Re: AltiVec [RFH] Overload resolution of vec_subs intrinsic fails

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Apr 8, 2004, at 20:33, Andrew Pinski wrote:

> On Apr 8, 2004, at 20:26, Fariborz Jahanian wrote:
>
> > In the following test case, gcc3.5, etc. incorrectly generate 
> > 'vsubuhs' instead of the
> > signed variety. It is because overload resolution of vec_subs selects
> > __builtin_altivec_vsubuhs, instead of __builtin_altivec_vsubshs. Our 
> > AltiVec
> > front-end guy is out on vacation and I appreciate help in resolving 
> > this issue. I
> > have spent enough time on this that I think I need to ask some one.
> >
> > - Thanks, Fariborz (fjahanian@apple.com)
> >
> > Test case:
> >
> > #include <altivec.h>
> > int main()
> > {
> >         __vector signed short  res;
> >         __vector signed short t8;
> >         __vector signed short t9;
> >         res = vec_subs(t8, t9);
> > }
>
> This is because __builtin_types_compatible_p (__vector bool short, 
> __vector signed short) is
> returning 1.

I forgot to mention here is the test case I used and say that it said 1:
int f()
{
  return __builtin_types_compatible_p (__vector bool short, __vector 
signed short);
}


Thanks,
Andrew Pinski