[glibc] powerpc: Update acosf ulps

[Paul Zimmermann]

thank you Adhemerval, this answers well my questions.
Ok to commit on my side.

Paul

> Date: Thu, 2 Jan 2025 11:07:46 -0300
> Cc: amonakov@ispras.ru, fweimer@redhat.com, libc-alpha@sourceware.org
> From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
> Organization: Linaro
> 
> 
> 
> On 02/01/25 11:05, Paul Zimmermann wrote:
> >        Hi Adhemerval,
> > 
> >> This seems to show no regression on gcc 11 and gcc 13 on x86_64, i686, aarch64,
> >> and powerpc64le:
> >>
> >> diff --git a/sysdeps/ieee754/flt-32/e_acosf.c b/sysdeps/ieee754/flt-32/e_acosf.c
> >> index cba01221dc..35e241f6c4 100644
> >> --- a/sysdeps/ieee754/flt-32/e_acosf.c
> >> +++ b/sysdeps/ieee754/flt-32/e_acosf.c
> >> @@ -28,6 +28,7 @@ SOFTWARE.
> >>  #include <math.h>
> >>  #include <math_private.h>
> >>  #include <libm-alias-finite.h>
> >> +#include <math-barriers.h>
> >>  #include "math_config.h"
> >>
> >>  static __attribute__ ((noinline)) float
> >> @@ -66,7 +67,7 @@ poly12 (double z, const double *c)
> >>  float
> >>  __ieee754_acosf (float x)
> >>  {
> >> -  const double pi2 = 0x1.921fb54442d18p+0;
> >> +  double pi2 = 0x1.921fb54442d18p+0;
> >>    static const double o[] = { 0, 0x1.921fb54442d18p+1 };
> >>    double xs = x;
> >>    double r;
> >> @@ -87,7 +88,10 @@ __ieee754_acosf (float x)
> >>         };
> >>        /* Avoid spurious underflow exception.  */
> >>        if (__glibc_unlikely (ax <= 0x40000000u)) /* |x| < 2^-63 */
> >> -       return (float) pi2;
> >> +       /* GCC <= 11 wrongly assumes the rounding is to nearest and
> >> +          performs a constant folding here:
> >> +          https://gcc.gnu.org/bugzilla/show_bug.cgi?id=57245 */
> >> +       return math_opt_barrier (pi2);
> >>        double z = xs;
> >>        double z2 = z * z;
> >>        double z4 = z2 * z2;
> > 
> > this looks good to me, with two questions:
> > 
> > 1) why did you remove "const" in pi2 ? This does not work with const ?
> 
> Because gcc complains due how math_opt_barrier is implemented:
> 
> ../sysdeps/ieee754/flt-32/e_acosf.c: In function ‘__ieee754_acosf’:
> ../sysdeps/aarch64/fpu/math-barriers.h:23:30: error: read-only variable ‘__x’ used as ‘asm’ output
>    23 |   ({ __typeof (x) __x = (x); __asm ("" : "+w" (__x)); __x; })
>       |                              ^~~~~
> ../sysdeps/ieee754/flt-32/e_acosf.c:94:16: note: in expansion of macro ‘math_opt_barrier’
>    94 |         return math_opt_barrier (pi2);
>       |                ^~~~~~~~~~~~~~~~
> 
> > 
> > 2) does math_opt_barrier() return a float? Or is the return value cast
> >    to float by the return?
> 
> The later, since math_opt_barrier aims to be type agnostic by using 
> __typeof.
> 
> > 
> > Paul
> 
>