diff options
Diffstat (limited to 'sysdeps/x86_64/fpu')
46 files changed, 3449 insertions, 0 deletions
diff --git a/sysdeps/x86_64/fpu/bits/fenv.h b/sysdeps/x86_64/fpu/bits/fenv.h new file mode 100644 index 0000000000..04576933f5 --- /dev/null +++ b/sysdeps/x86_64/fpu/bits/fenv.h @@ -0,0 +1,93 @@ +/* Copyright (C) 1997,1998,1999,2000,2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#ifndef _FENV_H +# error "Never use <bits/fenv.h> directly; include <fenv.h> instead." +#endif + + +/* Define bits representing the exception. We use the bit positions + of the appropriate bits in the FPU control word. */ +enum + { + FE_INVALID = 0x01, +#define FE_INVALID FE_INVALID + __FE_DENORM = 0x02, + FE_DIVBYZERO = 0x04, +#define FE_DIVBYZERO FE_DIVBYZERO + FE_OVERFLOW = 0x08, +#define FE_OVERFLOW FE_OVERFLOW + FE_UNDERFLOW = 0x10, +#define FE_UNDERFLOW FE_UNDERFLOW + FE_INEXACT = 0x20 +#define FE_INEXACT FE_INEXACT + }; + +#define FE_ALL_EXCEPT \ + (FE_INEXACT | FE_DIVBYZERO | FE_UNDERFLOW | FE_OVERFLOW | FE_INVALID) + +/* The ix87 FPU supports all of the four defined rounding modes. We + use again the bit positions in the FPU control word as the values + for the appropriate macros. */ +enum + { + FE_TONEAREST = 0, +#define FE_TONEAREST FE_TONEAREST + FE_DOWNWARD = 0x400, +#define FE_DOWNWARD FE_DOWNWARD + FE_UPWARD = 0x800, +#define FE_UPWARD FE_UPWARD + FE_TOWARDZERO = 0xc00 +#define FE_TOWARDZERO FE_TOWARDZERO + }; + + +/* Type representing exception flags. */ +typedef unsigned short int fexcept_t; + + +/* Type representing floating-point environment. This structure + corresponds to the layout of the block written by the `fstenv' + instruction and has additional fields for the contents of the MXCSR + register as written by the `stmxcsr' instruction. */ +typedef struct + { + unsigned short int __control_word; + unsigned short int __unused1; + unsigned short int __status_word; + unsigned short int __unused2; + unsigned short int __tags; + unsigned short int __unused3; + unsigned int __eip; + unsigned short int __cs_selector; + unsigned int __opcode:11; + unsigned int __unused4:5; + unsigned int __data_offset; + unsigned short int __data_selector; + unsigned short int __unused5; + unsigned int __mxcsr; + } +fenv_t; + +/* If the default argument is used we use this value. */ +#define FE_DFL_ENV ((__const fenv_t *) -1) + +#ifdef __USE_GNU +/* Floating-point environment where none of the exception is masked. */ +# define FE_NOMASK_ENV ((__const fenv_t *) -2) +#endif diff --git a/sysdeps/x86_64/fpu/bits/mathdef.h b/sysdeps/x86_64/fpu/bits/mathdef.h new file mode 100644 index 0000000000..4131d86565 --- /dev/null +++ b/sysdeps/x86_64/fpu/bits/mathdef.h @@ -0,0 +1,39 @@ +/* Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#if !defined _MATH_H && !defined _COMPLEX_H +# error "Never use <bits/mathdef.h> directly; include <math.h> instead" +#endif + +#if defined __USE_ISOC99 && defined _MATH_H && !defined _MATH_H_MATHDEF +# define _MATH_H_MATHDEF 1 + +/* The x86-64 architecture computes values with the precission of the + used type. */ +typedef float float_t; /* `float' expressions are evaluated as `float'. */ +typedef double double_t; /* `double' expressions are evaluated + as `double'. */ + +/* Define `INFINITY' as value of type `float'. */ +# define INFINITY HUGE_VALF + +/* The values returned by `ilogb' for 0 and NaN respectively. */ +# define FP_ILOGB0 (-2147483647 - 1) +# define FP_ILOGBNAN (-2147483647 - 1) + +#endif /* ISO C99 */ diff --git a/sysdeps/x86_64/fpu/e_acosl.c b/sysdeps/x86_64/fpu/e_acosl.c new file mode 100644 index 0000000000..1ef6d3c94a --- /dev/null +++ b/sysdeps/x86_64/fpu/e_acosl.c @@ -0,0 +1 @@ +#include "sysdeps/i386/fpu/e_acosl.c" diff --git a/sysdeps/x86_64/fpu/e_atan2l.c b/sysdeps/x86_64/fpu/e_atan2l.c new file mode 100644 index 0000000000..bbd549f307 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_atan2l.c @@ -0,0 +1,2 @@ +#include "sysdeps/i386/fpu/e_atan2l.c" + diff --git a/sysdeps/x86_64/fpu/e_exp2l.S b/sysdeps/x86_64/fpu/e_exp2l.S new file mode 100644 index 0000000000..336b989098 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_exp2l.S @@ -0,0 +1,38 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Adapted for exp2 by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + * Public domain. + */ + +#include <machine/asm.h> + +ENTRY(__ieee754_exp2l) + fldt 8(%rsp) +/* I added the following ugly construct because exp(+-Inf) resulted + in NaN. The ugliness results from the bright minds at Intel. + For the i686 the code can be written better. + -- drepper@cygnus.com. */ + fxam /* Is NaN or +-Inf? */ + fstsw %ax + movb $0x45, %dh + andb %ah, %dh + cmpb $0x05, %dh + je 1f /* Is +-Inf, jump. */ + fld %st + frndint /* int(x) */ + fsubr %st,%st(1) /* fract(x) */ + fxch + f2xm1 /* 2^(fract(x)) - 1 */ + fld1 + faddp /* 2^(fract(x)) */ + fscale /* e^x */ + fstp %st(1) + ret + +1: testl $0x200, %eax /* Test sign. */ + jz 2f /* If positive, jump. */ + fstp %st + fldz /* Set result to 0. */ +2: ret +END (__ieee754_exp2l) diff --git a/sysdeps/x86_64/fpu/e_expl.c b/sysdeps/x86_64/fpu/e_expl.c new file mode 100644 index 0000000000..5042e02db9 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_expl.c @@ -0,0 +1 @@ +#include "sysdeps/i386/fpu/e_expl.c" diff --git a/sysdeps/x86_64/fpu/e_fmodl.S b/sysdeps/x86_64/fpu/e_fmodl.S new file mode 100644 index 0000000000..2967bf224b --- /dev/null +++ b/sysdeps/x86_64/fpu/e_fmodl.S @@ -0,0 +1,22 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + +ENTRY(__ieee754_fmodl) + fldt 24(%rsp) + fldt 8(%rsp) +1: fprem + fstsw %ax + and $04,%ah + jnz 1b + fstp %st(1) + ret +END (__ieee754_fmodl) diff --git a/sysdeps/x86_64/fpu/e_log10l.S b/sysdeps/x86_64/fpu/e_log10l.S new file mode 100644 index 0000000000..b4343bef45 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_log10l.S @@ -0,0 +1,66 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * + * Changed to use fyl2xp1 for values near 1, <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(one,@object) +one: .double 1.0 + ASM_SIZE_DIRECTIVE(one) + /* It is not important that this constant is precise. It is only + a value which is known to be on the safe side for using the + fyl2xp1 instruction. */ + ASM_TYPE_DIRECTIVE(limit,@object) +limit: .double 0.29 + ASM_SIZE_DIRECTIVE(limit) + + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__ieee754_log10l) + fldlg2 // log10(2) + fldt 8(%rsp) // x : log10(2) + fxam + fnstsw + fld %st // x : x : log10(2) + andb $1,%ah + jnz 3f // in case x is NaN or ħInf +4: fsubl MO(one) // x-1 : x : log10(2) + fld %st // x-1 : x-1 : x : log10(2) + fabs // |x-1| : x-1 : x : log10(2) + fcompl MO(limit) // x-1 : x : log10(2) + fnstsw // x-1 : x : log10(2) + andb $0x45, %ah + jz 2f + fstp %st(1) // x-1 : log10(2) + fyl2xp1 // log10(x) + ret + +2: fstp %st(0) // x : log10(2) + fyl2x // log10(x) + ret + +3: jp 4b // in case x is ħInf + fstp %st(1) + fstp %st(1) + ret +END(__ieee754_log10l) diff --git a/sysdeps/x86_64/fpu/e_log2l.S b/sysdeps/x86_64/fpu/e_log2l.S new file mode 100644 index 0000000000..7a89b94d9f --- /dev/null +++ b/sysdeps/x86_64/fpu/e_log2l.S @@ -0,0 +1,63 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Adapted for use as log2 by Ulrich Drepper <drepper@cygnus.com>. + * Public domain. + * + * Changed to use fyl2xp1 for values near 1, <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(one,@object) +one: .double 1.0 + ASM_SIZE_DIRECTIVE(one) + /* It is not important that this constant is precise. It is only + a value which is known to be on the safe side for using the + fyl2xp1 instruction. */ + ASM_TYPE_DIRECTIVE(limit,@object) +limit: .double 0.29 + ASM_SIZE_DIRECTIVE(limit) + + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__ieee754_log2l) + fldl MO(one) + fldt 8(%rsp) // x : 1 + fxam + fnstsw + fld %st // x : x : 1 + andb $1,%ah + jnz 3f // in case x is NaN or ħInf +4: fsub %st(2), %st // x-1 : x : 1 + fld %st // x-1 : x-1 : x : 1 + fabs // |x-1| : x-1 : x : 1 + fcompl MO(limit) // x-1 : x : 1 + fnstsw // x-1 : x : 1 + andb $0x45, %ah + jz 2f + fstp %st(1) // x-1 : 1 + fyl2xp1 // log(x) + ret + +2: fstp %st(0) // x : 1 + fyl2x // log(x) + ret + +3: jp 4b // in case x is ħInf + fstp %st(1) + fstp %st(1) + ret +END (__ieee754_log2l) diff --git a/sysdeps/x86_64/fpu/e_logl.S b/sysdeps/x86_64/fpu/e_logl.S new file mode 100644 index 0000000000..a0bed663c8 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_logl.S @@ -0,0 +1,56 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(one,@object) +one: .double 1.0 + ASM_SIZE_DIRECTIVE(one) + /* It is not important that this constant is precise. It is only + a value which is known to be on the safe side for using the + fyl2xp1 instruction. */ + ASM_TYPE_DIRECTIVE(limit,@object) +limit: .double 0.29 + ASM_SIZE_DIRECTIVE(limit) + + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__ieee754_logl) + fldln2 // log(2) + fldt 8(%rsp) // x : log(2) + fld %st // x : x : log(2) + fsubl MO(one) // x-1 : x : log(2) + fld %st // x-1 : x-1 : x : log(2) + fabs // |x-1| : x-1 : x : log(2) + fcompl MO(limit) // x-1 : x : log(2) + fnstsw // x-1 : x : log(2) + andb $0x45, %ah + jz 2f + fstp %st(1) // x-1 : log(2) + fyl2xp1 // log(x) + ret + +2: fstp %st(0) // x : log(2) + fyl2x // log(x) + ret +END (__ieee754_logl) diff --git a/sysdeps/x86_64/fpu/e_powl.S b/sysdeps/x86_64/fpu/e_powl.S new file mode 100644 index 0000000000..8c690e16cc --- /dev/null +++ b/sysdeps/x86_64/fpu/e_powl.S @@ -0,0 +1,316 @@ +/* ix87 specific implementation of pow function. + Copyright (C) 1996, 1997, 1998, 1999, 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Ulrich Drepper <drepper@cygnus.com>, 1996. