Hal Finkel <[email protected]> wrote on 31.10.2012 19:11:46:

> > > The change looks good. Should we add a testcase?
> >
> > Ah, good point.  Like this?
>
> Yep, that should work.

OK, checked in as revision 167148.

> As nbjoerg points out on IRC, these functions
> should probably have __always_inline__ also. If you look at
> xmmintrin.h, all of the functions look like this:
> static __inline__ __m128 __attribute__((__always_inline__, __nodebug__))
> _mm_div_ss(__m128 a, __m128 b)
> ...
>
> We should probably do that same.

Oh, they already have __always_inline__, it's part
of the __ATTRS_o_ai define:

#define __ATTRS_o_ai __attribute__((__overloadable__, __always_inline__))

Bye,
Ulrich

_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to