Re: bootstrap/7589: libgcc.map file is empty; ld dies

[Phil Edwards <phil@jaj.com>]

The following reply was made to PR bootstrap/7589; it has been noted by GNATS.

From: Phil Edwards <phil@jaj.com>
To: gaedba@attbi.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: bootstrap/7589: libgcc.map file is empty; ld dies
Date: Tue, 13 Aug 2002 15:01:30 -0400

 On Tue, Aug 13, 2002 at 06:11:20PM -0000, gaedba@attbi.com wrote:
 > I started with a working gcc 2.96.  3.0.1 didn't bootstrap
 > until I configured it --disable-shared --enable-languages=c.
 > Then I tried 3.0.4, same problem, same workaround.
 > Then I tried 3.1, same problem, same workaround.
 > Now I'm trying 3.1.1.  I'll try --disable-shared on that.
 > I really want a working C++, not just C.
 
 Nothing comes to mind.  This must be some kind of local configuration error;
 people have been successfully building all those versions on your platform
 for quite some time.
 
 
 > I'm having the same sort of problem on Solaris.  Only on 
 > Solaris the .map file isn't empty; it doesn't parse correctly
 > for some reason.  Or, depending on configuration, it gets to a
 > place where it runs ./cc1 .... -o /tmp/xxxx.s and then ld
 > complains that /tmp/xxxx.s is an unknown file type and
 > isn't a valid linker command script.  Very strange.
 
 This, on the other hand, happens when GCC has been configured to use
 the native Sun binutils but finds the GNU ones at runtime instead.
 (Or vice versa, depending on the error message.)  If you're using
 the GNU linker/assembler/whatnot, then --with-gnu-as, --with-gnu-ld,
 --with-as=... and --with-ld=...  should all be used.
 
 
 Phil
 
 -- 
 I would therefore like to posit that computing's central challenge, viz. "How
 not to make a mess of it," has /not/ been met.
                                                  - Edsger Dijkstra, 1930-2002