msvcrt: Import cosf implementation from musl.

Signed-off-by: Piotr Caban <piotr@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
This commit is contained in:
Piotr Caban 2021-05-23 16:37:16 +02:00 committed by Alexandre Julliard
parent 73c0d36aa4
commit 8321ce14ad
3 changed files with 378 additions and 251 deletions

View File

@ -311,6 +311,244 @@ float CDECL _logbf(float x)
#endif
/* Copied from musl: src/math/scalbn.c */
static double __scalbn(double x, int n)
{
union {double f; UINT64 i;} u;
double y = x;
if (n > 1023) {
y *= 0x1p1023;
n -= 1023;
if (n > 1023) {
y *= 0x1p1023;
n -= 1023;
if (n > 1023)
n = 1023;
}
} else if (n < -1022) {
/* make sure final n < -53 to avoid double
rounding in the subnormal range */
y *= 0x1p-1022 * 0x1p53;
n += 1022 - 53;
if (n < -1022) {
y *= 0x1p-1022 * 0x1p53;
n += 1022 - 53;
if (n < -1022)
n = -1022;
}
}
u.i = (UINT64)(0x3ff + n) << 52;
x = y * u.f;
return x;
}
/* Copied from musl: src/math/__rem_pio2_large.c */
static int __rem_pio2_large(double *x, double *y, int e0, int nx, int prec)
{
static const int init_jk[] = {3, 4};
static const INT32 ipio2[] = {
0xA2F983, 0x6E4E44, 0x1529FC, 0x2757D1, 0xF534DD, 0xC0DB62,
0x95993C, 0x439041, 0xFE5163, 0xABDEBB, 0xC561B7, 0x246E3A,
0x424DD2, 0xE00649, 0x2EEA09, 0xD1921C, 0xFE1DEB, 0x1CB129,
0xA73EE8, 0x8235F5, 0x2EBB44, 0x84E99C, 0x7026B4, 0x5F7E41,
0x3991D6, 0x398353, 0x39F49C, 0x845F8B, 0xBDF928, 0x3B1FF8,
0x97FFDE, 0x05980F, 0xEF2F11, 0x8B5A0A, 0x6D1F6D, 0x367ECF,
0x27CB09, 0xB74F46, 0x3F669E, 0x5FEA2D, 0x7527BA, 0xC7EBE5,
0xF17B3D, 0x0739F7, 0x8A5292, 0xEA6BFB, 0x5FB11F, 0x8D5D08,
0x560330, 0x46FC7B, 0x6BABF0, 0xCFBC20, 0x9AF436, 0x1DA9E3,
0x91615E, 0xE61B08, 0x659985, 0x5F14A0, 0x68408D, 0xFFD880,
0x4D7327, 0x310606, 0x1556CA, 0x73A8C9, 0x60E27B, 0xC08C6B,
};
static const double PIo2[] = {
1.57079625129699707031e+00,
7.54978941586159635335e-08,
5.39030252995776476554e-15,
3.28200341580791294123e-22,
1.27065575308067607349e-29,
1.22933308981111328932e-36,
2.73370053816464559624e-44,
2.16741683877804819444e-51,
};
INT32 jz, jx, jv, jp, jk, carry, n, iq[20], i, j, k, m, q0, ih;
double z, fw, f[20], fq[20] = {0}, q[20];
/* initialize jk*/
jk = init_jk[prec];
jp = jk;
/* determine jx,jv,q0, note that 3>q0 */
jx = nx - 1;
jv = (e0 - 3) / 24;
if(jv < 0) jv = 0;
q0 = e0 - 24 * (jv + 1);
/* set up f[0] to f[jx+jk] where f[jx+jk] = ipio2[jv+jk] */
j = jv - jx;
m = jx + jk;
for (i = 0; i <= m; i++, j++)
f[i] = j < 0 ? 0.0 : (double)ipio2[j];
/* compute q[0],q[1],...q[jk] */
for (i = 0; i <= jk; i++) {
for (j = 0, fw = 0.0; j <= jx; j++)
fw += x[j] * f[jx + i - j];
q[i] = fw;
}
jz = jk;
recompute:
/* distill q[] into iq[] reversingly */
for (i = 0, j = jz, z = q[jz]; j > 0; i++, j--) {
fw = (double)(INT32)(0x1p-24 * z);
iq[i] = (INT32)(z - 0x1p24 * fw);
z = q[j - 1] + fw;
}
/* compute n */
z = __scalbn(z, q0); /* actual value of z */
z -= 8.0 * floor(z * 0.125); /* trim off integer >= 8 */
n = (INT32)z;
z -= (double)n;
