From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: can SFQ and perturb break paquet ordering ? Date: Wed, 10 Mar 2010 13:30:10 +0100 Message-ID: <4B9790D2.7030005@trash.net> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: Julien Vehent Return-path: Received: from stinky.trash.net ([213.144.137.162]:56536 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751147Ab0CJMaL (ORCPT ); Wed, 10 Mar 2010 07:30:11 -0500 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: Julien Vehent wrote: > Hi netdev folks, > > I was digging into the sch_sfq code and I was wondering the following (I > assume it's nothing new, I'm just trying to get it right): > > Let's consider one UDP connection that has two packets. Packet#1 goes to > bucket#1, then perturb modifies the hashing algorithms and thus Packet#2 > goes to bucket#800. > > If the round robin was positionned after bucket#1 but before bucket#800 > when perturb occured, then Packet#2 will be sent before Packet#1. And since > UDP can't reordered, it will be processed before on the received end. > > Am I getting this right or is there something in the code that I missed ? That's correct.