From owner-freebsd-stable@FreeBSD.ORG Sun Nov 12 07:29:39 2006 Return-Path: X-Original-To: freebsd-stable@freebsd.org Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id AF22216A407 for ; Sun, 12 Nov 2006 07:29:39 +0000 (UTC) (envelope-from jcw@highperformance.net) Received: from mx1.highperformance.net (dsl081-163-122.sea1.dsl.speakeasy.net [64.81.163.122]) by mx1.FreeBSD.org (Postfix) with ESMTP id 7955C43D45 for ; Sun, 12 Nov 2006 07:29:38 +0000 (GMT) (envelope-from jcw@highperformance.net) Received: from [192.168.1.16] (w16.stradamotorsports.com [192.168.1.16]) by mx1.highperformance.net (8.13.6/8.13.4) with ESMTP id kAC6SV7r016065; Sat, 11 Nov 2006 22:28:31 -0800 (PST) (envelope-from jcw@highperformance.net) Message-ID: <4556BF10.9030002@highperformance.net> Date: Sat, 11 Nov 2006 22:28:32 -0800 From: "Jason C. Wells" User-Agent: Thunderbird 1.5.0.4 (Windows/20060516) MIME-Version: 1.0 To: "O. Hartmann" References: <4554D43E.5010700@highperformance.net> <20061110222434.GA76724@icarus.home.lan> <200611111019.43944.doconnor@gsoft.com.au> <45560E17.6070306@mail.zedat.fu-berlin.de> In-Reply-To: <45560E17.6070306@mail.zedat.fu-berlin.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Status: No, score=-4.4 required=2.5 tests=ALL_TRUSTED,BAYES_00 autolearn=ham version=3.1.0 X-Spam-Checker-Version: SpamAssassin 3.1.0 (2005-09-13) on s4.stradamotorsports.com Cc: freebsd-stable@freebsd.org Subject: Re: Compiler Options X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 12 Nov 2006 07:29:39 -0000 O. Hartmann wrote: > Daniel O'Connor wrote: > >> On Saturday 11 November 2006 08:54, Jeremy Chadwick wrote: >> >> >>> The kernel itself _will not_ use any SSE or MMX operations when built. >>> This is because these optimisations are known to break the FreeBSD >>> kernel. This applies to all i386 architectures, and probably 64-bit >>> architectures too (not sure). >>> >>> >> I think this is mainly because the kernel has no FPU context so you can't >> actually use any FPU operation (including SSE & MMX) without potentially >> trashing userland data. >> >> > > This is a good question to ask why. Still a relict from FreeBSDs BSD4.4 > legacy root? > >> Also, the cost of saving/restoring the context is quite high so potential >> benefits are largely negated. >> >> > > Well, this is often subject of lectures even at universities today and > even it is so, modern hardware design should be aware of those > disadvantages. > Ans this leads to the question whether this is in FreeBSD so because it > is really still a disadvantage (what I do not believe, even on i386 > hardware) or it is said to be so due to nobody has taken care about > that. As I remember myself, modern Linux kernels use MMX/SSE registers > even for purposes of getting an advantage in speed. > Peter Wemm once had this to say about that. http://lists.freebsd.org/pipermail/freebsd-amd64/2005-February/003663.html