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, Michal Hocko <mhocko@suse.cz>,
Johannes Weiner <hannes@cmpxchg.org>, Mel Gorman <mel@csn.ul.ie>,
Minchan Kim <minchan@kernel.org>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Rik van Riel <riel@redhat.com>
Subject: [ 29/82] mm: consider all swapped back pages in used-once logic
Date: Fri, 08 Jun 2012 05:19:09 +0100 [thread overview]
Message-ID: <20120608041844.750619357@decadent.org.uk> (raw)
In-Reply-To: <20120608041840.861504477@decadent.org.uk>
3.2-stable review patch. If anyone has any objections, please let me know.
------------------
From: Michal Hocko <mhocko@suse.cz>
commit e48982734ea0500d1eba4f9d96195acc5406cad6 upstream.
Commit 645747462435 ("vmscan: detect mapped file pages used only once")
made mapped pages have another round in inactive list because they might
be just short lived and so we could consider them again next time. This
heuristic helps to reduce pressure on the active list with a streaming
IO worklods.
This patch fixes a regression introduced by this commit for heavy shmem
based workloads because unlike Anon pages, which are excluded from this
heuristic because they are usually long lived, shmem pages are handled
as a regular page cache.
This doesn't work quite well, unfortunately, if the workload is mostly
backed by shmem (in memory database sitting on 80% of memory) with a
streaming IO in the background (backup - up to 20% of memory). Anon
inactive list is full of (dirty) shmem pages when watermarks are hit.
Shmem pages are kept in the inactive list (they are referenced) in the
first round and it is hard to reclaim anything else so we reach lower
scanning priorities very quickly which leads to an excessive swap out.
Let's fix this by excluding all swap backed pages (they tend to be long
lived wrt. the regular page cache anyway) from used-once heuristic and
rather activate them if they are referenced.
The customer's workload is shmem backed database (80% of RAM) and they
are measuring transactions/s with an IO in the background (20%).
Transactions touch more or less random rows in the table. The
transaction rate fell by a factor of 3 (in the worst case) because of
commit 64574746. This patch restores the previous numbers.
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Minchan Kim <minchan@kernel.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Rik van Riel <riel@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 | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 44f0436..67a4fd4 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -657,7 +657,7 @@ static enum page_references page_check_references(struct page *page,
return PAGEREF_RECLAIM;
if (referenced_ptes) {
- if (PageAnon(page))
+ if (PageSwapBacked(page))
return PAGEREF_ACTIVATE;
/*
* All mapped pages start out with page table
next prev parent reply other threads:[~2012-06-08 5:14 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-06-08 4:18 [ 00/82] 3.2.20-stable review Ben Hutchings
2012-06-08 4:18 ` [ 01/82] iommu/amd: Add workaround for event log erratum Ben Hutchings
2012-06-08 4:18 ` [ 02/82] MIPS: BCM63XX: Add missing include for bcm63xx_gpio.h Ben Hutchings
2012-06-08 4:18 ` [ 03/82] cifs: Include backup intent search flags during searches {try #2) Ben Hutchings
2012-06-08 4:18 ` [ 04/82] sunrpc: fix loss of task->tk_status after rpc_delay call in xprt_alloc_slot Ben Hutchings
2012-06-08 4:18 ` [ 05/82] exofs: Fix CRASH on very early IO errors Ben Hutchings
2012-06-08 4:18 ` [ 06/82] cifs: fix oops while traversing open file list (try #4) Ben Hutchings
2012-06-08 4:18 ` [ 07/82] [SCSI] Fix dm-multipath starvation when scsi host is busy Ben Hutchings
