From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
To: Mel Gorman <mel@csn.ul.ie>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Christoph Lameter <cl@linux-foundation.org>,
Adam Litke <agl@us.ibm.com>, Avi Kivity <avi@redhat.com>,
David Rientjes <rientjes@google.com>,
Minchan Kim <minchan.kim@gmail.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Rik van Riel <riel@redhat.com>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH 10/11] Direct compact when a high-order allocation fails
Date: Fri, 26 Mar 2010 10:03:08 +0900 [thread overview]
Message-ID: <20100326100308.564ebb7b.kamezawa.hiroyu@jp.fujitsu.com> (raw)
In-Reply-To: <20100325101653.GN2024@csn.ul.ie>
On Thu, 25 Mar 2010 10:16:54 +0000
Mel Gorman <mel@csn.ul.ie> wrote:
> On Thu, Mar 25, 2010 at 06:50:21PM +0900, KAMEZAWA Hiroyuki wrote:
> > On Thu, 25 Mar 2010 09:48:26 +0000
> > Mel Gorman <mel@csn.ul.ie> wrote:
> >
> > > > In that case, compact_finished() can't
> > > > find there is a free chunk and do more work. How about using a function like
> > > > free_pcppages_bulk(zone, pcp->batch, pcp);
> > > > to bypass pcp list and freeing pages at once ?
> > > >
> > >
> > > I think you mean to drain the PCP lists while compaction is happening
> > > but is it justified? It's potentially a lot of IPI calls just to check
> > > if compaction can finish a little earlier. If the pages on the PCP lists
> > > are making that much of a difference to high-order page availability, it
> > > implies that the zone is pretty full and it's likely that compaction was
> > > avoided and we direct reclaimed.
> > >
> > Ah, sorry for my short word again. I mean draining "local" pcp list because
> > a thread which run direct-compaction freed pages. IPI is not necessary and
> > overkill.
> >
>
> Ah, I see now. There are two places that pages get freed. release_freepages()
> at the end of compaction when it's too late for compact_finished() to be
> helped and within migration itself. Migration frees with either
> free_page() or more commonly put_page() with put_page() being the most
> frequently used. As free_page() is called on failure to migrate (rare),
> there is little help in changing it and I'd rather not modify how
> put_page() works.
>
> I could add a variant of drain_local_pages() that drains just the local PCP of
> a given zone before compact_finished() is called. The cost would be a doubling
> of the number of times zone->lock is taken to do the drain. Is it
> justified? It seems overkill to me to take the zone->lock just in case
> compaction can finish a little earlier. It feels like it would be adding
> a guaranteed cost for a potential saving.
>
If you want to keep code comapct, I don't ask more.
I worried about that just because memory hot-unplug were suffered by pagevec
and pcp list before using MIGRATE_ISOLATE and proper lru_add_drain().
Thanks,
-Kame
next prev parent reply other threads:[~2010-03-26 1:06 UTC|newest]
Thread overview: 86+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-03-23 12:25 [PATCH 0/11] Memory Compaction v5 Mel Gorman
2010-03-23 12:25 ` [PATCH 01/11] mm,migration: Take a reference to the anon_vma before migrating Mel Gorman
2010-03-23 12:25 ` [PATCH 02/11] mm,migration: Do not try to migrate unmapped anonymous pages Mel Gorman
2010-03-23 17:22 ` Christoph Lameter
2010-03-23 18:04 ` Mel Gorman
2010-03-23 12:25 ` [PATCH 03/11] mm: Share the anon_vma ref counts between KSM and page migration Mel Gorman
2010-03-23 17:25 ` Christoph Lameter
2010-03-23 23:55 ` KAMEZAWA Hiroyuki
2010-03-23 12:25 ` [PATCH 04/11] Allow CONFIG_MIGRATION to be set without CONFIG_NUMA or memory hot-remove Mel Gorman
