From: Mel Gorman <mel@csn.ul.ie>
To: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: linux-mm@kvack.org, linux-fsdevel@vger.kernel.org,
linux-kernel@vger.kernel.org,
Chris Mason <chris.mason@oracle.com>,
Dave Chinner <david@fromorbit.com>,
Andi Kleen <andi@firstfloor.org>,
Johannes Weiner <hannes@cmpxchg.org>
Subject: Re: [PATCH 10/10] vmscan: Update isolated page counters outside of main path in shrink_inactive_list()
Date: Fri, 16 Apr 2010 15:35:21 +0100 [thread overview]
Message-ID: <20100416143521.GH19264@csn.ul.ie> (raw)
In-Reply-To: <20100416115315.27AA.A69D9226@jp.fujitsu.com>
On Fri, Apr 16, 2010 at 08:19:00PM +0900, KOSAKI Motohiro wrote:
> > When shrink_inactive_list() isolates pages, it updates a number of
> > counters using temporary variables to gather them. These consume stack
> > and it's in the main path that calls ->writepage(). This patch moves the
> > accounting updates outside of the main path to reduce stack usage.
> >
> > Signed-off-by: Mel Gorman <mel@csn.ul.ie>
> > ---
> > mm/vmscan.c | 63 +++++++++++++++++++++++++++++++++++-----------------------
> > 1 files changed, 38 insertions(+), 25 deletions(-)
> >
> > diff --git a/mm/vmscan.c b/mm/vmscan.c
> > index 2c22c83..4225319 100644
> > --- a/mm/vmscan.c
> > +++ b/mm/vmscan.c
> > @@ -1061,7 +1061,8 @@ static unsigned long clear_active_flags(struct list_head *page_list,
> > ClearPageActive(page);
> > nr_active++;
> > }
> > - count[lru]++;
> > + if (count)
> > + count[lru]++;
> > }
> >
> > return nr_active;
> > @@ -1141,12 +1142,13 @@ static int too_many_isolated(struct zone *zone, int file,
> > * TODO: Try merging with migrations version of putback_lru_pages
> > */
> > static noinline void putback_lru_pages(struct zone *zone,
> > - struct zone_reclaim_stat *reclaim_stat,
> > + struct scan_control *sc,
> > unsigned long nr_anon, unsigned long nr_file,
> > struct list_head *page_list)
> > {
> > struct page *page;
> > struct pagevec pvec;
> > + struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
>
> Seems unrelated change here.
> Otherwise looks good.
>
It was needed somewhat otherwise the split in accounting looked odd. It
could be done as two patches but it felt trickier to review.
> - kosaki
> >
> > pagevec_init(&pvec, 1);
> >
> > @@ -1185,6 +1187,37 @@ static noinline void putback_lru_pages(struct zone *zone,
> > pagevec_release(&pvec);
> > }
> >
> > +static noinline void update_isolated_counts(struct zone *zone,
> > + struct scan_control *sc,
> > + unsigned long *nr_anon,
> > + unsigned long *nr_file,
> > + struct list_head *isolated_list)
> > +{
> > + unsigned long nr_active;
> > + unsigned int count[NR_LRU_LISTS] = { 0, };
> > + struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
> > +
> > + nr_active = clear_active_flags(isolated_list, count);
> > + __count_vm_events(PGDEACTIVATE, nr_active);
> > +
> > + __mod_zone_page_state(zone, NR_ACTIVE_FILE,
> > + -count[LRU_ACTIVE_FILE]);
> > + __mod_zone_page_state(zone, NR_INACTIVE_FILE,
> > + -count[LRU_INACTIVE_FILE]);
> > + __mod_zone_page_state(zone, NR_ACTIVE_ANON,
> > + -count[LRU_ACTIVE_ANON]);
> > + __mod_zone_page_state(zone, NR_INACTIVE_ANON,
> > + -count[LRU_INACTIVE_ANON]);
> > +
> > + *nr_anon = count[LRU_ACTIVE_ANON] + count[LRU_INACTIVE_ANON];
> > + *nr_file = count[LRU_ACTIVE_FILE] + count[LRU_INACTIVE_FILE];
> > + __mod_zone_page_state(zone, NR_ISOLATED_ANON, *nr_anon);
> > + __mod_zone_page_state(zone, NR_ISOLATED_FILE, *nr_file);
> > +
> > + reclaim_stat->recent_scanned[0] += *nr_anon;
> > + reclaim_stat->recent_scanned[1] += *nr_file;
> > +}
> > +
> > /*
> > * shrink_inactive_list() is a helper for shrink_zone(). It returns the number
> > * of reclaimed pages
> > @@ -1196,11 +1229,9 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
> > LIST_HEAD(page_list);
> > unsigned long nr_scanned;
> > unsigned long nr_reclaimed = 0;
> > - struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
> > int lumpy_reclaim = 0;
> > unsigned long nr_taken;
> > unsigned long nr_active;
> > - unsigned int count[NR_LRU_LISTS] = { 0, };
> > unsigned long nr_anon;
> > unsigned long nr_file;
> >
> > @@ -1244,25 +1275,7 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
> > return 0;
> > }
> >
> > - nr_active = clear_active_flags(&page_list, count);
> > - __count_vm_events(PGDEACTIVATE, nr_active);
> > -
> > - __mod_zone_page_state(zone, NR_ACTIVE_FILE,
> > - -count[LRU_ACTIVE_FILE]);
> > - __mod_zone_page_state(zone, NR_INACTIVE_FILE,
> > - -count[LRU_INACTIVE_FILE]);
> > - __mod_zone_page_state(zone, NR_ACTIVE_ANON,
> > - -count[LRU_ACTIVE_ANON]);
> > - __mod_zone_page_state(zone, NR_INACTIVE_ANON,
> > - -count[LRU_INACTIVE_ANON]);
> > -
> > - nr_anon = count[LRU_ACTIVE_ANON] + count[LRU_INACTIVE_ANON];
> > - nr_file = count[LRU_ACTIVE_FILE] + count[LRU_INACTIVE_FILE];
> > - __mod_zone_page_state(zone, NR_ISOLATED_ANON, nr_anon);
> > - __mod_zone_page_state(zone, NR_ISOLATED_FILE, nr_file);
> > -
> > - reclaim_stat->recent_scanned[0] += nr_anon;
> > - reclaim_stat->recent_scanned[1] += nr_file;
> > + update_isolated_counts(zone, sc, &nr_anon, &nr_file, &page_list);
> >
> > spin_unlock_irq(&zone->lru_lock);
> >
> > @@ -1281,7 +1294,7 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
> > * The attempt at page out may have made some
> > * of the pages active, mark them inactive again.
