From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail172.messagelabs.com (mail172.messagelabs.com [216.82.254.3]) by kanga.kvack.org (Postfix) with ESMTP id C0CEF6B0022 for ; Fri, 13 May 2011 00:16:04 -0400 (EDT) Received: by qyk2 with SMTP id 2so155158qyk.14 for ; Thu, 12 May 2011 21:16:02 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: References: <1889981320.330808.1305081044822.JavaMail.root@zmail06.collab.prod.int.phx2.redhat.com> Date: Fri, 13 May 2011 13:16:02 +0900 Message-ID: Subject: Re: OOM Killer don't works at all if the system have >gigabytes memory (was Re: [PATCH] mm: check zone->all_unreclaimable in all_unreclaimable()) From: Minchan Kim Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-mm@kvack.org List-ID: To: David Rientjes Cc: CAI Qian , KOSAKI Motohiro , avagin@gmail.com, Andrey Vagin , Andrew Morton , Mel Gorman , linux-mm@kvack.org, linux-kernel@vger.kernel.org, Hugh Dickins , Oleg Nesterov On Fri, May 13, 2011 at 4:38 AM, David Rientjes wrote= : > On Thu, 12 May 2011, Minchan Kim wrote: > >> > processes a 1% bonus for every 30% of memory they use as proposed >> > earlier.) >> >> I didn't follow earlier your suggestion. >> But it's not formal patch so I expect if you send formal patch to >> merge, you would write down the rationale. >> > > Yes, I'm sure we'll still have additional discussion when KOSAKI-san > replies to my review of his patchset, so this quick patch was written onl= y > for CAI's testing at this point. > > In reference to the above, I think that giving root processes a 3% bonus > at all times may be a bit aggressive. =C2=A0As mentioned before, I don't = think > that all root processes using 4% of memory and the remainder of system > threads are using 1% should all be considered equal. =C2=A0At the same ti= me, I > do not believe that two threads using 50% of memory should be considered > equal if one is root and one is not. =C2=A0So my idea was to discount 1% = for > every 30% of memory that a root process uses rather than a strict 3%. > > That change can be debated and I think we'll probably settle on something > more aggressive like 1% for every 10% of memory used since oom scores are > only useful in comparison to other oom scores: in the above scenario wher= e > there are two threads, one by root and one not by root, using 50% of > memory each, I think it would be legitimate to give the root task a 5% > bonus so that it would only be selected if no other threads used more tha= n > 44% of memory (even though the root thread is truly using 50%). > > This is a heuristic within the oom killer badness scoring that can always > be debated back and forth, but I think a 1% bonus for root processes for > every 10% of memory used is plausible. > > Comments? Yes. Tend to agree. Apparently, absolute 3% bonus is a problem in CAI's case. Your approach which makes bonus with function of rss is consistent with current OOM heuristic. So In consistency POV, I like it as it could help deterministic OOM policy. About 30% or 10% things, I think it's hard to define a ideal magic value for handling for whole workloads. It would be very arguable. So we might need some standard method to measure it/or redhat/suse peoples. Anyway, I don't want to argue it until we get a number. > >> > diff --git a/mm/oom_kill.c b/mm/oom_kill.c >> > --- a/mm/oom_kill.c >> > +++ b/mm/oom_kill.c >> > @@ -160,7 +160,7 @@ unsigned int oom_badness(struct task_struct *p, st= ruct mem_cgroup *mem, >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 */ >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0if (p->flags & PF_OOM_ORIGIN) { >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0task_unlock(p); >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return 1000; >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 return 10000; >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0} >> > >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* >> > @@ -177,32 +177,32 @@ unsigned int oom_badness(struct task_struct *p, = struct mem_cgroup *mem, >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0points =3D get_mm_rss(p->mm) + p->mm->nr_pt= es; >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0points +=3D get_mm_counter(p->mm, MM_SWAPEN= TS); >> > >> > - =C2=A0 =C2=A0 =C2=A0 points *=3D 1000; >> > + =C2=A0 =C2=A0 =C2=A0 points *=3D 10000; >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0points /=3D totalpages; >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0task_unlock(p); >> > >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* Root processes get 3% bonus, just like = the __vm_enough_memory() >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* implementation used by LSMs. >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* Root processes get 1% bonus per 30% mem= ory used for a total of 3% >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* possible just like LSMs. >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 */ >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0if (has_capability_noaudit(p, CAP_SYS_ADMIN= )) >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 points -=3D 30; >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 points -=3D 100 * (= points / 3000); >> > >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * /proc/pid/oom_score_adj ranges from -100= 0 to +1000 such that it may >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * either completely disable oom killing or= always prefer a certain >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * task. >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 */ >> > - =C2=A0 =C2=A0 =C2=A0 points +=3D p->signal->oom_score_adj; >> > + =C2=A0 =C2=A0 =C2=A0 points +=3D p->signal->oom_score_adj * 10; >> > >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * Never return 0 for an eligible task that= may be killed since it's >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* possible that no single user task uses = more than 0.1% of memory and >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* possible that no single user task uses = more than 0.01% of memory and >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 * no single admin tasks uses more than 3.0= %. >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 */ >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0if (points <=3D 0) >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0return 1; >> > - =C2=A0 =C2=A0 =C2=A0 return (points < 1000) ? points : 1000; >> > + =C2=A0 =C2=A0 =C2=A0 return (points < 10000) ? points : 10000; >> > =C2=A0} >> > >> > =C2=A0/* >> > @@ -314,7 +314,7 @@ static struct task_struct *select_bad_process(unsi= gned int *ppoints, >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 */ >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0if (p =3D=3D current) { >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0chosen =3D p; >> > - =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 *ppoints =3D 1000; >> > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 *ppoints =3D 10000; >> >> Scattering constant value isn't good. >> You are proving it now. >> I think you did it since this is not a formal patch. >> I expect you will define new value (ex, OOM_INTERNAL_MAX_SCORE or whatev= er) >> > > Right, we could probably do something like > > =C2=A0 =C2=A0 =C2=A0 =C2=A0#define OOM_SCORE_MAX_FACTOR =C2=A0 =C2=A010 > =C2=A0 =C2=A0 =C2=A0 =C2=A0#define OOM_SCORE_MAX =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 (OOM_SCORE_ADJ_MAX * OOM_SCORE_MAX_FACTOR) > > in mm/oom_kill.c, which would then be used to replace all of the constant= s > above since OOM_SCORE_ADJ_MAX is already defined to be 1000 in > include/linux/oom.h. Looks good to me. Let's wait KOSAKI's opinion and CAI's test result. --=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