Sweden-Number/dlls/msvcrt/math.c

966 lines
23 KiB
C
Raw Normal View History

/*
* msvcrt.dll math functions
*
* Copyright 2000 Jon Griffiths
*/
#include "config.h"
#include "msvcrt.h"
#include "ms_errno.h"
#define __USE_ISOC9X 1
#define __USE_ISOC99 1
#include <math.h>
#ifdef HAVE_IEEEFP_H
#include <ieeefp.h>
#endif
#include "msvcrt/stdlib.h"
DEFAULT_DEBUG_CHANNEL(msvcrt);
#ifndef HAVE_FINITE
#ifndef finite /* Could be a macro */
#ifdef isfinite
#define finite(x) isfinite(x)
#else
#define finite(x) (!isnan(x)) /* At least catch some cases */
#endif
#endif
#endif
#ifndef signbit
#define signbit(x) 0
#endif
/* fpclass constants */
#define _FPCLASS_SNAN 1
#define _FPCLASS_QNAN 2
#define _FPCLASS_NINF 4
#define _FPCLASS_NN 8
#define _FPCLASS_ND 16
#define _FPCLASS_NZ 32
#define _FPCLASS_PZ 64
#define _FPCLASS_PD 128
#define _FPCLASS_PN 256
#define _FPCLASS_PINF 512
/* _statusfp bit flags */
#define _SW_INEXACT 0x1
#define _SW_UNDERFLOW 0x2
#define _SW_OVERFLOW 0x4
#define _SW_ZERODIVIDE 0x8
#define _SW_INVALID 0x10
#define _SW_DENORMAL 0x80000
/* _controlfp masks and bitflags - x86 only so far*/
#ifdef __i386__
#define _MCW_EM 0x8001f
#define _EM_INEXACT 0x1
#define _EM_UNDERFLOW 0x2
#define _EM_OVERFLOW 0x4
#define _EM_ZERODIVIDE 0x8
#define _EM_INVALID 0x10
#define _MCW_RC 0x300
#define _RC_NEAR 0x0
#define _RC_DOWN 0x100
#define _RC_UP 0x200
#define _RC_CHOP 0x300
#define _MCW_PC 0x30000
#define _PC_64 0x0
#define _PC_53 0x10000
#define _PC_24 0x20000
#define _MCW_IC 0x40000
#define _IC_AFFINE 0x40000
#define _IC_PROJECTIVE 0x0
#define _EM_DENORMAL 0x80000
#endif
typedef struct __MSVCRT_complex
{
double real;
double imaginary;
} MSVCRT_complex;
typedef struct __MSVCRT_exception
{
int type;
char *name;
double arg1;
double arg2;
double retval;
} MSVCRT_exception;
typedef int (*MSVCRT_matherr_func)(MSVCRT_exception *);
static MSVCRT_matherr_func MSVCRT_default_matherr_func = NULL;
#if defined(__GNUC__) && defined(__i386__)
#define FPU_DOUBLE(var) double var; \
__asm__ __volatile__( "fstpl %0;fwait" : "=m" (var) : )
#define FPU_DOUBLES(var1,var2) double var1,var2; \
__asm__ __volatile__( "fstpl %0;fwait" : "=m" (var2) : ); \
__asm__ __volatile__( "fstpl %0;fwait" : "=m" (var1) : )
/*********************************************************************
* _CIacos (MSVCRT.@)
*/
double _CIacos(void)
{
FPU_DOUBLE(x);
if (x < -1.0 || x > 1.0 || !finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return acos(x);
}
/*********************************************************************
* _CIasin (MSVCRT.@)
*/
double _CIasin(void)
{
FPU_DOUBLE(x);
if (x < -1.0 || x > 1.0 || !finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return asin(x);
}
/*********************************************************************
* _CIatan (MSVCRT.@)
*/
double _CIatan(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return atan(x);
}
/*********************************************************************
* _CIatan2 (MSVCRT.@)
*/
double _CIatan2(void)
{
FPU_DOUBLES(x,y);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return atan2(x,y);
}
/*********************************************************************
* _CIcos (MSVCRT.@)
*/
double _CIcos(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return cos(x);
}
/*********************************************************************
* _CIcosh (MSVCRT.@)
*/
double _CIcosh(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return cosh(x);
}
