From: Michal Hocko <mhocko@suse.cz>
To: Mel Gorman <mgorman@suse.de>
Cc: Linux-MM <linux-mm@kvack.org>, Jiri Slaby <jslaby@suse.cz>,
Valdis Kletnieks <Valdis.Kletnieks@vt.edu>,
Rik van Riel <riel@redhat.com>,
Zlatko Calusic <zcalusic@bitsync.net>,
Johannes Weiner <hannes@cmpxchg.org>,
dormando <dormando@rydia.net>,
Satoru Moriya <satoru.moriya@hds.com>,
LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 07/10 -v2r1] mm: vmscan: Block kswapd if it is encountering pages under writeback
Date: Thu, 21 Mar 2013 17:32:27 +0100 [thread overview]
Message-ID: <20130321163227.GT6094@dhcp22.suse.cz> (raw)
In-Reply-To: <1363525456-10448-8-git-send-email-mgorman@suse.de>
Here is what you have in your mm-vmscan-limit-reclaim-v2r1 branch:
> commit 0dae7d4be56e6a7fe3f128284679f5efc0cc2383
> Author: Mel Gorman <mgorman@suse.de>
> Date: Tue Mar 12 10:33:31 2013 +0000
>
> mm: vmscan: Block kswapd if it is encountering pages under writeback
>
> Historically, kswapd used to congestion_wait() at higher priorities if it
> was not making forward progress. This made no sense as the failure to make
> progress could be completely independent of IO. It was later replaced by
> wait_iff_congested() and removed entirely by commit 258401a6 (mm: don't
> wait on congested zones in balance_pgdat()) as it was duplicating logic
> in shrink_inactive_list().
>
> This is problematic. If kswapd encounters many pages under writeback and
> it continues to scan until it reaches the high watermark then it will
> quickly skip over the pages under writeback and reclaim clean young
> pages or push applications out to swap.
>
> The use of wait_iff_congested() is not suited to kswapd as it will only
> stall if the underlying BDI is really congested or a direct reclaimer was
> unable to write to the underlying BDI. kswapd bypasses the BDI congestion
> as it sets PF_SWAPWRITE but even if this was taken into account then it
> would cause direct reclaimers to stall on writeback which is not desirable.
>
> This patch sets a ZONE_WRITEBACK flag if direct reclaim or kswapd is
> encountering too many pages under writeback. If this flag is set and
> kswapd encounters a PageReclaim page under writeback then it'll assume
> that the LRU lists are being recycled too quickly before IO can complete
> and block waiting for some IO to complete.
>
> Signed-off-by: Mel Gorman <mgorman@suse.de>
Looks reasonable to me.
Reviewed-by: Michal Hocko <mhocko@suse.cz>
>
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index afedd1d..dd0d266 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -499,6 +499,9 @@ typedef enum {
> * many dirty file pages at the tail
> * of the LRU.
> */
> + ZONE_WRITEBACK, /* reclaim scanning has recently found
> + * many pages under writeback
> + */
> } zone_flags_t;
>
> static inline void zone_set_flag(struct zone *zone, zone_flags_t flag)
> @@ -526,6 +529,11 @@ static inline int zone_is_reclaim_dirty(const struct zone *zone)
> return test_bit(ZONE_TAIL_LRU_DIRTY, &zone->flags);
> }
>
> +static inline int zone_is_reclaim_writeback(const struct zone *zone)
> +{
> + return test_bit(ZONE_WRITEBACK, &zone->flags);
> +}
> +
> static inline int zone_is_reclaim_locked(const struct zone *zone)
> {
> return test_bit(ZONE_RECLAIM_LOCKED, &zone->flags);
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index a8b94fa..e87de90 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -723,25 +723,51 @@ static unsigned long shrink_page_list(struct list_head *page_list,
> may_enter_fs = (sc->gfp_mask & __GFP_FS) ||
> (PageSwapCache(page) && (sc->gfp_mask & __GFP_IO));
>
> + /*
> + * If a page at the tail of the LRU is under writeback, there
> + * are three cases to consider.
> + *
> + * 1) If reclaim is encountering an excessive number of pages
> + * under writeback and this page is both under writeback and
> + * PageReclaim then it indicates that pages are being queued
> + * for IO but are being recycled through the LRU before the
> + * IO can complete. In this case, wait on the IO to complete
> + * and then clear the ZONE_WRITEBACK flag to recheck if the
> + * condition exists.