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <machine/asm.h> + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(infinity,@object) +inf_zero: +infinity: + .byte 0, 0, 0, 0, 0, 0, 0xf0, 0x7f + ASM_SIZE_DIRECTIVE(infinity) + ASM_TYPE_DIRECTIVE(zero,@object) +zero: .double 0.0 + ASM_SIZE_DIRECTIVE(zero) + ASM_TYPE_DIRECTIVE(minf_mzero,@object) +minf_mzero: +minfinity: + .byte 0, 0, 0, 0, 0, 0, 0xf0, 0xff +mzero: + .byte 0, 0, 0, 0, 0, 0, 0, 0x80 + ASM_SIZE_DIRECTIVE(minf_mzero) + ASM_TYPE_DIRECTIVE(one,@object) +one: .double 1.0 + ASM_SIZE_DIRECTIVE(one) + ASM_TYPE_DIRECTIVE(limit,@object) +limit: .double 0.29 + ASM_SIZE_DIRECTIVE(limit) + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__ieee754_powl) + fldt 24(%rsp) // y + fxam + + + fnstsw + movb %ah, %dl + andb $0x45, %ah + cmpb $0x40, %ah // is y == 0 ? + je 11f + + cmpb $0x05, %ah // is y == ħinf ? + je 12f + + cmpb $0x01, %ah // is y == NaN ? + je 30f + + fldt 8(%rsp) // x : y + + fxam + fnstsw + movb %ah, %dh + andb $0x45, %ah + cmpb $0x40, %ah + je 20f // x is ħ0 + + cmpb $0x05, %ah + je 15f // x is ħinf + + fxch // y : x + + /* First see whether `y' is a natural number. In this case we + can use a more precise algorithm. */ + fld %st // y : y : x + fistpll -8(%rsp) // y : x + fildll -8(%rsp) // int(y) : y : x + fucomip %st(1),%st // y : x + jne 2f + + /* OK, we have an integer value for y. */ + mov -8(%rsp),%eax + mov -4(%rsp),%edx + orl $0, %edx + fstp %st(0) // x + jns 4f // y >= 0, jump + fdivrl MO(one) // 1/x (now referred to as x) + negl %eax + adcl $0, %edx + negl %edx +4: fldl MO(one) // 1 : x + fxch + +6: shrdl $1, %edx, %eax + jnc 5f + fxch + fmul %st(1) // x : ST*x + fxch +5: fmul %st(0), %st // x*x : ST*x + shrl $1, %edx + movl %eax, %ecx + orl %edx, %ecx + jnz 6b + fstp %st(0) // ST*x + ret + + /* y is ħNAN */ +30: fldt 8(%rsp) // x : y + fldl MO(one) // 1.0 : x : y + fucomip %st(1),%st // x : y + je 31f + fxch // y : x +31: fstp %st(1) + ret + + .align ALIGNARG(4) +2: /* y is a real number. */ + fxch // x : y + fldl MO(one) // 1.0 : x : y + fld %st(1) // x : 1.0 : x : y + fsub %st(1) // x-1 : 1.0 : x : y + fabs // |x-1| : 1.0 : x : y + fcompl MO(limit) // 1.0 : x : y + fnstsw + fxch // x : 1.0 : y + test $4500,%eax + jz 7f + fsub %st(1) // x-1 : 1.0 : y + fyl2xp1 // log2(x) : y + jmp 8f + +7: fyl2x // log2(x) : y +8: fmul %st(1) // y*log2(x) : y + fst %st(1) // y*log2(x) : y*log2(x) + frndint // int(y*log2(x)) : y*log2(x) + fsubr %st, %st(1) // int(y*log2(x)) : fract(y*log2(x)) + fxch // fract(y*log2(x)) : int(y*log2(x)) + f2xm1 // 2^fract(y*log2(x))-1 : int(y*log2(x)) + faddl MO(one) // 2^fract(y*log2(x)) : int(y*log2(x)) + fscale // 2^fract(y*log2(x))*2^int(y*log2(x)) : int(y*log2(x)) + fstp %st(1) // 2^fract(y*log2(x))*2^int(y*log2(x)) + ret + + + // pow(x,ħ0) = 1 + .align ALIGNARG(4) +11: fstp %st(0) // pop y + fldl MO(one) + ret + + // y == ħinf + .align ALIGNARG(4) +12: fstp %st(0) // pop y + fldt 8(%rsp) // x + fabs + fcompl MO(one) // < 1, == 1, or > 1 + fnstsw + andb $0x45, %ah + cmpb $0x45, %ah + je 13f // jump if x is NaN + + cmpb $0x40, %ah + je 14f // jump if |x| == 1 + + shlb $1, %ah + xorb %ah, %dl + andl $2, %edx +#ifdef PIC + lea inf_zero(%rip),%rcx + fldl (%rcx, %rdx, 4) +#else + fldl inf_zero(,%rdx, 4) +#endif + ret + + .align ALIGNARG(4) +14: fldl MO(one) + ret + + .align ALIGNARG(4) +13: fldt 8(%rsp) // load x == NaN + ret + + .align ALIGNARG(4) + // x is ħinf +15: fstp %st(0) // y + testb $2, %dh + jz 16f // jump if x == +inf + + // We must find out whether y is an odd integer. + fld %st // y : y + fistpll -8(%rsp) // y + fildll -8(%rsp) // int(y) : y + fucomip %st(1),%st + ffreep %st // <empty> + jne 17f + + // OK, the value is an integer, but is it odd? + mov -8(%rsp), %eax + mov -4(%rsp), %edx + andb $1, %al + jz 18f // jump if not odd + // It's an odd integer. + shrl $31, %edx +#ifdef PIC + lea minf_mzero(%rip),%rcx + fldl (%rcx, %rdx, 8) +#else + fldl minf_mzero(,%rdx, 8) +#endif + ret + + .align ALIGNARG(4) +16: fcompl MO(zero) + fnstsw + shrl $5, %eax + andl $8, %eax +#ifdef PIC + lea inf_zero(%rip),%rcx + fldl (%rcx, %rax, 1) +#else + fldl inf_zero(,%rax, 1) +#endif + ret + + .align ALIGNARG(4) +17: shll $30, %edx // sign bit for y in right position +18: shrl $31, %edx +#ifdef PIC + lea inf_zero(%rip),%rcx + fldl (%rcx, %rdx, 8) +#else + fldl inf_zero(,%rdx, 8) +#endif + ret + + .align ALIGNARG(4) + // x is ħ0 +20: fstp %st(0) // y + testb $2, %dl + jz 21f // y > 0 + + // x is ħ0 and y is < 0. We must find out whether y is an odd integer. + testb $2, %dh + jz 25f + + fld %st // y : y + fistpll -8(%rsp) // y + fildll -8(%rsp) // int(y) : y + fucomip %st(1),%st + ffreep %st // <empty> + jne 26f + + // OK, the value is an integer, but is it odd? + mov -8(%rsp),%eax + mov -4(%rsp),%edx + andb $1, %al + jz 27f // jump if not odd + // It's an odd integer. + // Raise divide-by-zero exception and get minus infinity value. + fldl MO(one) + fdivl MO(zero) + fchs + ret + +25: fstp %st(0) +26: +27: // Raise divide-by-zero exception and get infinity value. + fldl MO(one) + fdivl MO(zero) + ret + + .align ALIGNARG(4) + // x is ħ0 and y is > 0. We must find out whether y is an odd integer. +21: testb $2, %dh + jz 22f + + fld %st // y : y + fistpll -8(%rsp) // y + fildll -8(%rsp) // int(y) : y + fucomip %st(1),%st + ffreep %st // <empty> + jne 23f + + // OK, the value is an integer, but is it odd? + mov -8(%rsp),%eax + mov -4(%rsp),%edx + andb $1, %al + jz 24f // jump if not odd + // It's an odd integer. + fldl MO(mzero) + ret + +22: fstp %st(0) +23: +24: fldl MO(zero) + ret + +END(__ieee754_powl) diff --git a/sysdeps/x86_64/fpu/e_rem_pio2l.c b/sysdeps/x86_64/fpu/e_rem_pio2l.c new file mode 100644 index 0000000000..1347b0468c --- /dev/null +++ b/sysdeps/x86_64/fpu/e_rem_pio2l.c @@ -0,0 +1,3 @@ +/* Empty. This file is only meant to avoid compiling the file with the + same name in the libm-ieee754 directory. The code is not used since + there is an assembler version for all users of this file. */ diff --git a/sysdeps/x86_64/fpu/e_scalbl.S b/sysdeps/x86_64/fpu/e_scalbl.S new file mode 100644 index 0000000000..6b229705ba --- /dev/null +++ b/sysdeps/x86_64/fpu/e_scalbl.S @@ -0,0 +1,100 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de> + * + * Correct handling of y==-inf <drepper@gnu> + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(zero_nan,@object) +zero_nan: + .double 0.0 +nan: .byte 0, 0, 0, 0, 0, 0, 0xff, 0x7f +minus_zero: + .byte 0, 0, 0, 0, 0, 0, 0, 0x80 + .byte 0, 0, 0, 0, 0, 0, 0xff, 0x7f + ASM_SIZE_DIRECTIVE(zero_nan) + + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__ieee754_scalbl) + fldt 24(%rsp) + fxam + fnstsw + fldt 8(%rsp) + andl $0x4700, %eax + cmpl $0x0700, %eax + je 1f + andl $0x4500, %eax + cmpl $0x0100, %eax + je 2f + fxam + fnstsw + andl $0x4500, %eax + cmpl $0x0100, %eax + je 3f + fld %st(1) + frndint + fcomip %st(2), %st + jne 4f + fscale + fstp %st(1) + ret + + /* y is -inf */ +1: fxam + fnstsw + movl 16(%rsp), %edx + shrl $5, %eax + fstp %st + fstp %st + andl $0x8000, %edx + andl $8, %eax + jnz 4f + shrl $11, %edx + addl %edx, %eax +#ifdef PIC + lea zero_nan(%rip),%rdx + fldl (%rdx,%rax,1) +#else + fldl zero_nan(%rax, 1) +#endif + ret + + /* The result is NaN, but we must not raise an exception. + So use a variable. */ +2: fstp %st + fstp %st + fldl MO(nan) + ret + + /* The first parameter is a NaN. Return it. */ +3: fstp %st(1) + ret + + /* Return NaN and raise the invalid exception. */ +4: fstp %st + fstp %st + fldz + fdiv %st + ret +END(__ieee754_scalbl) diff --git a/sysdeps/x86_64/fpu/e_sqrtl.c b/sysdeps/x86_64/fpu/e_sqrtl.c new file mode 100644 index 0000000000..90e4e164e5 --- /dev/null +++ b/sysdeps/x86_64/fpu/e_sqrtl.c @@ -0,0 +1 @@ +#include "sysdeps/i386/fpu/e_sqrtl.c" diff --git a/sysdeps/x86_64/fpu/fclrexcpt.c b/sysdeps/x86_64/fpu/fclrexcpt.c new file mode 100644 index 0000000000..4fc3bfbdc7 --- /dev/null +++ b/sysdeps/x86_64/fpu/fclrexcpt.c @@ -0,0 +1,52 @@ +/* Clear given exceptions in current floating-point environment. