mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Daniel Phillips <phillips@bonn-fries.net>
To: Jan Harkes <jaharkes@cs.cmu.edu>,
	Linus Torvalds <torvalds@transmeta.com>
Cc: linux-kernel@vger.kernel.org
Subject: Re: broken VM in 2.4.10-pre9
Date: Mon, 17 Sep 2001 14:33:12 +0200	[thread overview]
Message-ID: <20010917122559Z16382-2758+129@humbolt.nl.linux.org> (raw)
In-Reply-To: <20010917003422Z16197-2757+375@humbolt.nl.linux.org> <Pine.LNX.4.33.0109161738110.1054-100000@penguin.transmeta.com> <20010917011157.A22989@cs.cmu.edu>
In-Reply-To: <20010917011157.A22989@cs.cmu.edu>

On September 17, 2001 07:11 am, Jan Harkes wrote:
> As far as I can understand the _original_ design on which the current VM
> is based, aging only occurs to pages on the active 'ring', the inactive
> lists are basically LRU-ordered victim caches. Pages are unmapped before
> they go to the inactive_dirty list and buffers are flushed before they
> can go to inactive_clean.
>
> Ofcourse both the used_once changes and -pre10 sort of flushed these
> designs down the toilet by putting mapped pages on the inactive_dirty
> list and turning the active list into an LRU.

The active list is *supposed* to approximate an LRU.  The inactive lists
are not LRUs but queues, and have always been.

The inactive queues have always had both mapped and unmapped pages on
them. The reason for unmapping a swap cache page page when putting it
on the inactive queue is to give it some time to be rescued, since we
otherwise have no information about its short-term activity because
we have no way of accessing the hardware dirty bit given the physical
page on the lru.  A second reason for unmapping it is, we don't have
any choice.  The point where we place it on the inactive queue is the
last point where we're able to find its userspace page table entry.

<paid advertisement>
We'd be able to avoid unmapping swap cache pages with Rik's rmap
patch because we can easily check the hardware referenced bit before
finally evicting the page.  Plus, and I hope I'm interpreting this
correctly, we can allocate the swap slot and perform swap clustering
at that time, greatly simplifying the swapout code.
</paid advertisment> ;-)

Drifting a little further offtopic.  As far as I can tell, there's no 
fundamental reason why we cannot make the current strategy work as 
well as Rik's rmaps probably will, with some more blood, sweat and
code study.  On the other hand, Matt Dillon, the reigning champion of
virtual memory managment, was quite firm in stating that we should
drop the current virtually scanning strategy in favor of 100%
physical scanning as BSD uses, relying on reverse mapping.

   http://mail.nl.linux.org/linux-mm/2000-05/msg00419.html
   (Matt Dillon holds forth on the design of BSD's memory manager)

--
Daniel

  reply	other threads:[~2001-09-17 12:26 UTC|newest]

