wine/include/msvcrt/math.h
Piotr Caban fd22e3b61b msvcrt: Import remainder implementation from musl.
Signed-off-by: Piotr Caban <piotr@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
2021-05-19 16:56:50 +02:00

341 lines
11 KiB
C

/*
* Math functions.
*
* Derived from the mingw header written by Colin Peters.
* Modified for Wine use by Hans Leidekker.
* This file is in the public domain.
*/
#ifndef __WINE_MATH_H
#define __WINE_MATH_H
#include <corecrt.h>
#include <pshpack8.h>
#ifdef __cplusplus
extern "C" {
#endif
#define _DOMAIN 1 /* domain error in argument */
#define _SING 2 /* singularity */
#define _OVERFLOW 3 /* range overflow */
#define _UNDERFLOW 4 /* range underflow */
#define _TLOSS 5 /* total loss of precision */
#define _PLOSS 6 /* partial loss of precision */
#ifndef _EXCEPTION_DEFINED
#define _EXCEPTION_DEFINED
struct _exception
{
int type;
char *name;
double arg1;
double arg2;
double retval;
};
#endif /* _EXCEPTION_DEFINED */
#ifndef _COMPLEX_DEFINED
#define _COMPLEX_DEFINED
struct _complex
{
double x; /* Real part */
double y; /* Imaginary part */
};
#endif /* _COMPLEX_DEFINED */
_ACRTIMP double __cdecl sin(double);
_ACRTIMP double __cdecl cos(double);
_ACRTIMP double __cdecl tan(double);
_ACRTIMP double __cdecl sinh(double);
_ACRTIMP double __cdecl cosh(double);
_ACRTIMP double __cdecl tanh(double);
_ACRTIMP double __cdecl asin(double);
_ACRTIMP double __cdecl acos(double);
_ACRTIMP double __cdecl atan(double);
_ACRTIMP double __cdecl atan2(double, double);
_ACRTIMP double __cdecl asinh(double);
_ACRTIMP double __cdecl acosh(double);
_ACRTIMP double __cdecl atanh(double);
_ACRTIMP double __cdecl exp(double);
_ACRTIMP double __cdecl log(double);
_ACRTIMP double __cdecl log10(double);
_ACRTIMP double __cdecl pow(double, double);
_ACRTIMP double __cdecl sqrt(double);
_ACRTIMP double __cdecl ceil(double);
_ACRTIMP double __cdecl floor(double);
_ACRTIMP double __cdecl fabs(double);
_ACRTIMP double __cdecl ldexp(double, int);
_ACRTIMP double __cdecl frexp(double, int*);
_ACRTIMP double __cdecl modf(double, double*);
_ACRTIMP double __cdecl fmod(double, double);
_ACRTIMP double __cdecl fmin(double, double);
_ACRTIMP double __cdecl fmax(double, double);
_ACRTIMP double __cdecl erf(double);
_ACRTIMP double __cdecl remquo(double, double, int*);
_ACRTIMP float __cdecl remquof(float, float, int*);
_ACRTIMP double __cdecl _hypot(double, double);
_ACRTIMP double __cdecl _j0(double);
_ACRTIMP double __cdecl _j1(double);
_ACRTIMP double __cdecl _jn(int, double);
_ACRTIMP double __cdecl _y0(double);
_ACRTIMP double __cdecl _y1(double);
_ACRTIMP double __cdecl _yn(int, double);
_ACRTIMP double __cdecl cbrt(double);
_ACRTIMP double __cdecl exp2(double);
_ACRTIMP double __cdecl log2(double);
_ACRTIMP double __cdecl rint(double);
_ACRTIMP double __cdecl round(double);
_ACRTIMP double __cdecl trunc(double);
_ACRTIMP float __cdecl cbrtf(float);
_ACRTIMP float __cdecl exp2f(float);
_ACRTIMP float __cdecl log2f(float);
_ACRTIMP float __cdecl rintf(float);
_ACRTIMP float __cdecl roundf(float);
_ACRTIMP float __cdecl truncf(float);
_ACRTIMP __msvcrt_long __cdecl lrint(double);
_ACRTIMP __msvcrt_long __cdecl lrintf(float);
_ACRTIMP __msvcrt_long __cdecl lround(double);
_ACRTIMP __msvcrt_long __cdecl lroundf(float);
_ACRTIMP double __cdecl scalbn(double,int);
_ACRTIMP float __cdecl scalbnf(float,int);
_ACRTIMP double __cdecl _copysign (double, double);
_ACRTIMP double __cdecl _chgsign (double);
_ACRTIMP double __cdecl _scalb(double, __msvcrt_long);
_ACRTIMP double __cdecl _logb(double);
_ACRTIMP double __cdecl _nextafter(double, double);
_ACRTIMP int __cdecl _finite(double);
_ACRTIMP int __cdecl _isnan(double);
_ACRTIMP int __cdecl _fpclass(double);
#ifndef __i386__
