From 2786c7d21611b9fa3b2fe356542cf213e7dd0ba4 Mon Sep 17 00:00:00 2001
From: nsz <nsz@port70.net>
Date: Mon, 19 Mar 2012 22:57:58 +0100
Subject: [PATCH] use scalbn or *2.0 instead of ldexp, fix fmal

Some code assumed ldexp(x, 1) is faster than 2.0*x,
but ldexp is a wrapper around scalbn which uses
multiplications inside, so this optimization is
wrong.

This commit also fixes fmal which accidentally
used ldexp instead of ldexpl loosing precision.

There are various additional changes from the
work-in-progress const cleanups.
---
 src/math/expl.c   |   8 ++--
 src/math/expm1l.c |   6 +--
 src/math/fma.c    |  12 +++---
 src/math/fmal.c   |  12 +++---
 src/math/log10l.c |  22 +++++-----
 src/math/log2l.c  |  20 ++++-----
 src/math/logl.c   |  20 ++++-----
 src/math/powl.c   | 103 +++++++++++++++++++++++-----------------------
 8 files changed, 102 insertions(+), 101 deletions(-)

diff --git a/src/math/expl.c b/src/math/expl.c
index 9507fd2e..b289ffec 100644
--- a/src/math/expl.c
+++ b/src/math/expl.c
@@ -102,13 +102,13 @@ long double expl(long double x)
 	if (x > MAXLOGL)
 		return INFINITY;
 	if (x < MINLOGL)
-		return 0.0L;
+		return 0.0;
 
 	/* Express e**x = e**g 2**n
 	 *   = e**g e**(n loge(2))
 	 *   = e**(g + n loge(2))
 	 */
-	px = floorl(LOG2EL * x + 0.5L); /* floor() truncates toward -infinity. */
+	px = floorl(LOG2EL * x + 0.5); /* floor() truncates toward -infinity. */
 	n = px;
 	x -= px * C1;
 	x -= px * C2;
@@ -120,8 +120,8 @@ long double expl(long double x)
 	xx = x * x;
 	px = x * __polevll(xx, P, 2);
 	x =  px/(__polevll(xx, Q, 3) - px);
-	x = 1.0L + ldexpl(x, 1);
-	x = ldexpl(x, n);
+	x = 1.0 + 2.0 * x;
+	x = scalbnl(x, n);
 	return x;
 }
 #endif
diff --git a/src/math/expm1l.c b/src/math/expm1l.c
index 2f94dfa2..ca0d141f 100644
--- a/src/math/expm1l.c
+++ b/src/math/expm1l.c
@@ -97,11 +97,11 @@ long double expm1l(long double x)
 		return x;
 	/* Minimum value.*/
 	if (x < minarg)
-		return -1.0L;
+		return -1.0;
 
 	xx = C1 + C2;
 	/* Express x = ln 2 (k + remainder), remainder not exceeding 1/2. */
-	px = floorl (0.5 + x / xx);
+	px = floorl(0.5 + x / xx);
 	k = px;
 	/* remainder times ln 2 */
 	x -= px * C1;
@@ -116,7 +116,7 @@ long double expm1l(long double x)
 	/* exp(x) = exp(k ln 2) exp(remainder ln 2) = 2^k exp(remainder ln 2).
 	 We have qx = exp(remainder ln 2) - 1, so
 	 exp(x) - 1  =  2^k (qx + 1) - 1  =  2^k qx + 2^k - 1.  */
-	px = ldexpl(1.0L, k);
+	px = scalbnl(1.0, k);
 	x = px * qx + (px - 1.0);
 	return x;
 }
diff --git a/src/math/fma.c b/src/math/fma.c
index 87d450c7..5fb95406 100644
--- a/src/math/fma.c
+++ b/src/math/fma.c
@@ -247,7 +247,7 @@ static inline double add_and_denormalize(double a, double b, int scale)
 			INSERT_WORD64(sum.hi, hibits);
 		}
 	}
-	return (ldexp(sum.hi, scale));
+	return scalbn(sum.hi, scale);
 }
 
 /*
@@ -364,7 +364,7 @@ double fma(double x, double y, double z)
 		}
 	}
 	if (spread <= DBL_MANT_DIG * 2)
-		zs = ldexp(zs, -spread);
+		zs = scalbn(zs, -spread);
 	else
 		zs = copysign(DBL_MIN, zs);
 
@@ -390,7 +390,7 @@ double fma(double x, double y, double z)
 		 */
 		fesetround(oround);
 		volatile double vzs = zs; /* XXX gcc CSE bug workaround */
-		return (xy.hi + vzs + ldexp(xy.lo, spread));
+		return xy.hi + vzs + scalbn(xy.lo, spread);
 	}
 
 	if (oround != FE_TONEAREST) {
@@ -400,13 +400,13 @@ double fma(double x, double y, double z)
 		 */
 		fesetround(oround);
 		adj = r.lo + xy.lo;
-		return (ldexp(r.hi + adj, spread));
+		return scalbn(r.hi + adj, spread);
 	}
 
 	adj = add_adjusted(r.lo, xy.lo);
 	if (spread + ilogb(r.hi) > -1023)
-		return (ldexp(r.hi + adj, spread));
+		return scalbn(r.hi + adj, spread);
 	else
-		return (add_and_denormalize(r.hi, adj, spread));
+		return add_and_denormalize(r.hi, adj, spread);
 }
 #endif
diff --git a/src/math/fmal.c b/src/math/fmal.c
index cbaf46eb..be64f145 100644
--- a/src/math/fmal.c
+++ b/src/math/fmal.c
@@ -115,7 +115,7 @@ static inline long double add_and_denormalize(long double a, long double b, int
 		if (bits_lost != 1 ^ (int)(u.bits.manl & 1))
 			sum.hi = nextafterl(sum.hi, INFINITY * sum.lo);
 	}
-	return (ldexp(sum.hi, scale));
+	return scalbnl(sum.hi, scale);
 }
 
