Date: Mon, 29 May 1995 10:42:08 -0700 From: James Binkley <jrb@cs.pdx.edu> To: hackers@FreeBSD.org Subject: Van Jacobson releases vat/sd/wb for linux Message-ID: <199505291742.KAA22513@sirius.cs.pdx.edu>
next in thread | raw e-mail | index | archive | help
Anything that could be done to get them to make vat/sd/wb binaries
for freebsd? The following forwarded from the rem-conf mbone list.
In any case, the discussion of multicast problems is interesting.
regards,
Jim Binkley
jrb@cs.pdx.edu
---------------------------------------------------------
------- Forwarded Message
Return-Path: rem-conf-request@es.net
Return-Path: <rem-conf-request@es.net>
Received: from osi-west.es.net (osi-west.es.net [128.55.32.32])
by cs.pdx.edu (8.6.10/CATastrophe-12/23/94-P) with SMTP
id BAA00659; Sat, 27 May 1995 01:47:45 -0700 for
Received: from rx7.ee.lbl.gov by osi-west.es.net with ESnet SMTP (PP);
Sat, 27 May 1995 00:08:46 -0700
Received: by rx7.ee.lbl.gov (8.6.12/1.43r) id AAA23238;
Sat, 27 May 1995 00:09:19 -0700
Message-Id: <199505270709.AAA23238@rx7.ee.lbl.gov>
To: mbone@isi.edu, rem-conf@es.net
Subject: linux vat, sd & wb available on ftp.ee.lbl.gov
Date: Sat, 27 May 95 00:09:17 PDT
From: Van Jacobson <van@ee.lbl.gov>
Two weeks ago I sent a message saying we had just purchased
a machine for linux development & it would probably take
about two weeks to port vat, sd & wb to linux. Thanks to
a lot of help from Craig Leres, it only took 9 days. We
then spent 4 days trying (unsuccessfully) to fix a problem
with the linux multicast code so we could offer a patch
with the tools. Since the tools are usable even without fixing
the kernel problem, we've decided to release them without a
kernel patch. They're available in the usual place. I.e.,
ftp://ftp.ee.lbl.gov/conferencing/vat/linux-vat-3.5.tar.Z
ftp://ftp.ee.lbl.gov/conferencing/wb/linux-wb-1.59.tar.Z
ftp://ftp.ee.lbl.gov/conferencing/sd/linux-sd-1.17.tar.Z
Everything has been build under the Slackware 1.2.8 distribution.
The tools were all linked '-static' in the hopes of minimizing
problems with different versions of various libraries. They were
also all compiled "-g" so interested parties could send us gdb
tracebacks with bug reports. Since -g makes the binaries huge,
you might want to strip them if your system is tight on disk space.
Let us know of any problems. Thanks.
- Van Jacobson & Steve McCanne
- ------------ This is the README.linux from vat:
Notes on vat for linux. Wed May 24 08:50:53 PDT 1995
This is the first release of vat for linux. It has had very
little testing so expect bugs. There are some minor problems
due to the way PC audio cards work & a major problem from
a missing part of the linux multicast code. The audio problems
are:
- Most PC audio is half-duplex (cannot play & record at the
same time) so none of the 'speakerphone' or full-duplex
audio modes of vat are available. Essentially you always
run in "mike mutes net" mode.
We have a PAS-16 on order (the one card we know of that does
full duplex audio) and hope to work on full-duplex linux audio
support whenever it shows up. That's probably at least a
month away.
- The linux audio driver doesn't include any way to set the
play or record volume so the volume sliders do nothing.
The major problem is the multicast code is that multicast through
the loopback interface does not work. This is a very serious
deficiency since local (loopback) multicast sockets are used to
implement "conference busses" that vat/vic/etc. use to do things
like pass the audio hardware between different vats, automatically
switch video windows on audio activity, etc. We, and others, also
have floor & conference control agents (for moderated meetings)
and key distribution agents (for private meetings with dynamic
membership control) in the works. None of these things will
work on linux without fixing the linux kernel. I was hoping to
come up with a patch for this but, since the loopback interface
is used (wrongly, in my view) to implement the IP_MULTICAST_LOOP
socket option, quite a lot of code needs to be changed & I haven't
yet figured out enough of the linux socket buffer & locking/ipl
conventions to make the changes.
Ignoring the future tools that won't work, right now you will
find that the lack of local multicast makes it *very* difficult
to switch the audio between multiple instances of vat. You will
probably want to run just one at a time.
There is another, minor, multicast bug that doesn't allow the
multicast ttl to be set to 0 (the spec specifically defines 0
as "this machine"). I think the code in linux/net/inet/ip.c
should be patched as follows:
diff -c -r1.1 ip.c
*** 1.1 1995/05/27 06:24:35
- --- ip.c 1995/05/27 06:25:28
***************
*** 2090,2096 ****
unsigned char ucval;
ucval=get_fs_byte((unsigned char *)optval);
! if(ucval<1||ucval>255)
return -EINVAL;
sk->ip_mc_ttl=(int)ucval;
return 0;
- --- 2090,2096 ----
unsigned char ucval;
ucval=get_fs_byte((unsigned char *)optval);
! if(ucval>255)
return -EINVAL;
sk->ip_mc_ttl=(int)ucval;
return 0;
This bug currently has little effect but will be annoying if the
loopback bug is fixed.
Good luck. Please send any bug reports to "vat@ee.lbl.gov". Thanks.
------- End of Forwarded Message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199505291742.KAA22513>