From mboxrd@z Thu Jan 1 00:00:00 1970 From: Lucas Meneghel Rodrigues Subject: Re: [PATCH] cpu hotplug issue Date: Thu, 21 Jul 2011 10:27:41 -0300 Message-ID: <4E28294D.6040605@redhat.com> References: <20110720083507.GS2400@redhat.com> <20110721113342.GB3044@redhat.com> <4E281090.9070300@siemens.com> <20110721115118.GD3044@redhat.com> <4E2813AE.7090102@siemens.com> <4E2818FB.5070302@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Jan Kiszka , Gleb Natapov , Vasilis Liaskovitis , "kvm@vger.kernel.org" , Anthony Liguori To: Avi Kivity Return-path: Received: from mx1.redhat.com ([209.132.183.28]:14592 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750802Ab1GUN1s (ORCPT ); Thu, 21 Jul 2011 09:27:48 -0400 In-Reply-To: <4E2818FB.5070302@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: On 07/21/2011 09:18 AM, Avi Kivity wrote: > On 07/21/2011 02:55 PM, Jan Kiszka wrote: >> >> >> > CPU hotplug for Linux suppose to be easy (with allow_hotplug patch >> > applied). But we have two bugs currently. One is that ACPI interrupt >> >> I bet we have multiple bugs for quite a while now, which accelerated bit >> rotting even further. >> > > kvm-autotest to the rescue! > > Lucas, do we have a cpu hotplug test in kvm-autotest? Hey Avi, I just read the thread, and let me tell you guys, I have worked on a cpu_set test, that was supposed to test CPU hotplug. I have created the entire test, however, I could never ever get the test to pass due to the fact that the infrastructure on qemu is broken. I have revisited the test late 2010, and I've verified that although cpu_set does not crash qemu anymore, the guest OS has no idea whatsoever that a new CPU was added. I remember that one of the problems we had was lack of functionality in Seabios... I can gladly get back to it and get it commited (I was reluctant in getting it upstream because I never ever got a successful test result), but I can put it into shape and get it commited. Cheers, Lucas > If so, I promise to add it to my autotest run once it is fixed, to > prevent the spread of the disease. > > btw, Anthony, are you running kvm-autotest these days? with qemu-kvm > diverging less and less from qemu.git these days, it's becoming less > effective to run it downstream. >