From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264315AbTLKB2s (ORCPT ); Wed, 10 Dec 2003 20:28:48 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264318AbTLKB2s (ORCPT ); Wed, 10 Dec 2003 20:28:48 -0500 Received: from holomorphy.com ([199.26.172.102]:1764 "EHLO holomorphy.com") by vger.kernel.org with ESMTP id S264315AbTLKB2r (ORCPT ); Wed, 10 Dec 2003 20:28:47 -0500 Date: Wed, 10 Dec 2003 17:28:17 -0800 From: William Lee Irwin III To: Andrea Arcangeli Cc: rl@hellgate.ch, Con Kolivas , Chris Vine , Rik van Riel , linux-kernel@vger.kernel.org, "Martin J. Bligh" Subject: Re: 2.6.0-test9 - poor swap performance on low end machines Message-ID: <20031211012817.GQ14258@holomorphy.com> Mail-Followup-To: William Lee Irwin III , Andrea Arcangeli , rl@hellgate.ch, Con Kolivas , Chris Vine , Rik van Riel , linux-kernel@vger.kernel.org, "Martin J. Bligh" References: <200311031148.40242.kernel@kolivas.org> <200311032113.14462.chris@cvine.freeserve.co.uk> <200311041355.08731.kernel@kolivas.org> <20031208135225.GT19856@holomorphy.com> <20031208194930.GA8667@k3.hellgate.ch> <20031208204817.GA19856@holomorphy.com> <20031210215235.GC11193@dualathlon.random> <20031210220525.GA28912@k3.hellgate.ch> <20031210224445.GE11193@dualathlon.random> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20031210224445.GE11193@dualathlon.random> Organization: The Domain of Holomorphy User-Agent: Mutt/1.5.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Dec 10, 2003 at 11:05:25PM +0100, Roger Luethi wrote: >> Also, the 2.6 core VM doesn't seem all that bad since it was introduced in >> 2.5.27 but most of the problems I measured were introduced after 2.5.40. >> Check out the graph I posted. On Wed, Dec 10, 2003 at 11:44:46PM +0100, Andrea Arcangeli wrote: > you're confusing rmap with core vm. rmap in no way can be defined as the > core vm, rmap is just a method used by the core vm to find some > information more efficiently at the expenses of all the fast paths > that now have to do the rmap bookkeeping. I've been maintaining one of the answers to this (anobjrmap, originally from hugh). I still haven't removed page->mapcount because keeping nr_mapped straight requires some care, though doing so should be feasible. I could probably use some helpers to untangle it from the highpmd, compile-time mapping->page_lock rwlock/spinlock switching, RCU mapping->i_shared_lock, and O(1) proc_pid_statm() bits. -- wli