Changes in the v8 patch: 1. Redirect target specific roundeven functions for aarch64, ldbl-128ibm and riscv to fix build for aarch64, power and riscv. Tested on i686 and x86-64. Passed build-many-glibcs.py. I will checked them in on Friday if there are no objections. H.J. --- H.J. Lu (1): Update math: redirect roundeven function Shen-Ta Hsieh (3): math: redirect roundeven function x86_64: roundeven with sse4.1 support Use GCC builtins for roundeven functions if desired. include/math.h | 2 +- sysdeps/aarch64/fpu/s_roundeven.c | 2 +- sysdeps/aarch64/fpu/s_roundevenf.c | 1 + sysdeps/generic/math-use-builtins-roundeven.h | 4 +++ sysdeps/generic/math-use-builtins.h | 1 + sysdeps/ieee754/dbl-64/s_roundeven.c | 9 +++++- sysdeps/ieee754/float128/float128_private.h | 2 ++ sysdeps/ieee754/float128/s_roundevenf128.c | 1 + sysdeps/ieee754/flt-32/s_roundevenf.c | 8 +++++ sysdeps/ieee754/ldbl-128/s_roundevenl.c | 6 ++++ sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c | 1 + sysdeps/ieee754/ldbl-96/s_roundevenl.c | 1 + sysdeps/riscv/rv64/rvd/s_roundeven.c | 2 +- sysdeps/riscv/rvf/s_roundevenf.c | 1 + sysdeps/x86_64/fpu/multiarch/Makefile | 5 +-- sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++ .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++ .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++ 21 files changed, 155 insertions(+), 6 deletions(-) create mode 100644 sysdeps/generic/math-use-builtins-roundeven.h create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c -- 2.31.1
From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> This patch redirect roundeven function for futhermore changes. Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> Reviewed-by: H.J. Lu <hjl.tools@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
Redirect target specific roundeven functions for aarch64, ldbl-128ibm and riscv. --- sysdeps/aarch64/fpu/s_roundeven.c | 2 +- sysdeps/aarch64/fpu/s_roundevenf.c | 1 + sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c | 1 + sysdeps/riscv/rv64/rvd/s_roundeven.c | 2 +- sysdeps/riscv/rvf/s_roundevenf.c | 1 + 5 files changed, 5 insertions(+), 2 deletions(-) diff --git a/sysdeps/aarch64/fpu/s_roundeven.c b/sysdeps/aarch64/fpu/s_roundeven.c index d74b40daf5..fbce9aaded 100644 --- a/sysdeps/aarch64/fpu/s_roundeven.c +++ b/sysdeps/aarch64/fpu/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 <libm-alias-double.h> @@ -25,5 +26,4 @@ __roundeven (double x) asm volatile ("frintn \t%d0, %d1" : "=w" (x) : "w" (x)); return x; } -hidden_def (__roundeven) libm_alias_double (__roundeven, roundeven) diff --git a/sysdeps/aarch64/fpu/s_roundevenf.c b/sysdeps/aarch64/fpu/s_roundevenf.c index dfc492c2f8..7985ca5f30 100644 --- a/sysdeps/aarch64/fpu/s_roundevenf.c +++ b/sysdeps/aarch64/fpu/s_roundevenf.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 <libm-alias-float.h> diff --git a/sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c b/sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c index 6701970f4a..90eecf496b 100644 --- a/sysdeps/ieee754/ldbl-128ibm/s_roundevenl.c +++ b/sysdeps/ieee754/ldbl-128ibm/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> diff --git a/sysdeps/riscv/rv64/rvd/s_roundeven.c b/sysdeps/riscv/rv64/rvd/s_roundeven.c index e77d8307d1..02dd0f7a9a 100644 --- a/sysdeps/riscv/rv64/rvd/s_roundeven.c +++ b/sysdeps/riscv/rv64/rvd/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 <fenv_private.h> #include <libm-alias-double.h> @@ -49,5 +50,4 @@ __roundeven (double x) return x; } -hidden_def (__roundeven) libm_alias_double (__roundeven, roundeven) diff --git a/sysdeps/riscv/rvf/s_roundevenf.c b/sysdeps/riscv/rvf/s_roundevenf.c index fa594d46bd..be22d047e2 100644 --- a/sysdeps/riscv/rvf/s_roundevenf.c +++ b/sysdeps/riscv/rvf/s_roundevenf.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 <fenv_private.h> #include <libm-alias-float.h> -- 2.31.1
From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> This patch adds support for the sse4.1 hardware floating point roundeven. Here is some benchmark results on my systems: =AMD Ryzen 9 3900X 12-Core Processor= * benchmark result before this commit | | roundeven | roundevenf | |------------|--------------|--------------| | duration | 3.75587e+09 | 3.75114e+09 | | iterations | 3.93053e+08 | 4.35402e+08 | | max | 52.592 | 58.71 | | min | 7.98 | 7.22 | | mean | 9.55563 | 8.61535 | * benchmark result after this commit | | roundeven | roundevenf | |------------|---------------|--------------| | duration | 3.73815e+09 | 3.73738e+09 | | iterations | 5.82692e+08 | 5.91498e+08 | | max | 56.468 | 51.642 | | min | 6.27 | 6.156 | | mean | 6.41532 | 6.3185 | =Intel(R) Pentium(R) CPU D1508 @ 2.20GHz= * benchmark result before this commit | | roundeven | roundevenf | |------------|--------------|--------------| | duration | 2.18208e+09 | 2.18258e+09 | | iterations | 2.39932e+08 | 2.46924e+08 | | max | 96.378 | 98.035 | | min | 6.776 | 5.94 | | mean | 9.09456 | 8.83907 | * benchmark result after this commit | | roundeven | roundevenf | |------------|--------------|--------------| | duration | 2.17415e+09 | 2.17005e+09 | | iterations | 3.56193e+08 | 4.09824e+08 | | max | 51.693 | 97.192 | | min | 5.926 | 5.093 | | mean | 6.10385 | 5.29507 | Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> Reviewed-by: H.J. Lu <hjl.tools@gmail.com> --- sysdeps/x86_64/fpu/multiarch/Makefile | 5 +-- sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++ .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++ .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++ sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++ 7 files changed, 118 insertions(+), 2 deletions(-) create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile index 57892c56bb..d425ffd6d3 100644 --- a/sysdeps/x86_64/fpu/multiarch/Makefile +++ b/sysdeps/x86_64/fpu/multiarch/Makefile @@ -1,11 +1,12 @@ ifeq ($(subdir),math) libm-sysdep_routines += s_floor-c s_ceil-c s_floorf-c s_ceilf-c \ s_rint-c s_rintf-c s_nearbyint-c s_nearbyintf-c \ - s_trunc-c s_truncf-c + s_roundeven-c s_roundevenf-c s_trunc-c s_truncf-c libm-sysdep_routines += s_ceil-sse4_1 s_ceilf-sse4_1 s_floor-sse4_1 \ s_floorf-sse4_1 s_nearbyint-sse4_1 \ - s_nearbyintf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \ + s_nearbyintf-sse4_1 s_roundeven-sse4_1 \ + s_roundevenf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \ s_trunc-sse4_1 s_truncf-sse4_1 libm-sysdep_routines += e_exp-fma e_log-fma e_pow-fma s_atan-fma \ diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c new file mode 100644 index 0000000000..c7be43cb22 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c @@ -0,0 +1,2 @@ +#define __roundeven __roundeven_c +#include <sysdeps/ieee754/dbl-64/s_roundeven.c> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S new file mode 100644 index 0000000000..6ae8f6b1d3 --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S @@ -0,0 +1,24 @@ +/* Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#include <sysdep.h> + + .section .text.sse4.1,"ax",@progbits +ENTRY(__roundeven_sse41) + roundsd $8, %xmm0, %xmm0 + ret +END(__roundeven_sse41) diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c new file mode 100644 index 0000000000..d92eda652a --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c @@ -0,0 +1,31 @@ +/* Multiple versions of __roundeven. + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#include <libm-alias-double.h> + +#define roundeven __redirect_roundeven +#define __roundeven __redirect___roundeven +#include <math.h> +#undef roundeven +#undef __roundeven + +#define SYMBOL_NAME roundeven +#include "ifunc-sse4_1.h" + +libc_ifunc_redirected (__redirect_roundeven, __roundeven, IFUNC_SELECTOR ()); +libm_alias_double (__roundeven, roundeven) diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c new file mode 100644 index 0000000000..72a6e7d1fb --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c @@ -0,0 +1,3 @@ +#undef __roundevenf +#define __roundevenf __roundevenf_c +#include <sysdeps/ieee754/flt-32/s_roundevenf.c> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S new file mode 100644 index 0000000000..a76e10807e --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S @@ -0,0 +1,24 @@ +/* Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#include <sysdep.h> + + .section .text.sse4.1,"ax",@progbits +ENTRY(__roundevenf_sse41) + roundss $8, %xmm0, %xmm0 + ret +END(__roundevenf_sse41) diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c new file mode 100644 index 0000000000..2ee196e68f --- /dev/null +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c @@ -0,0 +1,31 @@ +/* Multiple versions of __roundevenf. + Copyright (C) 2021 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + <https://www.gnu.org/licenses/>. */ + +#include <libm-alias-float.h> + +#define roundevenf __redirect_roundevenf +#define __roundevenf __redirect___roundevenf +#include <math.h> +#undef roundevenf +#undef __roundevenf + +#define SYMBOL_NAME roundevenf +#include "ifunc-sse4_1.h" + +libc_ifunc_redirected (__redirect_roundevenf, __roundevenf, IFUNC_SELECTOR ()); +libm_alias_float (__roundeven, roundeven) -- 2.31.1
From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> This patch is using the corresponding GCC builtin for roundevenf, roundeven and roundevenl if the USE_FUNCTION_BUILTIN macros are defined to one in math-use-builtins.h. These builtin functions is supported since GCC 10. The code of the generic implementation is not changed. Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> Reviewed-by: H.J. Lu <hjl.tools@gmail.com> --- sysdeps/generic/math-use-builtins-roundeven.h | 4 ++++ sysdeps/generic/math-use-builtins.h | 1 + sysdeps/ieee754/dbl-64/s_roundeven.c | 5 +++++ sysdeps/ieee754/float128/float128_private.h | 2 ++ sysdeps/ieee754/flt-32/s_roundevenf.c | 5 +++++ sysdeps/ieee754/ldbl-128/s_roundevenl.c | 5 +++++ 6 files changed, 22 insertions(+) create mode 100644 sysdeps/generic/math-use-builtins-roundeven.h diff --git a/sysdeps/generic/math-use-builtins-roundeven.h b/sysdeps/generic/math-use-builtins-roundeven.h new file mode 100644 index 0000000000..bd2e84c93f --- /dev/null +++ b/sysdeps/generic/math-use-builtins-roundeven.h @@ -0,0 +1,4 @@ +#define USE_ROUNDEVEN_BUILTIN 0 +#define USE_ROUNDEVENF_BUILTIN 0 +#define USE_ROUNDEVENL_BUILTIN 0 +#define USE_ROUNDEVENF128_BUILTIN 0 diff --git a/sysdeps/generic/math-use-builtins.h b/sysdeps/generic/math-use-builtins.h index cc3e9f2201..19d2d1cf3c 100644 --- a/sysdeps/generic/math-use-builtins.h +++ b/sysdeps/generic/math-use-builtins.h @@ -30,6 +30,7 @@ #include <math-use-builtins-ceil.h> #include <math-use-builtins-trunc.h> #include <math-use-builtins-round.h> +#include <math-use-builtins-roundeven.h> #include <math-use-builtins-copysign.h> #include <math-use-builtins-sqrt.h> #include <math-use-builtins-fma.h> diff --git a/sysdeps/ieee754/dbl-64/s_roundeven.c b/sysdeps/ieee754/dbl-64/s_roundeven.c index 5deff3bb8a..88c051e9fe 100644 --- a/sysdeps/ieee754/dbl-64/s_roundeven.c +++ b/sysdeps/ieee754/dbl-64/s_roundeven.c @@ -21,6 +21,7 @@ #include <math_private.h> #include <libm-alias-double.h> #include <stdint.h> +#include <math-use-builtins.h> #define BIAS 0x3ff #define MANT_DIG 53 @@ -29,6 +30,9 @@ double __roundeven (double x) { +#if USE_ROUNDEVEN_BUILTIN + return __builtin_roundeven (x); +#else uint64_t ix, ux; EXTRACT_WORDS64 (ix, x); ux = ix & 0x7fffffffffffffffULL; @@ -66,6 +70,7 @@ __roundeven (double x) ix &= 0x8000000000000000ULL; INSERT_WORDS64 (x, ix); return x; +#endif /* ! USE_ROUNDEVEN_BUILTIN */ } #ifndef __roundeven libm_alias_double (__roundeven, roundeven) diff --git a/sysdeps/ieee754/float128/float128_private.h b/sysdeps/ieee754/float128/float128_private.h index 979138851b..b6b6d3d5fd 100644 --- a/sysdeps/ieee754/float128/float128_private.h +++ b/sysdeps/ieee754/float128/float128_private.h @@ -152,6 +152,8 @@ #define USE_TRUNCL_BUILTIN USE_TRUNCF128_BUILTIN #undef USE_ROUNDL_BUILTIN #define USE_ROUNDL_BUILTIN USE_ROUNDF128_BUILTIN +#undef USE_ROUNDEVENL_BUILTIN +#define USE_ROUNDEVENL_BUILTIN USE_ROUNDEVENF128_BUILTIN #undef USE_COPYSIGNL_BUILTIN #define USE_COPYSIGNL_BUILTIN USE_COPYSIGNF128_BUILTIN #undef USE_FMAL_BUILTIN diff --git a/sysdeps/ieee754/flt-32/s_roundevenf.c b/sysdeps/ieee754/flt-32/s_roundevenf.c index 2cc0d7b001..4f568579d8 100644 --- a/sysdeps/ieee754/flt-32/s_roundevenf.c +++ b/sysdeps/ieee754/flt-32/s_roundevenf.c @@ -21,6 +21,7 @@ #include <math.h> #include <math_private.h> #include <libm-alias-float.h> +#include <math-use-builtins.h> #include <stdint.h> #define BIAS 0x7f @@ -30,6 +31,9 @@ float __roundevenf (float x) { +#if USE_ROUNDEVENF_BUILTIN + return __builtin_roundevenf (x); +#else uint32_t ix, ux; GET_FLOAT_WORD (ix, x); ux = ix & 0x7fffffff; @@ -67,6 +71,7 @@ __roundevenf (float x) ix &= 0x80000000; SET_FLOAT_WORD (x, ix); return x; +#endif /* ! USE_ROUNDEVENF_BUILTIN */ } #ifndef __roundevenf libm_alias_float (__roundeven, roundeven) diff --git a/sysdeps/ieee754/ldbl-128/s_roundevenl.c b/sysdeps/ieee754/ldbl-128/s_roundevenl.c index 7c9c11fad2..afbf13d4f2 100644 --- a/sysdeps/ieee754/ldbl-128/s_roundevenl.c +++ b/sysdeps/ieee754/ldbl-128/s_roundevenl.c @@ -21,6 +21,7 @@ #include <math.h> #include <math_private.h> #include <libm-alias-ldouble.h> +#include <math-use-builtins.h> #include <stdint.h> #define BIAS 0x3fff @@ -30,6 +31,9 @@ _Float128 __roundevenl (_Float128 x) { +#if USE_ROUNDEVENL_BUILTIN + return __builtin_roundevenl (x); +#else uint64_t hx, lx, uhx; GET_LDOUBLE_WORDS64 (hx, lx, x); uhx = hx & 0x7fffffffffffffffULL; @@ -101,5 +105,6 @@ __roundevenl (_Float128 x) } SET_LDOUBLE_WORDS64 (x, hx, lx); return x; +#endif /* ! USE_ROUNDEVENL_BUILTIN */ } libm_alias_ldouble (__roundeven, roundeven) -- 2.31.1
s390x fails to build. gcc ../sysdeps/s390/fpu/s_roundevenl.c -c -std=gnu11 -fgnu89-inline -O2 -g -m64 -fmessage-length=0 -funwind-tables -fasynchronous-unwind-tables -Wall -Wwrite-strings -Wundef -Werror -fmerge-all-constants -frounding-math -fstack-protector -Wstrict-prototypes -Wold-style-definition -fno-math-errno -mlong-double-128 -I../include -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base -I../sysdeps/unix/sysv/linux/s390/s390-64 -I../sysdeps/unix/sysv/linux/wordsize-64 -I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu -I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl -I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt -I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux -I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/s390/s390-64 -I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch -I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I. -D_LIBC_REENTRANT -include /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/libc-modules.h -DMODULE_NAME=libm -include ../include/libc-symbols.h -DPIC -DTOP_NAMESPACE=glibc -o /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o -MD -MP -MF /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o.dt -MT /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o /tmp/ccYVXAks.s: Assembler messages: /tmp/ccYVXAks.s:42: Error: symbol `__roundevenl' is already defined make[2]: *** [/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/sysd-rules:135: /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o] Error 1 Andreas. -- Andreas Schwab, schwab@linux-m68k.org GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 "And now for something completely different."