Thread overview: 133+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-09-16 15:19 Ricardo Galli
2001-09-16 15:23 ` Michael Rothwell
2001-09-16 16:33   ` Rik van Riel
2001-09-16 16:50     ` Andreas Steinmetz
2001-09-16 17:12       ` Ricardo Galli
2001-09-16 17:06     ` Ricardo Galli
2001-09-16 17:18       ` Jeremy Zawodny
2001-09-16 18:45       ` Stephan von Krawczynski
2001-09-21  3:16         ` Bill Davidsen
2001-09-21 10:21         ` Stephan von Krawczynski
2001-09-21 14:08           ` Bill Davidsen
2001-09-21 14:23             ` Rik van Riel
2001-09-23 13:13               ` Eric W. Biederman
2001-09-23 13:27                 ` Rik van Riel
2001-09-21 10:43         ` Stephan von Krawczynski
2001-09-21 12:13           ` Rik van Riel
2001-09-21 12:55           ` Stephan von Krawczynski
2001-09-21 13:01             ` Rik van Riel
2001-09-22 11:01           ` Daniel Phillips
2001-09-22 20:05             ` Rik van Riel
2001-09-24  9:36           ` Linux VM design VDA
2001-09-24 11:06             ` Dave Jones
2001-09-24 12:15               ` Kirill Ratkin
2001-09-24 13:29             ` Rik van Riel
2001-09-24 14:05               ` VDA
2001-09-24 14:37                 ` Rik van Riel
2001-09-24 14:42                 ` Rik van Riel
2001-09-24 18:37             ` Daniel Phillips
2001-09-24 19:32               ` Rik van Riel
2001-09-24 17:27                 ` Rob Landley
2001-09-24 21:48                   ` Rik van Riel
2001-09-25  9:58                 ` Daniel Phillips
2001-09-25 16:03               ` bill davidsen
2001-09-24 18:46             ` Jonathan Morton
2001-09-24 19:16               ` Daniel Phillips
2001-09-24 19:11             ` Dan Mann
2001-09-25 10:55             ` VDA
2001-09-16 18:16     ` broken VM in 2.4.10-pre9 Stephan von Krawczynski
2001-09-16 19:43     ` Linus Torvalds
2001-09-16 19:57       ` Rik van Riel
2001-09-16 20:17       ` Rik van Riel
2001-09-16 20:29       ` Andreas Steinmetz
2001-09-16 21:28         ` Linus Torvalds
2001-09-16 22:47           ` Alex Bligh - linux-kernel
2001-09-16 22:55             ` Linus Torvalds
2001-09-16 22:59           ` Stephan von Krawczynski
2001-09-16 22:14             ` Linus Torvalds
2001-09-16 23:29               ` Stephan von Krawczynski
2001-09-17 15:35             ` Stephan von Krawczynski
2001-09-17 15:51               ` Linus Torvalds
2001-09-17 16:34               ` Stephan von Krawczynski
2001-09-17 16:46                 ` Linus Torvalds
2001-09-17 17:20                 ` Stephan von Krawczynski
2001-09-17 17:37                   ` Linus Torvalds
2001-09-17  0:37       ` Daniel Phillips
2001-09-17  1:07         ` Linus Torvalds
2001-09-17  2:23           ` Daniel Phillips
2001-09-17  5:11           ` Jan Harkes
2001-09-17 12:33             ` Daniel Phillips [this message]
2001-09-17 12:41               ` Rik van Riel
2001-09-17 14:49                 ` Daniel Phillips
2001-09-17 16:14               ` Jan Harkes
2001-09-17 16:34                 ` Linus Torvalds
2001-09-17 15:38             ` Linus Torvalds
2001-09-17 12:26           ` Rik van Riel
2001-09-17 15:42             ` Linus Torvalds
2001-09-18 12:04               ` Rik van Riel
2001-09-17 17:33             ` Linus Torvalds
2001-09-17 18:07               ` Linus Torvalds
2001-09-18 12:09               ` Rik van Riel
2001-09-21  3:10       ` Bill Davidsen
2001-09-17  8:06     ` Eric W. Biederman
2001-09-17 12:12       ` Rik van Riel
2001-09-17 15:45         ` Eric W. Biederman
  -- strict thread matches above, loose matches on Subject: below --
2001-09-22 19:59 Peter Magnusson
2001-09-22 20:46 ` Jan Harkes
2001-09-22 21:46   ` Peter Magnusson
2001-09-22 19:59 Peter Magnusson
2001-09-22 20:18 ` Rik van Riel
2001-09-19 22:15 Rob Fuller
2001-09-19 22:21 ` David S. Miller
2001-09-19 22:26 ` Christoph Hellwig
2001-09-19 22:30 ` Alan Cox
2001-09-19 22:48 ` Eric W. Biederman
2001-09-19 22:51 ` Bryan O'Sullivan
2001-09-20  3:16 ` GOTO Masanori
2001-09-20  7:38   ` Christoph Hellwig
2001-09-17 15:40 Rob Fuller
2001-09-17 16:03 ` Eric W. Biederman
2001-09-19  9:45   ` Daniel Phillips
2001-09-19 19:45     ` Alan Cox
2001-09-19 21:03       ` Eric W. Biederman
2001-09-19 22:04         ` Alan Cox
2001-09-19 22:26           ` Eric W. Biederman
2001-09-19 23:05           ` Rik van Riel
2001-09-20 11:28           ` Daniel Phillips
2001-09-20 12:06             ` Rik van Riel
2001-09-21  8:13               ` Daniel Phillips
2001-09-21 12:10                 ` Rik van Riel
2001-09-21 15:27                 ` Jan Harkes
2001-09-22  7:09                   ` Daniel Phillips
2001-09-25 11:04                     ` Mike Fedyk
2001-09-20 12:57             ` Alan Cox
2001-09-20 13:40               ` Daniel Phillips
2001-09-24 22:50           ` Pavel Machek
2001-09-26 18:22             ` Marcelo Tosatti
2001-09-26 23:44               ` Pavel Machek
2001-09-27 13:52                 ` Eric W. Biederman
2001-10-01 11:37                 ` Marcelo Tosatti
2001-09-19 23:00         ` Rik van Riel
2001-09-21  8:23           ` Eric W. Biederman
2001-09-21 12:01             ` Rik van Riel
2001-09-22  2:14             ` Alexander Viro
2001-09-22  3:09               ` Rik van Riel
2001-09-21 14:29           ` Gábor Lénárt
2001-09-21 14:35             ` Horst von Brand
2001-09-19 21:37       ` Eric W. Biederman
2001-09-19 21:55       ` David S. Miller
2001-09-20 13:02         ` Rik van Riel
2001-09-16 19:07 vm rewrite ready [Re: broken VM in 2.4.10-pre9] Rik van Riel
2001-09-16 15:19 ` broken VM in 2.4.10-pre9 Phillip Susi
2001-09-16 19:33   ` Jeremy Zawodny
2001-09-16 19:52   ` Rik van Riel
     [not found] <fa.i95if5v.74un2p@ifi.uio.no>
     [not found] ` <fa.gu977tv.1b7u0g9@ifi.uio.no>
2001-09-16 18:06   ` Dan Maas
2001-09-15 22:43 Peter Magnusson
2001-09-15 23:50 ` Jan Harkes
2001-09-16  5:31 ` Linus Torvalds
2001-09-16  8:45   ` Eric W. Biederman
2001-09-17 10:25 ` Tonu Samuel
2001-09-16 16:47   ` Jeremy Zawodny
2001-09-16 18:36     ` Alan Cox
2001-09-16 19:38       ` Linus Torvalds
2001-09-16 19:37   ` Linus Torvalds
2001-09-17 14:04     ` Olaf Zaplinski

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=20010917122559Z16382-2758+129@humbolt.nl.linux.org \
    --to=phillips@bonn-fries.net \
    --cc=jaharkes@cs.cmu.edu \
    --cc=linux-kernel@vger.kernel.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

all inboxes | Powered by JetHome®