> > */
> > - nr_active = clear_active_flags(&page_list, count);
> > + nr_active = clear_active_flags(&page_list, NULL);
> > count_vm_events(PGDEACTIVATE, nr_active);
> >
> > nr_reclaimed += shrink_page_list(&page_list, sc,
> > @@ -1293,7 +1306,7 @@ static unsigned long shrink_inactive_list(unsigned long nr_to_scan,
> > __count_vm_events(KSWAPD_STEAL, nr_reclaimed);
> > __count_zone_vm_events(PGSTEAL, zone, nr_reclaimed);
> >
> > - putback_lru_pages(zone, reclaim_stat, nr_anon, nr_file, &page_list);
> > + putback_lru_pages(zone, sc, nr_anon, nr_file, &page_list);
> > return nr_reclaimed;
> > }
> >
> > --
> > 1.6.5
> >
>
>
>
--
Mel Gorman
Part-time Phd Student Linux Technology Center
University of Limerick IBM Dublin Software Lab
next prev parent reply other threads:[~2010-04-16 14:35 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-04-15 17:21 [RFC PATCH 00/10] Reduce stack usage used by page reclaim V1 Mel Gorman
2010-04-15 17:21 ` [PATCH 01/10] vmscan: kill prev_priority completely Mel Gorman
2010-04-16 22:37 ` Johannes Weiner
2010-04-15 17:21 ` [PATCH 02/10] vmscan: move priority variable into scan_control Mel Gorman
2010-04-16 22:48 ` Johannes Weiner
2010-05-26 10:23 ` Mel Gorman
2010-05-28 2:39 ` KOSAKI Motohiro
2010-04-15 17:21 ` [PATCH 03/10] vmscan: simplify shrink_inactive_list() Mel Gorman
2010-04-16 22:54 ` Johannes Weiner
2010-04-15 17:21 ` [PATCH 04/10] vmscan: Remove useless loop at end of do_try_to_free_pages Mel Gorman
2010-04-16 2:48 ` KOSAKI Motohiro
2010-04-16 22:56 ` Johannes Weiner
2010-04-15 17:21 ` [PATCH 05/10] vmscan: Remove unnecessary temporary vars in do_try_to_free_pages Mel Gorman
2010-04-16 2:47 ` KOSAKI Motohiro
2010-04-15 17:21 ` [PATCH 06/10] vmscan: Split shrink_zone to reduce stack usage Mel Gorman
2010-04-16 4:23 ` Dave Chinner
2010-04-16 14:27 ` Mel Gorman
2010-04-16 6:26 ` KOSAKI Motohiro
2010-04-16 23:14 ` Johannes Weiner
2010-04-15 17:21 ` [PATCH 07/10] vmscan: Remove unnecessary temporary variables in shrink_zone() Mel Gorman
2010-04-16 2:51 ` KOSAKI Motohiro
2010-04-16 23:03 ` Johannes Weiner
2010-05-26 11:21 ` Mel Gorman
2010-05-28 2:51 ` KOSAKI Motohiro
2010-04-15 17:21 ` [PATCH 08/10] vmscan: Setup pagevec as late as possible in shrink_inactive_list() Mel Gorman
2010-04-16 4:27 ` Dave Chinner
2010-04-16 6:30 ` KOSAKI Motohiro
2010-04-16 14:31 ` Mel Gorman
2010-04-16 23:28 ` Johannes Weiner
2010-04-15 17:21 ` [PATCH 09/10] vmscan: Setup pagevec as late as possible in shrink_page_list() Mel Gorman
2010-04-16 7:54 ` KOSAKI Motohiro
2010-04-16 14:34 ` Mel Gorman
2010-04-15 17:21 ` [PATCH 10/10] vmscan: Update isolated page counters outside of main path in shrink_inactive_list() Mel Gorman
2010-04-16 11:19 ` KOSAKI Motohiro
2010-04-16 14:35 ` Mel Gorman [this message]
2010-04-16 23:34 ` Johannes Weiner
2010-04-16 14:50 ` [RFC PATCH 00/10] Reduce stack usage used by page reclaim V1 Mel Gorman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100416143521.GH19264@csn.ul.ie \
--to=mel@csn.ul.ie \
--cc=andi@firstfloor.org \
--cc=chris.mason@oracle.com \
--cc=david@fromorbit.com \
--cc=hannes@cmpxchg.org \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).