_ACRTIMP float __cdecl sinf(float);
_ACRTIMP float __cdecl cosf(float);
_ACRTIMP float __cdecl tanf(float);
_ACRTIMP float __cdecl sinhf(float);
_ACRTIMP float __cdecl coshf(float);
_ACRTIMP float __cdecl tanhf(float);
_ACRTIMP float __cdecl asinf(float);
_ACRTIMP float __cdecl acosf(float);
_ACRTIMP float __cdecl atanf(float);
_ACRTIMP float __cdecl atan2f(float, float);
_ACRTIMP float __cdecl atanhf(float);
_ACRTIMP float __cdecl expf(float);
_ACRTIMP float __cdecl logf(float);
_ACRTIMP float __cdecl log10f(float);
_ACRTIMP float __cdecl powf(float, float);
_ACRTIMP float __cdecl sqrtf(float);
_ACRTIMP float __cdecl ceilf(float);
_ACRTIMP float __cdecl floorf(float);
_ACRTIMP float __cdecl frexpf(float, int*);
_ACRTIMP float __cdecl modff(float, float*);
_ACRTIMP float __cdecl fmodf(float, float);
_ACRTIMP int __cdecl _finitef(float);
_ACRTIMP int __cdecl _isnanf(float);
_ACRTIMP int __cdecl _fpclassf(float);
#else
static inline float sinf(float x) { return sin(x); }
static inline float cosf(float x) { return cos(x); }
static inline float tanf(float x) { return tan(x); }
static inline float sinhf(float x) { return sinh(x); }
static inline float coshf(float x) { return cosh(x); }
static inline float tanhf(float x) { return tanh(x); }
static inline float asinf(float x) { return asin(x); }
static inline float acosf(float x) { return acos(x); }
static inline float atanf(float x) { return atan(x); }
static inline float atan2f(float x, float y) { return atan2(x, y); }
static inline float expf(float x) { return exp(x); }
static inline float logf(float x) { return log(x); }
static inline float log10f(float x) { return log10(x); }
static inline float powf(float x, float y) { return pow(x, y); }
static inline float sqrtf(float x) { return sqrt(x); }
static inline float ceilf(float x) { return ceil(x); }
static inline float floorf(float x) { return floor(x); }
static inline float frexpf(float x, int *y) { return frexp(x, y); }
static inline float modff(float x, float *y) { double yd, ret = modf(x, &yd); *y = yd; return ret; }
static inline float fmodf(float x, float y) { return fmod(x, y); }
static inline int _finitef(float x) { return _finite(x); }
static inline int _isnanf(float x) { return _isnan(x); }
static inline int _fpclassf(float x)
{
unsigned int ix = *(int*)&x;
double d = x;
/* construct denormal double */
if (!(ix >> 23 & 0xff) && (ix << 1))
{
unsigned __int64 id = (((unsigned __int64)ix >> 31) << 63) | 1;
d = *(double*)&id;
}
return _fpclass(d);
}
#endif
#if !defined(__i386__) && !defined(__x86_64__) && (_MSVCR_VER == 0 || _MSVCR_VER >= 110)
_ACRTIMP float __cdecl fabsf(float);
#else
static inline float fabsf(float x) { return fabs(x); }
#endif
#if !defined(__i386__) || _MSVCR_VER>=120
_ACRTIMP float __cdecl _chgsignf(float);
_ACRTIMP float __cdecl _copysignf(float, float);
_ACRTIMP float __cdecl _logbf(float);
_ACRTIMP float __cdecl acoshf(float);
_ACRTIMP float __cdecl asinhf(float);
_ACRTIMP float __cdecl atanhf(float);
#else
static inline float _chgsignf(float x) { return _chgsign(x); }
static inline float _copysignf(float x, float y) { return _copysign(x, y); }
static inline float _logbf(float x) { return _logb(x); }
static inline float acoshf(float x) { return acosh(x); }
static inline float asinhf(float x) { return asinh(x); }
static inline float atanhf(float x) { return atanh(x); }
#endif
static inline float ldexpf(float x, int y) { return ldexp(x, y); }
#ifdef _UCRT
_ACRTIMP double __cdecl copysign(double, double);
_ACRTIMP float __cdecl copysignf(float, float);
#else
#define copysign(x,y) _copysign(x,y)
#define copysignf(x,y) _copysignf(x,y)
#endif
_ACRTIMP double __cdecl nearbyint(double);
_ACRTIMP float __cdecl nearbyintf(float);