ih = 0;
if (q0 > 0) { /* need iq[jz-1] to determine n */
i = iq[jz - 1] >> (24 - q0);
n += i;
iq[jz - 1] -= i << (24 - q0);
ih = iq[jz - 1] >> (23 - q0);
}
else if (q0 == 0) ih = iq[jz - 1] >> 23;
else if (z >= 0.5) ih = 2;
if (ih > 0) { /* q > 0.5 */
n += 1;
carry = 0;
for (i = 0; i < jz; i++) { /* compute 1-q */
j = iq[i];
if (carry == 0) {
if (j != 0) {
carry = 1;
iq[i] = 0x1000000 - j;
}
} else
iq[i] = 0xffffff - j;
}
if (q0 > 0) { /* rare case: chance is 1 in 12 */
switch(q0) {
case 1:
iq[jz - 1] &= 0x7fffff;
break;
case 2:
iq[jz - 1] &= 0x3fffff;
break;
}
}
if (ih == 2) {
z = 1.0 - z;
if (carry != 0)
z -= __scalbn(1.0, q0);
}
}
/* check if recomputation is needed */
if (z == 0.0) {
j = 0;
for (i = jz - 1; i >= jk; i--) j |= iq[i];
if (j == 0) { /* need recomputation */
for (k = 1; iq[jk - k] == 0; k++); /* k = no. of terms needed */
for (i = jz + 1; i <= jz + k; i++) { /* add q[jz+1] to q[jz+k] */
f[jx + i] = (double)ipio2[jv + i];
for (j = 0, fw = 0.0; j <= jx; j++)
fw += x[j] * f[jx + i - j];
q[i] = fw;
}
jz += k;
goto recompute;
}
}
/* chop off zero terms */
if (z == 0.0) {
jz -= 1;
q0 -= 24;
while (iq[jz] == 0) {
jz--;
q0 -= 24;
}
} else { /* break z into 24-bit if necessary */
z = __scalbn(z, -q0);
if (z >= 0x1p24) {
fw = (double)(INT32)(0x1p-24 * z);
iq[jz] = (INT32)(z - 0x1p24 * fw);
jz += 1;
q0 += 24;
iq[jz] = (INT32)fw;
} else
iq[jz] = (INT32)z;
}
/* convert integer "bit" chunk to floating-point value */
fw = __scalbn(1.0, q0);
for (i = jz; i >= 0; i--) {
q[i] = fw * (double)iq[i];
fw *= 0x1p-24;
}
/* compute PIo2[0,...,jp]*q[jz,...,0] */
for(i = jz; i >= 0; i--) {
for (fw = 0.0, k = 0; k <= jp && k <= jz - i; k++)
fw += PIo2[k] * q[i + k];
fq[jz - i] = fw;
}
/* compress fq[] into y[] */
switch(prec) {
case 0:
fw = 0.0;
for (i = jz; i >= 0; i--)
fw += fq[i];
y[0] = ih == 0 ? fw : -fw;
break;
case 1:
case 2:
fw = 0.0;
for (i = jz; i >= 0; i--)
fw += fq[i];
fw = (double)fw;
y[0] = ih==0 ? fw : -fw;
fw = fq[0] - fw;
for (i = 1; i <= jz; i++)
fw += fq[i];
y[1] = ih == 0 ? fw : -fw;
break;
case 3: /* painful */
for (i = jz; i > 0; i--) {
fw = fq[i - 1] + fq[i];
fq[i] += fq[i - 1] - fw;
fq[i - 1] = fw;
}
for (i = jz; i > 1; i--) {
fw = fq[i - 1] + fq[i];
fq[i] += fq[i - 1] - fw;
fq[i - 1] = fw;
}
for (fw = 0.0, i = jz; i >= 2; i--)
fw += fq[i];
if (ih == 0) {
y[0] = fq[0];
y[1] = fq[1];
y[2] = fw;
} else {
y[0] = -fq[0];
y[1] = -fq[1];
y[2] = -fw;
}
}
return n & 7;
}
#ifndef __i386__
/*********************************************************************
@ -609,14 +847,151 @@ float CDECL atan2f( float y, float x )
}
}
/* Copied from musl: src/math/__rem_pio2f.c */
static int __rem_pio2f(float x, double *y)
{
static const double toint = 1.5 / DBL_EPSILON,
pio4 = 0x1.921fb6p-1,
invpio2 = 6.36619772367581382433e-01,
pio2_1 = 1.57079631090164184570e+00,
pio2_1t = 1.58932547735281966916e-08;
union {float f; uint32_t i;} u = {x};
double tx[1], ty[1], fn;
UINT32 ix;
int n, sign, e0;
ix = u.i & 0x7fffffff;
