From owner-freebsd-amd64@FreeBSD.ORG Tue Mar 15 15:41:09 2005 Return-Path: Delivered-To: freebsd-amd64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 3A13C16A4CE for ; Tue, 15 Mar 2005 15:41:09 +0000 (GMT) Received: from mail.mcneil.com (mcneil.com [24.199.45.54]) by mx1.FreeBSD.org (Postfix) with ESMTP id BD3C043D3F for ; Tue, 15 Mar 2005 15:41:08 +0000 (GMT) (envelope-from sean@mcneil.com) Received: from localhost (localhost.mcneil.com [127.0.0.1]) by mail.mcneil.com (Postfix) with ESMTP id 69A8EF3CFE; Tue, 15 Mar 2005 07:41:06 -0800 (PST) Received: from mail.mcneil.com ([127.0.0.1]) by localhost (server.mcneil.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 35831-07; Tue, 15 Mar 2005 07:40:47 -0800 (PST) Received: from mcneil.com (mcneil.com [24.199.45.54]) by mail.mcneil.com (Postfix) with ESMTP id 631B5F3CFA; Tue, 15 Mar 2005 07:40:47 -0800 (PST) From: Sean McNeil To: Dag-Erling =?ISO-8859-1?Q?Sm=F8rgrav?= In-Reply-To: <86psy1krtz.fsf@xps.des.no> References: <200503150942.02118.kono@kth.se> <86psy1krtz.fsf@xps.des.no> Content-Type: text/plain; charset=ISO-8859-1 Date: Tue, 15 Mar 2005 07:40:46 -0800 Message-Id: <1110901246.78953.16.camel@server.mcneil.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.0 FreeBSD GNOME Team Port Content-Transfer-Encoding: 8bit X-Virus-Scanned: by amavisd-new at mcneil.com cc: kono@kth.se cc: freebsd-amd64@freebsd.org Subject: Re: make buildworld ERROR in RELENG_5 with WITH_LIB32 (amd64) X-BeenThere: freebsd-amd64@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the AMD64 platform List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 15 Mar 2005 15:41:09 -0000 On Tue, 2005-03-15 at 13:37 +0100, Dag-Erling Smørgrav wrote: > Sean McNeil writes: > > This has been discussed before. You are trying to compile with > > optimization -O2 and it isn't supported. > > It is supposed to be supported. The people who say it isn't are the > ones who keep introducing (and refusing to fix) bugs that break it. More accurately, there exist bugs that have not been fixed yet. These could have been there forever, or are introduced by a faulty compiler. Also, RELENG_5 is set up to compile as -O -pipe, not -O2 -fno-strict-aliasing -pipe. There is good reason for this IMHO. As long as the default settings are not -O2, then I would say it is not supported. > > You'll have to remove any > > custom settings of CFLAGS in your /etc/make.conf. Also, there are bugs > > in the current amd64 compiler version that prevent proper functionality > > at -O2. > > Please either document that claim, or retract it. I am not a maintainer, so I cannot insert any documentation to make this claim. You can, however, scan the -current list and possibly this one and discover several claims about it. Alexander's problem was actually that he used -O2 without using -fno-strict-aliasing. That is why he got his error. But I have personally encountered various SIGFPE signals with the tcl port when compiled with -O2 -fno-strict-aliasing -pipe. This first-hand knowledge plus various reports from other FreeBSD users gave me the courage to make this claim. Until such time as tcl 8.4 or gcc is updated so it works without issues on RELENG_5 when compiled with -O2, I will continue to make the claim that -O2 is broken on AMD64. Cheers, Sean