From mboxrd@z Thu Jan 1 00:00:00 1970 From: Changli Gao Subject: Re: [PATCH] rps: add flow director support Date: Tue, 13 Apr 2010 11:11:06 +0800 Message-ID: References: <1271022140-3917-1-git-send-email-xiaosuo@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: "David S. Miller" , netdev@vger.kernel.org To: Tom Herbert Return-path: Received: from mail-iw0-f197.google.com ([209.85.223.197]:50315 "EHLO mail-iw0-f197.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754683Ab0DMDL1 convert rfc822-to-8bit (ORCPT ); Mon, 12 Apr 2010 23:11:27 -0400 Received: by iwn35 with SMTP id 35so1623278iwn.21 for ; Mon, 12 Apr 2010 20:11:26 -0700 (PDT) In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On Tue, Apr 13, 2010 at 1:13 AM, Tom Herbert wrot= e: > On Mon, Apr 12, 2010 at 7:27 AM, Changli Gao wrot= e: >> On Mon, Apr 12, 2010 at 9:34 PM, Tom Herbert w= rote: >> > > Ideally, this should replace rps_cpus if it's a better interface.... > right now these would be conflicting interfaces. > How about sw-rxs and sw-rx-$ (SoftWare Receive queue). It is a software emulation of hardware receive queue, as softnet to NIC. sw-rx-$ is equivalent of /proc/irq/$/smp_affinity, only cpuid is instead of cpumask. Does anyone support this interface replacing the current rps_cpus? I do= =2E :) > > It's probably a little more work, but the CPU->weight mappings could > be implemented to cause minimal disruption in the rps_map. =C2=A0Also= , if > OOO is an issue, then the mitigation technique in RFS could be applie= d > (this will work best when hash table is larger I believe). > I am thinking about the cost of keeping packets in order. Is it really worth for this kind of random migration? --=20 Regards=EF=BC=8C Changli Gao(xiaosuo@gmail.com)