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +feclearexcept (int excepts) +{ + fenv_t temp; + unsigned int mxcsr; + + /* Mask out unsupported bits/exceptions. */ + excepts &= FE_ALL_EXCEPT; + + /* Bah, we have to clear selected exceptions. Since there is no + `fldsw' instruction we have to do it the hard way. */ + __asm__ ("fnstenv %0" : "=m" (*&temp)); + + /* Clear the relevant bits. */ + temp.__status_word &= excepts ^ FE_ALL_EXCEPT; + + /* Put the new data in effect. */ + __asm__ ("fldenv %0" : : "m" (*&temp)); + + /* And the same procedure for SSE. */ + __asm__ ("stmxcsr %0" : "=m" (*&mxcsr)); + + /* Clear the relevant bits. */ + mxcsr &= ~excepts; + + /* And put them into effect. */ + __asm__ ("ldmxcsr %0" : : "m" (*&mxcsr)); + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/fedisblxcpt.c b/sysdeps/x86_64/fpu/fedisblxcpt.c new file mode 100644 index 0000000000..e8157da126 --- /dev/null +++ b/sysdeps/x86_64/fpu/fedisblxcpt.c @@ -0,0 +1,49 @@ +/* Disable floating-point exceptions. + Copyright (C) 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Andreas Jaeger <aj@suse.de>, 2001. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fedisableexcept (int excepts) +{ + unsigned short int new_exc, old_exc; + unsigned int new, old; + + excepts &= FE_ALL_EXCEPT; + + /* Get the current control word of the x87 FPU. */ + __asm__ ("fstcw %0" : "=m" (*&new_exc)); + + old_exc = (~new_exc) & FE_ALL_EXCEPT; + + new_exc |= excepts; + __asm__ ("fldcw %0" : : "m" (*&new_exc)); + + /* And now the same for the SSE MXCSR register. */ + __asm__ ("stmxcsr %0" : "=m" (*&new)); + + /* The SSE exception masks are shifted by 7 bits. */ + old = (~new) & (FE_ALL_EXCEPT << 7); + + new |= excepts << 7; + __asm__ ("ldmxcsr %0" : : "m" (*&new)); + + return old_exc; +} diff --git a/sysdeps/x86_64/fpu/feenablxcpt.c b/sysdeps/x86_64/fpu/feenablxcpt.c new file mode 100644 index 0000000000..43259d0cfb --- /dev/null +++ b/sysdeps/x86_64/fpu/feenablxcpt.c @@ -0,0 +1,49 @@ +/* Enable floating-point exceptions. + Copyright (C) 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Andreas Jaeger <aj@suse.de>, 2001. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +feenableexcept (int excepts) +{ + unsigned short int new_exc, old_exc; + unsigned int new, old; + + excepts &= FE_ALL_EXCEPT; + + /* Get the current control word of the x87 FPU. */ + __asm__ ("fstcw %0" : "=m" (*&new_exc)); + + old_exc = (~new_exc) & FE_ALL_EXCEPT; + + new_exc &= ~excepts; + __asm__ ("fldcw %0" : : "m" (*&new_exc)); + + /* And now the same for the SSE MXCSR register. */ + __asm__ ("stmxcsr %0" : "=m" (*&new)); + + /* The SSE exception masks are shifted by 7 bits. */ + old = (~new) & (FE_ALL_EXCEPT << 7); + + new &= ~(excepts << 7); + __asm__ ("ldmxcsr %0" : : "m" (*&new)); + + return old_exc; +} diff --git a/sysdeps/x86_64/fpu/fegetenv.c b/sysdeps/x86_64/fpu/fegetenv.c new file mode 100644 index 0000000000..fa5a8dadcb --- /dev/null +++ b/sysdeps/x86_64/fpu/fegetenv.c @@ -0,0 +1,30 @@ +/* Store current floating-point environment. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fegetenv (fenv_t *envp) +{ + __asm__ ("fnstenv %0\n" + "stmxcsr %1" : "=m" (*envp), "=m" (envp->__mxcsr)); + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/fegetexcept.c b/sysdeps/x86_64/fpu/fegetexcept.c new file mode 100644 index 0000000000..04df4b66a2 --- /dev/null +++ b/sysdeps/x86_64/fpu/fegetexcept.c @@ -0,0 +1,32 @@ +/* Get enabled floating-point exceptions. + Copyright (C) 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Andreas Jaeger <aj@suse.de>, 2001. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fegetexcept (void) +{ + unsigned short int exc; + + /* Get the current control word. */ + __asm__ ("fstcw %0" : "=m" (*&exc)); + + return (~exc) & FE_ALL_EXCEPT; +} diff --git a/sysdeps/x86_64/fpu/fegetround.c b/sysdeps/x86_64/fpu/fegetround.c new file mode 100644 index 0000000000..f237b18d1d --- /dev/null +++ b/sysdeps/x86_64/fpu/fegetround.c @@ -0,0 +1,33 @@ +/* Return current rounding direction. + Copyright (C) 1997, 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fegetround (void) +{ + int cw; + /* We only check the x87 FPU unit. The SSE unit should be the same + - and if it's not the same there's no way to signal it. */ + + __asm__ ("fnstcw %0" : "=m" (*&cw)); + + return cw & 0xc00; +} diff --git a/sysdeps/x86_64/fpu/feholdexcpt.c b/sysdeps/x86_64/fpu/feholdexcpt.c new file mode 100644 index 0000000000..db53d0f07f --- /dev/null +++ b/sysdeps/x86_64/fpu/feholdexcpt.c @@ -0,0 +1,41 @@ +/* Store current floating-point environment and clear exceptions. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +feholdexcept (fenv_t *envp) +{ + unsigned short int work; + unsigned int mxcsr; + + /* Store the environment. */ + __asm__ ("fnstenv %0\n" + "stmxcsr %1" : "=m" (*envp), "=m" (envp->__mxcsr)); + + /* Now set all exceptions to non-stop, first the x87 FPU. */ + work = envp->__control_word | 0x3f; + __asm__ ("fldcw %0" : : "m" (*&work)); + + /* Set the SSE MXCSR register. */ + mxcsr = envp->__mxcsr | 0x1f80; + __asm__ ("ldmxcsr %0" : : "m" (*&mxcsr)); + + return 0; +} diff --git a/sysdeps/x86_64/fpu/fesetenv.c b/sysdeps/x86_64/fpu/fesetenv.c new file mode 100644 index 0000000000..4527b16ed9 --- /dev/null +++ b/sysdeps/x86_64/fpu/fesetenv.c @@ -0,0 +1,89 @@ +/* Install given floating-point environment. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> +#include <assert.h> + + +int +fesetenv (const fenv_t *envp) +{ + fenv_t temp; + + /* Install the environment specified by ENVP. But there are a few + values which we do not want to come from the saved environment. + Therefore, we get the current environment and replace the values + we want to use from the environment specified by the parameter. */ + __asm__ ("fnstenv %0\n" + "stmxcsr %1" : "=m" (*&temp), "=m" (*&temp.__mxcsr)); + + if (envp == FE_DFL_ENV) + { + temp.__control_word |= FE_ALL_EXCEPT; + temp.__control_word &= ~FE_TOWARDZERO; + temp.__status_word &= ~FE_ALL_EXCEPT; + temp.__eip = 0; + temp.__cs_selector = 0; + temp.__opcode = 0; + temp.__data_offset = 0; + temp.__data_selector = 0; + /* Set mask for SSE MXCSR. */ + temp.__mxcsr |= (FE_ALL_EXCEPT << 7); + /* Set rounding to FE_TOWARDZERO. */ + temp.__mxcsr &= ~(FE_TOWARDZERO << 3); + /* Disable exceptions. */ + temp.__mxcsr &= ~FE_ALL_EXCEPT; + } + else if (envp == FE_NOMASK_ENV) + { + temp.__control_word &= ~(FE_ALL_EXCEPT | FE_TOWARDZERO); + temp.__status_word &= ~FE_ALL_EXCEPT; + temp.__eip = 0; + temp.__cs_selector = 0; + temp.__opcode = 0; + temp.__data_offset = 0; + temp.__data_selector = 0; + /* Set mask for SSE MXCSR. */ + temp.__mxcsr &= ~(FE_ALL_EXCEPT << 7); + /* Set rounding to FE_TOWARDZERO. */ + temp.__mxcsr &= ~(FE_TOWARDZERO << 3); + /* Disable exceptions. */ + temp.__mxcsr &= ~FE_ALL_EXCEPT; + } + else + { + temp.__control_word &= ~(FE_ALL_EXCEPT | FE_TOWARDZERO); + temp.__control_word |= (envp->__control_word + & (FE_ALL_EXCEPT | FE_TOWARDZERO)); + temp.__status_word &= ~FE_ALL_EXCEPT; + temp.__status_word |= envp->__status_word & FE_ALL_EXCEPT; + temp.__eip = envp->__eip; + temp.__cs_selector = envp->__cs_selector; + temp.__opcode = envp->__opcode; + temp.__data_offset = envp->__data_offset; + temp.__data_selector = envp->__data_selector; + temp.__mxcsr = envp->__mxcsr; + } + + __asm__ ("fldenv %0\n" + "ldmxcsr %1" : : "m" (temp), "m" (temp.__mxcsr)); + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/fesetround.c b/sysdeps/x86_64/fpu/fesetround.c new file mode 100644 index 0000000000..429b7f1d0c --- /dev/null +++ b/sysdeps/x86_64/fpu/fesetround.c @@ -0,0 +1,46 @@ +/* Set current rounding direction. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fesetround (int round) +{ + unsigned short int cw; + int mxcsr; + + if ((round & ~0xc00) != 0) + /* ROUND is no valid rounding mode. */ + return 1; + + /* First set the x87 FPU. */ + asm ("fnstcw %0" : "=m" (*&cw)); + cw &= ~0xc00; + cw |= round; + asm ("fldcw %0" : : "m" (*&cw)); + + /* And now the MSCSR register for SSE, the precision is at different bit + positions in the different units, we need to shift it 3 bits. */ + asm ("stmxcsr %0" : "=m" (*&mxcsr)); + mxcsr &= ~ 0x6000; + mxcsr |= round << 3; + asm ("ldmxcsr %0" : : "m" (*&mxcsr)); + + return 0; +} diff --git a/sysdeps/x86_64/fpu/fgetexcptflg.c b/sysdeps/x86_64/fpu/fgetexcptflg.c new file mode 100644 index 0000000000..f0681dc9d4 --- /dev/null +++ b/sysdeps/x86_64/fpu/fgetexcptflg.c @@ -0,0 +1,36 @@ +/* Store current representation for exceptions. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fegetexceptflag (fexcept_t *flagp, int excepts) +{ + fexcept_t temp; + unsigned int mxscr; + + /* Get the current exceptions for the x87 FPU and SSE unit. */ + __asm__ ("fnstsw %0\n" + "stmxcsr %1" : "=m" (*&temp), "=m" (*&mxscr)); + + *flagp = (temp | mxscr) & FE_ALL_EXCEPT & excepts; + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/fraiseexcpt.c b/sysdeps/x86_64/fpu/fraiseexcpt.c new file mode 100644 index 0000000000..96ebb6dea2 --- /dev/null +++ b/sysdeps/x86_64/fpu/fraiseexcpt.c @@ -0,0 +1,119 @@ +/* Raise given exceptions. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> +#include <math.h> + +int +feraiseexcept (int excepts) +{ + /* Raise exceptions represented by EXPECTS. But we must raise only + one signal at a time. It is important that if the overflow/underflow + exception and the inexact exception are given at the same time, + the overflow/underflow exception follows the inexact exception. */ + + /* First: invalid exception. */ + if ((FE_INVALID & excepts) != 0) + { + /* One example of a invalid operation is 0.0 / 0.0. */ + float f = 0.0; + + __asm__ __volatile__ ("divss %0, %0 " : : "x" (f)); + (void) &f; + } + + /* Next: division by zero. */ + if ((FE_DIVBYZERO & excepts) != 0) + { + float f = 1.0; + float g = 0.0; + + __asm__ __volatile__ ("divss %1, %0" : : "x" (f), "x" (g)); + (void) &f; + } + + /* Next: overflow. */ + if ((FE_OVERFLOW & excepts) != 0) + { + /* XXX: Is it ok to only set the x87 FPU? */ + /* There is no way to raise only the overflow flag. Do it the + hard way. */ + fenv_t temp; + + /* Bah, we have to clear selected exceptions. Since there is no + `fldsw' instruction we have to do it the hard way. */ + __asm__ __volatile__ ("fnstenv %0" : "=m" (*&temp)); + + /* Set the relevant bits. */ + temp.__status_word |= FE_OVERFLOW; + + /* Put the new data in effect. */ + __asm__ __volatile__ ("fldenv %0" : : "m" (*&temp)); + + /* And raise the exception. */ + __asm__ __volatile__ ("fwait"); + } + + /* Next: underflow. */ + if ((FE_UNDERFLOW & excepts) != 0) + { + /* XXX: Is it ok to only set the x87 FPU? */ + /* There is no way to raise only the underflow flag. Do it the + hard way. */ + fenv_t temp; + + /* Bah, we have to clear selected exceptions. Since there is no + `fldsw' instruction we have to do it the hard way. */ + __asm__ __volatile__ ("fnstenv %0" : "=m" (*&temp)); + + /* Set the relevant bits. */ + temp.__status_word |= FE_UNDERFLOW; + + /* Put the new data in effect. */ + __asm__ __volatile__ ("fldenv %0" : : "m" (*&temp)); + + /* And raise the exception. */ + __asm__ __volatile__ ("fwait"); + } + + /* Last: inexact. */ + if ((FE_INEXACT & excepts) != 0) + { + /* XXX: Is it ok to only set the x87 FPU? */ + /* There is no way to raise only the inexact flag. Do it the + hard way. */ + fenv_t temp; + + /* Bah, we have to clear selected exceptions. Since there is no + `fldsw' instruction we have to do it the hard way. */ + __asm__ __volatile__ ("fnstenv %0" : "=m" (*&temp)); + + /* Set the relevant bits. */ + temp.__status_word |= FE_INEXACT; + + /* Put the new data in effect. */ + __asm__ __volatile__ ("fldenv %0" : : "m" (*&temp)); + + /* And raise the exception. */ + __asm__ __volatile__ ("fwait"); + } + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/fsetexcptflg.c b/sysdeps/x86_64/fpu/fsetexcptflg.c new file mode 100644 index 0000000000..fc4a42cf31 --- /dev/null +++ b/sysdeps/x86_64/fpu/fsetexcptflg.c @@ -0,0 +1,54 @@ +/* Set floating-point environment exception handling. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> +#include <math.h> + +int +fesetexceptflag (const fexcept_t *flagp, int excepts) +{ + fenv_t temp; + unsigned int mxcsr; + + /* XXX: Do we really need to set both the exception in both units? + Shouldn't it be enough to set only the SSE unit? */ + + /* Get the current x87 FPU environment. We have to do this since we + cannot separately set the status word. */ + __asm__ ("fnstenv %0" : "=m" (*&temp)); + + temp.__status_word &= ~(excepts & FE_ALL_EXCEPT); + temp.__status_word |= *flagp & excepts & FE_ALL_EXCEPT; + + /* Store the new status word (along with the rest of the environment. + Possibly new exceptions are set but they won't get executed unless + the next floating-point instruction. */ + __asm__ ("fldenv %0" : : "m" (*&temp)); + + /* And now the same for SSE. */ + __asm__ ("stmxcsr %0" : "=m" (*&mxcsr)); + + mxcsr &= (excepts & FE_ALL_EXCEPT); + mxcsr |= *flagp & excepts & FE_ALL_EXCEPT; + + __asm__ ("ldmxcsr %0" : : "m" (*&mxcsr)); + + /* Success. */ + return 0; +} diff --git a/sysdeps/x86_64/fpu/ftestexcept.c b/sysdeps/x86_64/fpu/ftestexcept.c new file mode 100644 index 0000000000..091c25157b --- /dev/null +++ b/sysdeps/x86_64/fpu/ftestexcept.c @@ -0,0 +1,33 @@ +/* Test exception in current environment. + Copyright (C) 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <fenv.h> + +int +fetestexcept (int excepts) +{ + int temp; + unsigned int mxscr; + + /* Get current exceptions. */ + __asm__ ("fnstsw %0\n" + "stmxcsr %1" : "=m" (*&temp), "=m" (*&mxscr)); + + return (temp | mxscr) & excepts & FE_ALL_EXCEPT; +} diff --git a/sysdeps/x86_64/fpu/libm-test-ulps b/sysdeps/x86_64/fpu/libm-test-ulps new file mode 100644 index 0000000000..f533d69c6e --- /dev/null +++ b/sysdeps/x86_64/fpu/libm-test-ulps @@ -0,0 +1,1493 @@ +# Begin of automatic generation + +# asin +Test "asin (-0.5) == -pi/6": +ildouble: 1 +ldouble: 1 +Test "asin (-1.0) == -pi/2": +ildouble: 1 +ldouble: 1 +Test "asin (0.5) == pi/6": +ildouble: 1 +ldouble: 1 +Test "asin (0.7) == 0.77539749661075306374035335271498708": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "asin (1.0) == pi/2": +ildouble: 1 +ldouble: 1 + +# asinh +Test "asinh (0.7) == 0.652666566082355786": +ildouble: 15 +ldouble: 15 + +# atan2 +Test "atan2 (-0.7, -1.0) == -2.530866689200584621918884506789267": +float: 3 +ifloat: 3 +Test "atan2 (0.7, -1.0) == 2.530866689200584621918884506789267": +float: 3 +ifloat: 3 +Test "atan2 (1.4, -0.93) == 2.1571487668237843754887415992772736": +float: 4 +ifloat: 4 + +# atanh +Test "atanh (0.7) == 0.8673005276940531944": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +# cabs +Test "cabs (-0.7 + 12.4 i) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "cabs (-0.7 - 12.4 i) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "cabs (-12.4 + 0.7 i) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "cabs (-12.4 - 0.7 i) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "cabs (0.7 + 1.2 i) == 1.3892443989449804508432547041028554": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "cabs (0.7 + 12.4 i) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 + +# cacos +Test "Real part of: cacos (0.7 + 1.2 i) == 1.1351827477151551088992008271819053 - 1.0927647857577371459105272080819308 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: cacos (0.7 + 1.2 i) == 1.1351827477151551088992008271819053 - 1.0927647857577371459105272080819308 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# cacosh +Test "Real part of: cacosh (-2 - 3 i) == -1.9833870299165354323470769028940395 + 2.1414491111159960199416055713254211 i": +double: 1 +float: 7 +idouble: 1 +ifloat: 7 +ildouble: 6 +ldouble: 6 +Test "Imaginary part of: cacosh (-2 - 3 i) == -1.9833870299165354323470769028940395 + 2.1414491111159960199416055713254211 i": +double: 1 +float: 3 +idouble: 1 +ifloat: 3 +ildouble: 1 +ldouble: 1 +Test "Real part of: cacosh (0.7 + 1.2 i) == 1.0927647857577371459105272080819308 + 1.1351827477151551088992008271819053 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: cacosh (0.7 + 1.2 i) == 1.0927647857577371459105272080819308 + 1.1351827477151551088992008271819053 i": +ildouble: 1 +ldouble: 1 + +# casin +Test "Real part of: casin (0.7 + 1.2 i) == 0.4356135790797415103321208644578462 + 1.0927647857577371459105272080819308 i": +double: 3 +float: 2 +idouble: 3 +ifloat: 2 +Test "Imaginary part of: casin (0.7 + 1.2 i) == 0.4356135790797415103321208644578462 + 1.0927647857577371459105272080819308 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# casinh +Test "Real part of: casinh (-2 - 3 i) == -1.9686379257930962917886650952454982 - 0.96465850440760279204541105949953237 i": +double: 5 +float: 1 +idouble: 5 +ifloat: 1 +ildouble: 5 +ldouble: 5 +Test "Imaginary part of: casinh (-2 - 3 i) == -1.9686379257930962917886650952454982 - 0.96465850440760279204541105949953237 i": +double: 3 +float: 6 +idouble: 3 +ifloat: 6 +ildouble: 5 +ldouble: 5 +Test "Real part of: casinh (0.7 + 1.2 i) == 0.97865459559367387689317593222160964 + 0.91135418953156011567903546856170941 i": +double: 1 +idouble: 1 +Test "Imaginary part of: casinh (0.7 + 1.2 i) == 0.97865459559367387689317593222160964 + 0.91135418953156011567903546856170941 i": +float: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +# catan +Test "Real part of: catan (-2 - 3 i) == -1.4099210495965755225306193844604208 - 0.22907268296853876629588180294200276 i": +float: 3 +ifloat: 3 +Test "Imaginary part of: catan (-2 - 3 i) == -1.4099210495965755225306193844604208 - 0.22907268296853876629588180294200276 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Real part of: catan (0.7 + 1.2 i) == 1.0785743834118921877443707996386368 + 0.57705737765343067644394541889341712 i": +float: 4 +ifloat: 4 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: catan (0.7 + 1.2 i) == 1.0785743834118921877443707996386368 + 0.57705737765343067644394541889341712 