From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail203.messagelabs.com (mail203.messagelabs.com [216.82.254.243]) by kanga.kvack.org (Postfix) with ESMTP id 44B756B00EE for ; Thu, 21 Jul 2011 20:30:46 -0400 (EDT) Received: by qwa26 with SMTP id 26so1294458qwa.14 for ; Thu, 21 Jul 2011 17:30:43 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: References: <1308926697-22475-1-git-send-email-mgorman@suse.de> <20110721153722.GD1713@barrios-desktop> <20110721160958.GT5349@suse.de> <20110721162417.GF1713@barrios-desktop> <20110721164238.GA3326@barrios-desktop> Date: Fri, 22 Jul 2011 09:30:43 +0900 Message-ID: Subject: Re: [PATCH 0/4] Stop kswapd consuming 100% CPU when highest zone is small From: Minchan Kim Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-mm@kvack.org List-ID: To: Andrew Lutomirski Cc: Mel Gorman , Andrew Morton , P?draig Brady , James Bottomley , Colin King , Rik van Riel , Johannes Weiner , linux-mm , linux-kernel On Fri, Jul 22, 2011 at 1:58 AM, Andrew Lutomirski wrote: > On Thu, Jul 21, 2011 at 12:42 PM, Minchan Kim wro= te: >> On Thu, Jul 21, 2011 at 12:36:11PM -0400, Andrew Lutomirski wrote: >>> On Thu, Jul 21, 2011 at 12:24 PM, Minchan Kim w= rote: >>> > On Thu, Jul 21, 2011 at 05:09:59PM +0100, Mel Gorman wrote: >>> >> On Fri, Jul 22, 2011 at 12:37:22AM +0900, Minchan Kim wrote: >>> >> > On Fri, Jun 24, 2011 at 03:44:53PM +0100, Mel Gorman wrote: >>> >> > > (Built this time and passed a basic sniff-test.) >>> >> > > >>> >> > > During allocator-intensive workloads, kswapd will be woken frequ= ently >>> >> > > causing free memory to oscillate between the high and min waterm= ark. >>> >> > > This is expected behaviour. =C2=A0Unfortunately, if the highest = zone is >>> >> > > small, a problem occurs. >>> >> > > >>> >> > > This seems to happen most with recent sandybridge laptops but it= 's >>> >> > > probably a co-incidence as some of these laptops just happen to = have >>> >> > > a small Normal zone. The reproduction case is almost always duri= ng >>> >> > > copying large files that kswapd pegs at 100% CPU until the file = is >>> >> > > deleted or cache is dropped. >>> >> > > >>> >> > > The problem is mostly down to sleeping_prematurely() keeping ksw= apd >>> >> > > awake when the highest zone is small and unreclaimable and compo= unded >>> >> > > by the fact we shrink slabs even when not shrinking zones causin= g a lot >>> >> > > of time to be spent in shrinkers and a lot of memory to be recla= imed. >>> >> > > >>> >> > > Patch 1 corrects sleeping_prematurely to check the zones matchin= g >>> >> > > =C2=A0 the classzone_idx instead of all zones. >>> >> > > >>> >> > > Patch 2 avoids shrinking slab when we are not shrinking a zone. >>> >> > > >>> >> > > Patch 3 notes that sleeping_prematurely is checking lower zones = against >>> >> > > =C2=A0 a high classzone which is not what allocators or balance_= pgdat() >>> >> > > =C2=A0 is doing leading to an artifical believe that kswapd shou= ld be >>> >> > > =C2=A0 still awake. >>> >> > > >>> >> > > Patch 4 notes that when balance_pgdat() gives up on a high zone = that the >>> >> > > =C2=A0 decision is not communicated to sleeping_prematurely() >>> >> > > >>> >> > > This problem affects 2.6.38.8 for certain and is expected to aff= ect >>> >> > > 2.6.39 and 3.0-rc4 as well. If accepted, they need to go to -sta= ble >>> >> > > to be picked up by distros and this series is against 3.0-rc4. I= 've >>> >> > > cc'd people that reported similar problems recently to see if th= ey >>> >> > > still suffer from the problem and if this fixes it. >>> >> > > >>> >> > >>> >> > Good! >>> >> > This patch solved the problem. >>> >> > But there is still a mystery. >>> >> > >>> >> > In log, we could see excessive shrink_slab calls. >>> >> >>> >> Yes, because shrink_slab() was called on each loop through >>> >> balance_pgdat() even if the zone was balanced. >>> >> >>> >> >>> >> > And as you know, we had merged patch which adds cond_resched where= last of the function >>> >> > in shrink_slab. So other task should get the CPU and we should not= see >>> >> > 100% CPU of kswapd, I think. >>> >> > >>> >> >>> >> cond_resched() is not a substitute for going to sleep. >>> > >>> > Of course, it's not equal with sleep but other task should get CPU an= d conusme their time slice >>> > So we should never see 100% CPU consumption of kswapd. >>> > No? >>> >>> If the rest of the system is idle, then kswapd will happily use 100% >>> CPU. =C2=A0(Or on a multi-core system, kswapd will use close to 100% of= one >> >> Of course. But at least, we have a test program and I think it's not idl= e. > > The test program I used was 'top', which is pretty close to idle. > >> >>> CPU even if another task is using the other one. =C2=A0This is bad enou= gh >>> on a desktop, but on a laptop you start to notice when your battery >>> dies.) >> >> Of course it's bad. :) >> What I want to know is just what's exact cause of 100% CPU usage. >> It might be not 100% but we might use the word sloppily. >> > > Well, if you want to pedantic, my laptop can, in theory, demonstrate > true 100% CPU usage. =C2=A0Trigger the bug, suspend every other thread, a= nd > listen to the laptop fan spin and feel the laptop get hot. =C2=A0(The fan > is controlled by the EC and takes no CPU.) > > In practice, the usage was close enough to 100% that it got rounded. > > The cond_resched was enough to at least make the system responsive > instead of the hard freeze I used to get. I don't want to be pedantic. :) What I have a thought about 100% CPU usage was that it doesn't yield CPU and spins on the CPU but as I heard your example(ie, cond_resched makes the system responsive), it's not the case. It was just to use most of time in kswapd, not 100%. It seems I was paranoid about the word, sorry for that. --=20 Kind regards, Minchan Kim -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: email@kvack.org