/*********************************************************************
* _CIexp (MSVCRT.@)
*/
double _CIexp(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return exp(x);
}
/*********************************************************************
* _CIfmod (MSVCRT.@)
*/
double _CIfmod(void)
{
FPU_DOUBLES(x,y);
if (!finite(x) || !finite(y)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return fmod(x,y);
}
/*********************************************************************
* _CIlog (MSVCRT.@)
*/
double _CIlog(void)
{
FPU_DOUBLE(x);
if (x < 0.0 || !finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
if (x == 0.0) SET_THREAD_VAR(errno,MSVCRT_ERANGE);
return log(x);
}
/*********************************************************************
* _CIlog10 (MSVCRT.@)
*/
double _CIlog10(void)
{
FPU_DOUBLE(x);
if (x < 0.0 || !finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
if (x == 0.0) SET_THREAD_VAR(errno,MSVCRT_ERANGE);
return log10(x);
}
/*********************************************************************
* _CIpow (MSVCRT.@)
*/
double _CIpow(void)
{
double z;
FPU_DOUBLES(x,y);
/* FIXME: If x < 0 and y is not integral, set EDOM */
z = pow(x,y);
if (!finite(z)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return z;
}
/*********************************************************************
* _CIsin (MSVCRT.@)
*/
double _CIsin(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return sin(x);
}
/*********************************************************************
* _CIsinh (MSVCRT.@)
*/
double _CIsinh(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return sinh(x);
}
/*********************************************************************
* _CIsqrt (MSVCRT.@)
*/
double _CIsqrt(void)
{
FPU_DOUBLE(x);
if (x < 0.0 || !finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return sqrt(x);
}
/*********************************************************************
* _CItan (MSVCRT.@)
*/
double _CItan(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return tan(x);
}
/*********************************************************************
* _CItanh (MSVCRT.@)
*/
double _CItanh(void)
{
FPU_DOUBLE(x);
if (!finite(x)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return tanh(x);
}
#else /* defined(__GNUC__) && defined(__i386__) */
/* The above cannot be called on non x86 platforms, stub them for linking */
#define IX86_ONLY(func) double MSVCRT_##func(void) { return 0.0; }
IX86_ONLY(_CIacos)
IX86_ONLY(_CIasin)
IX86_ONLY(_CIatan)
IX86_ONLY(_CIatan2)
IX86_ONLY(_CIcos)
IX86_ONLY(_CIcosh)
IX86_ONLY(_CIexp)
IX86_ONLY(_CIfmod)
IX86_ONLY(_CIlog)
IX86_ONLY(_CIlog10)
IX86_ONLY(_CIpow)
IX86_ONLY(_CIsin)
IX86_ONLY(_CIsinh)
IX86_ONLY(_CIsqrt)
IX86_ONLY(_CItan)
IX86_ONLY(_CItanh)
#endif /* defined(__GNUC__) && defined(__i386__) */
/*********************************************************************
* _fpclass (MSVCRT.@)
*/
int _fpclass(double num)
{
#if defined(HAVE_FPCLASS) || defined(fpclass)
switch (fpclass( num ))
{
case FP_SNAN: return _FPCLASS_SNAN;
case FP_QNAN: return _FPCLASS_QNAN;
case FP_NINF: return _FPCLASS_NINF;
case FP_PINF: return _FPCLASS_PINF;
case FP_NDENORM: return _FPCLASS_ND;
case FP_PDENORM: return _FPCLASS_PD;
case FP_NZERO: return _FPCLASS_NZ;
case FP_PZERO: return _FPCLASS_PZ;
case FP_NNORM: return _FPCLASS_NN;
}
return _FPCLASS_PN;
#elif defined (fpclassify)
switch (fpclassify( num ))
{
case FP_NAN: return _FPCLASS_QNAN;
case FP_INFINITE: return signbit(num) ? _FPCLASS_NINF : _FPCLASS_PINF;
