From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756962AbcK3NTP (ORCPT ); Wed, 30 Nov 2016 08:19:15 -0500 Received: from mail-wj0-f194.google.com ([209.85.210.194]:35197 "EHLO mail-wj0-f194.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755449AbcK3NTN (ORCPT ); Wed, 30 Nov 2016 08:19:13 -0500 Date: Wed, 30 Nov 2016 14:19:10 +0100 From: Michal Hocko To: "Paul E. McKenney" Cc: Donald Buczek , Paul Menzel , dvteam@molgen.mpg.de, linux-mm@kvack.org, linux-kernel@vger.kernel.org, Josh Triplett Subject: Re: INFO: rcu_sched detected stalls on CPUs/tasks with `kswapd` and `mem_cgroup_shrink_node` Message-ID: <20161130131910.GF18432@dhcp22.suse.cz> References: <20161121141818.GD18112@dhcp22.suse.cz> <20161121142901.GV3612@linux.vnet.ibm.com> <68025f6c-6801-ab46-b0fc-a9407353d8ce@molgen.mpg.de> <20161124101525.GB20668@dhcp22.suse.cz> <583AA50A.9010608@molgen.mpg.de> <20161128110449.GK14788@dhcp22.suse.cz> <109d5128-f3a4-4b6e-db17-7a1fcb953500@molgen.mpg.de> <29196f89-c35e-f79d-8e4d-2bf73fe930df@molgen.mpg.de> <20161130110944.GD18432@dhcp22.suse.cz> <20161130115320.GO3924@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20161130115320.GO3924@linux.vnet.ibm.com> User-Agent: Mutt/1.6.0 (2016-04-01) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Wed 30-11-16 03:53:20, Paul E. McKenney wrote: > On Wed, Nov 30, 2016 at 12:09:44PM +0100, Michal Hocko wrote: > > [CCing Paul] > > > > On Wed 30-11-16 11:28:34, Donald Buczek wrote: > > [...] > > > shrink_active_list gets and releases the spinlock and calls cond_resched(). > > > This should give other tasks a chance to run. Just as an experiment, I'm > > > trying > > > > > > --- a/mm/vmscan.c > > > +++ b/mm/vmscan.c > > > @@ -1921,7 +1921,7 @@ static void shrink_active_list(unsigned long > > > nr_to_scan, > > > spin_unlock_irq(&pgdat->lru_lock); > > > > > > while (!list_empty(&l_hold)) { > > > - cond_resched(); > > > + cond_resched_rcu_qs(); > > > page = lru_to_page(&l_hold); > > > list_del(&page->lru); > > > > > > and didn't hit a rcu_sched warning for >21 hours uptime now. We'll see. > > > > This is really interesting! Is it possible that the RCU stall detector > > is somehow confused? > > No, it is not confused. Again, cond_resched() is not a quiescent > state unless it does a context switch. Therefore, if the task running > in that loop was the only runnable task on its CPU, cond_resched() > would -never- provide RCU with a quiescent state. Sorry for being dense here. But why cannot we hide the QS handling into cond_resched()? I mean doesn't every current usage of cond_resched suffer from the same problem wrt RCU stalls? > In contrast, cond_resched_rcu_qs() unconditionally provides RCU > with a quiescent state (hence the _rcu_qs in its name), regardless > of whether or not a context switch happens. > > It is therefore expected behavior that this change might prevent > RCU CPU stall warnings. > > Thanx, Paul -- Michal Hocko SUSE Labs