/* 25+53 bit pi is good enough for medium size */
if (ix < 0x4dc90fdb) { /* |x| ~< 2^28*(pi/2), medium size */
/* Use a specialized rint() to get fn. */
fn = fp_barrier(x * invpio2 + toint) - toint;
n = (int)fn;
*y = x - fn * pio2_1 - fn * pio2_1t;
/* Matters with directed rounding. */
if (*y < -pio4) {
n--;
fn--;
*y = x - fn * pio2_1 - fn * pio2_1t;
} else if (*y > pio4) {
n++;
fn++;
*y = x - fn * pio2_1 - fn * pio2_1t;
}
return n;
}
if(ix >= 0x7f800000) { /* x is inf or NaN */
*y = x - x;
return 0;
}
/* scale x into [2^23, 2^24-1] */
sign = u.i >> 31;
e0 = (ix >> 23) - (0x7f + 23); /* e0 = ilogb(|x|)-23, positive */
u.i = ix - (e0 << 23);
tx[0] = u.f;
n = __rem_pio2_large(tx, ty, e0, 1, 0);
if (sign) {
*y = -ty[0];
return -n;
}
*y = ty[0];
return n;
}
/* Copied from musl: src/math/__sindf.c */
static float __sindf(double x)
{
static const double S1 = -0x15555554cbac77.0p-55,
S2 = 0x111110896efbb2.0p-59,
S3 = -0x1a00f9e2cae774.0p-65,
S4 = 0x16cd878c3b46a7.0p-71;
double r, s, w, z;
z = x * x;
w = z * z;
r = S3 + z * S4;
s = z * x;
return (x + s * (S1 + z * S2)) + s * w * r;
}
/* Copied from musl: src/math/__cosdf.c */
static float __cosdf(double x)
{
static const double C0 = -0x1ffffffd0c5e81.0p-54,
C1 = 0x155553e1053a42.0p-57,
C2 = -0x16c087e80f1e27.0p-62,
C3 = 0x199342e0ee5069.0p-68;
double r, w, z;
z = x * x;
w = z * z;
r = C2 + z * C3;
return ((1.0 + z * C0) + w * C1) + (w * z) * r;
}
/*********************************************************************
* cosf (MSVCRT.@)
*
* Copied from musl: src/math/cosf.c
*/
float CDECL cosf( float x )
{
float ret = unix_funcs->cosf( x );
if (!isfinite(x)) return math_error(_DOMAIN, "cosf", x, 0, ret);
return ret;
static const double c1pio2 = 1*M_PI_2,
c2pio2 = 2*M_PI_2,
c3pio2 = 3*M_PI_2,
c4pio2 = 4*M_PI_2;
double y;
UINT32 ix;
unsigned n, sign;
ix = *(UINT32*)&x;
sign = ix >> 31;
ix &= 0x7fffffff;
if (ix <= 0x3f490fda) { /* |x| ~<= pi/4 */
if (ix < 0x39800000) { /* |x| < 2**-12 */
/* raise inexact if x != 0 */
fp_barrierf(x + 0x1p120f);
return 1.0f;
}
return __cosdf(x);
}
if (ix <= 0x407b53d1) { /* |x| ~<= 5*pi/4 */
if (ix > 0x4016cbe3) /* |x| ~> 3*pi/4 */
return -__cosdf(sign ? x + c2pio2 : x - c2pio2);
else {
if (sign)
return __sindf(x + c1pio2);
else
return __sindf(c1pio2 - x);
}
}
if (ix <= 0x40e231d5) { /* |x| ~<= 9*pi/4 */
if (ix > 0x40afeddf) /* |x| ~> 7*pi/4 */
return __cosdf(sign ? x + c4pio2 : x - c4pio2);
else {
if (sign)
return __sindf(-x - c3pio2);
else
return __sindf(x - c3pio2);
}
}
/* cos(Inf or NaN) is NaN */
if (isinf(x)) return math_error(_DOMAIN, "cosf", x, 0, x - x);
if (ix >= 0x7f800000)
return x - x;
/* general argument reduction needed */
n = __rem_pio2f(x, &y);
switch (n & 3) {
case 0: return __cosdf(y);
case 1: return __sindf(-y);
case 2: return -__cosdf(y);
default: return __sindf(y);
}
}
/*********************************************************************
@ -1343,38 +1718,6 @@ double CDECL atan2( double y, double x )
}
}
