From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jane Li Subject: Re: [PATCH] cpufreq: Fix timer/workqueue corruption by protecting reading governor_enabled Date: Thu, 2 Jan 2014 11:15:44 +0800 Message-ID: <52C4D9E0.5070501@marvell.com> References: <1388136651-21883-1-git-send-email-jiel@marvell.com> <52C112CC.4040306@marvell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: cpufreq-owner@vger.kernel.org To: Viresh Kumar Cc: "Rafael J. Wysocki" , "cpufreq@vger.kernel.org" , "linux-pm@vger.kernel.org" , Linux Kernel Mailing List List-Id: linux-pm@vger.kernel.org >> Yes, I test it. After adding cpufreq_governor_lock in gov_queue_work() and >> running same test, there is no debugobjects warning. > But it really can't work at all.. There should be a separate copy of lock > in every file that includes cpufreq.h.. And so this shouldn't have worked. > Oh.. I understand what you mean now. My patch is not right and cannot fix this issue. By default, the debugobjects warning sometimes occurs after five minutes, and sometimes occurs after twenty hours. With this patch, I test more than fifty hours, and warning did not occurs. It shows that my test time is not long enough and miss the right one. I have updated PATCH v2, please review again. Thanks.