linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* Why isn't shrink_slab more zone oriented ?
@ 2011-09-26 11:51 kautuk.c @samsung.com
  2011-09-28  0:04 ` Dave Chinner
  0 siblings, 1 reply; 2+ messages in thread
From: kautuk.c @samsung.com @ 2011-09-26 11:51 UTC (permalink / raw)
  To: Andrew Morton, Mel Gorman, Minchan Kim, KAMEZAWA Hiroyuki,
	Rik van Riel
  Cc: linux-mm, linux-kernel

Hi,

I was going through the do_try_to_free_pages(), balance_pgdat(),
__zone_reclaim()
functions and I see that shrink_zone and shrink_slab are called for each zone.

But, shrink_slab() doesn't seem to bother about the zone from where it
is freeing
memory.

My questions are:

- Will this be a strain on the direct/indirect reclamation algorithm ?
  The loops involved expects to free something from that particular zone.
  shrink_slab might take more time than is required to free up pages from that
  particular zone which might not be optimal.
  Am I right about this conclusion ?

- If the above is correct, then is there any work happening on this
front, i.e., to
  make shrink_slab functionality and the shrinker callbacks more zone-centric ?
  Are there any patches that I could look at or download from
somewhere for this ?

- Are there any other considerations to be careful of for making
shrink_slab more
  zone centric ? Can there be other side effects that affects
performance if the
  shrink_slab and shrinker callbacks are made to free pages only from
a particular
  zone ?

Thanks,
Kautuk.

--
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>

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Why isn't shrink_slab more zone oriented ?
  2011-09-26 11:51 Why isn't shrink_slab more zone oriented ? kautuk.c @samsung.com
@ 2011-09-28  0:04 ` Dave Chinner
  0 siblings, 0 replies; 2+ messages in thread
From: Dave Chinner @ 2011-09-28  0:04 UTC (permalink / raw)
  To: kautuk.c @samsung.com
  Cc: Andrew Morton, Mel Gorman, Minchan Kim, KAMEZAWA Hiroyuki,
	Rik van Riel, linux-mm, linux-kernel

On Mon, Sep 26, 2011 at 05:21:12PM +0530, kautuk.c @samsung.com wrote:
> Hi,
> 
> I was going through the do_try_to_free_pages(), balance_pgdat(),
> __zone_reclaim()
> functions and I see that shrink_zone and shrink_slab are called for each zone.
> 
> But, shrink_slab() doesn't seem to bother about the zone from where it
> is freeing
> memory.

Work is in progress to do this.

http://lwn.net/Articles/456071/

Dirty slab objects are currently not tracked in a manner that makes
per-zone reclaim efficient to do, os that needs to be corrected
first.  Once we have generic per-zone LRU infrastructure, then we
can easily push zone reclaim hints down into the shrinkers for them
to scan the appropriate LRU....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

--
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>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2011-09-28  0:04 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-09-26 11:51 Why isn't shrink_slab more zone oriented ? kautuk.c @samsung.com
2011-09-28  0:04 ` Dave Chinner

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).