/* Copied from musl: src/math/scalbn.c */
static double __scalbn(double x, int n)
{
union {double f; UINT64 i;} u;
double y = x;
if (n > 1023) {
y *= 0x1p1023;
n -= 1023;
if (n > 1023) {
y *= 0x1p1023;
n -= 1023;
if (n > 1023)
n = 1023;
}
} else if (n < -1022) {
/* make sure final n < -53 to avoid double
rounding in the subnormal range */
y *= 0x1p-1022 * 0x1p53;
n += 1022 - 53;
if (n < -1022) {
y *= 0x1p-1022 * 0x1p53;
n += 1022 - 53;
if (n < -1022)
n = -1022;
}
}
u.i = (UINT64)(0x3ff + n) << 52;
x = y * u.f;
return x;
}
/* Copied from musl: src/math/rint.c */
static double __rint(double x)
{
@ -1402,212 +1745,6 @@ static double __rint(double x)
return y;
}
/* Copied from musl: src/math/__rem_pio2_large.c */
static int __rem_pio2_large(double *x, double *y, int e0, int nx, int prec)
{
static const int init_jk[] = {3, 4};
static const INT32 ipio2[] = {
0xA2F983, 0x6E4E44, 0x1529FC, 0x2757D1, 0xF534DD, 0xC0DB62,
0x95993C, 0x439041, 0xFE5163, 0xABDEBB, 0xC561B7, 0x246E3A,
0x424DD2, 0xE00649, 0x2EEA09, 0xD1921C, 0xFE1DEB, 0x1CB129,
0xA73EE8, 0x8235F5, 0x2EBB44, 0x84E99C, 0x7026B4, 0x5F7E41,
0x3991D6, 0x398353, 0x39F49C, 0x845F8B, 0xBDF928, 0x3B1FF8,
0x97FFDE, 0x05980F, 0xEF2F11, 0x8B5A0A, 0x6D1F6D, 0x367ECF,
0x27CB09, 0xB74F46, 0x3F669E, 0x5FEA2D, 0x7527BA, 0xC7EBE5,
0xF17B3D, 0x0739F7, 0x8A5292, 0xEA6BFB, 0x5FB11F, 0x8D5D08,
0x560330, 0x46FC7B, 0x6BABF0, 0xCFBC20, 0x9AF436, 0x1DA9E3,
0x91615E, 0xE61B08, 0x659985, 0x5F14A0, 0x68408D, 0xFFD880,
0x4D7327, 0x310606, 0x1556CA, 0x73A8C9, 0x60E27B, 0xC08C6B,
};
static const double PIo2[] = {
1.57079625129699707031e+00,
7.54978941586159635335e-08,
5.39030252995776476554e-15,
3.28200341580791294123e-22,
1.27065575308067607349e-29,
1.22933308981111328932e-36,
2.73370053816464559624e-44,
2.16741683877804819444e-51,
};
INT32 jz, jx, jv, jp, jk, carry, n, iq[20], i, j, k, m, q0, ih;
double z, fw, f[20], fq[20], q[20];
/* initialize jk*/
jk = init_jk[prec];
jp = jk;
/* determine jx,jv,q0, note that 3>q0 */
jx = nx - 1;
jv = (e0 - 3) / 24;
if(jv < 0) jv = 0;
q0 = e0 - 24 * (jv + 1);
/* set up f[0] to f[jx+jk] where f[jx+jk] = ipio2[jv+jk] */
j = jv - jx;
m = jx + jk;
for (i = 0; i <= m; i++, j++)
f[i] = j < 0 ? 0.0 : (double)ipio2[j];
/* compute q[0],q[1],...q[jk] */
for (i = 0; i <= jk; i++) {
for (j = 0, fw = 0.0; j <= jx; j++)
fw += x[j] * f[jx + i - j];
q[i] = fw;
}
jz = jk;
recompute:
/* distill q[] into iq[] reversingly */
for (i = 0, j = jz, z = q[jz]; j > 0; i++, j--) {
fw = (double)(INT32)(0x1p-24 * z);
iq[i] = (INT32)(z - 0x1p24 * fw);
z = q[j - 1] + fw;
}
/* compute n */
z = __scalbn(z, q0); /* actual value of z */
z -= 8.0 * floor(z * 0.125); /* trim off integer >= 8 */
n = (INT32)z;
z -= (double)n;
ih = 0;
if (q0 > 0) { /* need iq[jz-1] to determine n */
i = iq[jz - 1] >> (24 - q0);
n += i;
iq[jz - 1] -= i << (24 - q0);
ih = iq[jz - 1] >> (23 - q0);
}
else if (q0 == 0) ih = iq[jz - 1] >> 23;
else if (z >= 0.5) ih = 2;
if (ih > 0) { /* q > 0.5 */
n += 1;
carry = 0;
