From: Minchan Kim Subject: Re: [PATCH 3/3] mm: slub: Default slub_max_order to 0 Date: Fri, 13 May 2011 07:58:17 +0900 Message-ID: References: <1305127773-10570-4-git-send-email-mgorman@suse.de> <1305213359.2575.46.camel@mulgrave.site> <1305214993.2575.50.camel@mulgrave.site> <1305215742.27848.40.camel@jaguar> <1305225467.2575.66.camel@mulgrave.site> <1305229447.2575.71.camel@mulgrave.site> <1305230652.2575.72.camel@mulgrave.site> <1305237882.2575.100.camel@mulgrave.site> <20110512221506.GM16531@cmpxchg.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: James Bottomley , Pekka Enberg , Christoph Lameter , Mel Gorman , Andrew Morton , Colin King , Raghavendra D Prabhu , Jan Kara , Chris Mason , Rik van Riel , linux-fsdevel , linux-mm , linux-kernel , linux-ext4 To: Johannes Weiner Return-path: Received: from mail-qy0-f174.google.com ([209.85.216.174]:41267 "EHLO mail-qy0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756240Ab1ELW6T convert rfc822-to-8bit (ORCPT ); Thu, 12 May 2011 18:58:19 -0400 In-Reply-To: <20110512221506.GM16531@cmpxchg.org> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Fri, May 13, 2011 at 7:15 AM, Johannes Weiner w= rote: > On Thu, May 12, 2011 at 05:04:41PM -0500, James Bottomley wrote: >> On Thu, 2011-05-12 at 15:04 -0500, James Bottomley wrote: >> > Confirmed, I'm afraid ... I can trigger the problem with all three >> > patches under PREEMPT. =C2=A0It's not a hang this time, it's just = kswapd >> > taking 100% system time on 1 CPU and it won't calm down after I un= load >> > the system. >> >> Just on a "if you don't know what's wrong poke about and see" basis,= I >> sliced out all the complex logic in sleeping_prematurely() and, as f= ar >> as I can tell, it cures the problem behaviour. =C2=A0I've loaded up = the >> system, and taken the tar load generator through three runs without >> producing a spinning kswapd (this is PREEMPT). =C2=A0I'll try with a >> non-PREEMPT kernel shortly. >> >> What this seems to say is that there's a problem with the complex lo= gic >> in sleeping_prematurely(). =C2=A0I'm pretty sure hacking up >> sleeping_prematurely() just to dump all the calculations is the wron= g >> thing to do, but perhaps someone can see what the right thing is ... > > I think I see the problem: the boolean logic of sleeping_prematurely(= ) > is odd. =C2=A0If it returns true, kswapd will keep running. =C2=A0So = if > pgdat_balanced() returns true, kswapd should go to sleep. > > This? Yes. Good catch. > > diff --git a/mm/vmscan.c b/mm/vmscan.c > index 2b701e0..092d773 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -2261,7 +2261,7 @@ static bool sleeping_prematurely(pg_data_t *pgd= at, int order, long remaining, > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * must be balanced > =C2=A0 =C2=A0 =C2=A0 =C2=A0 */ > =C2=A0 =C2=A0 =C2=A0 =C2=A0if (order) > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return pgdat_balan= ced(pgdat, balanced, classzone_idx); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return !pgdat_bala= nced(pgdat, balanced, classzone_idx); > =C2=A0 =C2=A0 =C2=A0 =C2=A0else > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0return !all_zo= nes_ok; > =C2=A0} > > -- > To unsubscribe, send a message with 'unsubscribe linux-mm' in > the body to majordomo@kvack.org. =C2=A0For more info on Linux MM, > see: http://www.linux-mm.org/ . > Fight unfair telecom internet charges in Canada: sign http://stopthem= eter.ca/ > Don't email: email@kvack.org > --=20 Kind regards, Minchan Kim -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html