From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1764550AbXGaPCt (ORCPT ); Tue, 31 Jul 2007 11:02:49 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1761473AbXGaPCm (ORCPT ); Tue, 31 Jul 2007 11:02:42 -0400 Received: from e2.ny.us.ibm.com ([32.97.182.142]:42473 "EHLO e2.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1761228AbXGaPCl (ORCPT ); Tue, 31 Jul 2007 11:02:41 -0400 Date: Tue, 31 Jul 2007 08:02:34 -0700 From: "Paul E. McKenney" To: Greg KH Cc: linux-kernel@vger.kernel.org, stable@kernel.org, torvalds@linux-foundation.org, akpm@linux-foundation.org, Justin Forbes , Zwane Mwaikambo , "Theodore Ts'o" , Randy Dunlap , Dave Jones , Chuck Wolber , Chris Wedgwood , Michael Krufky , Chuck Ebbert , Domenico Andreoli , alan@lxorguk.ukuu.org.uk, vatsa@linux.vnet.ibm.com, chrisw@sous-sol.org, clameter@sgi.com, Ingo Molnar Subject: Re: [patch 14/26] sched: fix next_interval determination in idle_balance() Message-ID: <20070731150234.GA11956@linux.vnet.ibm.com> Reply-To: paulmck@linux.vnet.ibm.com References: <20070731042108.546594256@blue.kroah.org> <20070731043254.GO3975@kroah.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070731043254.GO3975@kroah.com> User-Agent: Mutt/1.5.13 (2006-08-11) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jul 30, 2007 at 09:32:54PM -0700, Greg KH wrote: > -stable review patch. If anyone has any objections, please let us know. Hello, Greg, No objections here -- on the contrary, very nice not to have to manually "taskset" my rcutorture runs. ;-) Thanx, Paul > ------------------ > > From: Christoph Lameter > > Fix massive SMP imbalance on NUMA nodes observed on 2.6.21.5 with CFS. > (and later on reproduced without CFS as well). > > The intervals of domains that do not have SD_BALANCE_NEWIDLE must be > considered for the calculation of the time of the next balance. > Otherwise we may defer rebalancing forever and nodes might stay idle for > very long times. > > Siddha also spotted that the conversion of the balance interval to > jiffies is missing. Fix that to. > > From: Srivatsa Vaddagiri > > also continue the loop if !(sd->flags & SD_LOAD_BALANCE). > > Tested-by: Paul E. McKenney > > It did in fact trigger under all three of mainline, CFS, and -rt > including CFS -- see below for a couple of emails from last Friday > giving results for these three on the AMD box (where it happened) and on > a single-quad NUMA-Q system (where it did not, at least not with such > severity). > > Signed-off-by: Christoph Lameter > Signed-off-by: Ingo Molnar > Signed-off-by: Chris Wright > Signed-off-by: Greg Kroah-Hartman > --- > kernel/sched.c | 22 +++++++++++++--------- > 1 file changed, 13 insertions(+), 9 deletions(-) > > --- linux-2.6.21.6.orig/kernel/sched.c > +++ linux-2.6.21.6/kernel/sched.c > @@ -2831,17 +2831,21 @@ static void idle_balance(int this_cpu, s > unsigned long next_balance = jiffies + 60 * HZ; > > for_each_domain(this_cpu, sd) { > - if (sd->flags & SD_BALANCE_NEWIDLE) { > + unsigned long interval; > + > + if (!(sd->flags & SD_LOAD_BALANCE)) > + continue; > + > + if (sd->flags & SD_BALANCE_NEWIDLE) > /* If we've pulled tasks over stop searching: */ > pulled_task = load_balance_newidle(this_cpu, > - this_rq, sd); > - if (time_after(next_balance, > - sd->last_balance + sd->balance_interval)) > - next_balance = sd->last_balance > - + sd->balance_interval; > - if (pulled_task) > - break; > - } > + this_rq, sd); > + > + interval = msecs_to_jiffies(sd->balance_interval); > + if (time_after(next_balance, sd->last_balance + interval)) > + next_balance = sd->last_balance + interval; > + if (pulled_task) > + break; > } > if (!pulled_task) > /* > > --