From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 13007 invoked by alias); 3 Aug 2002 19:28:07 -0000 Mailing-List: contact gcc-help-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-help-owner@gcc.gnu.org Received: (qmail 12998 invoked from network); 3 Aug 2002 19:28:06 -0000 Received: from unknown (HELO mail2.utc.com) (192.249.46.67) by sources.redhat.com with SMTP; 3 Aug 2002 19:28:06 -0000 Received: (from uucp@localhost) by mail2.utc.com (8.10.0/8.10.0) id g73JS5K08026 for ; Sat, 3 Aug 2002 15:28:05 -0400 (EDT) Received: from uusnwa0p.utc.com(159.82.80.106) by mail2.utc.com via csmap (V6.0) id srcAAAclayQp; Sat, 3 Aug 02 15:28:05 -0400 Received: from ronbo.sikorsky.com (ronbo.sikorsky.com [140.76.12.150]) by uusnwa0p.utc.com (Switch-2.2.0/Switch-2.2.0) with ESMTP id g73JS2r19460 for ; Sat, 3 Aug 2002 15:28:02 -0400 (EDT) Received: (from s29299@localhost) by ronbo.sikorsky.com (8.11.6+Sun/8.11.6) id g73JRxW10591 for gcc-help@gcc.gnu.org; Sat, 3 Aug 2002 15:27:59 -0400 (EDT) X-Authentication-Warning: ronbo.sikorsky.com: s29299 set sender to rmccall@sikorsky.com using -f Date: Sat, 03 Aug 2002 12:28:00 -0000 From: Ron McCall To: GCC Help Mailing List Subject: Re: Problem building Solaris hosted PowerPC EABI cross compiler Message-ID: <20020803152759.A10586@sikorsky.com> Mail-Followup-To: GCC Help Mailing List References: <20020801172743.A5998@sikorsky.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5.1i In-Reply-To: <20020801172743.A5998@sikorsky.com>; from rmccall@sikorsky.com on Thu, Aug 01, 2002 at 05:27:43PM -0400 X-SW-Source: 2002-08/txt/msg00019.txt.bz2 Following the instructions at the below URL, I was finally able to build and install a Solaris hosted PowerPC EABI targeted cross compiler tool chain (thanks Mumit!). This includes binutils 2.12.1, gcc 3.1.1 and newlib 1.10.0. It is interesting that these instructions seem to merely avoid running the particular test that was failing before when I followed Bill Gatliff's CrossGCC FAQ instructions. The resulting cross compiler still cannot build an executable by default. After having done more research into those undefined symbols, I found that the cross compiler does not by default link in any of the startup files nor the C library. I found that adding a compiler option such as -mads causes it to link in a particular set of startup files (and the C library) which resolves all those undefined symbols and that the result of the -mads option is controlled by the specs file. Now I can work on adding support for our board! http://www.nanotech.wisc.edu/~khan/software/gnu-win32/cygwin-to-newlib-cross-howto.txt Ron McCall