> + *
> + * 2) Global reclaim encounters a page, memcg encounters a
> + * page that is not marked for immediate reclaim or
> + * the caller does not have __GFP_IO. In this case mark
> + * the page for immediate reclaim and continue scanning.
> + *
> + * __GFP_IO is checked because a loop driver thread might
> + * enter reclaim, and deadlock if it waits on a page for
> + * which it is needed to do the write (loop masks off
> + * __GFP_IO|__GFP_FS for this reason); but more thought
> + * would probably show more reasons.
> + *
> + * Don't require __GFP_FS, since we're not going into the
> + * FS, just waiting on its writeback completion. Worryingly,
> + * ext4 gfs2 and xfs allocate pages with
> + * grab_cache_page_write_begin(,,AOP_FLAG_NOFS), so testing
> + * may_enter_fs here is liable to OOM on them.
> + *
> + * 3) memcg encounters a page that is not already marked
> + * PageReclaim. memcg does not have any dirty pages
> + * throttling so we could easily OOM just because too many
> + * pages are in writeback and there is nothing else to
> + * reclaim. Wait for the writeback to complete.
> + */
> if (PageWriteback(page)) {
> - /*
> - * memcg doesn't have any dirty pages throttling so we
> - * could easily OOM just because too many pages are in
> - * writeback and there is nothing else to reclaim.
> - *
> - * Check __GFP_IO, certainly because a loop driver
> - * thread might enter reclaim, and deadlock if it waits
> - * on a page for which it is needed to do the write
> - * (loop masks off __GFP_IO|__GFP_FS for this reason);
> - * but more thought would probably show more reasons.
> - *
> - * Don't require __GFP_FS, since we're not going into
> - * the FS, just waiting on its writeback completion.
> - * Worryingly, ext4 gfs2 and xfs allocate pages with
> - * grab_cache_page_write_begin(,,AOP_FLAG_NOFS), so
> - * testing may_enter_fs here is liable to OOM on them.
> - */
> - if (global_reclaim(sc) ||
> + /* Case 1 above */
> + if (current_is_kswapd() &&
> + PageReclaim(page) &&
> + zone_is_reclaim_writeback(zone)) {
> + wait_on_page_writeback(page);
> + zone_clear_flag(zone, ZONE_WRITEBACK);
> +
> + /* Case 2 above */
> + } else if (global_reclaim(sc) ||
> !PageReclaim(page) || !(sc->gfp_mask & __GFP_IO)) {
> /*
> * This is slightly racy - end_page_writeback()
> @@ -756,9 +782,13 @@ static unsigned long shrink_page_list(struct list_head *page_list,
> */
> SetPageReclaim(page);
> nr_writeback++;
> +
> goto keep_locked;
> +
> + /* Case 3 above */
> + } else {
> + wait_on_page_writeback(page);
> }
> - wait_on_page_writeback(page);
> }
>
> if (!force_reclaim)
> @@ -1373,8 +1403,10 @@ shrink_inactive_list(unsigned long nr_to_scan, struct lruvec *lruvec,
> * isolated page is PageWriteback
> */
> if (nr_writeback && nr_writeback >=
> - (nr_taken >> (DEF_PRIORITY - sc->priority)))
> + (nr_taken >> (DEF_PRIORITY - sc->priority))) {
> wait_iff_congested(zone, BLK_RW_ASYNC, HZ/10);
> + zone_set_flag(zone, ZONE_WRITEBACK);
> + }
>
> /*
> * Similarly, if many dirty pages are encountered that are not
> @@ -2639,8 +2671,8 @@ static bool prepare_kswapd_sleep(pg_data_t *pgdat, int order, long remaining,
> * kswapd shrinks the zone by the number of pages required to reach
> * the high watermark.
> *
> - * Returns true if kswapd scanned at least the requested number of
> - * pages to reclaim.
> + * Returns true if kswapd scanned at least the requested number of pages to
> + * reclaim or if the lack of process was due to pages under writeback.
