From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Sat, 22 Sep 2001 16:47:02 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Sat, 22 Sep 2001 16:46:52 -0400 Received: from RAVEL.CODA.CS.CMU.EDU ([128.2.222.215]:34200 "EHLO ravel.coda.cs.cmu.edu") by vger.kernel.org with ESMTP id ; Sat, 22 Sep 2001 16:46:39 -0400 Date: Sat, 22 Sep 2001 16:46:47 -0400 To: Peter Magnusson Cc: linux-kernel@vger.kernel.org Subject: Re: broken VM in 2.4.10-pre9 Message-ID: <20010922164645.D15681@cs.cmu.edu> Mail-Followup-To: Peter Magnusson , linux-kernel@vger.kernel.org In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.20i From: Jan Harkes Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Sep 22, 2001 at 09:59:59PM +0200, Peter Magnusson wrote: > Jan Harkes wrote: > > Because pages aren't 'aged' until there is swap allocated for them, your > > kernel should actually work better if it has a lot of pages backed by > > swap. The only thing is, we don't really make the right decision about > > It doesnt. It just gets slower. > If it really become faster i would not have written my orginal posting. Only when the wrong pages have been swapped out and need to be swapped in again. Swapped out pages should be relatively inactive, the amount of swap space allocated number that you see in 'free' is not the same as the amount of pages that are actually swapped out and removed from memory, any active ones should still be around in the pagecache. > Use the swap as little as possible == good. Nope, Use the swap-in as little as possible == good, I don't mind having 4GB of data in swap, as long as I typically don't need to load it back into in memory. And every page that is in swap that I don't really need means another page that can be used to avoid paging out an executable, or purging the dentry lookup caches, or dropping one of those files I access once every 5 minutes. Jan