From mboxrd@z Thu Jan 1 00:00:00 1970 From: Viresh Kumar Subject: Re: [PATCH v2] cpufreq: schedutil: Examine the correct CPU when we update util Date: Fri, 3 Nov 2017 09:10:22 +0530 Message-ID: <20171103034022.GD4240@vireshk-i7> References: <20171102111022.GB4240@vireshk-i7> <20171102113840.17439-1-chris.redpath@arm.com> <20171102114048.GC4240@vireshk-i7> <037169f3-61f9-b29e-9561-0f7dba213cd5@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mail-pg0-f50.google.com ([74.125.83.50]:49031 "EHLO mail-pg0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751297AbdKCDk1 (ORCPT ); Thu, 2 Nov 2017 23:40:27 -0400 Received: by mail-pg0-f50.google.com with SMTP id v78so1396528pgb.5 for ; Thu, 02 Nov 2017 20:40:27 -0700 (PDT) Content-Disposition: inline In-Reply-To: <037169f3-61f9-b29e-9561-0f7dba213cd5@arm.com> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Chris Redpath Cc: linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org, Morten Rasmussen , Dietmar Eggemann , "Rafael J . Wysocki" , Ingo Molnar , Peter Zijlstra On 02-11-17, 12:06, Chris Redpath wrote: > According to my diff, this was the commit which switched from assigning > the values directly (and not overwriting the cpu member, which was > introduced in the other commit you reference) to using a memset and > clearing the whole struct. I understand what you want to convey here but the log says something else according to me. It says: "Since: 4296f23ed cpufreq: schedutil: Fix per-CPU structure initialization in sugov_start(), We lost the value of sg_cpu->cpu which is assigned during sugov_register." Reading this line it looks like sg_cpu->cpu was screwed up by 4296f23ed, which happened in 4.9. But the sg_cpu->cpu field itself got added in 4.14 and so I think we should write it differently. If I were you, I wouldn't mention 4296f23ed here at all but just say that memset() is clearing the value of sg_cpu->cpu, fix that. Thanks. -- viresh