From mboxrd@z Thu Jan 1 00:00:00 1970 From: KOVACS Krisztian Subject: initial ct_sync port for Linux 2.6 Date: Fri, 17 Dec 2004 00:10:00 +0100 Message-ID: <41C215C8.1070905@sch.bme.hu> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-2; format=flowed Content-Transfer-Encoding: quoted-printable Return-path: To: Netfilter-failover list , netfilter-devel List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: netfilter-devel-bounces@lists.netfilter.org Errors-To: netfilter-devel-bounces@lists.netfilter.org List-Id: netfilter-devel.vger.kernel.org Hi, In the last few days I've started porting ct_sync to Linux 2.6 and=20 Pablo Neira's new notifier API. This evening I managed to get it compile=20 and work on my UML-based "test system", so I thought it's time to=20 publish the code. However, since the 2.6 port is still in heavy=20 development, and because of my difficulties with the Netfilter SVN the=20 code is aveilable in a public GNU Arch archive repository here: http://people.netfilter.org/hidden/archives/2004-public For those of you unexperienced with tla, you can get the code using=20 the following magic incantations: $ tla register-archive hidden@sch.bme.hu--2004-public \=20 http://people.netfilter.org/hidden/archives/2004-public $ tla get -A hidden@sch.bme.hu--2004-public netfilter-ha--mainline--1.0 The repository contains the complete patchset for Linux 2.6.9, plus=20 the ported ct_sync code. I've done some changes in the code itself as wel= l: - simplified conntrack entry manipulation code while trying to minimize=20 necessary core conntrack patches - ct_sync does not depend on ctnetlink anymore, only Pablo Neira's new=20 per-packet conntrack notifier patch is necessary - the patchtree for 2.6.9 still contains the connmark patch, which has=20 been included in the mainline kernel recently, so it won't be necessary=20 for 2.6.10 - polling support in the receive thread has also been included, this=20 feature is especially experimental (comments welcome) Unfortunately I did not have time to update the README yet, it's=20 slightly outdated when describing the patching process required to=20 compile ct_sync. However, with a little imagination everyone should be=20 able to guess which changes to apply to the commands in the README. :) Of course, there are a couple of known bugs in the code as well, and=20 the whole 2.6 tree is even more experimental than the 2.4 variant of the=20 code. --=20 KOV=C1CS, Kriszti=E1n