From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andy Furniss Subject: Re: [PATCH v2 2/2] [iproute2/tc] hfsc: add link layer overhead adaption Date: Tue, 24 Jun 2008 22:39:24 +0100 Message-ID: <4861698C.3040007@andyfurniss.entadsl.com> References: <20080624085514.22540.91038.stgit@fate.lan> <20080624085519.22540.40534.stgit@fate.lan> <4860C910.4050703@andyfurniss.entadsl.com> <20080624225850.nvasycyghwko8k8o@hayate.ip6> Reply-To: lists@andyfurniss.entadsl.com Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, Patrick McHardy , Stephen Hemminger To: Jussi Kivilinna Return-path: Received: from mx1.ukfsn.org ([77.75.108.10]:57978 "EHLO mail.ukfsn.org" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753147AbYFXVj3 (ORCPT ); Tue, 24 Jun 2008 17:39:29 -0400 In-Reply-To: <20080624225850.nvasycyghwko8k8o@hayate.ip6> Sender: netdev-owner@vger.kernel.org List-ID: Jussi Kivilinna wrote: > Quoting Andy Furniss : > >> Jussi Kivilinna wrote: >>> Patch adds 'mpu', 'mtu', 'overhead' and 'linklayer' options to hfsc. >>> These >>> options are used to create size table for sch_hfsc. Size table is >>> only used >>> and passed to kernel if these options are used. >> >>> + unsigned short overhead = 0, mpu = 0; >> >> overhead may need to be negative because skb->len on eth is payload + 14 >> >> Andy. > > Ok, I made size tables use signed short for overhead instead. But this > makes me think should htb/cfq/tfb be changed to support negative > overhead too? Rate table they use have unsigned overhead. Yea they should really, somehow - eventually :-) It was discussed here - http://news.gmane.org/find-root.php?group=gmane.linux.network&article=90386 Andy.