From owner-freebsd-current@FreeBSD.ORG Mon Apr 21 03:49:50 2003 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0497C37B404 for ; Mon, 21 Apr 2003 03:49:50 -0700 (PDT) Received: from HAL9000.homeunix.com (12-233-57-131.client.attbi.com [12.233.57.131]) by mx1.FreeBSD.org (Postfix) with ESMTP id E467B43FA3 for ; Mon, 21 Apr 2003 03:49:48 -0700 (PDT) (envelope-from das@freebsd.org) Received: from HAL9000.homeunix.com (localhost [127.0.0.1]) by HAL9000.homeunix.com (8.12.9/8.12.5) with ESMTP id h3LAnljC003717; Mon, 21 Apr 2003 03:49:47 -0700 (PDT) (envelope-from das@freebsd.org) Received: (from das@localhost) by HAL9000.homeunix.com (8.12.9/8.12.5/Submit) id h3LAnWbk003712; Mon, 21 Apr 2003 03:49:32 -0700 (PDT) (envelope-from das@freebsd.org) Date: Mon, 21 Apr 2003 03:49:32 -0700 From: David Schultz To: Matthew Dillon Message-ID: <20030421104932.GA3658@HAL9000.homeunix.com> Mail-Followup-To: Matthew Dillon , freebsd-current@freebsd.org References: <000501c30682$4e5e64b0$6601a8c0@VAIO650> <20030420002940.GB46590@HAL9000.homeunix.com> <20030420191744.G19683@gamplex.bde.org> <20030420101401.GA2821@HAL9000.homeunix.com> <200304201828.h3KISlKq090099@apollo.backplane.com> <20030420191029.GA4803@HAL9000.homeunix.com> <200304201924.h3KJOxWo090302@apollo.backplane.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200304201924.h3KJOxWo090302@apollo.backplane.com> cc: freebsd-current@freebsd.org Subject: Re: Broken memory management on system with no swap X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Apr 2003 10:49:50 -0000 On Sun, Apr 20, 2003, Matthew Dillon wrote: > note (1): normally act_count protects against thrashing. It is the > active queue's act_count algorithm which gives FreeBSD's such a nice > smooth degredation curve when memory loads become extreme by preventing > a frequently accessed page from being freed too early, so we don't > want to just turn it off. Maybe we need a test for 'too many active > pages', aka when > 80% of available pages are in the active queue > to temporarily disable the act_count test. Another possibility is to increase act_count's decay constant ACT_DECLINE in proportion to the size of the active queue. However, I don't know whether six bits would provide enough resolution for this strategy to have any advantage over your proposal.