From: Mike Galbraith <efault@gmx.de>
To: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Rik van Riel <riel@redhat.com>,
lwoodman@redhat.com, akpm@linux-foundation.org,
linux-mm@kvack.org, linux-kernel@vger.kernel.org,
minchan.kim@gmail.com
Subject: Re: [PATCH 4/8] Use prepare_to_wait_exclusive() instead prepare_to_wait()
Date: Wed, 16 Dec 2009 06:43:44 +0100 [thread overview]
Message-ID: <1260942224.5766.57.camel@marge.simson.net> (raw)
In-Reply-To: <20091216093533.CDF1.A69D9226@jp.fujitsu.com>
On Wed, 2009-12-16 at 09:48 +0900, KOSAKI Motohiro wrote:
> > On 12/15/2009 12:32 AM, Mike Galbraith wrote:
> > > On Tue, 2009-12-15 at 09:45 +0900, KOSAKI Motohiro wrote:
> > >>> On 12/14/2009 07:30 AM, KOSAKI Motohiro wrote:
> > >>>> if we don't use exclusive queue, wake_up() function wake _all_ waited
> > >>>> task. This is simply cpu wasting.
> > >>>>
> > >>>> Signed-off-by: KOSAKI Motohiro<kosaki.motohiro@jp.fujitsu.com>
> > >>>
> > >>>> if (zone_watermark_ok(zone, sc->order, low_wmark_pages(zone),
> > >>>> 0, 0)) {
> > >>>> - wake_up(wq);
> > >>>> + wake_up_all(wq);
> > >>>> finish_wait(wq,&wait);
> > >>>> sc->nr_reclaimed += sc->nr_to_reclaim;
> > >>>> return -ERESTARTSYS;
> > >>>
> > >>> I believe we want to wake the processes up one at a time
> > >>> here.
> >
> > >> Actually, wake_up() and wake_up_all() aren't different so much.
> > >> Although we use wake_up(), the task wake up next task before
> > >> try to alloate memory. then, it's similar to wake_up_all().
> >
> > That is a good point. Maybe processes need to wait a little
> > in this if() condition, before the wake_up(). That would give
> > the previous process a chance to allocate memory and we can
> > avoid waking up too many processes.
>
> if we really need wait a bit, Mike's wake_up_batch is best, I think.
> It mean
> - if another CPU is idle, wake up one process soon. iow, it don't
> make meaningless idle.
Along those lines, there's also NEWIDLE balancing considerations. That
idle may result in a task being pulled, which may or may not hurt a bit.
'course, if you're jamming up on memory allocation, that's the least of
your worries, but every idle avoided is potentially a pull avoided.
Just a thought.
-Mike
next prev parent reply other threads:[~2009-12-16 5:43 UTC|newest]
Thread overview: 66+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-12-11 21:46 [PATCH v2] vmscan: limit concurrent reclaimers in shrink_zone Rik van Riel
2009-12-14 0:14 ` Minchan Kim
2009-12-14 4:09 ` Rik van Riel
2009-12-14 4:19 ` Minchan Kim
2009-12-14 4:29 ` Rik van Riel
2009-12-14 5:00 ` Minchan Kim
2009-12-14 12:22 ` KOSAKI Motohiro
2009-12-14 12:23 ` [cleanup][PATCH 1/8] vmscan: Make shrink_zone_begin/end helper function KOSAKI Motohiro
2009-12-14 14:34 ` Rik van Riel
2009-12-14 22:39 ` Minchan Kim
2009-12-14 12:24 ` [PATCH 2/8] Mark sleep_on as deprecated KOSAKI Motohiro
2009-12-14 13:03 ` Christoph Hellwig
2009-12-14 16:04 ` Arjan van de Ven
2009-12-14 14:34 ` Rik van Riel
2009-12-14 22:44 ` Minchan Kim
2009-12-14 12:29 ` [PATCH 3/8] Don't use sleep_on() KOSAKI Motohiro
2009-12-14 14:35 ` Rik van Riel
2009-12-14 22:46 ` Minchan Kim
2009-12-14 12:30 ` [PATCH 4/8] Use prepare_to_wait_exclusive() instead prepare_to_wait() KOSAKI Motohiro
2009-12-14 14:33 ` Rik van Riel
2009-12-15 0:45 ` KOSAKI Motohiro
2009-12-15 5:32 ` Mike Galbraith
2009-12-15 8:28 ` Mike Galbraith
2009-12-15 14:36 ` Mike Galbraith
2009-12-15 14:58 ` Rik van Riel
2009-12-15 18:17 ` Mike Galbraith
2009-12-15 18:43 ` Mike Galbraith
2009-12-15 19:33 ` Rik van Riel
2009-12-16 0:48 ` KOSAKI Motohiro
2009-12-16 2:44 ` Rik van Riel
2009-12-16 5:43 ` Mike Galbraith [this message]
2009-12-14 23:03 ` Minchan Kim
2009-12-14 12:30 ` [PATCH 5/8] Use io_schedule() instead schedule() KOSAKI Motohiro
2009-12-14 14:37 ` Rik van Riel
2009-12-14 23:46 ` Minchan Kim
2009-12-15 0:56 ` KOSAKI Motohiro
2009-12-15 1:13 ` Minchan Kim
2009-12-14 12:31 ` [PATCH 6/8] Stop reclaim quickly when the task reclaimed enough lots pages KOSAKI Motohiro
2009-12-14 14:45 ` Rik van Riel
2009-12-14 23:51 ` KOSAKI Motohiro
2009-12-15 0:11 ` Minchan Kim
2009-12-15 0:35 ` KOSAKI Motohiro
2009-12-14 12:32 ` [PATCH 7/8] Use TASK_KILLABLE instead TASK_UNINTERRUPTIBLE KOSAKI Motohiro
2009-12-14 14:47 ` Rik van Riel
2009-12-14 23:52 ` Minchan Kim
2009-12-14 12:32 ` [PATCH 8/8] mm: Give up allocation if the task have fatal signal KOSAKI Motohiro
2009-12-14 14:48 ` Rik van Riel
2009-12-14 23:54 ` Minchan Kim
2009-12-15 0:50 ` KOSAKI Motohiro
2009-12-15 1:03 ` Minchan Kim
2009-12-15 1:16 ` KOSAKI Motohiro
2009-12-14 12:40 ` [PATCH v2] vmscan: limit concurrent reclaimers in shrink_zone KOSAKI Motohiro
2009-12-14 17:08 ` Larry Woodman
2009-12-15 0:49 ` KOSAKI Motohiro
[not found] ` <20091217193818.9FA9.A69D9226@jp.fujitsu.com>
2009-12-17 12:23 ` FWD: " Larry Woodman
2009-12-17 14:43 ` Rik van Riel
2009-12-17 19:55 ` Rik van Riel
2009-12-17 21:05 ` Hugh Dickins
2009-12-17 22:52 ` Rik van Riel
2009-12-18 16:23 ` Andrea Arcangeli
2009-12-18 17:43 ` Rik van Riel
2009-12-18 10:27 ` KOSAKI Motohiro
2009-12-18 14:09 ` Rik van Riel
2009-12-18 13:38 ` Avi Kivity
2009-12-18 14:12 ` Rik van Riel
2009-12-18 14:13 ` Avi Kivity
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=1260942224.5766.57.camel@marge.simson.net \
--to=efault@gmx.de \
--cc=akpm@linux-foundation.org \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=lwoodman@redhat.com \
--cc=minchan.kim@gmail.com \
--cc=riel@redhat.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).