On 28/06/2021 10:08, Andreas Schwab wrote:
> s390x fails to build.
>
> gcc ../sysdeps/s390/fpu/s_roundevenl.c -c -std=gnu11 -fgnu89-inline -O2 -g -m64 -fmessage-length=0 -funwind-tables -fasynchronous-unwind-tables -Wall -Wwrite-strings -Wundef -Werror -fmerge-all-constants -frounding-math -fstack-protector -Wstrict-prototypes -Wold-style-definition -fno-math-errno -mlong-double-128 -I../include -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base -I../sysdeps/unix/sysv/linux/s390/s390-64 -I../sysdeps/unix/sysv/linux/wordsize-64 -I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu -I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl -I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt -I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux -I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/s390/s390-64 -I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch -I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I. -D_LIBC_REENTRANT -include /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/libc-modules.h -DMODULE_NAME=libm -include ../include/libc-symbols.h -DPIC -DTOP_NAMESPACE=glibc -o /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o -MD -MP -MF /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o.dt -MT /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o
> /tmp/ccYVXAks.s: Assembler messages:
> /tmp/ccYVXAks.s:42: Error: symbol `__roundevenl' is already defined
> make[2]: *** [/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/sysd-rules:135: /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o] Error 1
>
> Andreas.
>
Hi Andreas,
I've also just recognized this issue and will post a patch later.
Thanks,
Stefan
On Mon, Jun 28, 2021 at 1:34 AM Stefan Liebler via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> On 28/06/2021 10:08, Andreas Schwab wrote:
> > s390x fails to build.
> >
> > gcc ../sysdeps/s390/fpu/s_roundevenl.c -c -std=gnu11 -fgnu89-inline -O2 -g -m64 -fmessage-length=0 -funwind-tables -fasynchronous-unwind-tables -Wall -Wwrite-strings -Wundef -Werror -fmerge-all-constants -frounding-math -fstack-protector -Wstrict-prototypes -Wold-style-definition -fno-math-errno -mlong-double-128 -I../include -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base -I../sysdeps/unix/sysv/linux/s390/s390-64 -I../sysdeps/unix/sysv/linux/wordsize-64 -I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu -I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl -I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt -I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux -I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/s390/s390-64 -I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch -I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I. -D_LIBC_REENTRANT -include /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/libc-modules.h -DMODULE_NAME=libm -include ../include/libc-symbols.h -DPIC -DTOP_NAMESPACE=glibc -o /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o -MD -MP -MF /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o.dt -MT /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o
> > /tmp/ccYVXAks.s: Assembler messages:
> > /tmp/ccYVXAks.s:42: Error: symbol `__roundevenl' is already defined
> > make[2]: *** [/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/sysd-rules:135: /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o] Error 1
> >
> > Andreas.
> >
>
> Hi Andreas,
>
> I've also just recognized this issue and will post a patch later.
>
Sorry for this. Why didn't scripts/build-many-glibcs.py catch it?
$ s390x-glibc-linux-gnu-gcc ../sysdeps/s390/fpu/s_roundevenl.c -c
-std=gnu11 -fgnu89-inline -g -O2 -Wall -Wwrite-strings -Wundef
-Werror -fmerge-all-constants -frounding-math -fno-stack-protector
-Wstrict-prototypes -Wold-style-definition -fno-math-errno
-mlong-double-128 -I../include
-I/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math
-I/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc
-I../sysdeps/unix/sysv/linux/s390/s390-64
-I../sysdeps/unix/sysv/linux/wordsize-64
-I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu
-I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl
-I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt
-I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux
-I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu
-I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix
-I../sysdeps/posix -I../sysdeps/s390/s390-64
-I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch
-I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128
-I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32
-I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I.
-D_LIBC_REENTRANT -include
/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/libc-modules.h
-DMODULE_NAME=libm -include ../include/libc-symbols.h
-DTOP_NAMESPACE=glibc -o
/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o
-MD -MP -MF /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o.dt
-MT /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o
compiled fine.
--
H.J.
On 28/06/2021 13:42, H.J. Lu wrote: > On Mon, Jun 28, 2021 at 1:34 AM Stefan Liebler via Libc-alpha > <libc-alpha@sourceware.org> wrote: >> >> On 28/06/2021 10:08, Andreas Schwab wrote: >>> s390x fails to build. >>> >>> gcc ../sysdeps/s390/fpu/s_roundevenl.c -c -std=gnu11 -fgnu89-inline -O2 -g -m64 -fmessage-length=0 -funwind-tables -fasynchronous-unwind-tables -Wall -Wwrite-strings -Wundef -Werror -fmerge-all-constants -frounding-math -fstack-protector -Wstrict-prototypes -Wold-style-definition -fno-math-errno -mlong-double-128 -I../include -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math -I/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base -I../sysdeps/unix/sysv/linux/s390/s390-64 -I../sysdeps/unix/sysv/linux/wordsize-64 -I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu -I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl -I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt -I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux -I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix -I../sysdeps/posix -I../sysdeps/s390/s390-64 -I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch -I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128 -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I. -D_LIBC_REENTRANT -include /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/libc-modules.h -DMODULE_NAME=libm -include ../include/libc-symbols.h -DPIC -DTOP_NAMESPACE=glibc -o /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o -MD -MP -MF /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o.dt -MT /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o >>> /tmp/ccYVXAks.s: Assembler messages: >>> /tmp/ccYVXAks.s:42: Error: symbol `__roundevenl' is already defined >>> make[2]: *** [/home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/sysd-rules:135: /home/abuild/rpmbuild/BUILD/glibc-2.33.9000.810.g5adda61f62/cc-base/math/s_roundevenl.o] Error 1 >>> >>> Andreas. >>> >> >> Hi Andreas, >> >> I've also just recognized this issue and will post a patch later. >> > > Sorry for this. Why didn't scripts/build-many-glibcs.py catch it? > > $ s390x-glibc-linux-gnu-gcc ../sysdeps/s390/fpu/s_roundevenl.c -c > -std=gnu11 -fgnu89-inline -g -O2 -Wall -Wwrite-strings -Wundef > -Werror -fmerge-all-constants -frounding-math -fno-stack-protector > -Wstrict-prototypes -Wold-style-definition -fno-math-errno > -mlong-double-128 -I../include > -I/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math > -I/export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc > -I../sysdeps/unix/sysv/linux/s390/s390-64 > -I../sysdeps/unix/sysv/linux/wordsize-64 > -I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu > -I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl > -I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt > -I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux > -I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu > -I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix > -I../sysdeps/posix -I../sysdeps/s390/s390-64 > -I../sysdeps/wordsize-64 -I../sysdeps/s390/multiarch > -I../sysdeps/s390 -I../sysdeps/ieee754/ldbl-128 > -I../sysdeps/ieee754/dbl-64 -I../sysdeps/ieee754/flt-32 > -I../sysdeps/ieee754 -I../sysdeps/generic -I.. -I../libio -I. > -D_LIBC_REENTRANT -include > /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/libc-modules.h > -DMODULE_NAME=libm -include ../include/libc-symbols.h > -DTOP_NAMESPACE=glibc -o > /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o > -MD -MP -MF /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o.dt > -MT /export/project/tools-build/glibc-many/build/glibcs/s390x-linux-gnu/glibc/math/s_roundevenl.o > > compiled fine. > I assume, that the default ALS of gcc is older than z196. Then sysdeps/ieee754/ldbl-128/s_roundevenl.c is included which builds fine. I've wanted to add some different ALS-levels in the past, but in the end, we only added -O3 due to more aggressive inlining on s390x: https://sourceware.org/pipermail/libc-alpha/2020-August/116839.html Bye, Stefan
On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
>
> This patch adds support for the sse4.1 hardware floating point
> roundeven.
