This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc-3.2.2 for armv5b - big endian not the default?
On Fri, Mar 07, 2003 at 10:20:38AM +0000, Richard Earnshaw wrote:
> >
> > hi,
> >
> > i'm building an armv5b cross toolchain, with the flags below, and
> > have found something odd. when i specify that i want the result
> > to be armv5b, hence big-endian, it seems that gcc is trying to link
> > as little-endian. it _builds_ correctly, but unless i specify the
> > '-mbig-endian' flag to everything... things just won't link.
> >
> > did i build it incorrectly or something? or is binutils not built
> > correctly? all the flags for building the toolchains are below, as
> > well as the broken output during a compile.
>
> You are not building a big-endian tool chain, you are building a
> little-endian one. There are no GCC/arm configurations that generate
> big-endian by default to my knowledge: the arm*-*-linux* configuration is
> certainly little-endian.
>
> Gcc doesn't take any account of letters in the cpu part of the
> configuration name after "arm", the rest is just window dressing.
I've got patches for a big-endian ARM/Linux port; I've just been lax in
submitting them for months now. I'll update them to HEAD and submit
them later today. It's just a matter of configury.
--
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer