From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH] htb: improved accuracy at high rates Date: Sat, 20 Oct 2012 09:26:34 +0200 Message-ID: <1350717994.13333.298.camel@edumazet-glaptop> References: <1350685582-65334-1-git-send-email-j.vimal@gmail.com> <5081E7A8.2080009@hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Cc: Vimalkumar , davem@davemloft.net, Jamal Hadi Salim , netdev@vger.kernel.org To: Rick Jones Return-path: Received: from mail-wi0-f172.google.com ([209.85.212.172]:60087 "EHLO mail-wi0-f172.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751653Ab2JTH1H (ORCPT ); Sat, 20 Oct 2012 03:27:07 -0400 Received: by mail-wi0-f172.google.com with SMTP id hq12so942512wib.1 for ; Sat, 20 Oct 2012 00:27:05 -0700 (PDT) In-Reply-To: <5081E7A8.2080009@hp.com> Sender: netdev-owner@vger.kernel.org List-ID: On Fri, 2012-10-19 at 16:52 -0700, Rick Jones wrote: > Did it really take 20 concurrent netperf UDP_STREAM tests to get to > those rates? And why UDP_STREAM rather than TCP_STREAM? > > I couldn't recall if GSO did anything for UDP, so did some quick and > dirty tests flipping GSO on and off on a 3.2.0 kernel, and the service > demands didn't seem to change. So, with 8000 bytes of user payload did > HTB actually see 8000ish byte packets, or did it actually see a series > of <= MTU sized IP datagram fragmentsOn 10/19/2012 03:26 PM, Vimalkumar wrote: UFO is quite a virtual feature for the moment ;) To my knowledge only one driver uses it (neterion/s2io.c) So most probably HTB sees the fragments built by the stack before queueing packets to the qdisc layer. tcpdump can be used to check this.