Re: [openss7] another SCTP implementation for linux

["Brian F. G. Bidulock" <bidulock@openss7.org>]

Michael,

There are actually many other freely available implementations, in
various states of maturity.  The implementation on sourceforge is the a
port of the SCTP user-space reference implementation that can be found
at http://www.sctp.org/  Michael Tuxex has a page at http://www.sctp.de/
There are about five or six freely available implementations.  I looked
at the lksctp implementation before writing OpenSS7's SCTP.  I consider
it an example of how not to write kernel code.

Nevertheless, when I went to write SCTP, it was for several reasons:

   o   There was no freely available SCTP implementation in a usable
       state for SIGTRAN protocols for the OpenSS7 stack.

   o   OpenSS7 needed experience in SCTP to support SIGTRAN protocols.

   o   OpenSS7 needed a STREAMS version of SCTP for SIGTRAN protocols.
       (I believe that our STREAMS SCTP due in the upcoming release will
       be the first freely available STREAMS version).

Writing it as an Linux Kernel Module under the socket abstraction was
the quickest path to target, following Alan Cox and others work on the
NET4 Linux TCP stack was fascile.  (I started 4 weeks before the SCTP
Bakeoff, and have not really touched much of the code since!)

If lksctp makes it into a 2.(2n) kernel it will be a sad day for Linux
indeed.  Differences between lksctp and OpenSS7 sctp are numerous.
lksctp attempts to take a user-space process written without
consideration for races and other kernel considerations and embeds it
inside the kernel with a socket wrapper.  It is a very inefficient
an possibly invalid approach in my mind.  OpenSS7 SCTP was written with
the Linux TCP stack as a template.  Because of this it fits as well as
TCP into other advanced Linux features such as transparent proxy,
firewalling, IPSEC, etc.

Merging the two would be impossible.

I was hoping that, due to it's GPL, that the Linux Kernel team may
consider OpenSS7's SCTP implementation as a starting point for Linux's
own implementation (that is, one with maintainers).  Certainly, there
will need to be more demand for SCTP in general before SCTP in the Linux
kernel becomes more than an EXPERIMENTAL module.

--Brian


Michael Schlueter wrote:                   Tue, 31 Jul 2001 20:52:19
> Hi,
> 
> I found another implementation of the SCTP on sourceforge 
> (http://sourceforge.net/projects/lksctp). It seems to be that it will be 
> included in the 2.5 Version of Linux. Maybe  it is possible to conflate
> both driver?
> 
> Bye, Michael
> -- 
> Michael Schlueter,Rheinische Strasse 91,44137 Dortmund,0231-7275720
> Mails: michael@johalla.de/schlue@debian.org
> GPG Fingerprint = 2B65 3B34 8212 337F B37E  585A 79D9 4694 00E0 D054
> PGP Fingerprint = AF DD 1F FA A6 C4 C0 65  F6 45 AD AC 2E F5 16 78
> 

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