2010-03-23 12:25 ` [PATCH 05/11] Export unusable free space index via /proc/unusable_index Mel Gorman
2010-03-23 17:31 ` Christoph Lameter
2010-03-23 18:14 ` Mel Gorman
2010-03-24 0:03 ` KAMEZAWA Hiroyuki
2010-03-24 0:16 ` Minchan Kim
2010-03-24 0:13 ` KAMEZAWA Hiroyuki
2010-03-24 10:25 ` Mel Gorman
2010-03-23 12:25 ` [PATCH 06/11] Export fragmentation index via /proc/extfrag_index Mel Gorman
2010-03-23 17:37 ` Christoph Lameter
2010-03-23 12:25 ` [PATCH 07/11] Memory compaction core Mel Gorman
2010-03-23 17:56 ` Christoph Lameter
2010-03-23 18:15 ` Mel Gorman
2010-03-23 18:33 ` Christoph Lameter
2010-03-23 18:58 ` Mel Gorman
2010-03-23 19:20 ` Christoph Lameter
2010-03-24 1:03 ` KAMEZAWA Hiroyuki
2010-03-24 1:47 ` Minchan Kim
2010-03-24 1:53 ` KAMEZAWA Hiroyuki
2010-03-24 2:10 ` Minchan Kim
2010-03-24 10:57 ` Mel Gorman
2010-03-24 20:33 ` Andrew Morton
2010-03-24 20:59 ` Jonathan Corbet
2010-03-24 21:14 ` Andrew Morton
2010-03-24 21:19 ` Christoph Lameter
2010-03-24 21:19 ` Andrea Arcangeli
2010-03-24 21:28 ` Jonathan Corbet
2010-03-24 21:47 ` Andrea Arcangeli
2010-03-24 21:54 ` Jonathan Corbet
2010-03-24 22:06 ` Andrea Arcangeli
2010-03-24 21:57 ` Andrea Arcangeli
2010-03-25 9:13 ` Mel Gorman
2010-03-23 12:25 ` [PATCH 08/11] Add /proc trigger for memory compaction Mel Gorman
2010-03-23 18:25 ` Christoph Lameter
2010-03-23 18:32 ` Mel Gorman
2010-03-24 20:33 ` Andrew Morton
2010-03-26 10:46 ` Mel Gorman
2010-03-23 12:25 ` [PATCH 09/11] Add /sys trigger for per-node " Mel Gorman
2010-03-23 18:27 ` Christoph Lameter
2010-03-23 22:45 ` Minchan Kim
2010-03-24 0:19 ` KAMEZAWA Hiroyuki
2010-03-23 12:25 ` [PATCH 10/11] Direct compact when a high-order allocation fails Mel Gorman
2010-03-23 23:10 ` Minchan Kim
2010-03-24 11:11 ` Mel Gorman
2010-03-24 11:59 ` Minchan Kim
2010-03-24 12:06 ` Minchan Kim
2010-03-24 12:10 ` Mel Gorman
2010-03-24 12:09 ` Mel Gorman
2010-03-24 12:25 ` Minchan Kim
2010-03-24 1:19 ` KAMEZAWA Hiroyuki
2010-03-24 11:40 ` Mel Gorman
2010-03-25 0:30 ` KAMEZAWA Hiroyuki
2010-03-25 9:48 ` Mel Gorman
2010-03-25 9:50 ` KAMEZAWA Hiroyuki
2010-03-25 10:16 ` Mel Gorman
2010-03-26 1:03 ` KAMEZAWA Hiroyuki [this message]
2010-03-26 9:40 ` Mel Gorman
2010-03-24 20:48 ` Andrew Morton
2010-03-25 0:57 ` KAMEZAWA Hiroyuki
2010-03-25 10:21 ` Mel Gorman
2010-03-23 12:25 ` [PATCH 11/11] Do not compact within a preferred zone after a compaction failure Mel Gorman
2010-03-23 18:31 ` Christoph Lameter
2010-03-23 18:39 ` Mel Gorman
2010-03-23 19:27 ` Christoph Lameter
2010-03-24 10:37 ` Mel Gorman
2010-03-24 19:54 ` Christoph Lameter
2010-03-24 20:53 ` Andrew Morton
2010-03-25 9:40 ` Mel Gorman
-- strict thread matches above, loose matches on Subject: below --
2010-03-12 16:41 [PATCH 0/11] Memory Compaction v4 Mel Gorman
2010-03-12 16:41 ` [PATCH 10/11] Direct compact when a high-order allocation fails Mel Gorman
2010-03-16 2:47 ` Minchan Kim
2010-03-19 6:21 ` KOSAKI Motohiro
2010-03-19 6:31 ` KOSAKI Motohiro
2010-03-19 10:10 ` Mel Gorman
2010-03-25 11:22 ` KOSAKI Motohiro
2010-03-19 10:09 ` Mel Gorman
2010-03-25 11:08 ` KOSAKI Motohiro
2010-03-25 15:11 ` Mel Gorman
2010-03-26 6:01 ` KOSAKI Motohiro
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=20100326100308.564ebb7b.kamezawa.hiroyu@jp.fujitsu.com \
--to=kamezawa.hiroyu@jp.fujitsu.com \
--cc=aarcange@redhat.com \
--cc=agl@us.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=avi@redhat.com \
--cc=cl@linux-foundation.org \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
--cc=minchan.kim@gmail.com \
--cc=riel@redhat.com \
--cc=rientjes@google.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).