From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261195AbUBWCCy (ORCPT ); Sun, 22 Feb 2004 21:02:54 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261227AbUBWCCy (ORCPT ); Sun, 22 Feb 2004 21:02:54 -0500 Received: from mail-03.iinet.net.au ([203.59.3.35]:36823 "HELO mail.iinet.net.au") by vger.kernel.org with SMTP id S261195AbUBWCCw (ORCPT ); Sun, 22 Feb 2004 21:02:52 -0500 Message-ID: <40395F49.1010702@cyberone.com.au> Date: Mon, 23 Feb 2004 13:02:49 +1100 From: Nick Piggin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040122 Debian/1.6-1 X-Accept-Language: en MIME-Version: 1.0 To: Andrew Morton CC: linux-kernel@vger.kernel.org Subject: Re: 2.6.3-mm3 References: <20040222172200.1d6bdfae.akpm@osdl.org> <40395ACE.4030203@cyberone.com.au> <20040222175507.558a5b3d.akpm@osdl.org> In-Reply-To: <20040222175507.558a5b3d.akpm@osdl.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Andrew Morton wrote: >Nick Piggin wrote: > >> >> >>Andrew Morton wrote: >> >> >>>ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.3/2.6.3-mm2/ >>> >>> >>> >>URL is of course, >>ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.3/2.6.3-mm3/ >> > >Yes, thanks. > > >>This still doesn't shrink slab correctly on highmem machines >>because you dropped my patch :( >> > >First, one needs to define "correctly". > >Certainly, it is not "solves the alleged updatedb problem". > > No, I think this is a non-problem. >The design behind the slab shrinking is to reclaim slab in response to >memory demand. Not in response to lowmem demand. With all the scaling, >accounting-for-seeks-and-locality, etc. > > That should come out in the wash with my patch anyway, because it causes lowmem LRU pressure to assert a *lot* more slab pressure. So highmem pressure should cause a similar amount of slab pressure with either patch, it just comes about in different ways. But allocations from lowmem will not shrink the slab nearly enough with your patch because it shrinks by a percent of all pages.