This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [gfortran,patch] bootstrap broken
- From: Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>
- To: Feng Wang <wf_cs at yahoo dot com>
- Cc: Hans-Peter Nilsson <hp at bitrange dot com>,Daniel Jacobowitz <drow at false dot org>, fortran <fortran at gcc dot gnu dot org>,patch <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 17 Sep 2004 18:35:54 +0200
- Subject: Re: [gfortran,patch] bootstrap broken
- References: <20040917152219.89605.qmail@web15604.mail.cnb.yahoo.com>
Feng Wang wrote:
>
> I have to say that I *did not* do cross-compile and the fortran did be
> compilered by system gcc. This is what I did:
>
> [ Build logs ]
>
> Please note the compiler is gcc, not *xgcc*. I think it is the system gcc.
>
> If it should be xgcc from the same tree, there must be something wrong in
> config or makefile.
>
> p.s. Indeed the libgfortran is compilerd by xgcc.
>
Indeed, my original assertion was wrong. The difference between libgfortran
and f951 is that libgfortran is code for the target, and f951 is code for the
host. The former needs to be compiled by a compiler for the target (=xgcc),
the latter by a compiler for the host (=system compiler in a non-bootstrapped
build). Of course in a native compiler this distinction makes no sense, but
I'm not volunteering to change the build machinery.
So indeed, we need to fix those appearances of code which doesn't compile with
GCC 2.95 (the documented minimum for building a cross-compiler).
Thanks to the people who explained this to me off-list.
- Tobi