From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753692Ab1G1AC5 (ORCPT ); Wed, 27 Jul 2011 20:02:57 -0400 Received: from e4.ny.us.ibm.com ([32.97.182.144]:56624 "EHLO e4.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751435Ab1G1AC4 (ORCPT ); Wed, 27 Jul 2011 20:02:56 -0400 Date: Wed, 27 Jul 2011 17:02:54 -0700 From: "Paul E. McKenney" To: Kyle Moffett Cc: kernelmail.jms@gmail.com, RKK , linux-kernel@vger.kernel.org, cmm@us.ibm.com Subject: Re: Fw: RCU boot debug patch Message-ID: <20110728000254.GB2318@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <20110719224142.GT2357@linux.vnet.ibm.com> <20110722022937.GB2408@linux.vnet.ibm.com> <201107272247.50613.kernelmail.jms@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Jul 27, 2011 at 07:56:10PM -0400, Kyle Moffett wrote: > On Wed, Jul 27, 2011 at 17:47, wrote: > >>A debug patch to verify if there is RCU callbacks before the rcu scheduler is active > > So something gives me the impression we may have an answer somewhere in this dmesg for this... ;-) > > > > Here is my dmesg (the whole thing): > > > > function free_object_rcu > > [    0.166707] RCU scheduler is not active yet, Calling _call_rcu() with this function free_object_rcu > > [....thousands of lines snipped...] > > [    0.225223] RCU scheduler is not active yet, Calling _call_rcu() with this function free_object_rcu Thank you, Julie! > Well this looks an awful lot like a smoking gun here... > > I can only find one function called "free_object_rcu", and it seems to be a > static function in mm/kmemleak.c, and it certainly does this: > call_rcu(&object->rcu, free_object_rcu); > > So the next question for you would be: Do all the messages go away if you > turn off the kmemleak config option? Hello, Kyle, This is not at all showing a problem in free_object_rcu(), but rather as a sanity-check for a shoot-from-the-hip fix. The fix seemed to work, but my paranoia dictated that I check that my guess was correct. Which it appears that it was. So again, these messages do not indicate a problem in free_object_rcu(). Thanx, Paul