From: Greg KH <gregkh@linuxfoundation.org>
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: [ 03/47] mm: consider all swapped back pages in used-once logic
Date: Thu, 07 Jun 2012 13:14:01 +0900 [thread overview]
Message-ID: <20120607041358.619939818@linuxfoundation.org> (raw)
In-Reply-To: <20120607041504.GA13819@kroah.com>
3.0-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: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
mm/vmscan.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -665,7 +665,7 @@ static enum page_references page_check_r
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:[~2012-06-07 4:14 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-06-07 4:15 [ 00/47] 3.0.34-stable review Greg KH
2012-06-07 4:13 ` [ 01/47] SCSI: fix scsi_wait_scan Greg KH
2012-06-07 4:14 ` [ 02/47] SCSI: Fix dm-multipath starvation when scsi host is busy Greg KH
2012-06-07 4:14 ` Greg KH [this message]
2012-06-07 4:14 ` [ 04/47] mm: pmd_read_atomic: fix 32bit PAE pmd walk vs pmd_populate SMP race condition Greg KH
2012-06-07 4:14 ` [ 05/47] iwlwifi: update BT traffic load states correctly Greg KH
2012-06-07 4:14 ` [ 06/47] cifs: fix oops while traversing open file list (try #4) Greg KH
2012-06-07 4:14 ` [ 07/47] PARISC: fix boot failure on 32-bit systems caused by branch stubs placed before .text Greg KH
2012-06-07 4:14 ` [ 08/47] PARISC: fix TLB fault path on PA2.0 narrow systems Greg KH
2012-06-07 4:14 ` [ 09/47] solos-pci: Fix DMA support Greg KH
2012-06-07 4:14 ` [ 10/47] mac80211: fix ADDBA declined after suspend with wowlan Greg KH
2012-06-07 4:14 ` [ 11/47] NFSv4: Map NFS4ERR_SHARE_DENIED into an EACCES error instead of EIO Greg KH
2012-06-07 4:14 ` [ 12/47] drm/radeon: fix XFX quirk Greg KH
2012-06-07 4:14 ` [ 13/47] drm/i915: properly handle interlaced bit for sdvo dtd conversion Greg KH
2012-06-07 4:14 ` [ 14/47] drm/i915: wait for a vblank to pass after tv detect Greg KH
2012-06-07 4:14 ` [ 15/47] Bluetooth: Add support for Broadcom BCM20702A0 Greg KH
2012-06-07 4:14 ` [ 16/47] Bluetooth: Add support for BCM20702A0 [0a5c:21e3] Greg KH
2012-06-07 4:14 ` [ 17/47] Bluetooth: btusb: Add vendor specific ID (0a5c 21f3) for BCM20702A0 Greg KH
2012-06-07 4:14 ` [ 18/47] Bluetooth: btusb: add support for BCM20702A0 [0a5c:21e6] Greg KH
2012-06-07 4:14 ` [ 19/47] Bluetooth: btusb: Add USB device ID "0a5c 21e8" Greg KH
2012-06-07 4:14 ` [ 20/47] Bluetooth: btusb: Add vendor specific ID (0489 e042) for BCM20702A0 Greg KH
2012-06-07 4:14 ` [ 21/47] Bluetooth: btusb: typo in Broadcom SoftSailing id Greg KH
2012-06-07 4:14 ` [ 22/47] Add Foxconn / Hon Hai IDs for btusb module Greg KH
2012-06-07 4:14 ` [ 23/47] Bluetooth: Add support for Foxconn/Hon Hai AR5BBU22 0489:E03C Greg KH
2012-06-07 4:14 ` [ 24/47] ALSA: usb-audio: fix rate_list memory leak Greg KH
2012-06-07 4:14 ` [ 25/47] vfs: umount_tree() might be called on subtree that had never made it Greg KH
2012-06-07 4:14 ` [ 26/47] mtd: nand: fix scan_read_raw_oob Greg KH
2012-06-07 4:14 ` [ 27/47] drm/radeon: properly program gart on rv740, juniper, cypress, barts, hemlock Greg KH
2012-06-07 4:14 ` [ 28/47] drm/radeon: fix HD6790, HD6570 backend programming Greg KH
2012-06-07 4:14 ` [ 29/47] drm/ttm: Fix spinlock imbalance Greg KH
2012-06-07 4:14 ` [ 30/47] ipv4: Do not use dead fib_info entries Greg KH
2012-06-07 4:14 ` [ 31/47] ipv4: fix the rcu race between free_fib_info and ip_route_output_slow Greg KH
2012-06-07 4:14 ` [ 32/47] ipv6: fix incorrect ipsec fragment Greg KH
2012-06-07 4:14 ` [ 33/47] l2tp: fix oops in L2TP IP sockets for connect() AF_UNSPEC case Greg KH
2012-06-07 4:14 ` [ 34/47] pktgen: fix crash at module unload Greg KH
2012-06-07 4:14 ` [ 35/47] pktgen: fix module unload for good Greg KH
2012-06-07 4:14 ` [ 36/47] Revert "net: maintain namespace isolation between vlan and real device" Greg KH
2012-06-07 4:14 ` [ 37/47] sctp: check cached dst before using it Greg KH
2012-06-07 4:14 ` [ 38/47] skb: avoid unnecessary reallocations in __skb_cow Greg KH
2012-06-07 4:14 ` [ 39/47] xfrm: take net hdr len into account for esp payload size calculation Greg KH
2012-06-07 4:14 ` [ 40/47] ext4: force ro mount if ext4_setup_super() fails Greg KH
2012-06-07 4:14 ` [ 41/47] ext4: add missing save_error_info() to ext4_error() Greg KH
2012-06-07 4:14 ` [ 42/47] ext4: dont trash state flags in EXT4_IOC_SETFLAGS Greg KH
2012-06-07 4:14 ` [ 43/47] ext4: add ext4_mb_unload_buddy in the error path Greg KH
2012-06-07 4:14 ` [ 44/47] ext4: remove mb_groups before tearing down the buddy_cache Greg KH
2012-06-07 4:14 ` [ 45/47] drm/radeon/kms: add new BTC PCI ids Greg KH
2012-06-07 4:14 ` [ 46/47] ACPI battery: only refresh the sysfs files when pertinent information changes Greg KH
2012-06-07 4:14 ` [ 47/47] wl1251: fix oops on early interrupt Greg KH
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=20120607041358.619939818@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--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;
as well as URLs for NNTP newsgroup(s).