2012-06-08 4:18 ` [ 08/82] ixp4xx: fix compilation by adding gpiolib support Ben Hutchings
2012-06-08 4:18 ` [ 09/82] drm/i915: properly handle interlaced bit for sdvo dtd conversion Ben Hutchings
2012-06-08 4:18 ` [ 10/82] drm/i915: enable vdd when switching off the eDP panel Ben Hutchings
2012-06-08 4:18 ` [ 11/82] drm/i915: Add Clientron E830 to the ignore LVDS list Ben Hutchings
2012-06-08 4:18 ` [ 12/82] drm/i915: Ignore LVDS on hp t5745 and hp st5747 thin client Ben Hutchings
2012-06-08 4:18 ` [ 13/82] drm/i915: no lvds quirk for HP t5740e Thin Client Ben Hutchings
2012-06-08 4:18 ` [ 14/82] drm/i915: wait for a vblank to pass after tv detect Ben Hutchings
2012-06-08 4:18 ` [ 15/82] drm/i915: Update GEN6_RP_CONTROL definitions Ben Hutchings
2012-06-08 4:18 ` [ 16/82] drm/i915: always use RPNSWREQ for turbo change requests Ben Hutchings
2012-06-08 4:18 ` [ 17/82] solos-pci: Fix DMA support Ben Hutchings
2012-06-08 4:18 ` [ 18/82] microblaze: Do not select GENERIC_GPIO by default Ben Hutchings
2012-06-08 4:18 ` [ 19/82] [PARISC] fix boot failure on 32-bit systems caused by branch stubs placed before .text Ben Hutchings
2012-06-08 4:19 ` [ 20/82] [PARISC] fix TLB fault path on PA2.0 narrow systems Ben Hutchings
2012-06-08 4:19 ` [ 21/82] iwlwifi: update BT traffic load states correctly Ben Hutchings
2012-06-08 4:19 ` [ 22/82] iwlwifi: do not use shadow registers by default Ben Hutchings
2012-06-08 4:19 ` [ 23/82] wl1251: fix oops on early interrupt Ben Hutchings
2012-06-08 4:19 ` [ 24/82] NFSv4: Map NFS4ERR_SHARE_DENIED into an EACCES error instead of EIO Ben Hutchings
2012-06-08 4:19 ` [ 25/82] drm/radeon: fix XFX quirk Ben Hutchings
2012-06-08 4:19 ` [ 26/82] ath9k: fix a use-after-free-bug when ath_tx_setup_buffer() fails Ben Hutchings
2012-06-08 4:19 ` [ 27/82] mac80211: fix ADDBA declined after suspend with wowlan Ben Hutchings
2012-06-08 4:19 ` [ 28/82] mm/fork: fix overflow in vma length when copying mmap on clone Ben Hutchings
2012-06-08 4:19 ` Ben Hutchings [this message]
2012-06-08 4:19 ` [ 30/82] hugetlb: fix resv_map leak in error path Ben Hutchings
2012-06-08 4:19 ` [ 31/82] mm/vmalloc.c: change void* into explict vm_struct* Ben Hutchings
2012-06-08 10:41 ` David Rientjes
2012-06-08 10:55 ` Minchan Kim
2012-06-08 12:16 ` Ben Hutchings
2012-06-08 4:19 ` [ 32/82] mm: fix faulty initialization in vmalloc_init() Ben Hutchings
2012-06-08 4:19 ` [ 33/82] [SCSI] fix scsi_wait_scan Ben Hutchings
2012-06-08 4:19 ` [ 34/82] mm: fix vma_resv_map() NULL pointer Ben Hutchings
2012-06-08 4:19 ` [ 35/82] x86, amd, xen: Avoid NULL pointer paravirt references Ben Hutchings
2012-06-08 4:19 ` [ 36/82] slub: fix a memory leak in get_partial_node() Ben Hutchings
2012-06-08 4:19 ` [ 37/82] ext4: force ro mount if ext4_setup_super() fails Ben Hutchings
2012-06-08 4:19 ` [ 38/82] ext4: disallow hard-linked directory in ext4_lookup Ben Hutchings
2012-06-08 4:19 ` [ 39/82] mtd: nand: fix scan_read_raw_oob Ben Hutchings
2012-06-08 4:19 ` [ 40/82] vfs: increment iversion when a file is truncated Ben Hutchings
2012-06-08 4:19 ` [ 41/82] vfs: umount_tree() might be called on subtree that had never made it Ben Hutchings
2012-06-08 4:19 ` [ 42/82] ext4: add missing save_error_info() to ext4_error() Ben Hutchings
2012-06-08 4:19 ` [ 43/82] ALSA: usb-audio: fix rate_list memory leak Ben Hutchings
2012-06-08 4:19 ` [ 44/82] ext4: add ext4_mb_unload_buddy in the error path Ben Hutchings
2012-06-08 4:19 ` [ 45/82] ext4: remove mb_groups before tearing down the buddy_cache Ben Hutchings
2012-06-08 4:19 ` [ 46/82] drm/radeon: fix bank information in tiling config Ben Hutchings
2012-06-08 4:19 ` [ 47/82] drm/radeon: properly program gart on rv740, juniper, cypress, barts, hemlock Ben Hutchings
