is configurable callsigns for ip
>broadcast packets.)
>
>For consistency you'd want it configurable with a tool like ifconfig or
>axparms maybe.
>
Interestingly, there is a SIOCSIFHWBROADCAST 0x8937 /* set hardware broadcast addr */
in /usr/src/linux/include/linux. From the source, I suspect this might just work
already, if you just
pass ioctl the callsign. Is there a shell command to do ioctls? If so, maybe we
have this
already, and just don't know it. If not, it's like two lines of C code.
I do know, having just looked at this, that the dev->broadcast varialbe is setting is
what
the kernel uses to match the callsign for received broadcasts, at least. Not sure
about
sending broadcasts, but in a sane world, that would be changed alsoo. :)
(Note, you need to weirdly mung the callsign by shifting the letters left by 1 bit.
-- 6 bytes
for the call padded with spaces and one byte for the number)
>> More interesting to me, is what happens if we turn on the MULTICAST flag with the
>> ax25 driver. Me, I'm not even sure exactly how to test MULTICAST, but I was
>thinking
>> that maybe, it might be fun to have some kind of HAMBONE (like M-bone) with news
>messages
>> or maybe audio or something.
>
>Same deal. Configurable multicast address would be nice. I think many
>ethernet drivers already support configurable multicast addresses, so
>we probably wouldn't have to break anything too much to implement it.
>
Yeah, having just thought about it, I thought I'd try setting up multicast on
my network here, but even just on ethernet, uh, this is a little bit tricky. Or
at least, I think I'm supposed to build mrouted and I find this requires fussing
with bad .h files to get it to compile in Linux on Redhat -- sigh. I see mrouted
is doing SOCK_RAW stuff, which I assume will work over ax25 -- right? It's
just AF_PACKET that's different between ethernet and ax25, isn't it, and
SOCK_RAW is the same for both?
Just for fun, I or-ed the flag in mkiss.c with ISS_MULTICAST, and yes, then
I could turn on multicast for ax25. But it'll take a lot of fuss to figure out if
it really works or not -- that is not knowing how the multicast stuff is supposed
to work in the first place very well. (I did a quick browse of some websites
on this.) There is a lot too this, it seems.
I suspect, that if MULTICAST does what I think it is supposed to do, that it might
be uniquely suited for ham radio links with low bandwidth. That is, say,
for doing something like a convers or irc type chat system, where the chat messages
were smart enough only to go down the routes where someone was listening,
and also only go down the route once, you know. That is maybe all those hard problems
are solved down in this code by clever PHd type people. Still, me I haven't
gotten it working yet with ethernet -- though I haven't put a lot of time into it, and
it's
not a high priority for me.
I did see an IOCTL for multicast address, though at this point, I can't say that I know
what that is.
>--
>[EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED]
>
