Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S965023AbXAGT4S (ORCPT ); Sun, 7 Jan 2007 14:56:18 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S965003AbXAGT4S (ORCPT ); Sun, 7 Jan 2007 14:56:18 -0500 Received: from tmailer.gwdg.de ([134.76.10.23]:34689 "EHLO tmailer.gwdg.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965023AbXAGT4R (ORCPT ); Sun, 7 Jan 2007 14:56:17 -0500 Date: Sun, 7 Jan 2007 20:55:31 +0100 (MET) From: Jan Engelhardt To: Linux Kernel Mailing List , Netfilter Mailing List , nmap-dev@insecure.org Subject: [announce] chaostables 0.4 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Spam-Report: Content analysis: 0.0 points, 6.0 required _SUMMARY_ Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1760 Lines: 48 [Not sure if netfilter/nmap-dev bounce when you are not subscribed, remove if in doubt.] Hello lists, chaostables 0.4 has been released. This is a package containing some netfilter modules to work against nmap and port scans. 'portscan' can match on stealth, syn, connect scans, also finds FIN-connect scans and simple banner grabbing. 'DELUDE' works like TARPIT, making it look like the port is open, but actually does not hold the connection like TARPIT, hence should not fill up conntrack. 'CHAOS' will deliver what it says: slows down nmap scans[1] and provides back bogus non-deterministic replies -- the more even with DELUDE/TARPIT. For Linux 2.6.18, .19, .20, iptables 1.3.6, 1.3.7. CHAOS currently requires TARPIT during build and runtime. 'portscan' and 'CHAOS' can also be synthesized by rule logic "only"[2], giving way to possibly implement it on BSD ipf* too. Details in doc/fw.html. Suggestions for improvement are welcome. Please discuss the usefulness and let me hear [LKML] any objections for inclusion in Linux mainline too. http://freshmeat.net/releases/244538/ - this release http://freshmeat.net/p/chaostables/ - generally Thanks, Jan [1] nmap 3.81 timings have been done back in 2005, but things have changed a little since 4.x. Redoing the timing benchmarks is already on TODO. [2] Requires: connmark && CONNMARK && (conntrack || state) && (random || nth || statistic || hashlimit) or equivalent. Depends on how you implement it. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/