| Message ID | 1490733875-20087-2-git-send-email-gftg@linux.vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
On Tue, 28 Mar 2017, Gabriel F. T. Gomes wrote: > The declarations of many functions in math_private.h are not required > since __MATHDECL and __MATHDECLX, in math.h, already provide the > declarations for these functions. This patch removes the declarations > from math_private.h. It also adds the inclusion of math.h to the files > which depended on the declaration of functions in math_private.h. > > Tested for powerpc64le and s390x. OK apart from the removal of the inline definition of __copysignl. That inline definition is still relevant in the absence of changing all calls to __copysign* to call copysign* directly so the compiler can inline them (and if changed to call copysign*, which I think would make sense, it would be desirable to run build-many-glibcs.py tests to make sure this doesn't introduce linknamespace issues).
On Wed, 29 Mar 2017 14:25:20 +0000 Joseph Myers <joseph@codesourcery.com> wrote: > OK apart from the removal of the inline definition of __copysignl. That > inline definition is still relevant in the absence of changing all calls > to __copysign* to call copysign* directly so the compiler can inline them > (and if changed to call copysign*, which I think would make sense, it > would be desirable to run build-many-glibcs.py tests to make sure this > doesn't introduce linknamespace issues). Thanks. Now committed without the removal of __copysignl (v3).
diff --git a/sysdeps/generic/math_private.h b/sysdeps/generic/math_private.h index be65b94..2d22552 100644 --- a/sysdeps/generic/math_private.h +++ b/sysdeps/generic/math_private.h @@ -318,32 +318,6 @@ extern long double __kernel_tanl (long double,long double,int); extern void __kernel_sincosl (long double,long double, long double *,long double *, int); -#ifndef NO_LONG_DOUBLE -/* prototypes required to compile the ldbl-96 support without warnings */ -extern int __finitel (long double); -extern int __ilogbl (long double); -extern int __isinfl (long double); -extern int __isnanl (long double); -extern long double __atanl (long double); -extern long double __copysignl (long double, long double); -extern long double __expm1l (long double); -extern long double __floorl (long double); -extern long double __frexpl (long double, int *); -extern long double __ldexpl (long double, int); -extern long double __log1pl (long double); -extern long double __nanl (const char *); -extern long double __rintl (long double); -extern long double __scalbnl (long double, int); -extern long double __sqrtl (long double x); -extern long double fabsl (long double x); -extern void __sincosl (long double, long double *, long double *); -extern long double __logbl (long double x); -extern long double __significandl (long double x); - -extern inline long double __copysignl (long double x, long double y) -{ return __builtin_copysignl (x, y); } - -#endif /* Prototypes for functions of the IBM Accurate Mathematical Library. */ extern double __exp1 (double __x, double __xx, double __error); diff --git a/sysdeps/ieee754/ldbl-128/e_logl.c b/sysdeps/ieee754/ldbl-128/e_logl.c index 78061a9..8672047 100644 --- a/sysdeps/ieee754/ldbl-128/e_logl.c +++ b/sysdeps/ieee754/ldbl-128/e_logl.c @@ -58,6 +58,7 @@ License along with this library; if not, see <http://www.gnu.org/licenses/>. */ +#include <math.h> #include <math_private.h> /* log(1+x) = x - .5 x^2 + x^3 l(x) diff --git a/sysdeps/ieee754/ldbl-128ibm/e_logl.c b/sysdeps/ieee754/ldbl-128ibm/e_logl.c index 14acfc2..c44feca 100644 --- a/sysdeps/ieee754/ldbl-128ibm/e_logl.c +++ b/sysdeps/ieee754/ldbl-128ibm/e_logl.c @@ -58,6 +58,7 @@ License along with this library; if not, see <http://www.gnu.org/licenses/>. */ +#include <math.h> #include <math_private.h> /* log(1+x) = x - .5 x^2 + x^3 l(x)