for (i = 0; i < jz; i++) { /* compute 1-q */
j = iq[i];
if (carry == 0) {
if (j != 0) {
carry = 1;
iq[i] = 0x1000000 - j;
}
} else
iq[i] = 0xffffff - j;
}
if (q0 > 0) { /* rare case: chance is 1 in 12 */
switch(q0) {
case 1:
iq[jz - 1] &= 0x7fffff;
break;
case 2:
iq[jz - 1] &= 0x3fffff;
break;
}
}
if (ih == 2) {
z = 1.0 - z;
if (carry != 0)
z -= __scalbn(1.0, q0);
}
}
/* check if recomputation is needed */
if (z == 0.0) {
j = 0;
for (i = jz - 1; i >= jk; i--) j |= iq[i];
if (j == 0) { /* need recomputation */
for (k = 1; iq[jk - k] == 0; k++); /* k = no. of terms needed */
for (i = jz + 1; i <= jz + k; i++) { /* add q[jz+1] to q[jz+k] */
f[jx + i] = (double)ipio2[jv + i];
for (j = 0, fw = 0.0; j <= jx; j++)
fw += x[j] * f[jx + i - j];
q[i] = fw;
}
jz += k;
goto recompute;
}
}
/* chop off zero terms */
if (z == 0.0) {
jz -= 1;
q0 -= 24;
while (iq[jz] == 0) {
jz--;
q0 -= 24;
}
} else { /* break z into 24-bit if necessary */
z = __scalbn(z, -q0);
if (z >= 0x1p24) {
fw = (double)(INT32)(0x1p-24 * z);
iq[jz] = (INT32)(z - 0x1p24 * fw);
jz += 1;
q0 += 24;
iq[jz] = (INT32)fw;
} else
iq[jz] = (INT32)z;
}
/* convert integer "bit" chunk to floating-point value */
fw = __scalbn(1.0, q0);
for (i = jz; i >= 0; i--) {
q[i] = fw * (double)iq[i];
fw *= 0x1p-24;
}
/* compute PIo2[0,...,jp]*q[jz,...,0] */
for(i = jz; i >= 0; i--) {
for (fw = 0.0, k = 0; k <= jp && k <= jz - i; k++)
fw += PIo2[k] * q[i + k];
fq[jz - i] = fw;
}
/* compress fq[] into y[] */
switch(prec) {
case 0:
fw = 0.0;
for (i = jz; i >= 0; i--)
fw += fq[i];
y[0] = ih == 0 ? fw : -fw;
break;
case 1:
case 2:
fw = 0.0;
for (i = jz; i >= 0; i--)
fw += fq[i];
fw = (double)fw;
y[0] = ih==0 ? fw : -fw;
fw = fq[0] - fw;
for (i = 1; i <= jz; i++)
fw += fq[i];
y[1] = ih == 0 ? fw : -fw;
break;
case 3: /* painful */
for (i = jz; i > 0; i--) {
fw = fq[i - 1] + fq[i];
fq[i] += fq[i - 1] - fw;
fq[i - 1] = fw;
}
for (i = jz; i > 1; i--) {
fw = fq[i - 1] + fq[i];
fq[i] += fq[i - 1] - fw;
fq[i - 1] = fw;
}
for (fw = 0.0, i = jz; i >= 2; i--)
fw += fq[i];
if (ih == 0) {
y[0] = fq[0];
y[1] = fq[1];
y[2] = fw;
} else {
y[0] = -fq[0];
y[1] = -fq[1];
y[2] = -fw;
}
}
return n & 7;
}
/* Copied from musl: src/math/__rem_pio2.c */
static int __rem_pio2(double x, double *y)
{

View File

@ -121,14 +121,6 @@ static float CDECL unix_atanhf(float x)
#endif
}
/*********************************************************************
* cosf
*/
static float CDECL unix_cosf( float x )
{
return cosf( x );
}
/*********************************************************************
* cosh
*/
@ -477,7 +469,6 @@ static const struct unix_funcs funcs =
unix_asinhf,
unix_atanh,
unix_atanhf,
unix_cosf,
unix_cosh,
unix_coshf,
unix_exp,

View File

@ -29,7 +29,6 @@ struct unix_funcs
float (CDECL *asinhf)(float x);
double (CDECL *atanh)(double x);
float (CDECL *atanhf)(float x);
float (CDECL *cosf)(float x);
double (CDECL *cosh)(double x);
float (CDECL *coshf)(float x);
double (CDECL *exp)(double x);