public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Ben Hutchings <ben@decadent.org.uk>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: torvalds@linux-foundation.org, akpm@linux-foundation.org,
	alan@lxorguk.ukuu.org.uk, Mel Gorman <mgorman@suse.de>,
	Rik van Riel <riel@redhat.com>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Minchan Kim <minchan.kim@gmail.com>,
	Dave Jones <davej@redhat.com>, Jan Kara <jack@suse.cz>,
	Andy Isaacson <adi@hexapodia.org>, Nai Xia <nai.xia@gmail.com>,
	Johannes Weiner <jweiner@redhat.com>
Subject: [ 09/73] mm: vmscan: check if reclaim should really abort even if compaction_ready() is true for one zone
Date: Tue, 31 Jul 2012 05:43:19 +0100	[thread overview]
Message-ID: <20120731044312.684965795@decadent.org.uk> (raw)
In-Reply-To: <20120731044310.013763753@decadent.org.uk>

3.2-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Mel Gorman <mgorman@suse.de>

commit 0cee34fd72c582b4f8ad8ce00645b75fb4168199 upstream.

Stable note: Not tracked on Bugzilla. THP and compaction was found to
	aggressively reclaim pages and stall systems under different
	situations that was addressed piecemeal over time.

If compaction can proceed for a given zone, shrink_zones() does not
reclaim any more pages from it.  After commit [e0c2327: vmscan: abort
reclaim/compaction if compaction can proceed], do_try_to_free_pages()
tries to finish as soon as possible once one zone can compact.

This was intended to prevent slabs being shrunk unnecessarily but there
are side-effects.  One is that a small zone that is ready for compaction
will abort reclaim even if the chances of successfully allocating a THP
from that zone is small.  It also means that reclaim can return too early
even though sc->nr_to_reclaim pages were not reclaimed.

This partially reverts the commit until it is proven that slabs are really
being shrunk unnecessarily but preserves the check to return 1 to avoid
OOM if reclaim was aborted prematurely.

[aarcange@redhat.com: This patch replaces a revert from Andrea]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
---
 mm/vmscan.c |   19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2173,7 +2173,8 @@ static inline bool compaction_ready(stru
  *
  * This function returns true if a zone is being reclaimed for a costly
  * high-order allocation and compaction is ready to begin. This indicates to
- * the caller that it should retry the allocation or fail.
+ * the caller that it should consider retrying the allocation instead of
+ * further reclaim.
  */
 static bool shrink_zones(int priority, struct zonelist *zonelist,
 					struct scan_control *sc)
@@ -2182,7 +2183,7 @@ static bool shrink_zones(int priority, s
 	struct zone *zone;
 	unsigned long nr_soft_reclaimed;
 	unsigned long nr_soft_scanned;
-	bool should_abort_reclaim = false;
+	bool aborted_reclaim = false;
 
 	for_each_zone_zonelist_nodemask(zone, z, zonelist,
 					gfp_zone(sc->gfp_mask), sc->nodemask) {
@@ -2208,7 +2209,7 @@ static bool shrink_zones(int priority, s
 				 * allocations.
 				 */
 				if (compaction_ready(zone, sc)) {
-					should_abort_reclaim = true;
+					aborted_reclaim = true;
 					continue;
 				}
 			}
@@ -2230,7 +2231,7 @@ static bool shrink_zones(int priority, s
 		shrink_zone(priority, zone, sc);
 	}
 
-	return should_abort_reclaim;
+	return aborted_reclaim;
 }
 
 static bool zone_reclaimable(struct zone *zone)
@@ -2284,7 +2285,7 @@ static unsigned long do_try_to_free_page
 	struct zoneref *z;
 	struct zone *zone;
 	unsigned long writeback_threshold;
-	bool should_abort_reclaim;
+	bool aborted_reclaim;
 
 	get_mems_allowed();
 	delayacct_freepages_start();
@@ -2296,9 +2297,7 @@ static unsigned long do_try_to_free_page
 		sc->nr_scanned = 0;
 		if (!priority)
 			disable_swap_token(sc->mem_cgroup);
-		should_abort_reclaim = shrink_zones(priority, zonelist, sc);
-		if (should_abort_reclaim)
-			break;
+		aborted_reclaim = shrink_zones(priority, zonelist, sc);
 
