From mboxrd@z Thu Jan 1 00:00:00 1970 From: Prarit Bhargava Subject: Re: cpupower reports uninitialized values for offline cpus Date: Thu, 01 Oct 2015 09:11:55 -0400 Message-ID: <560D311B.8090507@redhat.com> References: <560ADF55.6090901@redhat.com> <2397035.37ChXSYNtU@skinner> <560C292B.5030905@redhat.com> <3211183.5fvOAutzxZ@skinner> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Return-path: Received: from mx1.redhat.com ([209.132.183.28]:42313 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933358AbbJANL4 (ORCPT ); Thu, 1 Oct 2015 09:11:56 -0400 In-Reply-To: <3211183.5fvOAutzxZ@skinner> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Thomas Renninger Cc: Jacob Tanenbaum , linux-pm@vger.kernel.org On 10/01/2015 08:29 AM, Thomas Renninger wrote: > On Wednesday, September 30, 2015 02:25:47 PM Prarit Bhargava wrote: >> On 09/30/2015 12:23 PM, Thomas Renninger wrote: > ... >> Yeah, we'd have to add an additional state to the CPU hotplug callback state >> machine (CPU_ADD_PHYSICAL?) to identify when a cpu was hot-added. I >> suppose it could be done but it seems like an awful lot of churn for little >> gain. > > It shouldn't be that hard. > If I find some time I give it a try. > This is a rather central piece of kernel code and affects > more or less all architectures, so getting this finally mainline > may take a while... > > So it would be nice if somewhen in the future we have 2 independent > fixes: > > The cpupower one as you wrote before, ignoring cores without > topology sysfs info (but not simply checking for "is core offline"), > better check whether the directory or a specific file exists. > > And a kernel one which still shows topology info of soft offlined > cores which with above cpupower solution will result in previous > output (offlined cores listed, but with valid topology info). > > Makes sense? Yup, makes sense. Jacob (Jake) is going to handle the cpupower stuff right away. I'll dive into the topology stuff in a little bit and/or assist Jake with it. Thanks Thomas :) P. > > Thomas >