From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH 4/5] KVM: Lazify fpu activation and deactivation Date: Thu, 14 Jan 2010 17:04:45 +0200 Message-ID: <4B4F328D.6020103@redhat.com> References: <1262190342-18611-1-git-send-email-avi@redhat.com> <1262190342-18611-5-git-send-email-avi@redhat.com> <4B4F17F0.7010501@redhat.com> <4B4F1819.7040105@redhat.com> <1263479672.2300.5.camel@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: Marcelo Tosatti , Sheng Yang , Joerg Roedel , kvm@vger.kernel.org To: Lucas Meneghel Rodrigues Return-path: Received: from mx1.redhat.com ([209.132.183.28]:50787 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932206Ab0ANPEu (ORCPT ); Thu, 14 Jan 2010 10:04:50 -0500 In-Reply-To: <1263479672.2300.5.camel@localhost.localdomain> Sender: kvm-owner@vger.kernel.org List-ID: On 01/14/2010 04:34 PM, Lucas Meneghel Rodrigues wrote: > On Thu, 2010-01-14 at 15:11 +0200, Avi Kivity wrote: > >>> Strangely, autotest only caught this on AMD and even it took a while. >>> Lucas, can you integrate something like the following into autotest, >>> so we exercise the preemption code harder? >>> > All right Avi, just added it to our TODO list, thanks! > On second thoughts, this will cause a severe performance loss (which may also affect test results), due to cpu overcommit and excessive vcpu migrations. How about, instead: Launch a background process with nthreads == ncpus, bind each thread to a cpu, and execute a loop on each thread that sleeps for a random amount of time from 1 to 10 milliseconds. This will exercise the preemption code. In addition, every second or so, offline and online a random cpu. This will cause tasks to be migrated and exercise the task migration code. -- error compiling committee.c: too many arguments to function