 		/*
 		 * Don't shrink slabs when reclaiming memory from
@@ -2365,8 +2364,8 @@ out:
 	if (oom_killer_disabled)
 		return 0;
 
-	/* Aborting reclaim to try compaction? don't OOM, then */
-	if (should_abort_reclaim)
+	/* Aborted reclaim to try compaction? don't OOM, then */
+	if (aborted_reclaim)
 		return 1;
 
 	/* top priority shrink_zones still had more to do? don't OOM, then */



  parent reply	other threads:[~2012-07-31  5:06 UTC|newest]

Thread overview: 94+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-07-31  4:43 [ 00/73] 3.2.25-stable review Ben Hutchings
2012-07-31  4:43 ` [ 01/73] mm: reduce the amount of work done when updating min_free_kbytes Ben Hutchings
2012-07-31  4:43 ` [ 02/73] mm: compaction: allow compaction to isolate dirty pages Ben Hutchings
2012-07-31  4:43 ` [ 03/73] mm: compaction: determine if dirty pages can be migrated without blocking within ->migratepage Ben Hutchings
2012-07-31  4:43 ` [ 04/73] mm: page allocator: do not call direct reclaim for THP allocations while compaction is deferred Ben Hutchings
2012-07-31  4:43 ` [ 05/73] mm: compaction: make isolate_lru_page() filter-aware again Ben Hutchings
2012-07-31  4:43 ` [ 06/73] mm: compaction: introduce sync-light migration for use by compaction Ben Hutchings
2012-07-31 16:42   ` Herton Ronaldo Krzesinski
2012-07-31 17:00     ` Mel Gorman
2012-07-31 17:03       ` Mel Gorman
2012-07-31 23:12         ` Ben Hutchings
2012-07-31  4:43 ` [ 07/73] mm: vmscan: when reclaiming for compaction, ensure there are sufficient free pages available Ben Hutchings
2012-07-31  4:43 ` [ 08/73] mm: vmscan: do not OOM if aborting reclaim to start compaction Ben Hutchings
2012-07-31  4:43 ` Ben Hutchings [this message]
2012-07-31  4:43 ` [ 10/73] vmscan: promote shared file mapped pages Ben Hutchings
2012-07-31  4:43 ` [ 11/73] vmscan: activate executable pages after first usage Ben Hutchings
2012-07-31  4:43 ` [ 12/73] mm/vmscan.c: consider swap space when deciding whether to continue reclaim Ben Hutchings
2012-07-31  4:43 ` [ 13/73] mm: test PageSwapBacked in lumpy reclaim Ben Hutchings
2012-07-31  4:43 ` [ 14/73] mm: vmscan: convert global reclaim to per-memcg LRU lists Ben Hutchings
2012-07-31  4:43 ` [ 15/73] cpuset: mm: reduce large amounts of memory barrier related damage v3 Ben Hutchings
2012-07-31  4:43 ` [ 16/73] mm/hugetlb: fix warning in alloc_huge_page/dequeue_huge_page_vma Ben Hutchings
2012-07-31  4:43 ` [ 17/73] [SCSI] Fix NULL dereferences in scsi_cmd_to_driver Ben Hutchings
2012-07-31  4:43 ` [ 18/73] sched/nohz: Fix rq->cpu_load[] calculations Ben Hutchings
2012-07-31  4:43 ` [ 19/73] sched/nohz: Fix rq->cpu_load calculations some more Ben Hutchings
2012-07-31  4:43 ` [ 20/73] powerpc/ftrace: Fix assembly trampoline register usage Ben Hutchings
2012-07-31  4:43 ` [ 21/73] cx25821: Remove bad strcpy to read-only char* Ben Hutchings
2012-07-31  4:43 ` [ 22/73] x86: Fix boot on Twinhead H12Y Ben Hutchings
2012-07-31  4:43 ` [ 23/73] r8169: RxConfig hack for the 8168evl Ben Hutchings
