AltiVec [RFH] Overload resolution of vec_subs intrinsic fails

[Paolo Bonzini]

> > Anyway, unsigned types are compatible with signed ones, aren't them?
>
> Not for __builtin_types_compatible_p.  

I meant that Richard's suggestion though desirable would not have fixed this
one bug.

> I agree that, for simd-1.c, the test should not expect an error on
> sign difference, so perhaps the problem is just the one use of
> comptypes() for __builtin_types_compatible_p.

What about the attached (untested) patch?

Sorry for the breakage.

Paolo

2003-04-10  Paolo Bonzini  <bonzini@gnu.org>

        * c-parse.in (primary): Use COMPARE_SAME_VALUES for
        __builtin_types_compatible_p.
        * c-tree.h (COMPARE_SAME_VALUES): New value of enum.
        * c-typeck.c (comptypes): Check for matching signedness if
        the flags are COMPARE_SAME_VALUES.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: fix-altivec.patch
Type: application/octet-stream
Size: 1760 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20040410/3e08ff6f/attachment.obj>