From owner-freebsd-arch@FreeBSD.ORG Sat May 10 12:14:45 2003 Return-Path: Delivered-To: freebsd-arch@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id D70CA37B401; Sat, 10 May 2003 12:14:45 -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 DBADB43FB1; Sat, 10 May 2003 12:13:53 -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 h4AJDq7d000293; Sat, 10 May 2003 12:13:52 -0700 (PDT) (envelope-from das@FreeBSD.ORG) Received: (from das@localhost) by HAL9000.homeunix.com (8.12.9/8.12.5/Submit) id h4AJDq77000292; Sat, 10 May 2003 12:13:52 -0700 (PDT) (envelope-from das@FreeBSD.ORG) Date: Sat, 10 May 2003 12:13:52 -0700 From: David Schultz To: Terry Lambert Message-ID: <20030510191352.GA225@HAL9000.homeunix.com> Mail-Followup-To: Terry Lambert , John Baldwin , arch@FreeBSD.ORG References: <20030510172609.GA29039@HAL9000.homeunix.com> <3EBD4214.73CB8B7C@mindspring.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3EBD4214.73CB8B7C@mindspring.com> cc: arch@FreeBSD.ORG cc: John Baldwin Subject: Re: [Bikeshed] sigacts locking X-BeenThere: freebsd-arch@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussion related to FreeBSD architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 10 May 2003 19:14:46 -0000 On Sat, May 10, 2003, Terry Lambert wrote: > David Schultz wrote: > > It occurs to me that this leaves very little in the uarea. You > > have a struct pstats, which is less than 256 bytes, and you have > > the kinfo_proc, which shouldn't need to be there anyway. Perhaps > > now would also be a good time to get rid of uarea swapping and the > > associated complexity altogether. > > The swapping of the uarea doesn't really introduce a lot of > extra complexity, since all it does is allocate swappable > pages, just like the swappable pages in user space. > > Change that mode stuff out of the uarea are probably a bad > idea, since it increases KVA pressure by moving them to > wired kernel pages. No, they're already wired kernel pages that are unwired and unmapped when the process is swapped out. Moreover, there's probably *more* KVA pressure with upages swapping, because each tiny struct upages gets a 4K or larger page all to itself unless it's swapped out.