2012-07-31  4:43 ` [ 24/73] cifs: when CONFIG_HIGHMEM is set, serialize the read/write kmaps Ben Hutchings
2012-07-31  4:43 ` [ 25/73] wireless: rt2x00: rt2800usb add more devices ids Ben Hutchings
2012-07-31  4:43 ` [ 26/73] wireless: rt2x00: rt2800usb more devices were identified Ben Hutchings
2012-07-31  4:43 ` [ 27/73] rt2800usb: 2001:3c17 is an RT3370 device Ben Hutchings
2012-07-31  4:43 ` [ 28/73] ARM: OMAP2+: OPP: Fix to ensure check of right oppdef after bad one Ben Hutchings
2012-08-01  1:56   ` Herton Ronaldo Krzesinski
2012-08-01  2:36     ` Ben Hutchings
2012-07-31  4:43 ` [ 29/73] usb: gadget: Fix g_ether interface link status Ben Hutchings
2012-07-31  4:43 ` [ 30/73] ext4: pass a char * to ext4_count_free() instead of a buffer_head ptr Ben Hutchings
2012-07-31  4:43 ` [ 31/73] ftrace: Disable function tracing during suspend/resume and hibernation, again Ben Hutchings
2012-07-31  4:43 ` [ 32/73] x86, microcode: microcode_core.c simple_strtoul cleanup Ben Hutchings
2012-07-31  4:43 ` [ 33/73] x86, microcode: Sanitize per-cpu microcode reloading interface Ben Hutchings
2012-08-03  9:04   ` Sven Joachim
2012-08-03  9:43     ` Borislav Petkov
2012-08-03 12:27       ` Borislav Petkov
2012-08-04 15:41         ` Ben Hutchings
2012-08-04 16:07           ` Henrique de Moraes Holschuh
2012-08-04 17:23             ` Ben Hutchings
2012-08-05  9:21               ` Borislav Petkov
2012-08-05 18:56                 ` Ben Hutchings
2012-07-31  4:43 ` [ 34/73] usbdevfs: Correct amount of data copied to user in processcompl_compat Ben Hutchings
2012-07-31  4:43 ` [ 35/73] ASoC: dapm: Fix locking during codec shutdown Ben Hutchings
2012-07-31 16:11   ` Herton Ronaldo Krzesinski
2012-07-31 16:13     ` Mark Brown
2012-07-31 23:20       ` Ben Hutchings
2012-07-31  4:43 ` [ 36/73] ext4: fix overhead calculation used by ext4_statfs() Ben Hutchings
2012-07-31  4:43 ` [ 37/73] udf: Improve table length check to avoid possible overflow Ben Hutchings
2012-07-31  4:43 ` [ 38/73] powerpc: Add "memory" attribute for mfmsr() Ben Hutchings
2012-07-31  4:43 ` [ 39/73] mwifiex: correction in mcs index check Ben Hutchings
2012-07-31  4:43 ` [ 40/73] USB: option: Ignore ZTE (Vodafone) K3570/71 net interfaces Ben Hutchings
2012-07-31  4:43 ` [ 41/73] USB: option: add ZTE MF821D Ben Hutchings
2012-07-31  4:43 ` [ 42/73] target: Add generation of LOGICAL BLOCK ADDRESS OUT OF RANGE Ben Hutchings
2012-07-31  4:43 ` [ 43/73] target: Add range checking to UNMAP emulation Ben Hutchings
2012-07-31  4:43 ` [ 44/73] target: Fix reading of data length fields for UNMAP commands Ben Hutchings
2012-07-31  4:43 ` [ 45/73] target: Fix possible integer underflow in UNMAP emulation Ben Hutchings
2012-07-31  4:43 ` [ 46/73] target: Check number of unmap descriptors against our limit Ben Hutchings
2012-07-31  4:43 ` [ 47/73] s390/idle: fix sequence handling vs cpu hotplug Ben Hutchings
2012-07-31  4:43 ` [ 48/73] rtlwifi: rtl8192de: Fix phy-based version calculation Ben Hutchings
2012-07-31  4:43 ` [ 49/73] workqueue: perform cpu down operations from low priority cpu_notifier() Ben Hutchings
