From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pablo Neira Ayuso Subject: Re: conntrack-utils needs to autoload netlink Date: Sun, 19 Apr 2009 13:45:14 +0200 Message-ID: <49EB0ECA.1000405@netfilter.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Netfilter Developer Mailing List To: Jan Engelhardt Return-path: Received: from mail.us.es ([193.147.175.20]:51399 "EHLO mail.us.es" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756889AbZDSLpZ (ORCPT ); Sun, 19 Apr 2009 07:45:25 -0400 In-Reply-To: Sender: netfilter-devel-owner@vger.kernel.org List-ID: Jan Engelhardt wrote: > Hi, > > I observed that running `conntrack -E` does not autoload > nf_conntrack_netlink.ko, and as such, does not receive any events. > `conntrack -L` on the other hand does autoload it, and subsequent -E > works. Yes, that's a known issue. The problem is the following: with `conntrack -L', netlink loads nfnetlink when the socket is created, then nfnetlink receives a message requesting a dump, thus, nfnetlink loads nf_conntrack_netlink. However, with `conntrack -E', netlink loads nfnetlink, but nfnetlink does not receive any message from conntrack since listening to events doesn't require such operation. thus, nf_conntrack_netlink is not loaded. I wrote a patch time ago to load nf_conntrack_netlink from user-space when `conntrack -E' is invoked. I don't see a way to fix this in the kernel code, so that's the only idea that I can find to fix this problem. Still, I don't like too much the idea of doing the autoload from user-space. -- "Los honestos son inadaptados sociales" -- Les Luthiers