[PATCH v7 1/3] math: redirect roundeven function

[H.J. Lu]

On Tue, May 25, 2021 at 5:02 AM H.J. Lu <hjl.tools@gmail.com> wrote:
>
> On Sun, May 23, 2021 at 6:43 PM Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> wrote:
> >
> > This patch redirect roundeven function for futhermore changes.
> >
> > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > ---
> >  include/math.h                             | 2 +-
> >  sysdeps/ieee754/dbl-64/s_roundeven.c       | 4 +++-
> >  sysdeps/ieee754/float128/s_roundevenf128.c | 1 +
> >  sysdeps/ieee754/flt-32/s_roundevenf.c      | 3 +++
> >  sysdeps/ieee754/ldbl-128/s_roundevenl.c    | 1 +
> >  sysdeps/ieee754/ldbl-96/s_roundevenl.c     | 1 +
> >  6 files changed, 10 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/math.h b/include/math.h
> > index e1c2a4eb64..b4772d3d3b 100644
> > --- a/include/math.h
> > +++ b/include/math.h
> > @@ -38,7 +38,6 @@ libm_hidden_proto (__issignaling)
> >  libm_hidden_proto (__issignalingf)
> >  libm_hidden_proto (__exp)
> >  libm_hidden_proto (__expf)
> > -libm_hidden_proto (__roundeven)
> >
> >  #  if !defined __NO_LONG_DOUBLE_MATH \
> >        && __LDOUBLE_REDIRECTS_TO_FLOAT128_ABI == 0
> > @@ -159,6 +158,7 @@ fabsf128 (_Float128 x)
> >  MATH_REDIRECT (sqrt, "__ieee754_", MATH_REDIRECT_UNARY_ARGS)
> >  MATH_REDIRECT (ceil, "__", MATH_REDIRECT_UNARY_ARGS)
> >  MATH_REDIRECT (floor, "__", MATH_REDIRECT_UNARY_ARGS)
> > +MATH_REDIRECT (roundeven, "__", MATH_REDIRECT_UNARY_ARGS)
> >  MATH_REDIRECT (rint, "__", MATH_REDIRECT_UNARY_ARGS)
> >  MATH_REDIRECT (trunc, "__", MATH_REDIRECT_UNARY_ARGS)
> >  MATH_REDIRECT (round, "__", MATH_REDIRECT_UNARY_ARGS)
> > diff --git a/sysdeps/ieee754/dbl-64/s_roundeven.c b/sysdeps/ieee754/dbl-64/s_roundeven.c
> > index 943b2c634c..5deff3bb8a 100644
> > --- a/sysdeps/ieee754/dbl-64/s_roundeven.c
> > +++ b/sysdeps/ieee754/dbl-64/s_roundeven.c
> > @@ -16,6 +16,7 @@
> >     License along with the GNU C Library; if not, see
> >     <https://www.gnu.org/licenses/>.  */
> >
> > +#define NO_MATH_REDIRECT
> >  #include <math.h>
> >  #include <math_private.h>
> >  #include <libm-alias-double.h>
> > @@ -66,5 +67,6 @@ __roundeven (double x)
> >    INSERT_WORDS64 (x, ix);
> >    return x;
> >  }
> > -hidden_def (__roundeven)
> > +#ifndef __roundeven
> >  libm_alias_double (__roundeven, roundeven)
> > +#endif
> > diff --git a/sysdeps/ieee754/float128/s_roundevenf128.c b/sysdeps/ieee754/float128/s_roundevenf128.c
> > index 5a9b3f395f..e0faf727f6 100644
> > --- a/sysdeps/ieee754/float128/s_roundevenf128.c
> > +++ b/sysdeps/ieee754/float128/s_roundevenf128.c
> > @@ -1,2 +1,3 @@
> > +#define NO_MATH_REDIRECT
> >  #include <float128_private.h>
> >  #include "../ldbl-128/s_roundevenl.c"
> > diff --git a/sysdeps/ieee754/flt-32/s_roundevenf.c b/sysdeps/ieee754/flt-32/s_roundevenf.c
> > index ae54365bbe..2cc0d7b001 100644
> > --- a/sysdeps/ieee754/flt-32/s_roundevenf.c
> > +++ b/sysdeps/ieee754/flt-32/s_roundevenf.c
> > @@ -17,6 +17,7 @@
> >     License along with the GNU C Library; if not, see
> >     <https://www.gnu.org/licenses/>.  */
> >
> > +#define NO_MATH_REDIRECT
> >  #include <math.h>
> >  #include <math_private.h>
> >  #include <libm-alias-float.h>
> > @@ -67,4 +68,6 @@ __roundevenf (float x)
> >    SET_FLOAT_WORD (x, ix);
> >    return x;
> >  }
> > +#ifndef __roundevenf
> >  libm_alias_float (__roundeven, roundeven)
> > +#endif
> > diff --git a/sysdeps/ieee754/ldbl-128/s_roundevenl.c b/sysdeps/ieee754/ldbl-128/s_roundevenl.c
> > index 02765be0e0..7c9c11fad2 100644
> > --- a/sysdeps/ieee754/ldbl-128/s_roundevenl.c
> > +++ b/sysdeps/ieee754/ldbl-128/s_roundevenl.c
> > @@ -17,6 +17,7 @@
> >     License along with the GNU C Library; if not, see
> >     <https://www.gnu.org/licenses/>.  */
> >
> > +#define NO_MATH_REDIRECT
> >  #include <math.h>
> >  #include <math_private.h>
> >  #include <libm-alias-ldouble.h>
> > diff --git a/sysdeps/ieee754/ldbl-96/s_roundevenl.c b/sysdeps/ieee754/ldbl-96/s_roundevenl.c
> > index 8340116ed8..f826829c31 100644
> > --- a/sysdeps/ieee754/ldbl-96/s_roundevenl.c
> > +++ b/sysdeps/ieee754/ldbl-96/s_roundevenl.c
> > @@ -17,6 +17,7 @@
> >     License along with the GNU C Library; if not, see
> >     <https://www.gnu.org/licenses/>.  */
> >
> > +#define NO_MATH_REDIRECT
> >  #include <math.h>
> >  #include <math_private.h>
> >  #include <libm-alias-ldouble.h>
> > --
> > 2.31.1
> >
>
> LGTM.
>
> Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
>

I tried it with scripts/build-many-glibcs.py.  It failed with targets using:

sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c
sysdeps/aarch64/fpu/s_roundevenf.c
sysdeps/aarch64/fpu/s_roundeven.c
sysdeps/riscv/rv64/rvd/s_roundeven.c
sysdeps/riscv/rvf/s_roundevenf.c

-- 
H.J.