Date:	Thu, 4 Apr 2013 08:46:44 +0900
From:	Minchan Kim <minchan@...nel.org>
To:	Michael Kerrisk <mtk.manpages@...il.com>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Linux Kernel <linux-kernel@...r.kernel.org>,
	linux-mm <linux-mm@...ck.org>, Mel Gorman <mgorman@...e.de>,
	Rik van Riel <riel@...hat.com>,
	Johannes Weiner <hannes@...xchg.org>,
	Hugh Dickins <hughd@...gle.com>,
	Sangseok Lee <sangseok.lee@....com>
Subject: Re: [RFC 1/4] mm: Per process reclaim

On Wed, Apr 03, 2013 at 12:10:22PM +0200, Michael Kerrisk wrote:
> Hello Minchan,
> 
> On Mon, Mar 25, 2013 at 7:21 AM, Minchan Kim <minchan@...nel.org> wrote:
> > These day, there are many platforms avaiable in the embedded market
> > and they are smarter than kernel which has very limited information
> > about working set so they want to involve memory management more heavily
> > like android's lowmemory killer and ashmem or recent many lowmemory
> > notifier(there was several trial for various company NOKIA, SAMSUNG,
> > Linaro, Google ChromeOS, Redhat).
> >
> > One of the simple imagine scenario about userspace's intelligence is that
> > platform can manage tasks as forground and backgroud so it would be
> > better to reclaim background's task pages for end-user's *responsibility*
> > although it has frequent referenced pages.
> >
> > This patch adds new knob "reclaim under proc/<pid>/" so task manager
> > can reclaim any target process anytime, anywhere. It could give another
> > method to platform for using memory efficiently.
> >
> > It can avoid process killing for getting free memory, which was really
> > terrible experience because I lost my best score of game I had ever
> > after I switch the phone call while I enjoyed the game.
> >
> > Writing 1 to /proc/pid/reclaim reclaims only file pages.
> > Writing 2 to /proc/pid/reclaim reclaims only anonymous pages.
> > Writing 3 to /proc/pid/reclaim reclaims all pages from target process.
> 
> This interface seems to work as advertized, at least from some light
> testing that I've done.

Thanks for the testing!

> 
> However, the interface is a quite blunt instrument. Would there be any
> virtue in extending it so that an address range could be written to
> /proc/PID/reclaim? Used in conjunction with /proc/PID/maps, a manager
> process might then choose to trigger reclaim of just selected regions
> of a processes address space. Thus, one might reclaim file backed
> pages in a range using:
> 
>     echo '2 start-address end-address' > /proc/PID/reclaim
> 
> What do you think?

It is really nice idea because some platform use a address space with
mulitple object. Simply, multiple application could work in a address space
but they use separate virtual address range in a address space.
In such model, per-process reclaim isn't vaild any more so your idea
would be nice for it.

One nitpick is that I'd like to use (address, size) instead of (start_address
, end_address) because we always confuse that end_address itself is
inclusive or not in the range.

And I am thinking another options like this

RECLAIM_SOFT_[FILE|ANON] 

It reclaims pages which are not workset.

RECLAIM_HARD_[FILE|ANON]

It reclaims pages of range unconditionally although pages are shared by
several processes.

But before that, I'd like to hear other guys's opinion.

> 
> Thanks,
> 
> Michael
> 
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to majordomo@...ck.org.  For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"dont@...ck.org"> email@...ck.org </a>

-- 
Kind regards,
Minchan Kim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists