diff options
Diffstat (limited to 'sysdeps/x86')
| -rw-r--r-- | sysdeps/x86/bits/floatn.h | 96 | ||||
| -rw-r--r-- | sysdeps/x86/float128-abi.h | 2 | ||||
| -rw-r--r-- | sysdeps/x86/fpu/Makefile | 3 | ||||
| -rw-r--r-- | sysdeps/x86/fpu/e_sqrtf128.c | 47 | ||||
| -rw-r--r-- | sysdeps/x86/fpu/sfp-machine.h | 209 | ||||
| -rw-r--r-- | sysdeps/x86/math-tests.h | 25 |
6 files changed, 382 insertions, 0 deletions
diff --git a/sysdeps/x86/bits/floatn.h b/sysdeps/x86/bits/floatn.h new file mode 100644 index 0000000000..23f74782ea --- /dev/null +++ b/sysdeps/x86/bits/floatn.h @@ -0,0 +1,96 @@ +/* Macros to control TS 18661-3 glibc features on x86. + Copyright (C) 2017 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 + <http://www.gnu.org/licenses/>. */ + +#ifndef _BITS_FLOATN_H +#define _BITS_FLOATN_H + +#include <features.h> + +/* Defined to 1 if the current compiler invocation provides a + floating-point type with the IEEE 754 binary128 format, and this + glibc includes corresponding *f128 interfaces for it. The required + libgcc support was added some time after the basic compiler + support, for x86_64 and x86. */ +#if (defined __x86_64__ \ + ? __GNUC_PREREQ (4, 3) \ + : (defined __GNU__ ? __GNUC_PREREQ (4, 5) : __GNUC_PREREQ (4, 4))) +# define __HAVE_FLOAT128 1 +#else +# define __HAVE_FLOAT128 0 +#endif + +/* Defined to 1 if __HAVE_FLOAT128 is 1 and the type is ABI-distinct + from the default float, double and long double types in this glibc. */ +#if __HAVE_FLOAT128 +# define __HAVE_DISTINCT_FLOAT128 1 +#else +# define __HAVE_DISTINCT_FLOAT128 0 +#endif + +/* Defined to concatenate the literal suffix to be used with _Float128 + types, if __HAVE_FLOAT128 is 1. */ +#if __HAVE_FLOAT128 +# if !__GNUC_PREREQ (7, 0) || defined __cplusplus +/* The literal suffix f128 exists only since GCC 7.0. */ +# define __f128(x) x##q +# else +# define __f128(x) x##f128 +# endif +#endif + +/* Defined to a complex binary128 type if __HAVE_FLOAT128 is 1. */ +#if __HAVE_FLOAT128 +# if !__GNUC_PREREQ (7, 0) || defined __cplusplus +/* Add a typedef for older GCC compilers which don't natively support + _Complex _Float128. */ +typedef _Complex float __cfloat128 __attribute__ ((__mode__ (__TC__))); +# define __CFLOAT128 __cfloat128 +# else +# define __CFLOAT128 _Complex _Float128 +# endif +#endif + +/* The remaining of this file provides support for older compilers. */ +#if __HAVE_FLOAT128 + +/* The type _Float128 exists only since GCC 7.0. */ +# if !__GNUC_PREREQ (7, 0) || defined __cplusplus +typedef __float128 _Float128; +# endif + +/* __builtin_huge_valf128 doesn't exist before GCC 7.0. */ +# if !__GNUC_PREREQ (7, 0) +# define __builtin_huge_valf128() ((_Float128) __builtin_huge_val ()) +# endif + +/* Older GCC has only a subset of built-in functions for _Float128 on + x86, and __builtin_infq is not usable in static initializers. + Converting a narrower sNaN to _Float128 produces a quiet NaN, so + attempts to use _Float128 sNaNs will not work properly with older + compilers. */ +# if !__GNUC_PREREQ (7, 0) +# define __builtin_copysignf128 __builtin_copysignq +# define __builtin_fabsf128 __builtin_fabsq +# define __builtin_inff128() ((_Float128) __builtin_inf ()) +# define __builtin_nanf128(x) ((_Float128) __builtin_nan (x)) +# define __builtin_nansf128(x) ((_Float128) __builtin_nans (x)) +# endif + +#endif + +#endif /* _BITS_FLOATN_H */ diff --git a/sysdeps/x86/float128-abi.h b/sysdeps/x86/float128-abi.h new file mode 100644 index 0000000000..6b954cc673 --- /dev/null +++ b/sysdeps/x86/float128-abi.h @@ -0,0 +1,2 @@ +/* ABI version for _Float128 ABI introduction. */ +#define FLOAT128_VERSION GLIBC_2.26 diff --git a/sysdeps/x86/fpu/Makefile b/sysdeps/x86/fpu/Makefile index a8047a4504..600e42c3db 100644 --- a/sysdeps/x86/fpu/Makefile +++ b/sysdeps/x86/fpu/Makefile @@ -1,4 +1,7 @@ ifeq ($(subdir),math) +# sqrtf128 requires soft-fp. +CPPFLAGS += -I../soft-fp + libm-support += powl_helper tests += test-fenv-sse test-fenv-clear-sse test-fenv-x87 test-fenv-sse-2 \ test-flt-eval-method-387 test-flt-eval-method-sse diff --git a/sysdeps/x86/fpu/e_sqrtf128.c b/sysdeps/x86/fpu/e_sqrtf128.c new file mode 100644 index 0000000000..f6a62a83de --- /dev/null +++ b/sysdeps/x86/fpu/e_sqrtf128.c @@ -0,0 +1,47 @@ +/* soft-fp sqrt for _Float128 + Return sqrt(a) + Copyright (C) 2017 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. + + In addition to the permissions in the GNU Lesser General Public + License, the Free Software Foundation gives you unlimited + permission to link the compiled version of this file into + combinations with other programs, and to distribute those + combinations without any restriction coming from the use of this + file. (The Lesser General Public License restrictions do apply in + other respects; for example, they cover modification of the file, + and distribution when not linked into a combine executable.) + + 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 + <http://www.gnu.org/licenses/>. */ + +#include <soft-fp.h> +#include <quad.h> + +__float128 +__ieee754_sqrtf128 (__float128 a) +{ + FP_DECL_EX; + FP_DECL_Q (A); + FP_DECL_Q (R); + __float128 r; + + FP_INIT_ROUNDMODE; + FP_UNPACK_Q (A, a); + FP_SQRT_Q (R, A); + FP_PACK_Q (r, R); + FP_HANDLE_EXCEPTIONS; + return r; +} +strong_alias (__ieee754_sqrtf128, __sqrtf128_finite) diff --git a/sysdeps/x86/fpu/sfp-machine.h b/sysdeps/x86/fpu/sfp-machine.h new file mode 100644 index 0000000000..df8906acb4 --- /dev/null +++ b/sysdeps/x86/fpu/sfp-machine.h @@ -0,0 +1,209 @@ +/* Configure soft-fp for building sqrtf128. Based on sfp-machine.h in + libgcc, with soft-float and other irrelevant parts removed. */ + +/* The type of the result of a floating point comparison. This must + match `__libgcc_cmp_return__' in GCC for the target. */ +typedef int __gcc_CMPtype __attribute__ ((mode (__libgcc_cmp_return__))); +#define CMPtype __gcc_CMPtype + +#ifdef __x86_64__ +# define _FP_W_TYPE_SIZE 64 +# define _FP_W_TYPE unsigned long long +# define _FP_WS_TYPE signed long long +# define _FP_I_TYPE long long + +typedef int TItype __attribute__ ((mode (TI))); +typedef unsigned int UTItype __attribute__ ((mode (TI))); + +# define TI_BITS (__CHAR_BIT__ * (int)sizeof(TItype)) + +# define _FP_MUL_MEAT_Q(R,X,Y) \ + _FP_MUL_MEAT_2_wide(_FP_WFRACBITS_Q,R,X,Y,umul_ppmm) + +# define _FP_DIV_MEAT_Q(R,X,Y) _FP_DIV_MEAT_2_udiv(Q,R,X,Y) + +# define _FP_NANFRAC_S _FP_QNANBIT_S +# define _FP_NANFRAC_D _FP_QNANBIT_D +# define _FP_NANFRAC_E _FP_QNANBIT_E, 0 +# define _FP_NANFRAC_Q _FP_QNANBIT_Q, 0 + +# define FP_EX_SHIFT 7 + +# define _FP_DECL_EX \ + unsigned int _fcw __attribute__ ((unused)) = FP_RND_NEAREST; + +# define FP_RND_NEAREST 0 +# define FP_RND_ZERO 0x6000 +# define FP_RND_PINF 0x4000 +# define FP_RND_MINF 0x2000 + +# define FP_RND_MASK 0x6000 + +# define FP_INIT_ROUNDMODE \ + do { \ + __asm__ __volatile__ ("%vstmxcsr\t%0" : "=m" (_fcw)); \ + } while (0) +#else +# define _FP_W_TYPE_SIZE 32 +# define _FP_W_TYPE unsigned int +# define _FP_WS_TYPE signed int +# define _FP_I_TYPE int + +# define __FP_FRAC_ADD_4(r3,r2,r1,r0,x3,x2,x1,x0,y3,y2,y1,y0) \ + __asm__ ("add{l} {%11,%3|%3,%11}\n\t" \ + "adc{l} {%9,%2|%2,%9}\n\t" \ + "adc{l} {%7,%1|%1,%7}\n\t" \ + "adc{l} {%5,%0|%0,%5}" \ + : "=r" ((USItype) (r3)), \ + "=&r" ((USItype) (r2)), \ + "=&r" ((USItype) (r1)), \ + "=&r" ((USItype) (r0)) \ + : "%0" ((USItype) (x3)), \ + "g" ((USItype) (y3)), \ + "%1" ((USItype) (x2)), \ + "g" ((USItype) (y2)), \ + "%2" ((USItype) (x1)), \ + "g" ((USItype) (y1)), \ + "%3" ((USItype) (x0)), \ + "g" ((USItype) (y0))) +# define __FP_FRAC_ADD_3(r2,r1,r0,x2,x1,x0,y2,y1,y0) \ + __asm__ ("add{l} {%8,%2|%2,%8}\n\t" \ + "adc{l} {%6,%1|%1,%6}\n\t" \ + "adc{l} {%4,%0|%0,%4}" \ + : "=r" ((USItype) (r2)), \ + "=&r" ((USItype) (r1)), \ + "=&r" ((USItype) (r0)) \ + : "%0" ((USItype) (x2)), \ + "g" ((USItype) (y2)), \ + "%1" ((USItype) (x1)), \ + "g" ((USItype) (y1)), \ + "%2" ((USItype) (x0)), \ + "g" ((USItype) (y0))) +# define __FP_FRAC_SUB_4(r3,r2,r1,r0,x3,x2,x1,x0,y3,y2,y1,y0) \ + __asm__ ("sub{l} {%11,%3|%3,%11}\n\t" \ + "sbb{l} {%9,%2|%2,%9}\n\t" \ + "sbb{l} {%7,%1|%1,%7}\n\t" \ + "sbb{l} {%5,%0|%0,%5}" \ + : "=r" ((USItype) (r3)), \ + "=&r" ((USItype) (r2)), \ + "=&r" ((USItype) (r1)), \ + "=&r" ((USItype) (r0)) \ + : "0" ((USItype) (x3)), \ + "g" ((USItype) (y3)), \ + "1" ((USItype) (x2)), \ + "g" ((USItype) (y2)), \ + "2" ((USItype) (x1)), \ + "g" ((USItype) (y1)), \ + "3" ((USItype) (x0)), \ + "g" ((USItype) (y0))) +# define __FP_FRAC_SUB_3(r2,r1,r0,x2,x1,x0,y2,y1,y0) \ + __asm__ ("sub{l} {%8,%2|%2,%8}\n\t" \ + "sbb{l} {%6,%1|%1,%6}\n\t" \ + "sbb{l} {%4,%0|%0,%4}" \ + : "=r" ((USItype) (r2)), \ + "=&r" ((USItype) (r1)), \ + "=&r" ((USItype) (r0)) \ + : "0" ((USItype) (x2)), \ + "g" ((USItype) (y2)), \ + "1" ((USItype) (x1)), \ + "g" ((USItype) (y1)), \ + "2" ((USItype) (x0)), \ + "g" ((USItype) (y0))) +# define __FP_FRAC_ADDI_4(x3,x2,x1,x0,i) \ + __asm__ ("add{l} {%4,%3|%3,%4}\n\t" \ + "adc{l} {$0,%2|%2,0}\n\t" \ + "adc{l} {$0,%1|%1,0}\n\t" \ + "adc{l} {$0,%0|%0,0}" \ + : "+r" ((USItype) (x3)), \ + "+&r" ((USItype) (x2)), \ + "+&r" ((USItype) (x1)), \ + "+&r" ((USItype) (x0)) \ + : "g" ((USItype) (i))) + + +# define _FP_MUL_MEAT_S(R,X,Y) \ + _FP_MUL_MEAT_1_wide(_FP_WFRACBITS_S,R,X,Y,umul_ppmm) +# define _FP_MUL_MEAT_D(R,X,Y) \ + _FP_MUL_MEAT_2_wide(_FP_WFRACBITS_D,R,X,Y,umul_ppmm) +# define _FP_MUL_MEAT_Q(R,X,Y) \ + _FP_MUL_MEAT_4_wide(_FP_WFRACBITS_Q,R,X,Y,umul_ppmm) + +# define _FP_DIV_MEAT_S(R,X,Y) _FP_DIV_MEAT_1_loop(S,R,X,Y) +# define _FP_DIV_MEAT_D(R,X,Y) _FP_DIV_MEAT_2_udiv(D,R,X,Y) +# define _FP_DIV_MEAT_Q(R,X,Y) _FP_DIV_MEAT_4_udiv(Q,R,X,Y) + +# define _FP_NANFRAC_S _FP_QNANBIT_S +# define _FP_NANFRAC_D _FP_QNANBIT_D, 0 +/* Even if XFmode is 12byte, we have to pad it to + 16byte since soft-fp emulation is done in 16byte. */ +# define _FP_NANFRAC_E _FP_QNANBIT_E, 0, 0, 0 +# define _FP_NANFRAC_Q _FP_QNANBIT_Q, 0, 0, 0 + +# define FP_EX_SHIFT 0 + +# define _FP_DECL_EX \ + unsigned short _fcw __attribute__ ((unused)) = FP_RND_NEAREST; + +# define FP_RND_NEAREST 0 +# define FP_RND_ZERO 0xc00 +# define FP_RND_PINF 0x800 +# define FP_RND_MINF 0x400 + +# define FP_RND_MASK 0xc00 + +# define FP_INIT_ROUNDMODE \ + do { \ + __asm__ __volatile__ ("fnstcw\t%0" : "=m" (_fcw)); \ + } while (0) +#endif + +#define _FP_KEEPNANFRACP 1 +#define _FP_QNANNEGATEDP 0 + +#define _FP_NANSIGN_S 1 +#define _FP_NANSIGN_D 1 +#define _FP_NANSIGN_E 1 +#define _FP_NANSIGN_Q 1 + +/* Here is something Intel misdesigned: the specs don't define + the case where we have two NaNs with same mantissas, but + different sign. Different operations pick up different NaNs. */ +#define _FP_CHOOSENAN(fs, wc, R, X, Y, OP) \ + do { \ + if (_FP_FRAC_GT_##wc(X, Y) \ + || (_FP_FRAC_EQ_##wc(X,Y) && (OP == '+' || OP == '*'))) \ + { \ + R##_s = X##_s; \ + _FP_FRAC_COPY_##wc(R,X); \ + } \ + else \ + { \ + R##_s = Y##_s; \ + _FP_FRAC_COPY_##wc(R,Y); \ + } \ + R##_c = FP_CLS_NAN; \ + } while (0) + +#define FP_EX_INVALID 0x01 +#define FP_EX_DENORM 0x02 +#define FP_EX_DIVZERO 0x04 +#define FP_EX_OVERFLOW 0x08 +#define FP_EX_UNDERFLOW 0x10 +#define FP_EX_INEXACT 0x20 +#define FP_EX_ALL \ + (FP_EX_INVALID | FP_EX_DENORM | FP_EX_DIVZERO | FP_EX_OVERFLOW \ + | FP_EX_UNDERFLOW | FP_EX_INEXACT) + +void __sfp_handle_exceptions (int); + +#define FP_HANDLE_EXCEPTIONS \ + do { \ + if (__builtin_expect (_fex, 0)) \ + __sfp_handle_exceptions (_fex); \ + } while (0); + +#define FP_TRAPPING_EXCEPTIONS ((~_fcw >> FP_EX_SHIFT) & FP_EX_ALL) + +#define FP_ROUNDMODE (_fcw & FP_RND_MASK) + +#define _FP_TININESS_AFTER_ROUNDING 1 diff --git a/sysdeps/x86/math-tests.h b/sysdeps/x86/math-tests.h new file mode 100644 index 0000000000..c0ab95122b --- /dev/null +++ b/sysdeps/x86/math-tests.h @@ -0,0 +1,25 @@ +/* Configuration for math tests. x86 version. + Copyright (C) 2017 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 + <http://www.gnu.org/licenses/>. */ + +/* Before GCC 7, there is no built-in function to provide a __float128 + sNaN, so most sNaN tests for this type cannot work. */ +#if !__GNUC_PREREQ (7, 0) +# define SNAN_TESTS_float128 0 +#endif + +#include_next <math-tests.h> |