2012-07-31  4:44 ` [ 50/73] ALSA: hda - Add support for Realtek ALC282 Ben Hutchings
2012-07-31  4:44 ` [ 51/73] iommu/amd: Fix hotplug with iommu=pt Ben Hutchings
2012-07-31  4:44 ` [ 52/73] drm/radeon: Try harder to avoid HW cursor ending on a multiple of 128 columns Ben Hutchings
2012-07-31  4:44 ` [ 53/73] ALSA: hda - Turn on PIN_OUT from hdmi playback prepare Ben Hutchings
2012-07-31  4:44 ` [ 54/73] block: add blk_queue_dead() Ben Hutchings
2012-07-31  4:44 ` [ 55/73] [SCSI] Fix device removal NULL pointer dereference Ben Hutchings
2012-07-31  4:44 ` [ 56/73] [SCSI] Avoid dangling pointer in scsi_requeue_command() Ben Hutchings
2012-07-31  4:44 ` [ 57/73] [SCSI] fix hot unplug vs async scan race Ben Hutchings
2012-07-31  4:44 ` [ 58/73] [SCSI] fix eh wakeup (scsi_schedule_eh vs scsi_restart_operations) Ben Hutchings
2012-07-31  4:44 ` [ 59/73] [SCSI] libsas: continue revalidation Ben Hutchings
2012-07-31  4:44 ` [ 60/73] [SCSI] libsas: fix sas_discover_devices return code handling Ben Hutchings
2012-07-31  4:44 ` [ 61/73] iscsi-target: Drop bogus struct file usage for iSCSI/SCTP Ben Hutchings
2012-07-31  4:44 ` [ 62/73] mmc: sdhci-pci: CaFe has broken card detection Ben Hutchings
2012-07-31  4:44 ` [ 63/73] ext4: dont let i_reserved_meta_blocks go negative Ben Hutchings
2012-07-31  4:44 ` [ 64/73] ext4: undo ext4_calc_metadata_amount if we fail to claim space Ben Hutchings
2012-07-31  4:44 ` [ 65/73] ASoC: dapm: Fix _PRE and _POST events for DAPM performance improvements Ben Hutchings
2012-07-31  4:44 ` [ 66/73] locks: fix checking of fcntl_setlease argument Ben Hutchings
2012-07-31  4:44 ` [ 67/73] ACPI/AC: prevent OOPS on some boxes due to missing check power_supply_register() return value check Ben Hutchings
2012-07-31  4:44 ` [ 68/73] drm/radeon: fix bo creation retry path Ben Hutchings
2012-07-31  4:44 ` [ 69/73] drm/radeon: fix non revealent error message Ben Hutchings
2012-07-31  4:44 ` [ 70/73] drm/radeon: fix hotplug of DP to DVI|HDMI passive adapters (v2) Ben Hutchings
2012-07-31  4:44 ` [ 71/73] drm/radeon: on hotplug force link training to happen (v2) Ben Hutchings
2012-07-31  4:44 ` [ 72/73] Btrfs: call the ordered free operation without any locks held Ben Hutchings
2012-07-31  4:44 ` [ 73/73] nouveau: Fix alignment requirements on src and dst addresses Ben Hutchings
2012-07-31  5:00 ` [ 00/73] 3.2.25-stable review Ben Hutchings
2012-08-01 12:55 ` Steven Rostedt
2012-08-05 22:26   ` Ben Hutchings

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=20120731044312.684965795@decadent.org.uk \
    --to=ben@decadent.org.uk \
    --cc=aarcange@redhat.com \
    --cc=adi@hexapodia.org \
    --cc=akpm@linux-foundation.org \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=davej@redhat.com \
    --cc=jack@suse.cz \
    --cc=jweiner@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mgorman@suse.de \
    --cc=minchan.kim@gmail.com \
    --cc=nai.xia@gmail.com \
    --cc=riel@redhat.com \
    --cc=stable@vger.kernel.org \
    --cc=torvalds@linux-foundation.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