case FP_SUBNORMAL: return signbit(num) ?_FPCLASS_ND : _FPCLASS_PD;
case FP_ZERO: return signbit(num) ? _FPCLASS_NZ : _FPCLASS_PZ;
}
return signbit(num) ? _FPCLASS_NN : _FPCLASS_PN;
#else
if (!finite(num))
return _FPCLASS_QNAN;
return num == 0.0 ? _FPCLASS_PZ : (num < 0 ? _FPCLASS_NN : _FPCLASS_PN);
#endif
}
/*********************************************************************
* _rotl (MSVCRT.@)
*/
unsigned int _rotl(unsigned int num, int shift)
{
shift &= 31;
return (num << shift) | (num >> (32-shift));
}
/*********************************************************************
* _logb (MSVCRT.@)
*/
double _logb(double num)
{
if (!finite(num)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return logb(num);
}
/*********************************************************************
* _lrotl (MSVCRT.@)
*/
unsigned long _lrotl(unsigned long num, int shift)
{
shift &= 0x1f;
return (num << shift) | (num >> (32-shift));
}
/*********************************************************************
* _lrotr (MSVCRT.@)
*/
unsigned long _lrotr(unsigned long num, int shift)
{
shift &= 0x1f;
return (num >> shift) | (num << (32-shift));
}
/*********************************************************************
* _rotr (MSVCRT.@)
*/
unsigned int _rotr(unsigned int num, int shift)
{
shift &= 0x1f;
return (num >> shift) | (num << (32-shift));
}
/*********************************************************************
* _scalb (MSVCRT.@)
*/
double _scalb(double num, long power)
{
/* Note - Can't forward directly as libc expects y as double */
double dblpower = (double)power;
if (!finite(num)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
return scalb(num, dblpower);
}
/*********************************************************************
* _matherr (MSVCRT.@)
*/
int _matherr(MSVCRT_exception *e)
{
if (e)
TRACE("(%p = %d, %s, %g %g %g)\n",e, e->type, e->name, e->arg1, e->arg2,
e->retval);
else
TRACE("(null)\n");
if (MSVCRT_default_matherr_func)
return MSVCRT_default_matherr_func(e);
ERR(":Unhandled math error!\n");
return 0;
}
/*********************************************************************
* __setusermatherr (MSVCRT.@)
*/
void MSVCRT___setusermatherr(MSVCRT_matherr_func func)
{
MSVCRT_default_matherr_func = func;
TRACE(":new matherr handler %p\n", func);
}
/**********************************************************************
* _statusfp (MSVCRT.@)
*/
unsigned int _statusfp(void)
{
unsigned int retVal = 0;
#if defined(__GNUC__) && defined(__i386__)
unsigned int fpword;
__asm__ __volatile__( "fstsw %0" : "=m" (fpword) : );
if (fpword & 0x1) retVal |= _SW_INVALID;
if (fpword & 0x2) retVal |= _SW_DENORMAL;
if (fpword & 0x4) retVal |= _SW_ZERODIVIDE;
if (fpword & 0x8) retVal |= _SW_OVERFLOW;
if (fpword & 0x10) retVal |= _SW_UNDERFLOW;
if (fpword & 0x20) retVal |= _SW_INEXACT;
#else
FIXME(":Not implemented!\n");
#endif
return retVal;
}
/*********************************************************************
* _clearfp (MSVCRT.@)
*/
unsigned int _clearfp(void)
{
unsigned int retVal = _statusfp();
#if defined(__GNUC__) && defined(__i386__)
__asm__ __volatile__( "fnclex" );
#else
FIXME(":Not Implemented\n");
#endif
return retVal;
}
/*********************************************************************
* ldexp (MSVCRT.@)
*/
double MSVCRT_ldexp(double num, long exp)
{
double z = ldexp(num,exp);
if (!finite(z))
SET_THREAD_VAR(errno,MSVCRT_ERANGE);
else if (z == 0 && signbit(z))