>
> Here is some benchmark results on my systems:
>
> =AMD Ryzen 9 3900X 12-Core Processor=
>
> * benchmark result before this commit
> | | roundeven | roundevenf |
> |------------|--------------|--------------|
> | duration | 3.75587e+09 | 3.75114e+09 |
> | iterations | 3.93053e+08 | 4.35402e+08 |
> | max | 52.592 | 58.71 |
> | min | 7.98 | 7.22 |
> | mean | 9.55563 | 8.61535 |
>
> * benchmark result after this commit
> | | roundeven | roundevenf |
> |------------|---------------|--------------|
> | duration | 3.73815e+09 | 3.73738e+09 |
> | iterations | 5.82692e+08 | 5.91498e+08 |
> | max | 56.468 | 51.642 |
> | min | 6.27 | 6.156 |
> | mean | 6.41532 | 6.3185 |
>
> =Intel(R) Pentium(R) CPU D1508 @ 2.20GHz=
>
> * benchmark result before this commit
> | | roundeven | roundevenf |
> |------------|--------------|--------------|
> | duration | 2.18208e+09 | 2.18258e+09 |
> | iterations | 2.39932e+08 | 2.46924e+08 |
> | max | 96.378 | 98.035 |
> | min | 6.776 | 5.94 |
> | mean | 9.09456 | 8.83907 |
>
> * benchmark result after this commit
> | | roundeven | roundevenf |
> |------------|--------------|--------------|
> | duration | 2.17415e+09 | 2.17005e+09 |
> | iterations | 3.56193e+08 | 4.09824e+08 |
> | max | 51.693 | 97.192 |
> | min | 5.926 | 5.093 |
> | mean | 6.10385 | 5.29507 |
>
> Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
> ---
> sysdeps/x86_64/fpu/multiarch/Makefile | 5 +--
> sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++
> .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++
> sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++
> sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++
> .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++
> sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++
> 7 files changed, 118 insertions(+), 2 deletions(-)
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
>
> diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile
> index 57892c56bb..d425ffd6d3 100644
> --- a/sysdeps/x86_64/fpu/multiarch/Makefile
> +++ b/sysdeps/x86_64/fpu/multiarch/Makefile
> @@ -1,11 +1,12 @@
> ifeq ($(subdir),math)
> libm-sysdep_routines += s_floor-c s_ceil-c s_floorf-c s_ceilf-c \
> s_rint-c s_rintf-c s_nearbyint-c s_nearbyintf-c \
> - s_trunc-c s_truncf-c
> + s_roundeven-c s_roundevenf-c s_trunc-c s_truncf-c
>
> libm-sysdep_routines += s_ceil-sse4_1 s_ceilf-sse4_1 s_floor-sse4_1 \
> s_floorf-sse4_1 s_nearbyint-sse4_1 \
> - s_nearbyintf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> + s_nearbyintf-sse4_1 s_roundeven-sse4_1 \
> + s_roundevenf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> s_trunc-sse4_1 s_truncf-sse4_1
>
> libm-sysdep_routines += e_exp-fma e_log-fma e_pow-fma s_atan-fma \
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> new file mode 100644
> index 0000000000..c7be43cb22
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> @@ -0,0 +1,2 @@
> +#define __roundeven __roundeven_c
> +#include <sysdeps/ieee754/dbl-64/s_roundeven.c>
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> new file mode 100644
> index 0000000000..6ae8f6b1d3
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> @@ -0,0 +1,24 @@
> +/* Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <sysdep.h>
> +
> + .section .text.sse4.1,"ax",@progbits
> +ENTRY(__roundeven_sse41)
> + roundsd $8, %xmm0, %xmm0
> + ret
> +END(__roundeven_sse41)
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> new file mode 100644
> index 0000000000..d92eda652a
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> @@ -0,0 +1,31 @@
> +/* Multiple versions of __roundeven.
> + Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <libm-alias-double.h>
> +
> +#define roundeven __redirect_roundeven
> +#define __roundeven __redirect___roundeven
> +#include <math.h>
> +#undef roundeven
> +#undef __roundeven
> +
> +#define SYMBOL_NAME roundeven
> +#include "ifunc-sse4_1.h"
> +
> +libc_ifunc_redirected (__redirect_roundeven, __roundeven, IFUNC_SELECTOR ());
> +libm_alias_double (__roundeven, roundeven)
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> new file mode 100644
> index 0000000000..72a6e7d1fb
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> @@ -0,0 +1,3 @@
> +#undef __roundevenf
> +#define __roundevenf __roundevenf_c
> +#include <sysdeps/ieee754/flt-32/s_roundevenf.c>
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> new file mode 100644
> index 0000000000..a76e10807e
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> @@ -0,0 +1,24 @@
> +/* Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <sysdep.h>
> +
> + .section .text.sse4.1,"ax",@progbits
> +ENTRY(__roundevenf_sse41)
> + roundss $8, %xmm0, %xmm0
> + ret
> +END(__roundevenf_sse41)
> diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> new file mode 100644
> index 0000000000..2ee196e68f
> --- /dev/null
> +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> @@ -0,0 +1,31 @@
> +/* Multiple versions of __roundevenf.
> + Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <libm-alias-float.h>
> +
> +#define roundevenf __redirect_roundevenf
> +#define __roundevenf __redirect___roundevenf
> +#include <math.h>
> +#undef roundevenf
> +#undef __roundevenf
> +
> +#define SYMBOL_NAME roundevenf
> +#include "ifunc-sse4_1.h"
> +
> +libc_ifunc_redirected (__redirect_roundevenf, __roundevenf, IFUNC_SELECTOR ());
> +libm_alias_float (__roundeven, roundeven)
> --
> 2.31.1
>
I would like to backport this patch to release branches.
Any comments or objections?
--Sunil
On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
<libc-alpha@sourceware.org> wrote:
>
> From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
>
> This patch redirect roundeven function for futhermore changes.
>
> Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> Reviewed-by: H.J. Lu <hjl.tools@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
>
I would like to backport this patch to release branch 2.33.
Any comments/suggestions or objections on this.
commit 447954a206837b5f153869cfeeeab44631c3fac9
Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
Date: Mon May 24 09:43:10 2021 +0800
math: redirect roundeven function
This patch redirect roundeven function for futhermore changes.
Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
On Wed, Apr 27, 2022 at 5:11 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
>
> On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> >
> > This patch adds support for the sse4.1 hardware floating point
> > roundeven.
> >
> > Here is some benchmark results on my systems:
> >
> > =AMD Ryzen 9 3900X 12-Core Processor=
> >
> > * benchmark result before this commit
> > | | roundeven | roundevenf |
> > |------------|--------------|--------------|
> > | duration | 3.75587e+09 | 3.75114e+09 |
> > | iterations | 3.93053e+08 | 4.35402e+08 |
> > | max | 52.592 | 58.71 |
> > | min | 7.98 | 7.22 |
> > | mean | 9.55563 | 8.61535 |
> >
> > * benchmark result after this commit
> > | | roundeven | roundevenf |
> > |------------|---------------|--------------|
> > | duration | 3.73815e+09 | 3.73738e+09 |
> > | iterations | 5.82692e+08 | 5.91498e+08 |
> > | max | 56.468 | 51.642 |
> > | min | 6.27 | 6.156 |
> > | mean | 6.41532 | 6.3185 |
> >
> > =Intel(R) Pentium(R) CPU D1508 @ 2.20GHz=
> >
> > * benchmark result before this commit
> > | | roundeven | roundevenf |
> > |------------|--------------|--------------|
> > | duration | 2.18208e+09 | 2.18258e+09 |
> > | iterations | 2.39932e+08 | 2.46924e+08 |
> > | max | 96.378 | 98.035 |
> > | min | 6.776 | 5.94 |
> > | mean | 9.09456 | 8.83907 |
> >
> > * benchmark result after this commit
> > | | roundeven | roundevenf |
> > |------------|--------------|--------------|
> > | duration | 2.17415e+09 | 2.17005e+09 |
> > | iterations | 3.56193e+08 | 4.09824e+08 |
> > | max | 51.693 | 97.192 |
> > | min | 5.926 | 5.093 |
> > | mean | 6.10385 | 5.29507 |
> >
> > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
> > ---
> > sysdeps/x86_64/fpu/multiarch/Makefile | 5 +--
> > sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++
> > .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++
> > sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++
> > sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++
> > .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++
> > sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++
> > 7 files changed, 118 insertions(+), 2 deletions(-)
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> >
> > diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile
> > index 57892c56bb..d425ffd6d3 100644
> > --- a/sysdeps/x86_64/fpu/multiarch/Makefile
> > +++ b/sysdeps/x86_64/fpu/multiarch/Makefile
> > @@ -1,11 +1,12 @@
> > ifeq ($(subdir),math)
> > libm-sysdep_routines += s_floor-c s_ceil-c s_floorf-c s_ceilf-c \
> > s_rint-c s_rintf-c s_nearbyint-c s_nearbyintf-c \
> > - s_trunc-c s_truncf-c
> > + s_roundeven-c s_roundevenf-c s_trunc-c s_truncf-c
> >
> > libm-sysdep_routines += s_ceil-sse4_1 s_ceilf-sse4_1 s_floor-sse4_1 \
> > s_floorf-sse4_1 s_nearbyint-sse4_1 \
> > - s_nearbyintf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> > + s_nearbyintf-sse4_1 s_roundeven-sse4_1 \
> > + s_roundevenf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> > s_trunc-sse4_1 s_truncf-sse4_1
> >
> > libm-sysdep_routines += e_exp-fma e_log-fma e_pow-fma s_atan-fma \
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > new file mode 100644
> > index 0000000000..c7be43cb22
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > @@ -0,0 +1,2 @@
> > +#define __roundeven __roundeven_c
> > +#include <sysdeps/ieee754/dbl-64/s_roundeven.c>
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > new file mode 100644
> > index 0000000000..6ae8f6b1d3
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > @@ -0,0 +1,24 @@
> > +/* Copyright (C) 2021 Free Software Foundation, Inc.
> > + This file is part of the GNU C Library.
> > +
> > + The GNU C Library is free software; you can redistribute it and/or
> > + modify it under the terms of the GNU Lesser General Public
> > + License as published by the Free Software Foundation; either
> > + version 2.1 of the License, or (at your option) any later version.
> > +
> > + The GNU C Library is distributed in the hope that it will be useful,
> > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > + Lesser General Public License for more details.
> > +
> > + You should have received a copy of the GNU Lesser General Public
> > + License along with the GNU C Library; if not, see
> > + <https://www.gnu.org/licenses/>. */
> > +
> > +#include <sysdep.h>
> > +
> > + .section .text.sse4.1,"ax",@progbits
> > +ENTRY(__roundeven_sse41)
> > + roundsd $8, %xmm0, %xmm0
> > + ret
> > +END(__roundeven_sse41)
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > new file mode 100644
> > index 0000000000..d92eda652a
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > @@ -0,0 +1,31 @@
> > +/* Multiple versions of __roundeven.
> > + Copyright (C) 2021 Free Software Foundation, Inc.
> > + This file is part of the GNU C Library.
> > +
> > + The GNU C Library is free software; you can redistribute it and/or
> > + modify it under the terms of the GNU Lesser General Public
> > + License as published by the Free Software Foundation; either
> > + version 2.1 of the License, or (at your option) any later version.
> > +
> > + The GNU C Library is distributed in the hope that it will be useful,
> > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > + Lesser General Public License for more details.
> > +
> > + You should have received a copy of the GNU Lesser General Public
> > + License along with the GNU C Library; if not, see
> > + <https://www.gnu.org/licenses/>. */
> > +
> > +#include <libm-alias-double.h>
> > +
> > +#define roundeven __redirect_roundeven
> > +#define __roundeven __redirect___roundeven
> > +#include <math.h>
> > +#undef roundeven
> > +#undef __roundeven
> > +
> > +#define SYMBOL_NAME roundeven
> > +#include "ifunc-sse4_1.h"
> > +
> > +libc_ifunc_redirected (__redirect_roundeven, __roundeven, IFUNC_SELECTOR ());
> > +libm_alias_double (__roundeven, roundeven)
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > new file mode 100644
> > index 0000000000..72a6e7d1fb
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > @@ -0,0 +1,3 @@
> > +#undef __roundevenf
> > +#define __roundevenf __roundevenf_c
> > +#include <sysdeps/ieee754/flt-32/s_roundevenf.c>
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > new file mode 100644
> > index 0000000000..a76e10807e
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > @@ -0,0 +1,24 @@
> > +/* Copyright (C) 2021 Free Software Foundation, Inc.
> > + This file is part of the GNU C Library.
> > +
> > + The GNU C Library is free software; you can redistribute it and/or
> > + modify it under the terms of the GNU Lesser General Public
> > + License as published by the Free Software Foundation; either
> > + version 2.1 of the License, or (at your option) any later version.
> > +
> > + The GNU C Library is distributed in the hope that it will be useful,
> > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > + Lesser General Public License for more details.
> > +
> > + You should have received a copy of the GNU Lesser General Public
> > + License along with the GNU C Library; if not, see
> > + <https://www.gnu.org/licenses/>. */
> > +
> > +#include <sysdep.h>
> > +
> > + .section .text.sse4.1,"ax",@progbits
> > +ENTRY(__roundevenf_sse41)
> > + roundss $8, %xmm0, %xmm0
> > + ret
> > +END(__roundevenf_sse41)
> > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> > new file mode 100644
> > index 0000000000..2ee196e68f
> > --- /dev/null
> > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> > @@ -0,0 +1,31 @@
> > +/* Multiple versions of __roundevenf.
> > + Copyright (C) 2021 Free Software Foundation, Inc.
> > + This file is part of the GNU C Library.
> > +
> > + The GNU C Library is free software; you can redistribute it and/or
> > + modify it under the terms of the GNU Lesser General Public
> > + License as published by the Free Software Foundation; either
> > + version 2.1 of the License, or (at your option) any later version.
> > +
> > + The GNU C Library is distributed in the hope that it will be useful,
> > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > + Lesser General Public License for more details.
> > +
> > + You should have received a copy of the GNU Lesser General Public
> > + License along with the GNU C Library; if not, see
> > + <https://www.gnu.org/licenses/>. */
> > +
> > +#include <libm-alias-float.h>
> > +
> > +#define roundevenf __redirect_roundevenf
> > +#define __roundevenf __redirect___roundevenf
> > +#include <math.h>
> > +#undef roundevenf
> > +#undef __roundevenf
> > +
> > +#define SYMBOL_NAME roundevenf
> > +#include "ifunc-sse4_1.h"
> > +
> > +libc_ifunc_redirected (__redirect_roundevenf, __roundevenf, IFUNC_SELECTOR ());
> > +libm_alias_float (__roundeven, roundeven)
> > --
> > 2.31.1
> >
>
> I would like to backport this patch to release branches.
> Any comments or objections?
>
> --Sunil
I would like to backport this patch to release branch 2.33.
Any comments/suggestions or objections on this.
commit 1683249d17e14827b6579529742eb895027dfa84
Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
Date: Mon May 24 09:43:11 2021 +0800
x86_64: roundeven with sse4.1 support
This patch adds support for the sse4.1 hardware floating point
roundeven.
Here is some benchmark results on my systems:
On Sun, Sep 11, 2022 at 12:57 PM Sunil Pandey via Libc-stable
<libc-stable@sourceware.org> wrote:
>
> On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
> <libc-alpha@sourceware.org> wrote:
> >
> > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> >
> > This patch redirect roundeven function for futhermore changes.
> >
> > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > Reviewed-by: H.J. Lu <hjl.tools@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
> >
>
> I would like to backport this patch to release branch 2.33.
