linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: kanoj@google.engr.sgi.com (Kanoj Sarcar)
To: Andrea Arcangeli <andrea@suse.de>
Cc: Linus Torvalds <torvalds@transmeta.com>,
	Ingo Molnar <mingo@chiara.csoma.elte.hu>,
	Alan Cox <alan@lxorguk.ukuu.org.uk>,
	Rik van Riel <riel@nl.linux.org>,
	linux-mm@kvack.org, linux-kernel@vger.rutgers.edu
Subject: Re: [RFC] 2.3.39 zone balancing
Date: Thu, 13 Jan 2000 17:13:05 -0800 (PST)	[thread overview]
Message-ID: <200001140113.RAA62584@google.engr.sgi.com> (raw)
In-Reply-To: <Pine.LNX.4.21.0001140128510.3816-100000@alpha.random> from "Andrea Arcangeli" at Jan 14, 2000 01:33:54 AM

> 
> On Thu, 13 Jan 2000, Linus Torvalds wrote:
> 
> >Basically, my argument is that there is no way "swap_out()" can really
> >target any special zone, except by avoiding to do the final stage in a
> >long sequence of stages that it has already done. I think that's just
> >completely wasteful - doing all the work, and then at the last minute
> >deciding to not use the work after all. Especially as we don't really have
> >any good reason to believe that it's the right thing in the first place.
> 
> The only problem in what you are suggesting is that you may end swapping
> out also the wrong pages. Suppose you want to allocate 4k of DMA
> memory. Why should the machine swapout lots of mbytes of data while it
> could only swapout 4k? And after each swapout we have to restart from the
> vma because to swapout we have to drop the pagetable lock and so the
> mappings can be changed from under us.

Yes, I am worried about this a little too. Specially, when you are
hunting for a dma page, and the machine happens to have gigs of
highmem and regular pages, chances are that you will end up stealing
a lot of pages unneccesarily.

But as Linus points out, recovering from that is not that costly
(the page will be in the swapcache mostly, its just the cost of 
the page fault).

What about stealing the page only if the corresponding zone is
also running unbalanced?

Kanoj

> 
> >So that's why I think the page table walker should be completely
> >zone-blind, and just not care. It's likely to be more "balanced" that way
> >anyway.
> 
> The swapout will be definitely more balanced but we may end doing not
> necesary swapouts.
> 
> Andrea
> 
> --
> 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.nl.linux.org/Linux-MM/
> 

--
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.nl.linux.org/Linux-MM/

  parent reply	other threads:[~2000-01-14  1:13 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2000-01-12 21:11 Kanoj Sarcar
2000-01-13 13:40 ` Rik van Riel
2000-01-13 17:06   ` Andrea Arcangeli
2000-01-13 17:18   ` Alan Cox
2000-01-13 18:37     ` Rik van Riel
2000-01-13 20:13       ` Andrea Arcangeli
2000-01-13 21:12         ` Rik van Riel
2000-01-13 21:40         ` Kanoj Sarcar
2000-01-14 12:25           ` Jamie Lokier
2000-01-14 13:43             ` Andrea Arcangeli
2000-01-13 18:52   ` Kanoj Sarcar
2000-01-13 19:59     ` Andrea Arcangeli
2000-01-13 21:02       ` Kanoj Sarcar
2000-01-13 21:34         ` Benjamin C.R. LaHaise
2000-01-13 21:48           ` Kanoj Sarcar
2000-01-13 21:42         ` Alan Cox
2000-01-13 21:50           ` Kanoj Sarcar
2000-01-13 21:53             ` Alan Cox
2000-01-13 22:01           ` Linus Torvalds
2000-01-13 22:13             ` Kanoj Sarcar
2000-01-13 22:28               ` Rik van Riel
2000-01-13 22:30               ` Linus Torvalds
2000-01-13 23:53                 ` Ingo Molnar
2000-01-13 23:29                   ` Linus Torvalds
2000-01-14  0:33                     ` Andrea Arcangeli
2000-01-14  0:52                       ` Linus Torvalds
2000-01-14  1:08                         ` Rik van Riel
2000-01-14  2:13                         ` Ingo Molnar
2000-01-14  1:17                           ` Kanoj Sarcar
2000-01-14  2:36                             ` Ingo Molnar
2000-01-14 20:33                               ` Peter Rival
2000-01-14  1:13                       ` Kanoj Sarcar [this message]
2000-01-14  2:27                         ` Ingo Molnar
2000-01-14  2:46                         ` Ingo Molnar
2000-01-14  6:22                           ` Kanoj Sarcar
2000-01-15  2:03                     ` Reworked 2.3.39 zone balancing - v1 Kanoj Sarcar
2000-01-14  0:28                 ` [RFC] 2.3.39 zone balancing Andrea Arcangeli
2000-01-13 17:12 ` Andrea Arcangeli
2000-01-13 18:30   ` Kanoj Sarcar
2000-01-13 19:22     ` Andrea Arcangeli

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=200001140113.RAA62584@google.engr.sgi.com \
    --to=kanoj@google.engr.sgi.com \
    --cc=alan@lxorguk.ukuu.org.uk \
    --cc=andrea@suse.de \
    --cc=linux-kernel@vger.rutgers.edu \
    --cc=linux-mm@kvack.org \
    --cc=mingo@chiara.csoma.elte.hu \
    --cc=riel@nl.linux.org \
    --cc=torvalds@transmeta.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