[patch] location for unary operators

[Andrew Pinski]

On Tue, Sep 30, 2008 at 1:38 PM, Aldy Hernandez <aldyh@redhat.com> wrote:
> Well, the main goal was adding location to unary operators, but there
> were some supporting things that needed to be fixed and cleaned up.
>
> As an aside, now build_unary_op and build_binary_op set the
> EXPR_LCOATION so the caller doesn't have to.
>
> Nothing major here, just the usual fixes for locations.
>
> Tested on x86-64 Linux.  GCC tests done.  GDB tests are still going.
>
> OK pending GDB tests?

And this broke PowerPC because of the use of build_unary_op and
build_indirect_ref in rs6000-c.c

The following patch fixes the issue.
Applied as obvious after a bootstrap/test on ppc64-linux-gnu with no
regressions.

Note resolve_overloaded_builtin should really take a location_t but
that I assume will be worked on with the patch to explicitly pass the
location_t for call_exprs.

Thanks,
Andrew Pinski

ChangeLog:
* config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
Update calls to build_unary_op and build_indirect_ref for location
changes.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: fixbootstrap.diff.txt
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20081006/da47ae21/attachment.txt>