z = 0.0; /* Convert -0 -> +0 */
return z;
}
/*********************************************************************
* _cabs (MSVCRT.@)
*/
double _cabs(MSVCRT_complex num)
{
return sqrt(num.real * num.real + num.imaginary * num.imaginary);
}
/*********************************************************************
* _chgsign (MSVCRT.@)
*/
double _chgsign(double num)
{
/* FIXME: +-infinity,Nan not tested */
return -num;
}
/*********************************************************************
* _control87 (MSVCRT.@)
*/
unsigned int _control87(unsigned int newval, unsigned int mask)
{
#if defined(__GNUC__) && defined(__i386__)
unsigned int fpword = 0;
unsigned int flags = 0;
TRACE("(%08x, %08x): Called\n", newval, mask);
/* Get fp control word */
__asm__ __volatile__( "fstcw %0" : "=m" (fpword) : );
TRACE("Control word before : %08x\n", fpword);
/* Convert into mask constants */
if (fpword & 0x1) flags |= _EM_INVALID;
if (fpword & 0x2) flags |= _EM_DENORMAL;
if (fpword & 0x4) flags |= _EM_ZERODIVIDE;
if (fpword & 0x8) flags |= _EM_OVERFLOW;
if (fpword & 0x10) flags |= _EM_UNDERFLOW;
if (fpword & 0x20) flags |= _EM_INEXACT;
switch(fpword & 0xC00) {
case 0xC00: flags |= _RC_UP|_RC_DOWN; break;
case 0x800: flags |= _RC_UP; break;
case 0x400: flags |= _RC_DOWN; break;
}
switch(fpword & 0x300) {
case 0x0: flags |= _PC_24; break;
case 0x200: flags |= _PC_53; break;
case 0x300: flags |= _PC_64; break;
}
if (fpword & 0x1000) flags |= _IC_AFFINE;
/* Mask with parameters */
flags = (flags & ~mask) | (newval & mask);
/* Convert (masked) value back to fp word */
fpword = 0;
if (flags & _EM_INVALID) fpword |= 0x1;
if (flags & _EM_DENORMAL) fpword |= 0x2;
if (flags & _EM_ZERODIVIDE) fpword |= 0x4;
if (flags & _EM_OVERFLOW) fpword |= 0x8;
if (flags & _EM_UNDERFLOW) fpword |= 0x10;
if (flags & _EM_INEXACT) fpword |= 0x20;
switch(flags & (_RC_UP | _RC_DOWN)) {
case _RC_UP|_RC_DOWN: fpword |= 0xC00; break;
case _RC_UP: fpword |= 0x800; break;
case _RC_DOWN: fpword |= 0x400; break;
}
switch (flags & (_PC_24 | _PC_53)) {
case _PC_64: fpword |= 0x300; break;
case _PC_53: fpword |= 0x200; break;
case _PC_24: fpword |= 0x0; break;
}
if (flags & _IC_AFFINE) fpword |= 0x1000;
TRACE("Control word after : %08x\n", fpword);
/* Put fp control word */
__asm__ __volatile__( "fldcw %0" : : "m" (fpword) );
return flags;
#else
FIXME(":Not Implemented!\n");
return 0;
#endif
}
/*********************************************************************
* _controlfp (MSVCRT.@)
*/
unsigned int _controlfp(unsigned int newval, unsigned int mask)
{
return _control87( newval, mask & ~_EM_DENORMAL );
}
/*********************************************************************
* _copysign (MSVCRT.@)
*/
double _copysign(double num, double sign)
{
/* FIXME: Behaviour for Nan/Inf? */
if (sign < 0.0)
return num < 0.0 ? num : -num;
return num < 0.0 ? -num : num;
}
/*********************************************************************
* _finite (MSVCRT.@)
*/
int _finite(double num)
{
return (finite(num)?1:0); /* See comment for _isnan() */
}
/*********************************************************************
* _fpreset (MSVCRT.@)
*/
void _fpreset(void)
{
#if defined(__GNUC__) && defined(__i386__)
__asm__ __volatile__( "fninit" );
#else
FIXME(":Not Implemented!\n");
#endif
}
/*********************************************************************
* _isnan (MSVCRT.@)
*/
INT _isnan(double num)
{
/* Some implementations return -1 for true(glibc), msvcrt/crtdll return 1.