>
> Any comments/suggestions or objections on this.
>
> commit 447954a206837b5f153869cfeeeab44631c3fac9
> Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> Date: Mon May 24 09:43:10 2021 +0800
>
> math: redirect roundeven function
>
> This patch redirect roundeven function for futhermore changes.
>
> Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
Fine by me.
On Sun, Sep 11, 2022 at 1:09 PM Sunil Pandey via Libc-stable <libc-stable@sourceware.org> wrote: > > On Wed, Apr 27, 2022 at 5:11 PM Sunil Pandey <skpgkp2@gmail.com> wrote: > > > > On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha > > <libc-alpha@sourceware.org> wrote: > > > > > > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> > > > > > > This patch adds support for the sse4.1 hardware floating point > > > roundeven. > > > > > > Here is some benchmark results on my systems: > > > > > > =AMD Ryzen 9 3900X 12-Core Processor= > > > > > > * benchmark result before this commit > > > | | roundeven | roundevenf | > > > |------------|--------------|--------------| > > > | duration | 3.75587e+09 | 3.75114e+09 | > > > | iterations | 3.93053e+08 | 4.35402e+08 | > > > | max | 52.592 | 58.71 | > > > | min | 7.98 | 7.22 | > > > | mean | 9.55563 | 8.61535 | > > > > > > * benchmark result after this commit > > > | | roundeven | roundevenf | > > > |------------|---------------|--------------| > > > | duration | 3.73815e+09 | 3.73738e+09 | > > > | iterations | 5.82692e+08 | 5.91498e+08 | > > > | max | 56.468 | 51.642 | > > > | min | 6.27 | 6.156 | > > > | mean | 6.41532 | 6.3185 | > > > > > > =Intel(R) Pentium(R) CPU D1508 @ 2.20GHz= > > > > > > * benchmark result before this commit > > > | | roundeven | roundevenf | > > > |------------|--------------|--------------| > > > | duration | 2.18208e+09 | 2.18258e+09 | > > > | iterations | 2.39932e+08 | 2.46924e+08 | > > > | max | 96.378 | 98.035 | > > > | min | 6.776 | 5.94 | > > > | mean | 9.09456 | 8.83907 | > > > > > > * benchmark result after this commit > > > | | roundeven | roundevenf | > > > |------------|--------------|--------------| > > > | duration | 2.17415e+09 | 2.17005e+09 | > > > | iterations | 3.56193e+08 | 4.09824e+08 | > > > | max | 51.693 | 97.192 | > > > | min | 5.926 | 5.093 | > > > | mean | 6.10385 | 5.29507 | > > > > > > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> > > > Reviewed-by: H.J. Lu <hjl.tools@gmail.com> > > > --- > > > sysdeps/x86_64/fpu/multiarch/Makefile | 5 +-- > > > sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++ > > > .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++ > > > sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++ > > > sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++ > > > .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++ > > > sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++ > > > 7 files changed, 118 insertions(+), 2 deletions(-) > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c > > > > > > diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile > > > index 57892c56bb..d425ffd6d3 100644 > > > --- a/sysdeps/x86_64/fpu/multiarch/Makefile > > > +++ b/sysdeps/x86_64/fpu/multiarch/Makefile > > > @@ -1,11 +1,12 @@ > > > ifeq ($(subdir),math) > > > libm-sysdep_routines += s_floor-c s_ceil-c s_floorf-c s_ceilf-c \ > > > s_rint-c s_rintf-c s_nearbyint-c s_nearbyintf-c \ > > > - s_trunc-c s_truncf-c > > > + s_roundeven-c s_roundevenf-c s_trunc-c s_truncf-c > > > > > > libm-sysdep_routines += s_ceil-sse4_1 s_ceilf-sse4_1 s_floor-sse4_1 \ > > > s_floorf-sse4_1 s_nearbyint-sse4_1 \ > > > - s_nearbyintf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \ > > > + s_nearbyintf-sse4_1 s_roundeven-sse4_1 \ > > > + s_roundevenf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \ > > > s_trunc-sse4_1 s_truncf-sse4_1 > > > > > > libm-sysdep_routines += e_exp-fma e_log-fma e_pow-fma s_atan-fma \ > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c > > > new file mode 100644 > > > index 0000000000..c7be43cb22 > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c > > > @@ -0,0 +1,2 @@ > > > +#define __roundeven __roundeven_c > > > +#include <sysdeps/ieee754/dbl-64/s_roundeven.c> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S > > > new file mode 100644 > > > index 0000000000..6ae8f6b1d3 > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S > > > @@ -0,0 +1,24 @@ > > > +/* Copyright (C) 2021 Free Software Foundation, Inc. > > > + This file is part of the GNU C Library. > > > + > > > + The GNU C Library is free software; you can redistribute it and/or > > > + modify it under the terms of the GNU Lesser General Public > > > + License as published by the Free Software Foundation; either > > > + version 2.1 of the License, or (at your option) any later version. > > > + > > > + The GNU C Library is distributed in the hope that it will be useful, > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU > > > + Lesser General Public License for more details. > > > + > > > + You should have received a copy of the GNU Lesser General Public > > > + License along with the GNU C Library; if not, see > > > + <https://www.gnu.org/licenses/>. */ > > > + > > > +#include <sysdep.h> > > > + > > > + .section .text.sse4.1,"ax",@progbits > > > +ENTRY(__roundeven_sse41) > > > + roundsd $8, %xmm0, %xmm0 > > > + ret > > > +END(__roundeven_sse41) > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c > > > new file mode 100644 > > > index 0000000000..d92eda652a > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c > > > @@ -0,0 +1,31 @@ > > > +/* Multiple versions of __roundeven. > > > + Copyright (C) 2021 Free Software Foundation, Inc. > > > + This file is part of the GNU C Library. > > > + > > > + The GNU C Library is free software; you can redistribute it and/or > > > + modify it under the terms of the GNU Lesser General Public > > > + License as published by the Free Software Foundation; either > > > + version 2.1 of the License, or (at your option) any later version. > > > + > > > + The GNU C Library is distributed in the hope that it will be useful, > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU > > > + Lesser General Public License for more details. > > > + > > > + You should have received a copy of the GNU Lesser General Public > > > + License along with the GNU C Library; if not, see > > > + <https://www.gnu.org/licenses/>. */ > > > + > > > +#include <libm-alias-double.h> > > > + > > > +#define roundeven __redirect_roundeven > > > +#define __roundeven __redirect___roundeven > > > +#include <math.h> > > > +#undef roundeven > > > +#undef __roundeven > > > + > > > +#define SYMBOL_NAME roundeven > > > +#include "ifunc-sse4_1.h" > > > + > > > +libc_ifunc_redirected (__redirect_roundeven, __roundeven, IFUNC_SELECTOR ()); > > > +libm_alias_double (__roundeven, roundeven) > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c > > > new file mode 100644 > > > index 0000000000..72a6e7d1fb > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c > > > @@ -0,0 +1,3 @@ > > > +#undef __roundevenf > > > +#define __roundevenf __roundevenf_c > > > +#include <sysdeps/ieee754/flt-32/s_roundevenf.c> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S > > > new file mode 100644 > > > index 0000000000..a76e10807e > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S > > > @@ -0,0 +1,24 @@ > > > +/* Copyright (C) 2021 Free Software Foundation, Inc. > > > + This file is part of the GNU C Library. > > > + > > > + The GNU C Library is free software; you can redistribute it and/or > > > + modify it under the terms of the GNU Lesser General Public > > > + License as published by the Free Software Foundation; either > > > + version 2.1 of the License, or (at your option) any later version. > > > + > > > + The GNU C Library is distributed in the hope that it will be useful, > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU > > > + Lesser General Public License for more details. > > > + > > > + You should have received a copy of the GNU Lesser General Public > > > + License along with the GNU C Library; if not, see > > > + <https://www.gnu.org/licenses/>. */ > > > + > > > +#include <sysdep.h> > > > + > > > + .section .text.sse4.1,"ax",@progbits > > > +ENTRY(__roundevenf_sse41) > > > + roundss $8, %xmm0, %xmm0 > > > + ret > > > +END(__roundevenf_sse41) > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c > > > new file mode 100644 > > > index 0000000000..2ee196e68f > > > --- /dev/null > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c > > > @@ -0,0 +1,31 @@ > > > +/* Multiple versions of __roundevenf. > > > + Copyright (C) 2021 Free Software Foundation, Inc. > > > + This file is part of the GNU C Library. > > > + > > > + The GNU C Library is free software; you can redistribute it and/or > > > + modify it under the terms of the GNU Lesser General Public > > > + License as published by the Free Software Foundation; either > > > + version 2.1 of the License, or (at your option) any later version. > > > + > > > + The GNU C Library is distributed in the hope that it will be useful, > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU > > > + Lesser General Public License for more details. > > > + > > > + You should have received a copy of the GNU Lesser General Public > > > + License along with the GNU C Library; if not, see > > > + <https://www.gnu.org/licenses/>. */ > > > + > > > +#include <libm-alias-float.h> > > > + > > > +#define roundevenf __redirect_roundevenf > > > +#define __roundevenf __redirect___roundevenf > > > +#include <math.h> > > > +#undef roundevenf > > > +#undef __roundevenf > > > + > > > +#define SYMBOL_NAME roundevenf > > > +#include "ifunc-sse4_1.h" > > > + > > > +libc_ifunc_redirected (__redirect_roundevenf, __roundevenf, IFUNC_SELECTOR ()); > > > +libm_alias_float (__roundeven, roundeven) > > > -- > > > 2.31.1 > > > > > > > I would like to backport this patch to release branches. > > Any comments or objections? > > > > --Sunil > > I would like to backport this patch to release branch 2.33. > > Any comments/suggestions or objections on this. Fine by me. > > commit 1683249d17e14827b6579529742eb895027dfa84 > Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com> > Date: Mon May 24 09:43:11 2021 +0800 > > x86_64: roundeven with sse4.1 support > > This patch adds support for the sse4.1 hardware floating point > roundeven. > > Here is some benchmark results on my systems:
Fine by me, too
Noah Goldstein <goldstein.w.n@gmail.com> 於 2022年9月14日 週三 上午9:25寫道:
>
> On Sun, Sep 11, 2022 at 12:57 PM Sunil Pandey via Libc-stable
> <libc-stable@sourceware.org> wrote:
> >
> > On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
> > <libc-alpha@sourceware.org> wrote:
> > >
> > > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > >
> > > This patch redirect roundeven function for futhermore changes.
