From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 5170 invoked by alias); 13 May 2003 22:14:33 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 5159 invoked from network); 13 May 2003 22:14:33 -0000 Received: from unknown (HELO monty-python.gnu.org) (199.232.76.173) by sources.redhat.com with SMTP; 13 May 2003 22:14:33 -0000 Received: from port-212-202-172-137.reverse.qdsl-home.de ([212.202.172.137] helo=jackson.localnet) by monty-python.gnu.org with esmtp (Exim 4.10.13) id 19Fhmj-0008M0-00 for gcc@gcc.gnu.org; Tue, 13 May 2003 17:58:09 -0400 Received: from localhost ([127.0.0.1]) by jackson.localnet with esmtp (Exim 4.14) id 19Fhmp-00037s-KH; Tue, 13 May 2003 23:58:15 +0200 Date: Tue, 13 May 2003 22:14:00 -0000 Message-Id: <20030513.235807.838022001.rene.rebe@gmx.net> To: eager@eagercon.com Cc: pkoning@equallogic.com, drow@mvista.com, gcc@gcc.gnu.org, rock-linux@rocklinux.org Subject: Re: building a gcc-3.3-prerelease cross-compiler From: Rene Rebe In-Reply-To: <3EC156DF.2020201@eagercon.com> References: <3EC13D95.6080108@eagercon.com> <20030513.214707.412778972.rene.rebe@gmx.net> <3EC156DF.2020201@eagercon.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-Spam-Score: -26.7 (--------------------------) X-Scanner: exiscan for exim4 (http://duncanthrax.net/exiscan/) *19Fhmp-00037s-KH*zzYo3x6MuE.* X-SW-Source: 2003-05/txt/msg01374.txt.bz2 Hi, On: Tue, 13 May 2003 13:34:39 -0700, Michael Eager wrote: > Rene Rebe wrote: > > Could you elaborate on the problem that appears to be overlooked? >=20 > I occasionally have a need to build a toolchain for a new architecture > or new variant. After building binutils, the first step is to build > a bootstrap compiler. Once I have that compiler, I can build glibc. > After building glibc, I build a complete gcc. Both versions of gcc, > the bootstrap and the final, are built from the same sources. Yes we do so. > In your scheme, you assume that the bootstrap has been done. Also > the compiler you use to build glibc is potentially one revision back > from your final compiler. Nope. Our stage-0 is only creating the bootstrap toolchain, after that the glibc and other packages are build. > When you are working with one architecture, and repeatedly building > for that architecture, then this not a problem. But if you are building > tools for different architectures, or a new architecture, or you want to > make the entire process explicit, the bootstrap process should be explici= t. Yes - but this is how we do ;-) I even currently configure an regression tester to log and ensure those cross builds are working correctly. Might also be useful for the gcc folks, when I also log cvs HEAD checkouts. Sincerely, Ren=E9 Rebe - ROCK Linux stable release maintainer --=20=20 Ren=E9 Rebe - Europe/Germany/Berlin rene@rocklinux.org rene.rebe@gmx.net http://www.rocklinux.org http://www.rocklinux.org/people/rene=20=20=20=20= =20=20=20 http://gsmp.tfh-berlin.de/gsmp http://gsmp.tfh-berlin.de/rene