From mboxrd@z Thu Jan 1 00:00:00 1970 From: Dave Jones Subject: Re: cpufreq/ondemand: unpinning an unpinned lock. Date: Thu, 16 Jul 2015 08:42:41 -0400 Message-ID: <20150716124241.GA20320@codemonkey.org.uk> References: <20150715220422.GA13789@codemonkey.org.uk> <3682267.1lcDzeZuDs@vostro.rjw.lan> <20150716051150.GA15404@linux> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from arcturus.aphlor.org ([188.246.204.175]:58607 "EHLO arcturus.aphlor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754180AbbGPMmt (ORCPT ); Thu, 16 Jul 2015 08:42:49 -0400 Content-Disposition: inline In-Reply-To: <20150716051150.GA15404@linux> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Viresh Kumar Cc: "Rafael J. Wysocki" , Linux Kernel , linux-pm@vger.kernel.org On Thu, Jul 16, 2015 at 10:41:50AM +0530, Viresh Kumar wrote: > > > WARNING: CPU: 1 PID: 29529 at kernel/locking/lockdep.c:3497 lock_unpin_lock+0x109/0x110() > > > unpinning an unpinned lock > > > CPU: 1 PID: 29529 Comm: kworker/1:1 Not tainted 4.2.0-rc2-think+ #3 > > > Workqueue: events od_dbs_timer > > > 0000000000000009 ffff880094d5baa8 ffffffffae7f5e6f 0000000000000007 > > > ffff880094d5baf8 ffff880094d5bae8 ffffffffae07b91a 0000000000000118 > > > 00000000000000e0 ffff880507bd5c58 0000000000000092 0000000000000004 > > > Call Trace: > > > [] dump_stack+0x4f/0x7b > > > [] warn_slowpath_common+0x8a/0xc0 > > > [] warn_slowpath_fmt+0x46/0x50 > > > [] lock_unpin_lock+0x109/0x110 > > > [] __schedule+0x3ac/0xb60 > > > [] schedule+0x41/0x90 > > > [] schedule_preempt_disabled+0x18/0x30 > > > [] mutex_lock_nested+0x16f/0x3e0 > > > [] ? gov_queue_work+0x2f/0xf0 > > > [] ? od_check_cpu+0x57/0xd0 > > > [] ? gov_queue_work+0x2f/0xf0 > > > [] gov_queue_work+0x2f/0xf0 > > > [] od_dbs_timer+0xbd/0x150 > > I don't know why this will happen. Just to confirm, you are getting > this over 4.2-rc(1 or 2)? And you weren't getting these on 4.1 at all? > And its always reproducible? How ? I've hit it once, on rc2. I didn't have a lot of time for testing rc1. > There have been races in cpufreq core since sometime and what got > pushed in 4.2-rc1 is just half of the fix. The other half is present > here: > > http://marc.info/?i=cover.1434713657.git.viresh.kumar%40linaro.org > > Please try this and let us know if things work well or not. Sure. Though proving the absense of a bug like this is always fun. Dave