> > >
> > > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > Reviewed-by: H.J. Lu <hjl.tools@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
> > >
> >
> > I would like to backport this patch to release branch 2.33.
> >
> > Any comments/suggestions or objections on this.
> >
> > commit 447954a206837b5f153869cfeeeab44631c3fac9
> > Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > Date: Mon May 24 09:43:10 2021 +0800
> >
> > math: redirect roundeven function
> >
> > This patch redirect roundeven function for futhermore changes.
> >
> > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
>
> Fine by me.
Fine by me, too
Noah Goldstein <goldstein.w.n@gmail.com> 於 2022年9月14日 週三 上午9:25寫道:
>
> On Sun, Sep 11, 2022 at 1:09 PM Sunil Pandey via Libc-stable
> <libc-stable@sourceware.org> wrote:
> >
> > On Wed, Apr 27, 2022 at 5:11 PM Sunil Pandey <skpgkp2@gmail.com> wrote:
> > >
> > > On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
> > > <libc-alpha@sourceware.org> wrote:
> > > >
> > > > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > >
> > > > This patch adds support for the sse4.1 hardware floating point
> > > > roundeven.
> > > >
> > > > Here is some benchmark results on my systems:
> > > >
> > > > =AMD Ryzen 9 3900X 12-Core Processor=
> > > >
> > > > * benchmark result before this commit
> > > > | | roundeven | roundevenf |
> > > > |------------|--------------|--------------|
> > > > | duration | 3.75587e+09 | 3.75114e+09 |
> > > > | iterations | 3.93053e+08 | 4.35402e+08 |
> > > > | max | 52.592 | 58.71 |
> > > > | min | 7.98 | 7.22 |
> > > > | mean | 9.55563 | 8.61535 |
> > > >
> > > > * benchmark result after this commit
> > > > | | roundeven | roundevenf |
> > > > |------------|---------------|--------------|
> > > > | duration | 3.73815e+09 | 3.73738e+09 |
> > > > | iterations | 5.82692e+08 | 5.91498e+08 |
> > > > | max | 56.468 | 51.642 |
> > > > | min | 6.27 | 6.156 |
> > > > | mean | 6.41532 | 6.3185 |
> > > >
> > > > =Intel(R) Pentium(R) CPU D1508 @ 2.20GHz=
> > > >
> > > > * benchmark result before this commit
> > > > | | roundeven | roundevenf |
> > > > |------------|--------------|--------------|
> > > > | duration | 2.18208e+09 | 2.18258e+09 |
> > > > | iterations | 2.39932e+08 | 2.46924e+08 |
> > > > | max | 96.378 | 98.035 |
> > > > | min | 6.776 | 5.94 |
> > > > | mean | 9.09456 | 8.83907 |
> > > >
> > > > * benchmark result after this commit
> > > > | | roundeven | roundevenf |
> > > > |------------|--------------|--------------|
> > > > | duration | 2.17415e+09 | 2.17005e+09 |
> > > > | iterations | 3.56193e+08 | 4.09824e+08 |
> > > > | max | 51.693 | 97.192 |
> > > > | min | 5.926 | 5.093 |
> > > > | mean | 6.10385 | 5.29507 |
> > > >
> > > > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > > Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
> > > > ---
> > > > sysdeps/x86_64/fpu/multiarch/Makefile | 5 +--
> > > > sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c | 2 ++
> > > > .../x86_64/fpu/multiarch/s_roundeven-sse4_1.S | 24 ++++++++++++++
> > > > sysdeps/x86_64/fpu/multiarch/s_roundeven.c | 31 +++++++++++++++++++
> > > > sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c | 3 ++
> > > > .../fpu/multiarch/s_roundevenf-sse4_1.S | 24 ++++++++++++++
> > > > sysdeps/x86_64/fpu/multiarch/s_roundevenf.c | 31 +++++++++++++++++++
> > > > 7 files changed, 118 insertions(+), 2 deletions(-)
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > > > create mode 100644 sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> > > >
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/Makefile b/sysdeps/x86_64/fpu/multiarch/Makefile
> > > > index 57892c56bb..d425ffd6d3 100644
> > > > --- a/sysdeps/x86_64/fpu/multiarch/Makefile
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/Makefile
> > > > @@ -1,11 +1,12 @@
> > > > ifeq ($(subdir),math)
> > > > libm-sysdep_routines += s_floor-c s_ceil-c s_floorf-c s_ceilf-c \
> > > > s_rint-c s_rintf-c s_nearbyint-c s_nearbyintf-c \
> > > > - s_trunc-c s_truncf-c
> > > > + s_roundeven-c s_roundevenf-c s_trunc-c s_truncf-c
> > > >
> > > > libm-sysdep_routines += s_ceil-sse4_1 s_ceilf-sse4_1 s_floor-sse4_1 \
> > > > s_floorf-sse4_1 s_nearbyint-sse4_1 \
> > > > - s_nearbyintf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> > > > + s_nearbyintf-sse4_1 s_roundeven-sse4_1 \
> > > > + s_roundevenf-sse4_1 s_rint-sse4_1 s_rintf-sse4_1 \
> > > > s_trunc-sse4_1 s_truncf-sse4_1
> > > >
> > > > libm-sysdep_routines += e_exp-fma e_log-fma e_pow-fma s_atan-fma \
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > > > new file mode 100644
> > > > index 0000000000..c7be43cb22
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-c.c
> > > > @@ -0,0 +1,2 @@
> > > > +#define __roundeven __roundeven_c
> > > > +#include <sysdeps/ieee754/dbl-64/s_roundeven.c>
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > > > new file mode 100644
> > > > index 0000000000..6ae8f6b1d3
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven-sse4_1.S
> > > > @@ -0,0 +1,24 @@
> > > > +/* Copyright (C) 2021 Free Software Foundation, Inc.
> > > > + This file is part of the GNU C Library.
