From: Avi Kivity <avi@redhat.com>
To: Rik van Riel <riel@redhat.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Andrea Arcangeli <aarcange@redhat.com>,
"Dike, Jeffrey G" <jeffrey.g.dike@intel.com>,
"Yu, Wilfred" <wilfred.yu@intel.com>,
"Kleen, Andi" <andi.kleen@intel.com>,
Hugh Dickins <hugh.dickins@tiscali.co.uk>,
Andrew Morton <akpm@linux-foundation.org>,
Christoph Lameter <cl@linux-foundation.org>,
Mel Gorman <mel@csn.ul.ie>, LKML <linux-kernel@vger.kernel.org>,
linux-mm <linux-mm@kvack.org>
Subject: Re: [RFC] respect the referenced bit of KVM guest pages?
Date: Thu, 13 Aug 2009 22:12:01 +0300 [thread overview]
Message-ID: <4A846581.2020304@redhat.com> (raw)
In-Reply-To: <4A843EAE.6070200@redhat.com>
On 08/13/2009 07:26 PM, Rik van Riel wrote:
>> Why do we need to ignore the referenced bit in such cases? To avoid
>> overscanning?
>
>
> Because swapping out anonymous pages tends to be a relatively
> rare operation, we'll have many gigabytes of anonymous pages
> that all have the referenced bit set (because there was lots
> of time between swapout bursts).
>
> Ignoring the referenced bit on active anon pages makes no
> difference on these systems, because all active anon pages
> have the referenced bit set, anyway.
>
> All we need to do is put the pages on the inactive list and
> give them a chance to get referenced.
>
> However, on smaller systems (and cgroups!), the speed at
> which we can do pageout IO is larger, compared to the amount
> of memory. This means we can cycle through the pages more
> quickly and we may want to count references on the active
> list, too.
>
> Yes, on smaller systems we'll also often end up with bursty
> swapout loads and all pages referenced - but since we have
> fewer pages to begin with, it won't hurt as much.
>
> I suspect that an inactive_ratio of 3 or 4 might make a
> good cutoff value.
>
Thanks for the explanation. I think my earlier idea of
- do not ignore the referenced bit
- if you see a run of N pages which all have the referenced bit set, do
swap one
has merit. It means we cycle more quickly (by a factor of N) through
the list, looking for unreferenced pages. If we don't find any we've
spent a some more cpu, but if we do find an unreferenced page, we win by
swapping a truly unneeded page.
Cycling faster also means reducing the time between examinations of any
particular page, so it increases the meaningfulness of the check on
large systems (otherwise even rarely used pages will always show up as
referenced).
--
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2009-08-13 19:12 UTC|newest]
Thread overview: 122+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-08-05 2:40 [RFC] respect the referenced bit of KVM guest pages? Wu Fengguang
2009-08-05 4:15 ` KOSAKI Motohiro
2009-08-05 4:41 ` Wu Fengguang
2009-08-05 7:58 ` Avi Kivity
2009-08-05 8:17 ` Avi Kivity
2009-08-05 14:33 ` Rik van Riel
2009-08-05 15:37 ` Avi Kivity
2009-08-05 14:15 ` Rik van Riel
2009-08-05 15:12 ` Avi Kivity
2009-08-05 15:15 ` Rik van Riel
2009-08-05 15:25 ` Avi Kivity
2009-08-05 16:35 ` Andrea Arcangeli
2009-08-05 16:31 ` Andrea Arcangeli
2009-08-05 17:25 ` Rik van Riel
2009-08-05 15:45 ` Dike, Jeffrey G
2009-08-05 16:05 ` Andrea Arcangeli
2009-08-05 16:12 ` Dike, Jeffrey G
2009-08-05 16:19 ` Andrea Arcangeli
2009-08-05 15:58 ` Andrea Arcangeli
2009-08-05 17:20 ` Rik van Riel
2009-08-05 17:42 ` Rik van Riel
2009-08-06 10:15 ` Andrea Arcangeli
2009-08-06 10:08 ` Andrea Arcangeli
2009-08-06 10:18 ` Avi Kivity
2009-08-06 10:20 ` Andrea Arcangeli
2009-08-06 10:59 ` Wu Fengguang
2009-08-06 11:44 ` Avi Kivity
2009-08-06 13:06 ` Wu Fengguang
