linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Minchan Kim <minchan.kim@gmail.com>
To: Shaohua Li <shaohua.li@intel.com>
Cc: Mel Gorman <mgorman@suse.de>,
	Andrew Morton <akpm@linux-foundation.org>,
	linux-mm <linux-mm@kvack.org>
Subject: Re: [patch 1/3]vmscan: clear ZONE_CONGESTED for zone with good watermark
Date: Fri, 29 Jul 2011 18:13:58 +0900	[thread overview]
Message-ID: <20110729091358.GD1843@barrios-desktop> (raw)
In-Reply-To: <1311899725.15392.416.camel@sli10-conroe>

On Fri, Jul 29, 2011 at 08:35:25AM +0800, Shaohua Li wrote:
> On Thu, 2011-07-28 at 18:56 +0800, Mel Gorman wrote:
> > On Thu, Jul 28, 2011 at 04:13:01PM +0800, Shaohua Li wrote:
> > > correctly clear ZONE_CONGESTED. If a zone watermark is ok, we
> > > should clear ZONE_CONGESTED regardless if this is a high order
> > > allocation, because pages can be reclaimed in other tasks but
> > > ZONE_CONGESTED is only cleared in kswapd.
> > > 
> > 
> > What problem does this solve?
> > 
> > As it is, for high order allocations it takes the following steps
> > 
> > If reclaiming at high order {
> > 	for each zone {
> > 		if all_unreclaimable
> > 			skip
> > 		if watermark is not met
> > 			order = 0
> > 			loop again
> > 		
> > 		/* watermark is met */
> > 		clear congested
> > 	}
> > }
> > 
> > If high orders are failing, kswapd balances for order-0 where there
> > is already a cleaning of ZONE_CONGESTED if the zone was shrunk and
> > became balanced. I see the case for hunk 1 of the patch because now
> > it'll clear ZONE_CONGESTED for zones that are already balanced which
> > might have a noticable effect on wait_iff_congested. Is this what
> > you see? Even if it is, it does not explain hunk 2 of the patch.
> I first looked at the hunk 2 place and thought we don't clear
> ZONE_CONGESTED there. I then figured out we need do the same thing for
> the hunk 1. But you are correct, with hunk 1, hunk 2 isn't required.
> updated patch.
> 
> 
> 
> correctly clear ZONE_CONGESTED. If a zone watermark is ok, we
> should clear ZONE_CONGESTED because pages can be reclaimed in
> other tasks but ZONE_CONGESTED is only cleared in kswapd.
> 
> Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>

Even it will fix that when kswapd wakes up lately by order-0 and look at zones,
all zones would become okay so it jumps out with "if (i < 0) goto out" with missing
clearing ZONE_CONGESTED.

-- 
Kind regards,
Minchan Kim

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

      parent reply	other threads:[~2011-07-29  9:14 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-07-28  8:13 Shaohua Li
2011-07-28 10:56 ` Mel Gorman
2011-07-29  0:35   ` Shaohua Li
2011-07-29  8:50     ` Mel Gorman
2011-07-29 11:01       ` Shaohua Li
2011-07-29  9:13     ` Minchan Kim [this message]

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=20110729091358.GD1843@barrios-desktop \
    --to=minchan.kim@gmail.com \
    --cc=akpm@linux-foundation.org \
    --cc=linux-mm@kvack.org \
    --cc=mgorman@suse.de \
    --cc=shaohua.li@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