From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932146AbaICJ47 (ORCPT ); Wed, 3 Sep 2014 05:56:59 -0400 Received: from mx1.redhat.com ([209.132.183.28]:52272 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756283AbaICJ44 (ORCPT ); Wed, 3 Sep 2014 05:56:56 -0400 Date: Wed, 3 Sep 2014 12:59:39 +0300 From: "Michael S. Tsirkin" To: Peter Zijlstra Cc: Jason Wang , Eliezer Tamir , Ingo Molnar , Mike Galbraith , davem@davemloft.net, netdev@vger.kernel.org, linux-kernel@vger.kernel.org, Ingo Molnar Subject: Re: [PATCH net-next 2/2] net: exit busy loop when another process is runnable Message-ID: <20140903095939.GA8929@redhat.com> References: <53F6F14B.1030609@redhat.com> <20140822074224.GB7372@gmail.com> <53FFEEC0.4000304@redhat.com> <540414AB.9000004@linux.intel.com> <54053998.4040604@redhat.com> <54056076.2030603@linux.intel.com> <20140902083124.GB10356@redhat.com> <5406B9E6.6050103@linux.intel.com> <5406C443.5030109@redhat.com> <20140903093605.GD4783@worktop.ger.corp.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20140903093605.GD4783@worktop.ger.corp.intel.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Sep 03, 2014 at 11:36:05AM +0200, Peter Zijlstra wrote: > On Wed, Sep 03, 2014 at 03:33:23PM +0800, Jason Wang wrote: > > A new issue is for virt users. I implement busy polling for virtio-net > > but we don't want one vcpu monopoly the cpu if there's some tasks on > > other vcpus. We may need some hint from host to guest to let it exit the > > loop if needed. > > Aw god.. virt nonsense is wrecking things again. I don't see a reason to bring virt up here. Original patch shows improvement with a simple loopback netperf test. Some people like the linux kernel so much, they want to run many instances of it on the same CPU. Understandable. So of course everything is then magnified x2 as you are running two schedulers and two networking stacks, but let's focus on the important thing which is better interaction between busy polling and the linux scheduler. -- MST