Module: wine Branch: master Commit: 3171aa2b2d8db28e5bfde1cad92e0ef36ee7ccaa URL: https://source.winehq.org/git/wine.git/?a=commit;h=3171aa2b2d8db28e5bfde1cad...
Author: Piotr Caban piotr@codeweavers.com Date: Fri Jun 4 18:25:49 2021 +0200
msvcrt: Import lgamma implementation from musl.
Signed-off-by: Piotr Caban piotr@codeweavers.com Signed-off-by: Alexandre Julliard julliard@winehq.org
---
configure | 1 - configure.ac | 1 - dlls/msvcrt/math.c | 196 +++++++++++++++++++++++++++++++++++++++++++++++++- dlls/msvcrt/unixlib.c | 14 ---- dlls/msvcrt/unixlib.h | 1 - include/config.h.in | 3 - 6 files changed, 195 insertions(+), 21 deletions(-)
diff --git a/configure b/configure index edc15165750..6c890d0e36b 100755 --- a/configure +++ b/configure @@ -19620,7 +19620,6 @@ for ac_func in \ exp2 \ exp2f \ fmaf \ - lgamma \ lgammaf \ tgamma \ tgammaf diff --git a/configure.ac b/configure.ac index cb2b0cbeb17..f5bf9bc071f 100644 --- a/configure.ac +++ b/configure.ac @@ -2659,7 +2659,6 @@ AC_CHECK_FUNCS(\ exp2 \ exp2f \ fmaf \ - lgamma \ lgammaf \ tgamma \ tgammaf diff --git a/dlls/msvcrt/math.c b/dlls/msvcrt/math.c index ae9debce20c..6f0b0c1ba4f 100644 --- a/dlls/msvcrt/math.c +++ b/dlls/msvcrt/math.c @@ -8644,12 +8644,206 @@ end: return sx ? -x : x; }
+/* sin(pi*x) assuming x > 2^-100, if sin(pi*x)==0 the sign is arbitrary */ +static double sin_pi(double x) +{ + static const double pi = 3.14159265358979311600e+00; + int n; + + /* spurious inexact if odd int */ + x = 2.0 * (x * 0.5 - floor(x * 0.5)); /* x mod 2.0 */ + + n = (int)(x * 4.0); + n = (n + 1) / 2; + x -= n * 0.5f; + x *= pi; + + switch (n) { + default: /* case 4: */ + case 0: return __sin(x, 0.0, 0); + case 1: return __cos(x, 0.0); + case 2: return __sin(-x, 0.0, 0); + case 3: return -__cos(x, 0.0); + } +} + /********************************************************************* * lgamma (MSVCR120.@) + * + * Copied from musl: src/math/lgamma_r.c */ double CDECL lgamma(double x) { - return unix_funcs->lgamma( x ); + static const double pi = 3.14159265358979311600e+00, + a0 = 7.72156649015328655494e-02, + a1 = 3.22467033424113591611e-01, + a2 = 6.73523010531292681824e-02, + a3 = 2.05808084325167332806e-02, + a4 = 7.38555086081402883957e-03, + a5 = 2.89051383673415629091e-03, + a6 = 1.19270763183362067845e-03, + a7 = 5.10069792153511336608e-04, + a8 = 2.20862790713908385557e-04, + a9 = 1.08011567247583939954e-04, + a10 = 2.52144565451257326939e-05, + a11 = 4.48640949618915160150e-05, + tc = 1.46163214496836224576e+00, + tf = -1.21486290535849611461e-01, + tt = -3.63867699703950536541e-18, + t0 = 4.83836122723810047042e-01, + t1 = -1.47587722994593911752e-01, + t2 = 6.46249402391333854778e-02, + t3 = -3.27885410759859649565e-02, + t4 = 1.79706750811820387126e-02, + t5 = -1.03142241298341437450e-02, + t6 = 6.10053870246291332635e-03, + t7 = -3.68452016781138256760e-03, + t8 = 2.25964780900612472250e-03, + t9 = -1.40346469989232843813e-03, + t10 = 8.81081882437654011382e-04, + t11 = -5.38595305356740546715e-04, + t12 = 3.15632070903625950361e-04, + t13 = -3.12754168375120860518e-04, + t14 = 3.35529192635519073543e-04, + u0 = -7.72156649015328655494e-02, + u1 = 6.32827064025093366517e-01, + u2 = 1.45492250137234768737e+00, + u3 = 9.77717527963372745603e-01, + u4 = 2.28963728064692451092e-01, + u5 = 1.33810918536787660377e-02, + v1 = 2.45597793713041134822e+00, + v2 = 2.12848976379893395361e+00, + v3 = 7.69285150456672783825e-01, + v4 = 1.04222645593369134254e-01, + v5 = 3.21709242282423911810e-03, + s0 = -7.72156649015328655494e-02, + s1 = 2.14982415960608852501e-01, + s2 = 3.25778796408930981787e-01, + s3 = 1.46350472652464452805e-01, + s4 = 2.66422703033638609560e-02, + s5 = 1.84028451407337715652e-03, + s6 = 3.19475326584100867617e-05, + r1 = 1.39200533467621045958e+00, + r2 = 7.21935547567138069525e-01, + r3 = 1.71933865632803078993e-01, + r4 = 1.86459191715652901344e-02, + r5 = 7.77942496381893596434e-04, + r6 = 7.32668430744625636189e-06, + w0 = 4.18938533204672725052e-01, + w1 = 8.33333333333329678849e-02, + w2 = -2.77777777728775536470e-03, + w3 = 7.93650558643019558500e-04, + w4 = -5.95187557450339963135e-04, + w5 = 8.36339918996282139126e-04, + w6 = -1.63092934096575273989e-03; + + union {double f; UINT64 i;} u = {x}; + double t, y, z, nadj, p, p1, p2, p3, q, r, w; + UINT32 ix; + int sign,i; + + /* purge off +-inf, NaN, +-0, tiny and negative arguments */ + sign = u.i >> 63; + ix = u.i >> 32 & 0x7fffffff; + if (ix >= 0x7ff00000) + return x * x; + if (ix < (0x3ff - 70) << 20) { /* |x|<2**-70, return -log(|x|) */ + if(sign) + x = -x; + return -log(x); + } + if (sign) { + x = -x; + t = sin_pi(x); + if (t == 0.0) { /* -integer */ + *_errno() = ERANGE; + return 1.0 / (x - x); + } + if (t <= 0.0) + t = -t; + nadj = log(pi / (t * x)); + } + + /* purge off 1 and 2 */ + if ((ix == 0x3ff00000 || ix == 0x40000000) && (UINT32)u.i == 0) + r = 0; + /* for x < 2.0 */ + else if (ix < 0x40000000) { + if (ix <= 0x3feccccc) { /* lgamma(x) = lgamma(x+1)-log(x) */ + r = -log(x); + if (ix >= 0x3FE76944) { + y = 1.0 - x; + i = 0; + } else if (ix >= 0x3FCDA661) { + y = x - (tc - 1.0); + i = 1; + } else { + y = x; + i = 2; + } + } else { + r = 0.0; + if (ix >= 0x3FFBB4C3) { /* [1.7316,2] */ + y = 2.0 - x; + i = 0; + } else if(ix >= 0x3FF3B4C4) { /* [1.23,1.73] */ + y = x - tc; + i = 1; + } else { + y = x - 1.0; + i = 2; + } + } + switch (i) { + case 0: + z = y * y; + p1 = a0 + z * (a2 + z * (a4 + z * (a6 + z * (a8 + z * a10)))); + p2 = z * (a1 + z * (a3 + z * (a5 + z * (a7 + z * (a9 + z * a11))))); + p = y * p1 + p2; + r += (p - 0.5 * y); + break; + case 1: + z = y * y; + w = z * y; + p1 = t0 + w * (t3 + w * (t6 + w * (t9 + w * t12))); /* parallel comp */ + p2 = t1 + w * (t4 + w * (t7 + w * (t10 + w * t13))); + p3 = t2 + w * (t5 + w * (t8 + w * (t11 + w * t14))); + p = z * p1 - (tt - w * (p2 + y * p3)); + r += tf + p; + break; + case 2: + p1 = y * (u0 + y * (u1 + y * (u2 + y * (u3 + y * (u4 + y * u5))))); + p2 = 1.0 + y * (v1 + y * (v2 + y * (v3 + y * (v4 + y * v5)))); + r += -0.5 * y + p1 / p2; + } + } else if (ix < 0x40200000) { /* x < 8.0 */ + i = (int)x; + y = x - (double)i; + p = y * (s0 + y * (s1 + y * (s2 + y * (s3 + y * (s4 + y * (s5 + y * s6)))))); + q = 1.0 + y * (r1 + y * (r2 + y * (r3 + y * (r4 + y * (r5 + y * r6))))); + r = 0.5 * y + p / q; + z = 1.0; /* lgamma(1+s) = log(s) + lgamma(s) */ + switch (i) { + case 7: z *= y + 6.0; /* fall through */ + case 6: z *= y + 5.0; /* fall through */ + case 5: z *= y + 4.0; /* fall through */ + case 4: z *= y + 3.0; /* fall through */ + case 3: + z *= y + 2.0; + r += log(z); + break; + } + } else if (ix < 0x43900000) { /* 8.0 <= x < 2**58 */ + t = log(x); + z = 1.0 / x; + y = z * z; + w = w0 + z * (w1 + y * (w2 + y * (w3 + y * (w4 + y * (w5 + y * w6))))); + r = (x - 0.5) * (t - 1.0) + w; + } else /* 2**58 <= x <= inf */ + r = x * (log(x) - 1.0); + if (sign) + r = nadj - r; + return r; }
/********************************************************************* diff --git a/dlls/msvcrt/unixlib.c b/dlls/msvcrt/unixlib.c index 440d002869c..14aeef4c1e7 100644 --- a/dlls/msvcrt/unixlib.c +++ b/dlls/msvcrt/unixlib.c @@ -94,19 +94,6 @@ static float CDECL unix_fmaf( float x, float y, float z ) #endif }
-/********************************************************************* - * lgamma - */ -static double CDECL unix_lgamma(double x) -{ -#ifdef HAVE_LGAMMA - return lgamma(x); -#else - FIXME( "not implemented\n" ); - return 0; -#endif -} - /********************************************************************* * lgammaf */ @@ -169,7 +156,6 @@ static const struct unix_funcs funcs = unix_exp2, unix_exp2f, unix_fmaf, - unix_lgamma, unix_lgammaf, unix_pow, unix_powf, diff --git a/dlls/msvcrt/unixlib.h b/dlls/msvcrt/unixlib.h index fc2ceb7798d..4309ff835a0 100644 --- a/dlls/msvcrt/unixlib.h +++ b/dlls/msvcrt/unixlib.h @@ -28,7 +28,6 @@ struct unix_funcs double (CDECL *exp2)(double x); float (CDECL *exp2f)(float x); float (CDECL *fmaf)(float x, float y, float z); - double (CDECL *lgamma)(double x); float (CDECL *lgammaf)(float x); double (CDECL *pow)(double x, double y); float (CDECL *powf)(float x, float y); diff --git a/include/config.h.in b/include/config.h.in index 252c2819fcc..dd34230b268 100644 --- a/include/config.h.in +++ b/include/config.h.in @@ -297,9 +297,6 @@ /* Define to 1 if you have the `ldap_parse_vlv_control' function. */ #undef HAVE_LDAP_PARSE_VLV_CONTROL
-/* Define to 1 if you have the `lgamma' function. */ -#undef HAVE_LGAMMA - /* Define to 1 if you have the `lgammaf' function. */ #undef HAVE_LGAMMAF
-
Alexandre Julliard