From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262400AbVAPCiU (ORCPT ); Sat, 15 Jan 2005 21:38:20 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262393AbVAPCiU (ORCPT ); Sat, 15 Jan 2005 21:38:20 -0500 Received: from mx.freeshell.org ([192.94.73.21]:737 "EHLO sdf.lonestar.org") by vger.kernel.org with ESMTP id S262400AbVAPCiD (ORCPT ); Sat, 15 Jan 2005 21:38:03 -0500 Date: Sat, 15 Jan 2005 20:37:39 -0600 (CST) From: John Dahlstrom To: linux-kernel@vger.kernel.org Subject: NETWORK: udp_port_rover reluctant to rove (unlike tcp_port_rover) Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Regarding udp_port_rover (of linux/net/ipv4/udp.c): In Linux 2.4 or 2.6, I noticed that selected local port numbers for UDP resist roaming, unlike TCP ports numbers (tcp_port_rover) that appear to steadily increase irrespective of concurrent local port usage. What is the advantage of this lack of parallel behavior? (Other than exacerbating broken behavior of certain firewalls that insist on fixed UDP source port blocking?) Aesthetically peculiar it seems, that the kernel reins in port roving for the connectionless protocol unless a UDP stampede unleashes itself, while allowing the TCP ports to range free across the local ports regardless. Kind regards, - John -- http://jodarom.sdf1.org/