> */
> static bool kswapd_shrink_zone(struct zone *zone,
> struct scan_control *sc,
> @@ -2663,6 +2695,8 @@ static bool kswapd_shrink_zone(struct zone *zone,
> if (nr_slab == 0 && !zone_reclaimable(zone))
> zone->all_unreclaimable = 1;
>
> + zone_clear_flag(zone, ZONE_WRITEBACK);
> +
> return sc->nr_scanned >= sc->nr_to_reclaim;
> }
--
Michal Hocko
SUSE Labs
next prev parent reply other threads:[~2013-03-21 16:32 UTC|newest]
Thread overview: 120+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-03-17 13:04 [RFC PATCH 0/8] Reduce system disruption due to kswapd Mel Gorman
2013-03-17 13:04 ` [PATCH 01/10] mm: vmscan: Limit the number of pages kswapd reclaims at each priority Mel Gorman
2013-03-18 23:53 ` Simon Jeons
2013-03-19 9:55 ` Mel Gorman
2013-03-19 10:16 ` Simon Jeons
2013-03-19 10:59 ` Mel Gorman
2013-03-20 16:18 ` Michal Hocko
2013-03-21 0:52 ` Rik van Riel
2013-03-22 0:08 ` Will Huck
2013-03-21 9:47 ` Mel Gorman
2013-03-21 12:59 ` Michal Hocko
2013-03-21 0:51 ` Rik van Riel
2013-03-21 15:57 ` Johannes Weiner
2013-03-21 16:47 ` Mel Gorman
2013-03-22 0:05 ` Will Huck
2013-03-22 3:52 ` Rik van Riel
2013-03-22 3:56 ` Will Huck
2013-03-22 4:59 ` Will Huck
2013-03-22 13:01 ` Rik van Riel
2013-04-05 0:05 ` Will Huck
2013-04-07 7:32 ` Will Huck
2013-04-07 7:35 ` Will Huck
2013-04-11 5:54 ` Will Huck
2013-04-11 5:58 ` Will Huck
2013-04-12 5:46 ` Ric Mason
2013-04-12 9:34 ` Mel Gorman
2013-04-12 13:40 ` Rik van Riel
2013-03-25 9:07 ` Michal Hocko
2013-03-25 9:13 ` Jiri Slaby
2013-03-28 22:31 ` Jiri Slaby
2013-03-29 8:22 ` Michal Hocko
2013-03-30 22:07 ` Jiri Slaby
2013-04-02 11:15 ` Mel Gorman
2013-03-17 13:04 ` [PATCH 02/10] mm: vmscan: Obey proportional scanning requirements for kswapd Mel Gorman
2013-03-17 14:39 ` Andi Kleen
2013-03-17 15:08 ` Mel Gorman
2013-03-21 1:10 ` Rik van Riel
2013-03-21 9:54 ` Mel Gorman
2013-03-21 14:01 ` Michal Hocko
2013-03-21 14:31 ` Mel Gorman
2013-03-21 15:07 ` Michal Hocko
2013-03-21 15:34 ` Mel Gorman
2013-03-22 7:54 ` Michal Hocko
2013-03-22 8:37 ` Mel Gorman
2013-03-22 10:04 ` Michal Hocko
2013-03-22 10:47 ` Michal Hocko
2013-03-21 16:25 ` Johannes Weiner
2013-03-21 18:02 ` Mel Gorman
2013-03-22 16:53 ` Johannes Weiner
2013-03-22 18:25 ` Mel Gorman
2013-03-22 19:09 ` Johannes Weiner
2013-03-22 19:46 ` Mel Gorman
2013-03-17 13:04 ` [PATCH 03/10] mm: vmscan: Flatten kswapd priority loop Mel Gorman
2013-03-17 14:36 ` Andi Kleen
2013-03-17 15:09 ` Mel Gorman
2013-03-18 23:58 ` Simon Jeons
2013-03-19 10:12 ` Mel Gorman
2013-03-19 3:08 ` Simon Jeons
2013-03-19 8:23 ` Michal Hocko
2013-03-19 10:14 ` Mel Gorman
2013-03-19 10:26 ` Simon Jeons
2013-03-19 11:01 ` Mel Gorman
2013-03-21 14:54 ` Michal Hocko
2013-03-21 15:26 ` Mel Gorman
2013-03-21 15:38 ` Michal Hocko
2013-03-17 13:04 ` [PATCH 04/10] mm: vmscan: Decide whether to compact the pgdat based on reclaim progress Mel Gorman