* Do the same, as the result may be used in calculations
*/
return isnan(num) ? 1 : 0;
}
/*********************************************************************
* _y0 (MSVCRT.@)
*/
double _y0(double num)
{
double retval;
if (!finite(num)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = y0(num);
if (_fpclass(retval) == _FPCLASS_NINF)
{
SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = sqrt(-1);
}
return retval;
}
/*********************************************************************
* _y1 (MSVCRT.@)
*/
double _y1(double num)
{
double retval;
if (!finite(num)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = y1(num);
if (_fpclass(retval) == _FPCLASS_NINF)
{
SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = sqrt(-1);
}
return retval;
}
/*********************************************************************
* _yn (MSVCRT.@)
*/
double _yn(int order, double num)
{
double retval;
if (!finite(num)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = yn(order,num);
if (_fpclass(retval) == _FPCLASS_NINF)
{
SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = sqrt(-1);
}
return retval;
}
/*********************************************************************
* _nextafter (MSVCRT.@)
*/
double _nextafter(double num, double next)
{
double retval;
if (!finite(num) || !finite(next)) SET_THREAD_VAR(errno,MSVCRT_EDOM);
retval = nextafter(num,next);
return retval;
}
#include <stdlib.h> /* div_t, ldiv_t */
/*********************************************************************
* div (MSVCRT.@)
* VERSION
* [i386] Windows binary compatible - returns the struct in eax/edx.
*/
#ifdef __i386__
LONGLONG MSVCRT_div(int num, int denom)
{
LONGLONG retval;
div_t dt = div(num,denom);
retval = ((LONGLONG)dt.rem << 32) | dt.quot;
return retval;
}
#else
/*********************************************************************
* div (MSVCRT.@)
* VERSION
* [!i386] Non-x86 can't run win32 apps so we don't need binary compatibility
*/
MSVCRT_div_t MSVCRT_div(int num, int denom)
{
return div(num,denom);
}
#endif /* ifdef __i386__ */
/*********************************************************************
* ldiv (MSVCRT.@)
* VERSION
* [i386] Windows binary compatible - returns the struct in eax/edx.
*/
#ifdef __i386__
ULONGLONG MSVCRT_ldiv(long num, long denom)
{
ULONGLONG retval;
ldiv_t ldt = ldiv(num,denom);
retval = ((ULONGLONG)ldt.rem << 32) | (ULONG)ldt.quot;
return retval;
}
#else
/*********************************************************************
* ldiv (MSVCRT.@)
* VERSION
* [!i386] Non-x86 can't run win32 apps so we don't need binary compatibility
*/
MSVCRT_ldiv_t MSVCRT_ldiv(long num, long denom)
{
return ldiv(num,denom);
}
#endif /* ifdef __i386__ */
/***********************************************************************
* _adj_fdiv_m16i (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdiv_m16i(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdiv_m32 (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdiv_m32(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdiv_m32i (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdiv_m32i(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdiv_m64 (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdiv_m64(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdiv_r (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdiv_r(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdivr_m16i (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdivr_m16i(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdivr_m32 (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdivr_m32(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdivr_m32i (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdivr_m32i(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fdivr_m64 (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fdivr_m64(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fpatan (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fpatan(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fprem (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fprem(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fprem1 (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fprem1(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adj_fptan (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _adj_fptan(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _adjust_fdiv (MSVCRT.@)
* FIXME
* I _think_ this function should be a variable indicating whether
* Pentium fdiv bug safe code should be used.
*/
void _adjust_fdiv(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _safe_fdiv (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _safe_fdiv(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _safe_fdivr (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _safe_fdivr(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _safe_fprem (MSVCRT.@)
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _safe_fprem(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}
/***********************************************************************
* _safe_fprem1 (MSVCRT.@)
*
* FIXME
* This function is likely to have the wrong number of arguments.
*
* NOTE
* I _think_ this function is intended to work around the Pentium
* fdiv bug.
*/
void _safe_fprem1(void)
{
2001-05-09 19:31:31 +02:00
TRACE("(): stub\n");
}