> > > > +
> > > > + The GNU C Library is free software; you can redistribute it and/or
> > > > + modify it under the terms of the GNU Lesser General Public
> > > > + License as published by the Free Software Foundation; either
> > > > + version 2.1 of the License, or (at your option) any later version.
> > > > +
> > > > + The GNU C Library is distributed in the hope that it will be useful,
> > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > > > + Lesser General Public License for more details.
> > > > +
> > > > + You should have received a copy of the GNU Lesser General Public
> > > > + License along with the GNU C Library; if not, see
> > > > + <https://www.gnu.org/licenses/>. */
> > > > +
> > > > +#include <sysdep.h>
> > > > +
> > > > + .section .text.sse4.1,"ax",@progbits
> > > > +ENTRY(__roundeven_sse41)
> > > > + roundsd $8, %xmm0, %xmm0
> > > > + ret
> > > > +END(__roundeven_sse41)
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundeven.c b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > > > new file mode 100644
> > > > index 0000000000..d92eda652a
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundeven.c
> > > > @@ -0,0 +1,31 @@
> > > > +/* Multiple versions of __roundeven.
> > > > + Copyright (C) 2021 Free Software Foundation, Inc.
> > > > + This file is part of the GNU C Library.
> > > > +
> > > > + The GNU C Library is free software; you can redistribute it and/or
> > > > + modify it under the terms of the GNU Lesser General Public
> > > > + License as published by the Free Software Foundation; either
> > > > + version 2.1 of the License, or (at your option) any later version.
> > > > +
> > > > + The GNU C Library is distributed in the hope that it will be useful,
> > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > > > + Lesser General Public License for more details.
> > > > +
> > > > + You should have received a copy of the GNU Lesser General Public
> > > > + License along with the GNU C Library; if not, see
> > > > + <https://www.gnu.org/licenses/>. */
> > > > +
> > > > +#include <libm-alias-double.h>
> > > > +
> > > > +#define roundeven __redirect_roundeven
> > > > +#define __roundeven __redirect___roundeven
> > > > +#include <math.h>
> > > > +#undef roundeven
> > > > +#undef __roundeven
> > > > +
> > > > +#define SYMBOL_NAME roundeven
> > > > +#include "ifunc-sse4_1.h"
> > > > +
> > > > +libc_ifunc_redirected (__redirect_roundeven, __roundeven, IFUNC_SELECTOR ());
> > > > +libm_alias_double (__roundeven, roundeven)
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > > > new file mode 100644
> > > > index 0000000000..72a6e7d1fb
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-c.c
> > > > @@ -0,0 +1,3 @@
> > > > +#undef __roundevenf
> > > > +#define __roundevenf __roundevenf_c
> > > > +#include <sysdeps/ieee754/flt-32/s_roundevenf.c>
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > > > new file mode 100644
> > > > index 0000000000..a76e10807e
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf-sse4_1.S
> > > > @@ -0,0 +1,24 @@
> > > > +/* Copyright (C) 2021 Free Software Foundation, Inc.
> > > > + This file is part of the GNU C Library.
> > > > +
> > > > + The GNU C Library is free software; you can redistribute it and/or
> > > > + modify it under the terms of the GNU Lesser General Public
> > > > + License as published by the Free Software Foundation; either
> > > > + version 2.1 of the License, or (at your option) any later version.
> > > > +
> > > > + The GNU C Library is distributed in the hope that it will be useful,
> > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > > > + Lesser General Public License for more details.
> > > > +
> > > > + You should have received a copy of the GNU Lesser General Public
> > > > + License along with the GNU C Library; if not, see
> > > > + <https://www.gnu.org/licenses/>. */
> > > > +
> > > > +#include <sysdep.h>
> > > > +
> > > > + .section .text.sse4.1,"ax",@progbits
> > > > +ENTRY(__roundevenf_sse41)
> > > > + roundss $8, %xmm0, %xmm0
> > > > + ret
> > > > +END(__roundevenf_sse41)
> > > > diff --git a/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> > > > new file mode 100644
> > > > index 0000000000..2ee196e68f
> > > > --- /dev/null
> > > > +++ b/sysdeps/x86_64/fpu/multiarch/s_roundevenf.c
> > > > @@ -0,0 +1,31 @@
> > > > +/* Multiple versions of __roundevenf.
> > > > + Copyright (C) 2021 Free Software Foundation, Inc.
> > > > + This file is part of the GNU C Library.
> > > > +
> > > > + The GNU C Library is free software; you can redistribute it and/or
> > > > + modify it under the terms of the GNU Lesser General Public
> > > > + License as published by the Free Software Foundation; either
> > > > + version 2.1 of the License, or (at your option) any later version.
> > > > +
> > > > + The GNU C Library is distributed in the hope that it will be useful,
> > > > + but WITHOUT ANY WARRANTY; without even the implied warranty of
> > > > + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> > > > + Lesser General Public License for more details.
> > > > +
> > > > + You should have received a copy of the GNU Lesser General Public
> > > > + License along with the GNU C Library; if not, see
> > > > + <https://www.gnu.org/licenses/>. */
> > > > +
> > > > +#include <libm-alias-float.h>
> > > > +
> > > > +#define roundevenf __redirect_roundevenf
> > > > +#define __roundevenf __redirect___roundevenf
> > > > +#include <math.h>
> > > > +#undef roundevenf
> > > > +#undef __roundevenf
> > > > +
> > > > +#define SYMBOL_NAME roundevenf
> > > > +#include "ifunc-sse4_1.h"
> > > > +
> > > > +libc_ifunc_redirected (__redirect_roundevenf, __roundevenf, IFUNC_SELECTOR ());
> > > > +libm_alias_float (__roundeven, roundeven)
> > > > --
> > > > 2.31.1
> > > >
> > >
> > > I would like to backport this patch to release branches.
> > > Any comments or objections?
> > >
> > > --Sunil
> >
> > I would like to backport this patch to release branch 2.33.
> >
> > Any comments/suggestions or objections on this.
>
> Fine by me.
> >
> > commit 1683249d17e14827b6579529742eb895027dfa84
> > Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > Date: Mon May 24 09:43:11 2021 +0800
> >
> > x86_64: roundeven with sse4.1 support
> >
> > This patch adds support for the sse4.1 hardware floating point
> > roundeven.
> >
> > Here is some benchmark results on my systems:
On Fri, Sep 23, 2022 at 3:50 AM 謝昇達(Shen-Ta Hsieh)
<ibmibmibm.tw@gmail.com> wrote:
>
> Fine by me, too
>
> Noah Goldstein <goldstein.w.n@gmail.com> 於 2022年9月14日 週三 上午9:25寫道:
> >
> > On Sun, Sep 11, 2022 at 12:57 PM Sunil Pandey via Libc-stable
> > <libc-stable@sourceware.org> wrote:
> > >
> > > On Wed, Jun 23, 2021 at 3:32 PM H.J. Lu via Libc-alpha
> > > <libc-alpha@sourceware.org> wrote:
> > > >
> > > > From: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > >
> > > > This patch redirect roundeven function for futhermore changes.
> > > >
> > > > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > > Reviewed-by: H.J. Lu <hjl.tools@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
> > > >
> > >
> > > I would like to backport this patch to release branch 2.33.
> > >
> > > Any comments/suggestions or objections on this.
> > >
> > > commit 447954a206837b5f153869cfeeeab44631c3fac9
> > > Author: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > Date: Mon May 24 09:43:10 2021 +0800
> > >
> > > math: redirect roundeven function
> > >
> > > This patch redirect roundeven function for futhermore changes.
> > >
> > > Signed-off-by: Shen-Ta Hsieh <ibmibmibm.tw@gmail.com>
> > > Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
> >
> > Fine by me.
Quick git grep shows that this patch has dependency on architectures other
than x86_64.
$ git log --grep roundeven --oneline
259a17cc98 s390x: Update math: redirect roundeven function
3213ed770c Update math: redirect roundeven function
eb9066203f Use GCC builtins for roundeven functions if desired.
1683249d17 x86_64: roundeven with sse4.1 support
447954a206 math: redirect roundeven function
6a86bc0992 AArch64: Add support for roundeven[f]
This may require backporting these patches to other architectures too,
where build may potentially break.
Not sure if other architecture owners like to backport these patches to 2.33.