2013-03-18 11:35 ` Hillf Danton
2013-03-19 10:27 ` Mel Gorman
[not found] ` <20130318111130.GA7245@hacker.(null)>
2013-03-19 10:19 ` Mel Gorman
2013-03-21 15:32 ` Michal Hocko
2013-03-21 15:47 ` Mel Gorman
2013-03-21 15:50 ` Michal Hocko
2013-03-17 13:04 ` [PATCH 05/10] mm: vmscan: Do not allow kswapd to scan at maximum priority Mel Gorman
2013-03-21 1:20 ` Rik van Riel
2013-03-21 10:12 ` Mel Gorman
2013-03-21 12:30 ` Rik van Riel
2013-03-21 15:48 ` Michal Hocko
2013-03-17 13:04 ` [PATCH 06/10] mm: vmscan: Have kswapd writeback pages based on dirty pages encountered, not priority Mel Gorman
2013-03-17 14:42 ` Andi Kleen
2013-03-17 15:11 ` Mel Gorman
2013-03-21 17:53 ` Rik van Riel
2013-03-21 18:15 ` Mel Gorman
2013-03-21 18:21 ` Rik van Riel
[not found] ` <20130318110850.GA7144@hacker.(null)>
2013-03-19 10:35 ` Mel Gorman
2013-03-17 13:04 ` [PATCH 07/10] mm: vmscan: Block kswapd if it is encountering pages under writeback Mel Gorman
2013-03-17 14:49 ` Andi Kleen
2013-03-17 15:19 ` Mel Gorman
2013-03-17 15:40 ` Andi Kleen
2013-03-19 11:06 ` Mel Gorman
2013-03-18 11:37 ` Simon Jeons
2013-03-19 10:57 ` Mel Gorman
[not found] ` <20130318115827.GB7245@hacker.(null)>
2013-03-19 10:58 ` Mel Gorman
2013-03-21 16:32 ` Michal Hocko [this message]
2013-03-21 18:42 ` Rik van Riel
2013-03-22 8:27 ` Mel Gorman
2013-03-17 13:04 ` [PATCH 08/10] mm: vmscan: Have kswapd shrink slab only once per priority Mel Gorman
2013-03-17 14:53 ` Andi Kleen
2013-03-21 16:47 ` Michal Hocko
2013-03-21 19:47 ` Rik van Riel
2013-04-09 6:53 ` Joonsoo Kim
2013-04-09 8:41 ` Simon Jeons
2013-04-09 11:13 ` Mel Gorman
2013-04-10 1:07 ` Dave Chinner
2013-04-10 5:23 ` Joonsoo Kim
2013-04-11 9:53 ` Mel Gorman
2013-04-10 5:21 ` Joonsoo Kim
2013-04-11 10:01 ` Mel Gorman
2013-04-11 10:29 ` Ric Mason
2013-03-17 13:04 ` [PATCH 09/10] mm: vmscan: Check if kswapd should writepage " Mel Gorman
2013-03-21 16:58 ` Michal Hocko
2013-03-21 18:07 ` Mel Gorman
2013-03-21 19:52 ` Rik van Riel
2013-03-17 13:04 ` [PATCH 10/10] mm: vmscan: Move logic from balance_pgdat() to kswapd_shrink_zone() Mel Gorman
2013-03-17 14:55 ` Andi Kleen
2013-03-17 15:25 ` Mel Gorman
2013-03-21 17:18 ` Michal Hocko
2013-03-21 18:13 ` Mel Gorman
2013-03-22 14:37 ` [RFC PATCH 0/8] Reduce system disruption due to kswapd Mel Gorman
2013-03-24 19:00 ` Jiri Slaby
2013-03-25 8:17 ` Michal Hocko
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=20130321163227.GT6094@dhcp22.suse.cz \
--to=mhocko@suse.cz \
--cc=Valdis.Kletnieks@vt.edu \
--cc=dormando@rydia.net \
--cc=hannes@cmpxchg.org \
--cc=jslaby@suse.cz \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mgorman@suse.de \
--cc=riel@redhat.com \
--cc=satoru.moriya@hds.com \
--cc=zcalusic@bitsync.net \
/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).