From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161149AbXDSG4x (ORCPT ); Thu, 19 Apr 2007 02:56:53 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161151AbXDSG4x (ORCPT ); Thu, 19 Apr 2007 02:56:53 -0400 Received: from mx10.go2.pl ([193.17.41.74]:59529 "EHLO poczta.o2.pl" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1161149AbXDSG4w (ORCPT ); Thu, 19 Apr 2007 02:56:52 -0400 Date: Thu, 19 Apr 2007 09:02:28 +0200 From: Jarek Poplawski To: Ingo Molnar Cc: linux-kernel@vger.kernel.org Subject: Re: [PATCH -mm] workqueue: debug possible lockups in flush_workqueue Message-ID: <20070419070228.GC1782@ff.dom.local> References: <20070419055247.GA1782@ff.dom.local> <20070419061416.GB18894@elte.hu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070419061416.GB18894@elte.hu> User-Agent: Mutt/1.4.2.2i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Apr 19, 2007 at 08:14:16AM +0200, Ingo Molnar wrote: > > * Jarek Poplawski wrote: > > > Here is my patch proposal for detecting possible lockups, when > > flush_workqueue caller holds a lock (e.g. rtnl_lock) also used in work > > functions. > > looks good in principle - did you test it and it caught a bug that wasnt > caught before? Yes, but it was only my own testing bug... (I'm not a good tester, sorry). > > > +#ifdef CONFIG_PROVE_LOCKING > > +/* Detect possible flush_workqueue() lockup with circular dependency check. */ > > +static struct lockdep_map flush_dep_map = { .name = "flush_dep_map" }; > > +#endif > > > +#ifdef CONFIG_PROVE_LOCKING > > + /* lockdep dependency: flush_dep_map (read) before any lock: */ > > + lock_acquire(&flush_dep_map, 0, 0, 1, 2, _THIS_IP_); > > +#endif > > i think the #ifdef should only be needed for the .name initialization - > both lock_acquire() and lock_release() maps to NOP if PROVE_LOCKING is > off. There is also DEBUG_LOCK_ALLOC without PROVE_LOCKING possibility, which isn't usable here. Jarek P.