From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from psmtp.com (na3sys010amx191.postini.com [74.125.245.191]) by kanga.kvack.org (Postfix) with SMTP id 175EA6B004A for ; Wed, 29 Feb 2012 04:49:11 -0500 (EST) Received: by bkwq16 with SMTP id q16so133546bkw.14 for ; Wed, 29 Feb 2012 01:49:09 -0800 (PST) MIME-Version: 1.0 In-Reply-To: <1329929337-16648-13-git-send-email-m.szyprowski@samsung.com> References: <1329929337-16648-1-git-send-email-m.szyprowski@samsung.com> <1329929337-16648-13-git-send-email-m.szyprowski@samsung.com> From: Barry Song <21cnbao@gmail.com> Date: Wed, 29 Feb 2012 17:48:49 +0800 Message-ID: Subject: Re: [Linaro-mm-sig] [PATCHv23 12/16] mm: trigger page reclaim in alloc_contig_range() to stabilise watermarks Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-mm@kvack.org List-ID: To: Marek Szyprowski Cc: linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org, linux-media@vger.kernel.org, linux-mm@kvack.org, linaro-mm-sig@lists.linaro.org, Ohad Ben-Cohen , Daniel Walker , Russell King , Arnd Bergmann , Jonathan Corbet , Mel Gorman , Michal Nazarewicz , Dave Hansen , Jesse Barker , Kyungmin Park , Andrew Morton , Rob Clark , KAMEZAWA Hiroyuki 2012/2/23 Marek Szyprowski : > alloc_contig_range() performs memory allocation so it also should keep > track on keeping the correct level of memory watermarks. This commit adds > a call to *_slowpath style reclaim to grab enough pages to make sure that > the final collection of contiguous pages from freelists will not starve > the system. > > Signed-off-by: Marek Szyprowski > Signed-off-by: Kyungmin Park > CC: Michal Nazarewicz > Tested-by: Rob Clark > Tested-by: Ohad Ben-Cohen > Tested-by: Benjamin Gaignard > Tested-by: Robert Nelson > --- > =C2=A0include/linux/mmzone.h | =C2=A0 =C2=A09 +++++++ > =C2=A0mm/page_alloc.c =C2=A0 =C2=A0 =C2=A0 =C2=A0| =C2=A0 62 ++++++++++++= ++++++++++++++++++++++++++++++++++++ > =C2=A02 files changed, 71 insertions(+), 0 deletions(-) > > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h > index 4781f30..77db8c0 100644 > --- a/include/linux/mmzone.h > +++ b/include/linux/mmzone.h > @@ -63,8 +63,10 @@ enum { > > =C2=A0#ifdef CONFIG_CMA > =C2=A0# =C2=A0define is_migrate_cma(migratetype) unlikely((migratetype) = =3D=3D MIGRATE_CMA) > +# =C2=A0define cma_wmark_pages(zone) =C2=A0 =C2=A0 =C2=A0 =C2=A0zone->mi= n_cma_pages > =C2=A0#else > =C2=A0# =C2=A0define is_migrate_cma(migratetype) false > +# =C2=A0define cma_wmark_pages(zone) 0 > =C2=A0#endif > > =C2=A0#define for_each_migratetype_order(order, type) \ > @@ -371,6 +373,13 @@ struct zone { > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* see spanned/present_pages for more descript= ion */ > =C2=A0 =C2=A0 =C2=A0 =C2=A0seqlock_t =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 span_seqlock; > =C2=A0#endif > +#ifdef CONFIG_CMA > + =C2=A0 =C2=A0 =C2=A0 /* > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* CMA needs to increase watermark levels dur= ing the allocation > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* process to make sure that the system is no= t starved. > + =C2=A0 =C2=A0 =C2=A0 =C2=A0*/ > + =C2=A0 =C2=A0 =C2=A0 unsigned long =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 m= in_cma_pages; > +#endif > =C2=A0 =C2=A0 =C2=A0 =C2=A0struct free_area =C2=A0 =C2=A0 =C2=A0 =C2=A0fr= ee_area[MAX_ORDER]; > > =C2=A0#ifndef CONFIG_SPARSEMEM > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 7a0d286..39cd74f 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -5092,6 +5092,11 @@ static void __setup_per_zone_wmarks(void) > =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=A0low + (mi= n >> 2); > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0zone->watermark[WM= ARK_HIGH] =3D min_wmark_pages(zone) + > =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=A0low + (mi= n >> 1); > + > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 zone->watermark[WMARK_= MIN] +=3D cma_wmark_pages(zone); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 zone->watermark[WMARK_= LOW] +=3D cma_wmark_pages(zone); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 zone->watermark[WMARK_= HIGH] +=3D cma_wmark_pages(zone); > + > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0setup_zone_migrate= _reserve(zone); > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0spin_unlock_irqres= tore(&zone->lock, flags); > =C2=A0 =C2=A0 =C2=A0 =C2=A0} > @@ -5695,6 +5700,56 @@ static int __alloc_contig_migrate_range(unsigned l= ong start, unsigned long end) > =C2=A0 =C2=A0 =C2=A0 =C2=A0return ret > 0 ? 0 : ret; > =C2=A0} > > +/* > + * Update zone's cma pages counter used for watermark level calculation. > + */ > +static inline void __update_cma_watermarks(struct zone *zone, int count) > +{ > + =C2=A0 =C2=A0 =C2=A0 unsigned long flags; > + =C2=A0 =C2=A0 =C2=A0 spin_lock_irqsave(&zone->lock, flags); > + =C2=A0 =C2=A0 =C2=A0 zone->min_cma_pages +=3D count; > + =C2=A0 =C2=A0 =C2=A0 spin_unlock_irqrestore(&zone->lock, flags); > + =C2=A0 =C2=A0 =C2=A0 setup_per_zone_wmarks(); > +} > + > +/* > + * Trigger memory pressure bump to reclaim some pages in order to be abl= e to > + * allocate 'count' pages in single page units. Does similar work as > + *__alloc_pages_slowpath() function. > + */ > +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count) > +{ > + =C2=A0 =C2=A0 =C2=A0 enum zone_type high_zoneidx =3D gfp_zone(gfp_mask)= ; > + =C2=A0 =C2=A0 =C2=A0 struct zonelist *zonelist =3D node_zonelist(0, gfp= _mask); > + =C2=A0 =C2=A0 =C2=A0 int did_some_progress =3D 0; > + =C2=A0 =C2=A0 =C2=A0 int order =3D 1; > + =C2=A0 =C2=A0 =C2=A0 unsigned long watermark; > + > + =C2=A0 =C2=A0 =C2=A0 /* > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* Increase level of watermarks to force kswa= pd do his job > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* to stabilise at new watermark level. > + =C2=A0 =C2=A0 =C2=A0 =C2=A0*/ > + =C2=A0 =C2=A0 =C2=A0 __update_cma_watermarks(zone, count); > + > + =C2=A0 =C2=A0 =C2=A0 /* Obey watermarks as if the page was being alloca= ted */ > + =C2=A0 =C2=A0 =C2=A0 watermark =3D low_wmark_pages(zone) + count; > + =C2=A0 =C2=A0 =C2=A0 while (!zone_watermark_ok(zone, 0, watermark, 0, 0= )) { > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 wake_all_kswapd(order,= zonelist, high_zoneidx, zone_idx(zone)); > + > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 did_some_progress =3D = __perform_reclaim(gfp_mask, order, zonelist, > + =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=A0 =C2=A0 =C2=A0 NULL); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 if (!did_some_progress= ) { > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 /* Exhausted what can be done so it's blamo time */ > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 out_of_memory(zonelist, gfp_mask, order, NULL); out_of_memory() has got another param in the newest next/master tree, out_of_memory(zonelist, gfp_mask, order, NULL, false) should be OK. -barry -- 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