2009-08-06 13:16 ` Rik van Riel
2009-08-16 3:28 ` Wu Fengguang
2009-08-16 3:56 ` Rik van Riel
2009-08-16 4:43 ` Balbir Singh
2009-08-16 4:55 ` Wu Fengguang
2009-08-16 5:59 ` Balbir Singh
2009-08-17 19:47 ` Dike, Jeffrey G
2009-08-21 18:24 ` Balbir Singh
2009-08-31 19:43 ` Dike, Jeffrey G
2009-08-31 19:52 ` Rik van Riel
2009-08-31 20:06 ` Dike, Jeffrey G
2009-08-31 20:09 ` Rik van Riel
2009-08-31 20:11 ` Dike, Jeffrey G
2009-08-31 20:42 ` Balbir Singh
2009-08-06 13:46 ` Avi Kivity
2009-08-06 21:09 ` Jeff Dike
2009-08-16 3:18 ` Wu Fengguang
2009-08-16 3:53 ` Rik van Riel
2009-08-16 5:15 ` Wu Fengguang
2009-08-16 11:29 ` Wu Fengguang
2009-08-17 14:33 ` Minchan Kim
2009-08-18 2:34 ` Wu Fengguang
2009-08-18 4:17 ` Minchan Kim
2009-08-18 9:31 ` Wu Fengguang
2009-08-18 9:52 ` Minchan Kim
2009-08-18 10:00 ` Wu Fengguang
2009-08-18 11:00 ` Minchan Kim
2009-08-18 11:11 ` Wu Fengguang
2009-08-18 14:03 ` Minchan Kim
2009-08-18 16:27 ` KOSAKI Motohiro
2009-08-18 15:57 ` KOSAKI Motohiro
2009-08-19 12:01 ` Wu Fengguang
2009-08-19 12:05 ` KOSAKI Motohiro
2009-08-19 12:10 ` Wu Fengguang
2009-08-19 12:25 ` Minchan Kim
2009-08-19 13:19 ` KOSAKI Motohiro
2009-08-19 13:28 ` Minchan Kim
2009-08-21 11:17 ` KOSAKI Motohiro
2009-08-19 13:24 ` Wu Fengguang
2009-08-19 13:38 ` Minchan Kim
2009-08-19 14:00 ` Wu Fengguang
2009-08-06 13:13 ` Rik van Riel
2009-08-06 13:49 ` Avi Kivity
2009-08-07 3:11 ` KOSAKI Motohiro
2009-08-07 7:54 ` Balbir Singh
2009-08-07 8:24 ` KAMEZAWA Hiroyuki
2009-08-06 13:11 ` Rik van Riel
2009-08-06 13:08 ` Rik van Riel
2009-08-07 3:17 ` KOSAKI Motohiro
2009-08-12 7:48 ` Wu Fengguang
2009-08-12 14:31 ` Rik van Riel
2009-08-13 1:03 ` Wu Fengguang
2009-08-13 15:46 ` Rik van Riel
2009-08-13 16:12 ` Avi Kivity
2009-08-13 16:26 ` Rik van Riel
2009-08-13 19:12 ` Avi Kivity [this message]
2009-08-13 21:16 ` Johannes Weiner
2009-08-14 7:16 ` Avi Kivity
2009-08-14 9:10 ` Johannes Weiner
2009-08-14 9:51 ` Wu Fengguang
2009-08-14 13:19 ` Rik van Riel
2009-08-15 5:45 ` Wu Fengguang
2009-08-16 5:09 ` Balbir Singh
2009-08-16 5:41 ` Wu Fengguang
2009-08-16 5:50 ` Wu Fengguang
2009-08-18 15:57 ` KOSAKI Motohiro
2009-08-17 18:04 ` Dike, Jeffrey G
2009-08-18 2:26 ` Wu Fengguang
2009-09-02 19:30 ` Dike, Jeffrey G
2009-09-03 2:04 ` Wu Fengguang
2009-09-04 20:06 ` Dike, Jeffrey G
2009-09-04 20:57 ` Rik van Riel
2009-08-18 15:57 ` KOSAKI Motohiro
2009-08-19 12:08 ` Wu Fengguang
2009-08-19 13:40 ` [RFC] memcg: move definitions to .h and inline some functions Wu Fengguang
2009-08-19 14:18 ` KAMEZAWA Hiroyuki
2009-08-19 14:27 ` Balbir Singh
2009-08-20 1:34 ` Wu Fengguang
2009-08-14 21:42 ` [RFC] respect the referenced bit of KVM guest pages? Dike, Jeffrey G
2009-08-14 22:37 ` Rik van Riel
2009-08-15 5:32 ` Wu Fengguang
2009-09-13 16:23 ` KOSAKI Motohiro
2009-08-05 17:53 ` Rik van Riel
2009-08-05 19:00 ` Dike, Jeffrey G
2009-08-05 19:07 ` Rik van Riel
2009-08-05 19:18 ` Dike, Jeffrey G
2009-08-06 9:22 ` Avi Kivity
2009-08-06 9:25 ` Wu Fengguang
2009-08-06 9:35 ` Avi Kivity
2009-08-06 9:35 ` Wu Fengguang
2009-08-06 9:59 ` Avi Kivity
2009-08-06 9:59 ` Wu Fengguang
2009-08-06 10:14 ` Avi Kivity
2009-08-07 1:25 ` KAMEZAWA Hiroyuki
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=4A846581.2020304@redhat.com \
--to=avi@redhat.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=andi.kleen@intel.com \
--cc=cl@linux-foundation.org \
--cc=fengguang.wu@intel.com \
--cc=hugh.dickins@tiscali.co.uk \
--cc=jeffrey.g.dike@intel.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
--cc=riel@redhat.com \
--cc=wilfred.yu@intel.com \
/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).