_ACRTIMP float __cdecl _hypotf(float, float);
_ACRTIMP int __cdecl _matherr(struct _exception*);
_ACRTIMP double __cdecl _cabs(struct _complex);
#if (defined(__GNUC__) && ((__GNUC__ > 3) || ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 3)))) || defined(__clang__)
# define INFINITY __builtin_inff()
# define NAN __builtin_nanf("")
# define HUGE_VAL __builtin_huge_val()
#else
static const union {
unsigned int __i;
float __f;
} __inff = { 0x7f800000 }, __nanf = { 0x7fc00000 };
# define INFINITY (__inff.__f)
# define NAN (__nanf.__f)
# define HUGE_VAL ((double)INFINITY)
#endif
#define FP_INFINITE 1
#define FP_NAN 2
#define FP_NORMAL -1
#define FP_SUBNORMAL -2
#define FP_ZERO 0
#define _C2 1
#define FP_ILOGB0 (-0x7fffffff - _C2)
#define FP_ILOGBNAN 0x7fffffff
_ACRTIMP short __cdecl _dclass(double);
_ACRTIMP short __cdecl _fdclass(float);
_ACRTIMP int __cdecl _dsign(double);
_ACRTIMP int __cdecl _fdsign(float);
static inline int __isnanf(float x)
{
union { float x; unsigned int i; } u = { x };
return (u.i & 0x7fffffff) > 0x7f800000;
}
static inline int __isnan(double x)
{
union { double x; unsigned __int64 i; } u = { x };
return (u.i & ~0ull >> 1) > 0x7ffull << 52;
}
static inline int __isinff(float x)
{
union { float x; unsigned int i; } u = { x };
return (u.i & 0x7fffffff) == 0x7f800000;
}
static inline int __isinf(double x)
{
union { double x; unsigned __int64 i; } u = { x };
return (u.i & ~0ull >> 1) == 0x7ffull << 52;
}
static inline int __isnormalf(float x)
{
union { float x; unsigned int i; } u = { x };
return ((u.i + 0x00800000) & 0x7fffffff) >= 0x01000000;
}
static inline int __isnormal(double x)
{
union { double x; unsigned __int64 i; } u = { x };
return ((u.i + (1ull << 52)) & ~0ull >> 1) >= 1ull << 53;
}
static inline int __signbitf(float x)
{
union { float x; unsigned int i; } u = { x };
return (int)(u.i >> 31);
}
static inline int __signbit(double x)
{
union { double x; unsigned __int64 i; } u = { x };
return (int)(u.i >> 63);
}
#define isinf(x) (sizeof(x) == sizeof(float) ? __isinff(x) : __isinf(x))
#define isnan(x) (sizeof(x) == sizeof(float) ? __isnanf(x) : __isnan(x))
#define isnormal(x) (sizeof(x) == sizeof(float) ? __isnormalf(x) : __isnormal(x))
#define signbit(x) (sizeof(x) == sizeof(float) ? __signbitf(x) : __signbit(x))
#define isfinite(x) (!isinf(x) && !isnan(x))
#ifdef __cplusplus
}
#endif
#include <poppack.h>
#if !defined(__STRICT_ANSI__) || defined(_POSIX_C_SOURCE) || defined(_POSIX_SOURCE) || defined(_XOPEN_SOURCE) || defined(_GNU_SOURCE) || defined(_BSD_SOURCE) || defined(_USE_MATH_DEFINES)
#ifndef _MATH_DEFINES_DEFINED
#define _MATH_DEFINES_DEFINED
#define M_E 2.71828182845904523536
#define M_LOG2E 1.44269504088896340736
#define M_LOG10E 0.434294481903251827651
#define M_LN2 0.693147180559945309417
#define M_LN10 2.30258509299404568402
#define M_PI 3.14159265358979323846
#define M_PI_2 1.57079632679489661923
#define M_PI_4 0.785398163397448309616
#define M_1_PI 0.318309886183790671538
#define M_2_PI 0.636619772367581343076
#define M_2_SQRTPI 1.12837916709551257390
#define M_SQRT2 1.41421356237309504880
#define M_SQRT1_2 0.707106781186547524401
#endif /* !_MATH_DEFINES_DEFINED */
#endif /* _USE_MATH_DEFINES */
static inline double hypot( double x, double y ) { return _hypot( x, y ); }
static inline double j0( double x ) { return _j0( x ); }
static inline double j1( double x ) { return _j1( x ); }
static inline double jn( int n, double x ) { return _jn( n, x ); }
static inline double y0( double x ) { return _y0( x ); }
static inline double y1( double x ) { return _y1( x ); }
static inline double yn( int n, double x ) { return _yn( n, x ); }
static inline float hypotf( float x, float y ) { return _hypotf( x, y ); }
#endif /* __WINE_MATH_H */