i": +double: 1 +idouble: 1 + +# catanh +Test "Real part of: catanh (-2 - 3 i) == -0.14694666622552975204743278515471595 - 1.3389725222944935611241935759091443 i": +double: 4 +idouble: 4 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: catanh (-2 - 3 i) == -0.14694666622552975204743278515471595 - 1.3389725222944935611241935759091443 i": +float: 4 +ifloat: 4 +Test "Real part of: catanh (0.7 + 1.2 i) == 0.2600749516525135959200648705635915 + 0.97024030779509898497385130162655963 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: catanh (0.7 + 1.2 i) == 0.2600749516525135959200648705635915 + 0.97024030779509898497385130162655963 i": +float: 6 +ifloat: 6 + +# cbrt +Test "cbrt (-0.001) == -0.1": +ildouble: 102 +ldouble: 102 +Test "cbrt (-27.0) == -3.0": +double: 1 +idouble: 1 +ildouble: 948 +ldouble: 948 +Test "cbrt (0.7) == 0.8879040017426007084": +ildouble: 345 +ldouble: 345 +Test "cbrt (0.970299) == 0.99": +double: 1 +idouble: 1 +ildouble: 142 +ldouble: 142 +Test "cbrt (8) == 2": +ildouble: 191 +ldouble: 191 + +# ccos +Test "Imaginary part of: ccos (-2 - 3 i) == -4.1896256909688072301 - 9.1092278937553365979 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Real part of: ccos (0.7 + 1.2 i) == 1.3848657645312111080 - 0.97242170335830028619 i": +double: 1 +idouble: 1 +Test "Imaginary part of: ccos (0.7 + 1.2 i) == 1.3848657645312111080 - 0.97242170335830028619 i": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +# ccosh +Test "Real part of: ccosh (-2 - 3 i) == -3.7245455049153225654 + 0.5118225699873846088 i": +float: 1 +ifloat: 1 +Test "Imaginary part of: ccosh (-2 - 3 i) == -3.7245455049153225654 + 0.5118225699873846088 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Real part of: ccosh (0.7 + 1.2 i) == 0.4548202223691477654 + 0.7070296600921537682 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: ccosh (0.7 + 1.2 i) == 0.4548202223691477654 + 0.7070296600921537682 i": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +# cexp +Test "Real part of: cexp (-2.0 - 3.0 i) == -0.13398091492954261346140525546115575 - 0.019098516261135196432576240858800925 i": +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: cexp (-2.0 - 3.0 i) == -0.13398091492954261346140525546115575 - 0.019098516261135196432576240858800925 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Real part of: cexp (0.7 + 1.2 i) == 0.72969890915032360123451688642930727 + 1.8768962328348102821139467908203072 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "Imaginary part of: cexp (0.7 + 1.2 i) == 0.72969890915032360123451688642930727 + 1.8768962328348102821139467908203072 i": +float: 1 +ifloat: 1 + +# clog +Test "Imaginary part of: clog (-2 - 3 i) == 1.2824746787307683680267437207826593 - 2.1587989303424641704769327722648368 i": +float: 3 +ifloat: 3 + +# clog10 +Test "Imaginary part of: clog10 (-0 + inf i) == inf + pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-0 - inf i) == inf - pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-2 - 3 i) == 0.5569716761534183846 - 0.9375544629863747085 i": +double: 1 +float: 5 +idouble: 1 +ifloat: 5 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: clog10 (-3 + inf i) == inf + pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-3 - inf i) == inf - pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-inf + 0 i) == inf + pi*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-inf + 1 i) == inf + pi*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-inf + inf i) == inf + 3/4 pi*log10(e) i": +double: 1 +idouble: 1 +Test "Imaginary part of: clog10 (-inf - 0 i) == inf - pi*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (-inf - 1 i) == inf - pi*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (0 + inf i) == inf + pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (0 - inf i) == inf - pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Real part of: clog10 (0.7 + 1.2 i) == 0.1427786545038868803 + 0.4528483579352493248 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: clog10 (0.7 + 1.2 i) == 0.1427786545038868803 + 0.4528483579352493248 i": +double: 1 +idouble: 1 +ildouble: 3 +ldouble: 3 +Test "Imaginary part of: clog10 (3 + inf i) == inf + pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (3 - inf i) == inf - pi/2*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (inf + inf i) == inf + pi/4*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: clog10 (inf - inf i) == inf - pi/4*log10(e) i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +# cos +Test "cos (0.7) == 0.76484218728448842625585999019186495": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "cos (M_PI_6l * 2.0) == 0.5": +double: 1 +float: 0.5 +idouble: 1 +ifloat: 0.5 +Test "cos (M_PI_6l * 4.0) == -0.5": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 0.5 +ldouble: 0.5 +Test "cos (pi/2) == 0": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 0.25 +ldouble: 0.25 + +# cosh +Test "cosh (0.7) == 1.255169005630943018": +ildouble: 2 +ldouble: 2 + +# cpow +Test "Real part of: cpow (2 + 3 i, 4 + 0 i) == -119.0 - 120.0 i": +double: 1 +float: 4 +idouble: 1 +ifloat: 4 +Test "Imaginary part of: cpow (2 + 3 i, 4 + 0 i) == -119.0 - 120.0 i": +float: 2 +ifloat: 2 +ildouble: 2 +ldouble: 2 +Test "Imaginary part of: cpow (e + 0 i, 0 + 2 * M_PIl i) == 1.0 + 0.0 i": +double: 2 +float: 2 +idouble: 2 +ifloat: 2 +ildouble: 1 +ldouble: 1 +Test "Real part of: cpow (2 + 0 i, 10 + 0 i) == 1024.0 + 0.0 i": +ildouble: 1 +ldouble: 1 + +# csin +Test "Imaginary part of: csin (0.7 + 1.2 i) == 1.1664563419657581376 + 1.1544997246948547371 i": +float: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +# csinh +Test "Imaginary part of: csinh (-2 - 3 i) == 3.5905645899857799520 - 0.5309210862485198052 i": +double: 1 +idouble: 1 +ildouble: 2 +ldouble: 2 +Test "Real part of: csinh (0.7 + 1.2 i) == 0.27487868678117583582 + 1.1698665727426565139 i": +float: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "Imaginary part of: csinh (0.7 + 1.2 i) == 0.27487868678117583582 + 1.1698665727426565139 i": +float: 1 +ifloat: 1 + +# csqrt +Test "Real part of: csqrt (-2 + 3 i) == 0.89597747612983812471573375529004348 + 1.6741492280355400404480393008490519 i": +float: 1 +ifloat: 1 +Test "Real part of: csqrt (-2 - 3 i) == 0.89597747612983812471573375529004348 - 1.6741492280355400404480393008490519 i": +float: 1 +ifloat: 1 +Test "Real part of: csqrt (0.7 + 1.2 i) == 1.022067610030026450706487883081139 + 0.58704531296356521154977678719838035 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "Imaginary part of: csqrt (0.7 + 1.2 i) == 1.022067610030026450706487883081139 + 0.58704531296356521154977678719838035 i": +float: 1 +ifloat: 1 + +# ctan +Test "Real part of: ctan (-2 - 3 i) == 0.0037640256415042482 - 1.0032386273536098014 i": +double: 1 +idouble: 1 +ildouble: 439 +ldouble: 439 +Test "Imaginary part of: ctan (-2 - 3 i) == 0.0037640256415042482 - 1.0032386273536098014 i": +ildouble: 2 +ldouble: 2 +Test "Real part of: ctan (0.7 + 1.2 i) == 0.1720734197630349001 + 0.9544807059989405538 i": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: ctan (0.7 + 1.2 i) == 0.1720734197630349001 + 0.9544807059989405538 i": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +# ctanh +Test "Real part of: ctanh (-2 - 3 i) == -0.9653858790221331242 + 0.0098843750383224937 i": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 5 +ldouble: 5 +Test "Imaginary part of: ctanh (-2 - 3 i) == -0.9653858790221331242 + 0.0098843750383224937 i": +ildouble: 25 +ldouble: 25 +Test "Imaginary part of: ctanh (0 + pi/4 i) == 0.0 + 1.0 i": +float: 1 +ifloat: 1 +Test "Real part of: ctanh (0.7 + 1.2 i) == 1.3472197399061191630 + 0.4778641038326365540 i": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "Imaginary part of: ctanh (0.7 + 1.2 i) == 1.3472197399061191630 + 0.4778641038326365540 i": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# erfc +Test "erfc (0.7) == 0.32219880616258152702": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "erfc (1.2) == 0.089686021770364619762": +double: 2 +float: 2 +idouble: 2 +ifloat: 2 +ildouble: 3 +ldouble: 3 +Test "erfc (2.0) == 0.0046777349810472658379": +double: 1 +idouble: 1 +Test "erfc (4.1) == 0.67000276540848983727e-8": +double: 24 +float: 12 +idouble: 24 +ifloat: 12 +ildouble: 12 +ldouble: 12 +Test "erfc (9) == 0.41370317465138102381e-36": +ildouble: 36 +ldouble: 36 + +# exp10 +Test "exp10 (-1) == 0.1": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "exp10 (0.7) == 5.0118723362727228500155418688494574": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "exp10 (3) == 1000": +double: 6 +float: 2 +idouble: 6 +ifloat: 2 +ildouble: 3 +ldouble: 3 + +# expm1 +Test "expm1 (0.7) == 1.0137527074704765216": +ildouble: 1 +ldouble: 1 +Test "expm1 (1) == M_El - 1.0": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +# fmod +Test "fmod (-6.5, -2.3) == -1.9": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "fmod (-6.5, 2.3) == -1.9": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "fmod (6.5, -2.3) == 1.9": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "fmod (6.5, 2.3) == 1.9": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# gamma +Test "gamma (-0.5) == log(2*sqrt(pi))": +ildouble: 1 +ldouble: 1 + +# hypot +Test "hypot (-0.7, -12.4) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (-0.7, 12.4) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (-12.4, -0.7) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (-12.4, 0.7) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (0.7, -12.4) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (0.7, 1.2) == 1.3892443989449804508432547041028554": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "hypot (0.7, 12.4) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (12.4, -0.7) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 +Test "hypot (12.4, 0.7) == 12.419742348374220601176836866763271": +float: 1 +ifloat: 1 + +# j0 +Test "j0 (10.0) == -0.24593576445134833520": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "j0 (2.0) == 0.22389077914123566805": +float: 2 +ifloat: 2 +Test "j0 (8.0) == 0.17165080713755390609": +float: 1 +ifloat: 1 + +# j1 +Test "j1 (10.0) == 0.043472746168861436670": +float: 2 +ifloat: 2 +ildouble: 2 +ldouble: 2 +Test "j1 (2.0) == 0.57672480775687338720": +double: 1 +idouble: 1 +Test "j1 (8.0) == 0.23463634685391462438": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +# jn +Test "jn (0, 10.0) == -0.24593576445134833520": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "jn (0, 2.0) == 0.22389077914123566805": +float: 2 +ifloat: 2 +Test "jn (0, 8.0) == 0.17165080713755390609": +float: 1 +ifloat: 1 +Test "jn (1, 10.0) == 0.043472746168861436670": +float: 2 +ifloat: 2 +ildouble: 2 +ldouble: 2 +Test "jn (1, 2.0) == 0.57672480775687338720": +double: 1 +idouble: 1 +Test "jn (1, 8.0) == 0.23463634685391462438": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "jn (10, -1.0) == 0.26306151236874532070e-9": +ildouble: 1 +ldouble: 1 +Test "jn (10, 0.1) == 0.26905328954342155795e-19": +double: 6 +float: 4 +idouble: 6 +ifloat: 4 +ildouble: 1 +ldouble: 1 +Test "jn (10, 0.7) == 0.75175911502153953928e-11": +double: 3 +float: 1 +idouble: 3 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "jn (10, 1.0) == 0.26306151236874532070e-9": +ildouble: 1 +ldouble: 1 +Test "jn (10, 10.0) == 0.20748610663335885770": +double: 4 +float: 3 +idouble: 4 +ifloat: 3 +ildouble: 2 +ldouble: 2 +Test "jn (10, 2.0) == 0.25153862827167367096e-6": +float: 4 +ifloat: 4 +ildouble: 1 +ldouble: 1 +Test "jn (3, -1.0) == -0.019563353982668405919": +ildouble: 1 +ldouble: 1 +Test "jn (3, 0.1) == 0.000020820315754756261429": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "jn (3, 0.7) == 0.0069296548267508408077": +float: 1 +ifloat: 1 +Test "jn (3, 1.0) == 0.019563353982668405919": +ildouble: 1 +ldouble: 1 +Test "jn (3, 10.0) == 0.058379379305186812343": +double: 3 +float: 1 +idouble: 3 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "jn (3, 2.0) == 0.12894324947440205110": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 + +# lgamma +Test "lgamma (-0.5) == log(2*sqrt(pi))": +ildouble: 1 +ldouble: 1 +Test "lgamma (0.7) == 0.26086724653166651439": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "lgamma (1.2) == -0.853740900033158497197e-1": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 1 +ldouble: 1 + +# log +Test "log (0.7) == -0.35667494393873237891263871124118447": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# log10 +Test "log10 (0.7) == -0.15490195998574316929": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "log10 (e) == log10(e)": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# log1p +Test "log1p (-0.3) == -0.35667494393873237891263871124118447": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# log2 +Test "log2 (0.7) == -0.51457317282975824043": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +# sin +Test "sin (0.7) == 0.64421768723769105367261435139872014": +ildouble: 1 +ldouble: 1 + +# sincos +Test "sincos (0.7, &sin_res, &cos_res) puts 0.64421768723769105367261435139872014 in sin_res": +ildouble: 1 +ldouble: 1 +Test "sincos (0.7, &sin_res, &cos_res) puts 0.76484218728448842625585999019186495 in cos_res": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "sincos (M_PI_6l*2.0, &sin_res, &cos_res) puts 0.5 in cos_res": +double: 1 +float: 0.5 +idouble: 1 +ifloat: 0.5 +Test "sincos (M_PI_6l*2.0, &sin_res, &cos_res) puts 0.86602540378443864676372317075293616 in sin_res": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "sincos (pi/2, &sin_res, &cos_res) puts 0 in cos_res": +double: 0.2758 +float: 0.3667 +idouble: 0.2758 +ifloat: 0.3667 +ildouble: 0.25 +ldouble: 0.25 +Test "sincos (pi/6, &sin_res, &cos_res) puts 0.86602540378443864676372317075293616 in cos_res": +float: 1 +ifloat: 1 + +# sinh +Test "sinh (0.7) == 0.75858370183953350346": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# tan +Test "tan (0.7) == 0.84228838046307944812813500221293775": +ildouble: 1 +ldouble: 1 +Test "tan (pi/4) == 1": +double: 0.5 +idouble: 0.5 + +# tanh +Test "tanh (0.7) == 0.60436777711716349631": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Test "tanh (-0.7) == -0.60436777711716349631": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# tgamma +Test "tgamma (-0.5) == -2 sqrt (pi)": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "tgamma (0.5) == sqrt (pi)": +float: 1 +ifloat: 1 +Test "tgamma (0.7) == 1.29805533264755778568": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "tgamma (4) == 6": +ildouble: 2 +ldouble: 2 + +# y0 +Test "y0 (0.7) == -0.19066492933739506743": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "y0 (1.0) == 0.088256964215676957983": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "y0 (1.5) == 0.38244892379775884396": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "y0 (10.0) == 0.055671167283599391424": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "y0 (8.0) == 0.22352148938756622053": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +# y1 +Test "y1 (0.1) == -6.4589510947020269877": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "y1 (0.7) == -1.1032498719076333697": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "y1 (1.5) == -0.41230862697391129595": +float: 1 +ifloat: 1 +Test "y1 (10.0) == 0.24901542420695388392": +double: 3 +float: 1 +idouble: 3 +ifloat: 1 +Test "y1 (2.0) == -0.10703243154093754689": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "y1 (8.0) == -0.15806046173124749426": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 2 +ldouble: 2 + +# yn +Test "yn (0, 0.7) == -0.19066492933739506743": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "yn (0, 1.0) == 0.088256964215676957983": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "yn (0, 1.5) == 0.38244892379775884396": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +Test "yn (0, 10.0) == 0.055671167283599391424": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "yn (0, 8.0) == 0.22352148938756622053": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "yn (1, 0.1) == -6.4589510947020269877": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 +Test "yn (1, 0.7) == -1.1032498719076333697": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "yn (1, 1.5) == -0.41230862697391129595": +float: 1 +ifloat: 1 +Test "yn (1, 10.0) == 0.24901542420695388392": +double: 3 +float: 1 +idouble: 3 +ifloat: 1 +Test "yn (1, 2.0) == -0.10703243154093754689": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 +Test "yn (1, 8.0) == -0.15806046173124749426": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 2 +ldouble: 2 +Test "yn (10, 0.1) == -0.11831335132045197885e19": +double: 2 +float: 2 +idouble: 2 +ifloat: 2 +ildouble: 2 +ldouble: 2 +Test "yn (10, 0.7) == -0.42447194260703866924e10": +double: 3 +idouble: 3 +ildouble: 7 +ldouble: 7 +Test "yn (10, 1.0) == -0.12161801427868918929e9": +double: 1 +idouble: 1 +Test "yn (10, 10.0) == -0.35981415218340272205": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "yn (10, 2.0) == -129184.54220803928264": +double: 2 +idouble: 2 +ildouble: 1 +ldouble: 1 +Test "yn (3, 0.1) == -5099.3323786129048894": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "yn (3, 0.7) == -15.819479052819633505": +double: 3 +float: 1 +idouble: 3 +ifloat: 1 +ildouble: 2 +ldouble: 2 +Test "yn (3, 10.0) == -0.25136265718383732978": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +Test "yn (3, 2.0) == -1.1277837768404277861": +double: 1 +idouble: 1 + +# Maximal error of functions: +Function: "asin": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +Function: "asinh": +ildouble: 15 +ldouble: 15 + +Function: "atan2": +float: 4 +ifloat: 4 + +Function: "atanh": +double: 1 +idouble: 1 +ildouble: 1 +ldouble: 1 + +Function: "cabs": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Real