2012-06-08 4:19 ` [ 48/82] drm/radeon: fix HD6790, HD6570 backend programming Ben Hutchings
2012-06-08 4:19 ` [ 49/82] drm/ttm: Fix spinlock imbalance Ben Hutchings
2012-06-08 4:19 ` [ 50/82] drm/vmwgfx: Fix nasty write past alloced memory area Ben Hutchings
2012-06-08 4:19 ` [ 51/82] mtd: of_parts: fix breakage in Kconfig Ben Hutchings
2012-06-08 4:19 ` [ 52/82] fec_mpc52xx: fix timestamp filtering Ben Hutchings
2012-06-08 4:19 ` [ 53/82] Bluetooth: btusb: Add vendor specific ID (0a5c 21f3) for BCM20702A0 Ben Hutchings
2012-06-08 4:19 ` [ 54/82] Bluetooth: btusb: add support for BCM20702A0 [0a5c:21e6] Ben Hutchings
2012-06-08 4:19 ` [ 55/82] Bluetooth: btusb: Add USB device ID "0a5c 21e8" Ben Hutchings
2012-06-08 4:19 ` [ 56/82] Bluetooth: btusb: typo in Broadcom SoftSailing id Ben Hutchings
2012-06-08 4:19 ` [ 57/82] Bluetooth: btusb: Add vendor specific ID (0489 e042) for BCM20702A0 Ben Hutchings
2012-06-08 4:19 ` [ 58/82] Bluetooth: Add support for Atheros [13d3:3362] Ben Hutchings
2012-06-08 4:19 ` [ 59/82] Bluetooth: Add support for AR3012 [0cf3:e004] Ben Hutchings
2012-06-08 4:19 ` [ 60/82] Add Foxconn / Hon Hai IDs for btusb module Ben Hutchings
2012-06-08 4:19 ` [ 61/82] Bluetooth: Add support for Foxconn/Hon Hai AR5BBU22 0489:E03C Ben Hutchings
2012-06-08 4:19 ` [ 62/82] drm/i915:: Disable FBC on SandyBridge Ben Hutchings
2012-06-08 4:19 ` [ 63/82] ipv4: Do not use dead fib_info entries Ben Hutchings
2012-06-08 4:19 ` [ 64/82] ipv4: fix the rcu race between free_fib_info and ip_route_output_slow Ben Hutchings
2012-06-08 4:19 ` [ 65/82] set fake_rtables dst to NULL to avoid kernel Oops Ben Hutchings
2012-06-08 4:19 ` [ 66/82] ipv6: fix incorrect ipsec fragment Ben Hutchings
2012-06-08 4:19 ` [ 67/82] l2tp: fix oops in L2TP IP sockets for connect() AF_UNSPEC case Ben Hutchings
2012-06-08 4:19 ` [ 68/82] r8169: missing barriers Ben Hutchings
2012-06-08 4:19 ` [ 69/82] r8169: fix early queue wake-up Ben Hutchings
2012-06-08 4:19 ` [ 70/82] r8169: fix unsigned int wraparound with TSO Ben Hutchings
2012-06-08 4:19 ` [ 71/82] Revert "net: maintain namespace isolation between vlan and real device" Ben Hutchings
2012-06-08 4:19 ` [ 72/82] sctp: check cached dst before using it Ben Hutchings
2012-06-08 4:19 ` [ 73/82] skb: avoid unnecessary reallocations in __skb_cow Ben Hutchings
2012-06-08 4:19 ` [ 74/82] xfrm: take net hdr len into account for esp payload size calculation Ben Hutchings
2012-06-08 4:19 ` [ 75/82] ACPI battery: only refresh the sysfs files when pertinent information changes Ben Hutchings
2012-06-08 4:19 ` [ 76/82] target/file: Use O_DSYNC by default for FILEIO backends Ben Hutchings
2012-06-08 4:19 ` [ 77/82] iommu/amd: Cache pdev pointer to root-bridge Ben Hutchings
2012-06-08 4:19 ` [ 78/82] drm/radeon/kms: add new Palm, Sumo PCI ids Ben Hutchings
2012-06-08 4:19 ` [ 79/82] drm/radeon/kms: add new BTC " Ben Hutchings
2012-06-08 4:20 ` [ 80/82] btree: fix tree corruption in btree_get_prev() Ben Hutchings
2012-06-08 4:20 ` [ 81/82] kbuild: install kernel-page-flags.h Ben Hutchings
2012-06-08 4:20 ` [ 82/82] asix: allow full size 8021Q frames to be received Ben Hutchings
2012-06-08 5:09 ` [ 00/82] 3.2.20-stable review Ben Hutchings
2012-06-08 13:42 ` Maarten Lankhorst
2012-06-08 14:04 ` Ben Hutchings
2012-06-08 22:54 ` Ben Hutchings
2012-06-10 16:54 ` Maarten Lankhorst
2012-06-17 14:04 ` 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=20120608041844.750619357@decadent.org.uk \
--to=ben@decadent.org.uk \
--cc=akpm@linux-foundation.org \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=hannes@cmpxchg.org \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mel@csn.ul.ie \
--cc=mhocko@suse.cz \
--cc=minchan@kernel.org \
--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