This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PING]^N Fix ppc64 bootstrap with a -m32 host compiler
> Because we deliberately only build the "build" directories once, for
> the "build" system. There's no code to bootstrap it. If build were
> treated as host stage 0, this would happen, but it isn't.
>
> The stage 1 compiler was used to build the stage 2 host libiberty,
> host gcc,
Ok.
> and "build" tools.
Blam! The toplevel stuff has failed us, then. If it changes
build_cc, it needs to arrange for a new build libiberty et al also,
either through building them, symlinks, or new make variables, so that
$build_cc finds the right build libraries.
I haven't convinced myself that using host libiberty for stage2's
build tools is wrong in our usual case, but I don't think it's right
for gcc's Makefile to be responsible for that. It confuses the
concept of TOPLEVEL bootstrap ;-)
No reason why the toplevel Makefile can't set $build_objdir to match
whatever build_cc it's passing.