part of "cacos": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "cacos": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Real part of "cacosh": +double: 1 +float: 7 +idouble: 1 +ifloat: 7 +ildouble: 6 +ldouble: 6 + +Function: Imaginary part of "cacosh": +double: 1 +float: 3 +idouble: 1 +ifloat: 3 +ildouble: 1 +ldouble: 1 + +Function: Real part of "casin": +double: 3 +float: 2 +idouble: 3 +ifloat: 2 + +Function: Imaginary part of "casin": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Real part of "casinh": +double: 5 +float: 1 +idouble: 5 +ifloat: 1 +ildouble: 5 +ldouble: 5 + +Function: Imaginary part of "casinh": +double: 3 +float: 6 +idouble: 3 +ifloat: 6 +ildouble: 5 +ldouble: 5 + +Function: Real part of "catan": +float: 4 +ifloat: 4 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "catan": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +Function: Real part of "catanh": +double: 4 +float: 1 +idouble: 4 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "catanh": +float: 6 +ifloat: 6 + +Function: "cbrt": +double: 1 +idouble: 1 +ildouble: 948 +ldouble: 948 + +Function: Real part of "ccos": +double: 1 +idouble: 1 + +Function: Imaginary part of "ccos": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Real part of "ccosh": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "ccosh": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Real part of "cexp": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: Imaginary part of "cexp": +float: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "clog": +float: 3 +ifloat: 3 + +Function: Real part of "clog10": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "clog10": +double: 1 +float: 5 +idouble: 1 +ifloat: 5 +ildouble: 3 +ldouble: 3 + +Function: "cos": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 0.5 +ldouble: 0.5 + +Function: "cosh": +ildouble: 2 +ldouble: 2 + +Function: Real part of "cpow": +double: 1 +float: 4 +idouble: 1 +ifloat: 4 +ildouble: 1 +ldouble: 1 + +Function: Imaginary part of "cpow": +double: 1.1031 +float: 2 +idouble: 1.1031 +ifloat: 2 +ildouble: 2 +ldouble: 2 + +Function: Imaginary part of "csin": +float: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: Real part of "csinh": +float: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: Imaginary part of "csinh": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: Real part of "csqrt": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +Function: Imaginary part of "csqrt": +float: 1 +ifloat: 1 + +Function: Real part of "ctan": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 439 +ldouble: 439 + +Function: Imaginary part of "ctan": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: Real part of "ctanh": +double: 2 +float: 2 +idouble: 2 +ifloat: 2 +ildouble: 5 +ldouble: 5 + +Function: Imaginary part of "ctanh": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 25 +ldouble: 25 + +Function: "erfc": +double: 24 +float: 12 +idouble: 24 +ifloat: 12 +ildouble: 36 +ldouble: 36 + +Function: "exp10": +double: 6 +float: 2 +idouble: 6 +ifloat: 2 +ildouble: 3 +ldouble: 3 + +Function: "expm1": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "fmod": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "gamma": +ildouble: 1 +ldouble: 1 + +Function: "hypot": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "j0": +double: 2 +float: 2 +idouble: 2 +ifloat: 2 + +Function: "j1": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 2 +ldouble: 2 + +Function: "jn": +double: 6 +float: 4 +idouble: 6 +ifloat: 4 +ildouble: 2 +ldouble: 2 + +Function: "lgamma": +double: 1 +float: 2 +idouble: 1 +ifloat: 2 +ildouble: 1 +ldouble: 1 + +Function: "log": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "log10": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "log1p": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "log2": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 + +Function: "sin": +ildouble: 1 +ldouble: 1 + +Function: "sincos": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "sinh": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "tan": +double: 0.5 +idouble: 0.5 +ildouble: 1 +ldouble: 1 + +Function: "tanh": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 1 +ldouble: 1 + +Function: "tgamma": +double: 1 +float: 1 +idouble: 1 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: "y0": +double: 2 +float: 1 +idouble: 2 +ifloat: 1 +ildouble: 2 +ldouble: 2 + +Function: "y1": +double: 3 +float: 2 +idouble: 3 +ifloat: 2 +ildouble: 2 +ldouble: 2 + +Function: "yn": +double: 3 +float: 2 +idouble: 3 +ifloat: 2 +ildouble: 7 +ldouble: 7 + +# end of automatic generation diff --git a/sysdeps/x86_64/fpu/math_ldbl.h b/sysdeps/x86_64/fpu/math_ldbl.h new file mode 100644 index 0000000000..b9ff8dadaf --- /dev/null +++ b/sysdeps/x86_64/fpu/math_ldbl.h @@ -0,0 +1,79 @@ +#ifndef _MATH_PRIVATE_H_ +#error "Never use <math_ldbl.h> directly; include <math_private.h> instead." +#endif + +/* A union which permits us to convert between a long double and + three 32 bit ints. */ + +typedef union +{ + long double value; + struct + { + u_int32_t lsw; + u_int32_t msw; + int sign_exponent:16; + unsigned int empty1:16; + unsigned int empty0:32; + } parts; +} ieee_long_double_shape_type; + +/* Get three 32 bit ints from a double. */ + +#define GET_LDOUBLE_WORDS(exp,ix0,ix1,d) \ +do { \ + ieee_long_double_shape_type ew_u; \ + ew_u.value = (d); \ + (exp) = ew_u.parts.sign_exponent; \ + (ix0) = ew_u.parts.msw; \ + (ix1) = ew_u.parts.lsw; \ +} while (0) + +/* Set a double from two 32 bit ints. */ + +#define SET_LDOUBLE_WORDS(d,exp,ix0,ix1) \ +do { \ + ieee_long_double_shape_type iw_u; \ + iw_u.parts.sign_exponent = (exp); \ + iw_u.parts.msw = (ix0); \ + iw_u.parts.lsw = (ix1); \ + (d) = iw_u.value; \ +} while (0) + +/* Get the more significant 32 bits of a long double mantissa. */ + +#define GET_LDOUBLE_MSW(v,d) \ +do { \ + ieee_long_double_shape_type sh_u; \ + sh_u.value = (d); \ + (v) = sh_u.parts.msw; \ +} while (0) + +/* Set the more significant 32 bits of a long double mantissa from an int. */ + +#define SET_LDOUBLE_MSW(d,v) \ +do { \ + ieee_long_double_shape_type sh_u; \ + sh_u.value = (d); \ + sh_u.parts.msw = (v); \ + (d) = sh_u.value; \ +} while (0) + +/* Get int from the exponent of a long double. */ + +#define GET_LDOUBLE_EXP(exp,d) \ +do { \ + ieee_long_double_shape_type ge_u; \ + ge_u.value = (d); \ + (exp) = ge_u.parts.sign_exponent; \ +} while (0) + +/* Set exponent of a long double from an int. */ + +#define SET_LDOUBLE_EXP(d,exp) \ +do { \ + ieee_long_double_shape_type se_u; \ + se_u.value = (d); \ + se_u.parts.sign_exponent = (exp); \ + (d) = se_u.value; \ +} while (0) diff --git a/sysdeps/x86_64/fpu/printf_fphex.c b/sysdeps/x86_64/fpu/printf_fphex.c new file mode 100644 index 0000000000..d6ca102a88 --- /dev/null +++ b/sysdeps/x86_64/fpu/printf_fphex.c @@ -0,0 +1,92 @@ +/* Print floating point number in hexadecimal notation according to ISO C99. + Copyright (C) 1997, 1998, 1999, 2000, 2001 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#ifndef LONG_DOUBLE_DENORM_BIAS +# define LONG_DOUBLE_DENORM_BIAS (IEEE854_LONG_DOUBLE_BIAS - 1) +#endif + +#define PRINT_FPHEX_LONG_DOUBLE \ +do { \ + /* The "strange" 80 bit format on ix86 and m68k has an explicit \ + leading digit in the 64 bit mantissa. */ \ + unsigned long long int num; \ + \ + \ + num = (((unsigned long long int) fpnum.ldbl.ieee.mantissa0) << 32 \ + | fpnum.ldbl.ieee.mantissa1); \ + \ + zero_mantissa = num == 0; \ + \ + if (sizeof (unsigned long int) > 6) \ + { \ + numstr = _itoa_word (num, numbuf + sizeof numbuf, 16, \ + info->spec == 'A'); \ + wnumstr = _itowa_word (num, \ + wnumbuf + sizeof (wnumbuf) / sizeof (wchar_t),\ + 16, info->spec == 'A'); \ + } \ + else \ + { \ + numstr = _itoa (num, numbuf + sizeof numbuf, 16, info->spec == 'A');\ + wnumstr = _itowa (num, \ + wnumbuf + sizeof (wnumbuf) / sizeof (wchar_t), \ + 16, info->spec == 'A'); \ + } \ + \ + /* Fill with zeroes. */ \ + while (numstr > numbuf + (sizeof numbuf - 64 / 4)) \ + { \ + *--numstr = '0'; \ + *--wnumstr = L'0'; \ + } \ + \ + /* We use a full nibble for the leading digit. */ \ + leading = *numstr++; \ + \ + /* We have 3 bits from the mantissa in the leading nibble. \ + Therefore we are here using `IEEE854_LONG_DOUBLE_BIAS + 3'. */ \ + exponent = fpnum.ldbl.ieee.exponent; \ + \ + if (exponent == 0) \ + { \ + if (zero_mantissa) \ + expnegative = 0; \ + else \ + { \ + /* This is a denormalized number. */ \ + expnegative = 1; \ + /* This is a hook for the m68k long double format, where the \ + exponent bias is the same for normalized and denormalized \ + numbers. */ \ + exponent = LONG_DOUBLE_DENORM_BIAS + 3; \ + } \ + } \ + else if (exponent >= IEEE854_LONG_DOUBLE_BIAS + 3) \ + { \ + expnegative = 0; \ + exponent -= IEEE854_LONG_DOUBLE_BIAS + 3; \ + } \ + else \ + { \ + expnegative = 1; \ + exponent = -(exponent - (IEEE854_LONG_DOUBLE_BIAS + 3)); \ + } \ +} while (0) + +#include <sysdeps/generic/printf_fphex.c> diff --git a/sysdeps/x86_64/fpu/s_atanl.c b/sysdeps/x86_64/fpu/s_atanl.c new file mode 100644 index 0000000000..fd4a455b55 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_atanl.c @@ -0,0 +1 @@ +#include "sysdeps/i386/fpu/s_atanl.c" diff --git a/sysdeps/x86_64/fpu/s_cosl.S b/sysdeps/x86_64/fpu/s_cosl.S new file mode 100644 index 0000000000..6636fb5ec6 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_cosl.S @@ -0,0 +1,32 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + +ENTRY(__cosl) + fldt 8(%rsp) + fcos + fnstsw %ax + testl $0x400,%eax + jnz 1f + ret + .align ALIGNARG(4) +1: fldpi + fadd %st(0) + fxch %st(1) +2: fprem1 + fnstsw %ax + testl $0x400,%eax + jnz 2b + fstp %st(1) + fcos + ret +END (__cosl) +weak_alias (__cosl, cosl) diff --git a/sysdeps/x86_64/fpu/s_expm1l.S b/sysdeps/x86_64/fpu/s_expm1l.S new file mode 100644 index 0000000000..fa40e05572 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_expm1l.S @@ -0,0 +1,83 @@ +/* ix87 specific implementation of exp(x)-1. + Copyright (C) 1996, 1997, 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Ulrich Drepper <drepper@cygnus.com>, 1996. + Based on code by John C. Bowman <bowman@ipp-garching.mpg.de>. + Corrections by H.J. Lu (hjl@gnu.ai.mit.edu), 1997. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + + /* Using: e^x - 1 = 2^(x * log2(e)) - 1 */ + +#include <machine/asm.h> + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + .align ALIGNARG(4) + ASM_TYPE_DIRECTIVE(minus1,@object) +minus1: .double -1.0 + ASM_SIZE_DIRECTIVE(minus1) + ASM_TYPE_DIRECTIVE(one,@object) +one: .double 1.0 + ASM_SIZE_DIRECTIVE(one) + ASM_TYPE_DIRECTIVE(l2e,@object) +l2e: .tfloat 1.442695040888963407359924681002 + ASM_SIZE_DIRECTIVE(l2e) + +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__expm1l) + fldt 8(%rsp) // x + fxam // Is NaN or +-Inf? + fstsw %ax + movb $0x45, %ch + andb %ah, %ch + cmpb $0x40, %ch + je 3f // If +-0, jump. + cmpb $0x05, %ch + je 2f // If +-Inf, jump. + + fldt MO(l2e) // log2(e) : x + fmulp // log2(e)*x + fld %st // log2(e)*x : log2(e)*x + frndint // int(log2(e)*x) : log2(e)*x + fsubr %st, %st(1) // int(log2(e)*x) : fract(log2(e)*x) + fxch // fract(log2(e)*x) : int(log2(e)*x) + f2xm1 // 2^fract(log2(e)*x)-1 : int(log2(e)*x) + fscale // 2^(log2(e)*x)-2^int(log2(e)*x) : int(log2(e)*x) + fxch // int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x) + fldl MO(one) // 1 : int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x) + fscale // 2^int(log2(e)*x) : int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x) + fsubrl MO(one) // 1-2^int(log2(e)*x) : int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x) + fstp %st(1) // 1-2^int(log2(e)*x) : 2^(log2(e)*x)-2^int(log2(e)*x) + fsubrp %st, %st(1) // 2^(log2(e)*x) + ret + +2: testl $0x200, %eax // Test sign. + jz 3f // If positive, jump. + fstp %st + fldl MO(minus1) // Set result to -1.0. +3: ret +END(__expm1l) +weak_alias (__expm1l, expm1l) diff --git a/sysdeps/x86_64/fpu/s_fpclassifyl.c b/sysdeps/x86_64/fpu/s_fpclassifyl.c new file mode 100644 index 0000000000..856854b0f5 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_fpclassifyl.c @@ -0,0 +1,2 @@ +#include <sysdeps/i386/fpu/s_fpclassifyl.c> + diff --git a/sysdeps/x86_64/fpu/s_isinfl.c b/sysdeps/x86_64/fpu/s_isinfl.c new file mode 100644 index 0000000000..ca818b5e90 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_isinfl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_isinfl.c> diff --git a/sysdeps/x86_64/fpu/s_isnanl.c b/sysdeps/x86_64/fpu/s_isnanl.c new file mode 100644 index 0000000000..06e69c3aeb --- /dev/null +++ b/sysdeps/x86_64/fpu/s_isnanl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_isnanl.c> diff --git a/sysdeps/x86_64/fpu/s_log1pl.S b/sysdeps/x86_64/fpu/s_log1pl.S new file mode 100644 index 0000000000..7fbd0e5aaa --- /dev/null +++ b/sysdeps/x86_64/fpu/s_log1pl.S @@ -0,0 +1,71 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: s_log1p.S,v 1.7 1995/05/09 00:10:58 jtc Exp $") + +#ifdef __ELF__ + .section .rodata +#else + .text +#endif + .align ALIGNARG(4) + /* The fyl2xp1 can only be used for values in + -1 + sqrt(2) / 2 <= x <= 1 - sqrt(2) / 2 + 0.29 is a safe value. + */ +limit: .tfloat 0.29 + /* Please note: we use a double value here. Since 1.0 has + an exact representation this does not effect the accuracy + but it helps to optimize the code. */ +one: .double 1.0 + +/* + * Use the fyl2xp1 function when the argument is in the range -0.29 to 0.29, + * otherwise fyl2x with the needed extra computation. + */ +#ifdef PIC +#define MO(op) op##(%rip) +#else +#define MO(op) op +#endif + + .text +ENTRY(__log1pl) + fldln2 + + fldt 8(%rsp) + + fxam + fnstsw + fld %st + andb $1,%ah + jnz 3f // in case x is NaN or ħInf +4: + fabs + fldt MO(limit) + fcompp + fnstsw + andb $1,%ah + jz 2f + + faddl MO(one) + fyl2x + ret + +2: fyl2xp1 + ret + +3: jp 4b // in case x is ħInf + fstp %st(1) + fstp %st(1) + ret + +END (__log1pl) +weak_alias (__log1pl, log1pl) diff --git a/sysdeps/x86_64/fpu/s_logbl.c b/sysdeps/x86_64/fpu/s_logbl.c new file mode 100644 index 0000000000..4791ba64e8 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_logbl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_logbl.c> diff --git a/sysdeps/x86_64/fpu/s_nextafterl.c b/sysdeps/x86_64/fpu/s_nextafterl.c new file mode 100644 index 0000000000..f59f16848f --- /dev/null +++ b/sysdeps/x86_64/fpu/s_nextafterl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_nextafterl.c> diff --git a/sysdeps/x86_64/fpu/s_nexttoward.c b/sysdeps/x86_64/fpu/s_nexttoward.c new file mode 100644 index 0000000000..aee2bb5895 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_nexttoward.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_nexttoward.c> diff --git a/sysdeps/x86_64/fpu/s_nexttowardf.c b/sysdeps/x86_64/fpu/s_nexttowardf.c new file mode 100644 index 0000000000..55e95f6916 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_nexttowardf.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_nexttowardf.c> diff --git a/sysdeps/x86_64/fpu/s_rintl.c b/sysdeps/x86_64/fpu/s_rintl.c new file mode 100644 index 0000000000..1cad42e921 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_rintl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_rintl.c> diff --git a/sysdeps/x86_64/fpu/s_significandl.c b/sysdeps/x86_64/fpu/s_significandl.c new file mode 100644 index 0000000000..a4ad986164 --- /dev/null +++ b/sysdeps/x86_64/fpu/s_significandl.c @@ -0,0 +1 @@ +#include <sysdeps/i386/fpu/s_significandl.c> diff --git a/sysdeps/x86_64/fpu/s_sincosl.S b/sysdeps/x86_64/fpu/s_sincosl.S new file mode 100644 index 0000000000..9a3025ab5f --- /dev/null +++ b/sysdeps/x86_64/fpu/s_sincosl.S @@ -0,0 +1,61 @@ +/* Compute sine and cosine of argument. + Copyright (C) 1997, 2000, 2001 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Ulrich Drepper <drepper@cygnus.com>, 1997. + + 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, write to the Free + Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA + 02111-1307 USA. */ + +#include <machine/asm.h> +#include "bp-sym.h" +#include "bp-asm.h" + +#define PARMS LINKAGE /* no space for saved regs */ +#define ANGLE PARMS +#define SINP ANGLE+12 +#define COSP SINP+PTR_SIZE + + .text +ENTRY (BP_SYM (__sincosl)) + ENTER + + fldt 8(%rsp) + fsincos + fnstsw %ax + testl $0x400,%eax + jnz 1f + fstpt (%rsi) + fstpt (%rdi) + + LEAVE + ret + + .align ALIGNARG(4) +1: fldpi + fadd %st(0) + fxch %st(1) +2: fprem1 + fnstsw %ax + testl $0x400,%eax + jnz 2b + fstp %st(1) + fsincos + fstpt (%rsi) + fstpt (%rdi) + + LEAVE + ret +END (BP_SYM (__sincosl)) +weak_alias (BP_SYM (__sincosl), BP_SYM (sincosl)) diff --git a/sysdeps/x86_64/fpu/s_sinl.S b/sysdeps/x86_64/fpu/s_sinl.S new file mode 100644 index 0000000000..181f112f4f --- /dev/null +++ b/sysdeps/x86_64/fpu/s_sinl.S @@ -0,0 +1,30 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +ENTRY(__sinl) + fldt 8(%rsp) + fsin + fnstsw %ax + testl $0x400,%eax + jnz 1f + ret + .align ALIGNARG(4) +1: fldpi + fadd %st(0) + fxch %st(1) +2: fprem1 + fnstsw %ax + testl $0x400,%eax + jnz 2b + fstp %st(1) + fsin + ret +END (__sinl) +weak_alias (__sinl, sinl) diff --git a/sysdeps/x86_64/fpu/s_tanl.S b/sysdeps/x86_64/fpu/s_tanl.S new file mode 100644 index 0000000000..674e908acc --- /dev/null +++ b/sysdeps/x86_64/fpu/s_tanl.S @@ -0,0 +1,33 @@ +/* + * Written by J.T. Conklin <jtc@netbsd.org>. + * Public domain. + * + * Adapted for `long double' by Ulrich Drepper <drepper@cygnus.com>. + * Adapted for x86-64 by Andreas Jaeger <aj@suse.de>. + */ + +#include <machine/asm.h> + +RCSID("$NetBSD: $") + +ENTRY(__tanl) + fldt 8(%rsp) + fptan + fnstsw %ax + testl $0x400,%eax + jnz 1f + fstp %st(0) + ret +1: fldpi + fadd %st(0) + fxch %st(1) +2: fprem1 + fstsw %ax + testl $0x400,%eax + jnz 2b + fstp %st(1) + fptan + fstp %st(0) + ret +END (__tanl) +weak_alias (__tanl, tanl) |