On Fri, 21 Oct 2022 at 08:29, Jakub Jelinek <ja...@redhat.com> wrote: > > Hi! > > The following patch isn't for immediate commit, as it has several > dependencies, in particular: > https://gcc.gnu.org/pipermail/gcc-patches/2022-October/603665.html > https://gcc.gnu.org/pipermail/gcc-patches/2022-October/604080.html > https://gcc.gnu.org/pipermail/libstdc++/2022-October/054849.html > On top of those, this patch > 1) enables the std::float128_t overloads for x86 with glibc 2.26+ > 2) makes std::nextafter(std::float16_t, std::float16_t) and > std::nextafter(std::bfloat16_t, std::bfloat16_t) constexpr > 3) adds (small) testsuite coverage for that > > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk > if/when the above dependencies are in?
Yes, thanks. > > 2022-10-21 Jakub Jelinek <ja...@redhat.com> > > * config/os/gnu-linux/os_defines.h (_GLIBCXX_HAVE_FLOAT128_MATH): > Uncomment. > * include/c_global/cmath (nextafter(_Float16, _Float16)): Make it > constexpr. > If std::__is_constant_evaluated() call __builtin_nextafterf16. > (nextafter(__gnu_cxx::__bfloat16_t, __gnu_cxx::__bfloat16_t): > Similarly > but call __builtin_nextafterf16b. > * testsuite/26_numerics/headers/cmath/nextafter_c++23.cc (test): Add > static assertions to test constexpr nextafter. > > --- libstdc++-v3/config/os/gnu-linux/os_defines.h.jj 2022-10-18 > 11:35:55.514865483 +0200 > +++ libstdc++-v3/config/os/gnu-linux/os_defines.h 2022-10-20 > 16:57:59.715681664 +0200 > @@ -57,7 +57,7 @@ > || (defined(__powerpc__) && defined(_ARCH_PWR8) \ > && defined(__LITTLE_ENDIAN__) && (_CALL_ELF == 2) \ > && defined(__FLOAT128__))) > -//# define _GLIBCXX_HAVE_FLOAT128_MATH 1 > +# define _GLIBCXX_HAVE_FLOAT128_MATH 1 > #endif > > #if __GLIBC_PREREQ(2, 27) > --- libstdc++-v3/include/c_global/cmath.jj 2022-10-19 11:23:51.484488161 > +0200 > +++ libstdc++-v3/include/c_global/cmath 2022-10-20 17:03:56.760805581 +0200 > @@ -2755,9 +2755,11 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION > nearbyint(_Float16 __x) > { return _Float16(__builtin_nearbyintf(__x)); } > > - inline _Float16 > + constexpr _Float16 > nextafter(_Float16 __x, _Float16 __y) > { > + if (std::__is_constant_evaluated()) > + return __builtin_nextafterf16(__x, __y); > #ifdef __INT16_TYPE__ > using __float16_int_type = __INT16_TYPE__; > #else > @@ -3471,9 +3473,11 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION > nearbyint(__gnu_cxx::__bfloat16_t __x) > { return __gnu_cxx::__bfloat16_t(__builtin_nearbyintf(__x)); } > > - inline __gnu_cxx::__bfloat16_t > + constexpr __gnu_cxx::__bfloat16_t > nextafter(__gnu_cxx::__bfloat16_t __x, __gnu_cxx::__bfloat16_t __y) > { > + if (std::__is_constant_evaluated()) > + return __builtin_nextafterf16b(__x, __y); > #ifdef __INT16_TYPE__ > using __bfloat16_int_type = __INT16_TYPE__; > #else > --- libstdc++-v3/testsuite/26_numerics/headers/cmath/nextafter_c++23.cc.jj > 2022-10-20 16:57:29.940088318 +0200 > +++ libstdc++-v3/testsuite/26_numerics/headers/cmath/nextafter_c++23.cc > 2022-10-20 17:19:40.141923257 +0200 > @@ -100,6 +100,8 @@ test () > VERIFY( std::fpclassify(t36) == FP_NAN ); > T t37 = std::nextafter(T(-0.0), T()); > VERIFY( t37 == T() && !std::signbit(t37) ); > + static_assert(std::nextafter(T(1.0), T(2.0)) > T(1.0)); > + static_assert(std::nextafter(std::nextafter(T(1.0), T(5.0)), T(0.0)) == > T(1.0)); > } > > int > > Jakub >