From owner-freebsd-hackers  Thu Mar 30 10:39: 6 2000
Delivered-To: freebsd-hackers@freebsd.org
Received: from griffin.aciri.org (griffin.aciri.org [192.150.187.12])
	by hub.freebsd.org (Postfix) with ESMTP
	id 411D737BDF2; Thu, 30 Mar 2000 10:38:59 -0800 (PST)
	(envelope-from wilbertdg@hetnet.nl)
Received: from hetnet.nl (localhost.aciri.org [127.0.0.1])
	by griffin.aciri.org (8.9.3/8.9.3) with ESMTP id KAA45055;
	Thu, 30 Mar 2000 10:38:58 -0800 (PST)
	(envelope-from wilbertdg@hetnet.nl)
Message-ID: <38E39F42.823A681B@hetnet.nl>
Date: Thu, 30 Mar 2000 10:38:58 -0800
From: Wilbert de Graaf <wilbertdg@hetnet.nl>
X-Mailer: Mozilla 4.7 [en] (X11; U; Linux 2.0.36 i386)
X-Accept-Language: en
MIME-Version: 1.0
To: free-bsd-net@freebsd.org, freebsd-hackers@freebsd.org
Subject: The socket structure & igmpv3
Content-Type: multipart/mixed;
 boundary="------------CE6875120F7242AFB36C4312"
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

This is a multi-part message in MIME format.
--------------CE6875120F7242AFB36C4312
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit


Hello all,

I'm working on an implementation of the IGMPv3 protocol in the Net/3
stack, and several questions popped up. The first one is about the
socket structure.

The IGMPv3 protocol allows a process to set (ip)source filters, which is
a list of ip addresses of sources which indicates the process is not
(exclude mode) or just (include mode) interested to receive data from
those sources, on that particular socket. The igmpv3 extensions to the
socket api allow you to add source filters to sockets subscribed to one
or more multicast groups.

So a process identifies a socket with a descriptor, which is a index
into an array of pointers to file structures. These file structures
reside in the kernel and could be shared among processes.  Now I'm
thinking about where to keep this sourcfilter-list. If it's possible to
share a socket between processes it could be that one process is
interested, but another is not. On the other hand, I believe the filter
is a per socket and if it's shared ... so is the filter.

So my question is ...  Is it true that such a filter should be attached
/ stored within the socket datastructure ?

Thanks in advance, Wilbert

Btw.
Besides the fact that a sourcefilter-list per socket should be
implemented, there should also be a sourcefilter-list per
(the in_multi{} structure). This in order to generate IGMPv3 reports, or
maybe in later stage to do filtering at a lower level in order to
improve performance.


--------------CE6875120F7242AFB36C4312
Content-Type: text/x-vcard; charset=us-ascii;
 name="wilbertdg.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Wilbert de Graaf
Content-Disposition: attachment;
 filename="wilbertdg.vcf"

begin:vcard 
n:de Graaf;Wilbert
x-mozilla-html:FALSE
url:http://home.hetnet.nl/~wilbertdg/
org:KPN Research
adr:;;;;;;
version:2.1
email;internet:wilbertdg@hetnet.nl
x-mozilla-cpt:;0
fn:Wilbert de Graaf
end:vcard

--------------CE6875120F7242AFB36C4312--



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