 /*
@@ -228,7 +228,7 @@ long double fmal(long double x, long double y, long double z)
 		}
 	}
 	if (spread <= LDBL_MANT_DIG * 2)
-		zs = ldexpl(zs, -spread);
+		zs = scalbnl(zs, -spread);
 	else
 		zs = copysignl(LDBL_MIN, zs);
 
@@ -254,7 +254,7 @@ long double fmal(long double x, long double y, long double z)
 		 */
 		fesetround(oround);
 		volatile long double vzs = zs; /* XXX gcc CSE bug workaround */
-		return (xy.hi + vzs + ldexpl(xy.lo, spread));
+		return xy.hi + vzs + scalbnl(xy.lo, spread);
 	}
 
 	if (oround != FE_TONEAREST) {
@@ -264,13 +264,13 @@ long double fmal(long double x, long double y, long double z)
 		 */
 		fesetround(oround);
 		adj = r.lo + xy.lo;
-		return (ldexpl(r.hi + adj, spread));
+		return scalbnl(r.hi + adj, spread);
 	}
 
 	adj = add_adjusted(r.lo, xy.lo);
 	if (spread + ilogbl(r.hi) > -16383)
-		return (ldexpl(r.hi + adj, spread));
+		return scalbnl(r.hi + adj, spread);
 	else
-		return (add_and_denormalize(r.hi, adj, spread));
+		return add_and_denormalize(r.hi, adj, spread);
 }
 #endif
diff --git a/src/math/log10l.c b/src/math/log10l.c
index b954cc77..f0eeeafb 100644
--- a/src/math/log10l.c
+++ b/src/math/log10l.c
@@ -123,9 +123,9 @@ long double log10l(long double x)
 
 	if (isnan(x))
 		return x;
-	if(x <= 0.0L) {
-		if(x == 0.0L)
-			return -1.0L / (x - x);
+	if(x <= 0.0) {
+		if(x == 0.0)
+			return -1.0 / (x - x);
 		return (x - x) / (x - x);
 	}
 	if (x == INFINITY)
@@ -142,12 +142,12 @@ long double log10l(long double x)
 	if (e > 2 || e < -2) {
 		if (x < SQRTH) {  /* 2(2x-1)/(2x+1) */
 			e -= 1;
-			z = x - 0.5L;
-			y = 0.5L * z + 0.5L;
+			z = x - 0.5;
+			y = 0.5 * z + 0.5;
 		} else {  /*  2 (x-1)/(x+1)   */
-			z = x - 0.5L;
-			z -= 0.5L;
-			y = 0.5L * x  + 0.5L;
+			z = x - 0.5;
+			z -= 0.5;
+			y = 0.5 * x  + 0.5;
 		}
 		x = z / y;
 		z = x*x;
@@ -158,13 +158,13 @@ long double log10l(long double x)
 	/* logarithm using log(1+x) = x - .5x**2 + x**3 P(x)/Q(x) */
 	if (x < SQRTH) {
 		e -= 1;
-		x = ldexpl(x, 1) - 1.0L; /*  2x - 1  */
+		x = 2.0*x - 1.0;
 	} else {
-		x = x - 1.0L;
+		x = x - 1.0;
 	}
 	z = x*x;
 	y = x * (z * __polevll(x, P, 6) / __p1evll(x, Q, 7));
-	y = y - ldexpl(z, -1);   /* -0.5x^2 + ... */
+	y = y - 0.5*z;
 
 done:
 	/* Multiply log of fraction by log10(e)
diff --git a/src/math/log2l.c b/src/math/log2l.c
index 4339c033..8ebce9c4 100644
--- a/src/math/log2l.c
+++ b/src/math/log2l.c
@@ -121,8 +121,8 @@ long double log2l(long double x)
 		return x;
 	if (x == INFINITY)
 		return x;
-	if (x <= 0.0L) {
-		if (x == 0.0L)
+	if (x <= 0.0) {
+		if (x == 0.0)
 			return -INFINITY;
 		return NAN;
 	}
@@ -139,12 +139,12 @@ long double log2l(long double x)
 	if (e > 2 || e < -2) {
 		if (x < SQRTH) {  /* 2(2x-1)/(2x+1) */
 			e -= 1;
-			z = x - 0.5L;
-			y = 0.5L * z + 0.5L;
+			z = x - 0.5;
+			y = 0.5 * z + 0.5;
 		} else {  /*  2 (x-1)/(x+1)   */
-			z = x - 0.5L;
-			z -= 0.5L;
-			y = 0.5L * x  + 0.5L;
+			z = x - 0.5;
+			z -= 0.5;
+			y = 0.5 * x + 0.5;
 		}
 		x = z / y;
 		z = x*x;
@@ -155,13 +155,13 @@ long double log2l(long double x)
 	/* logarithm using log(1+x) = x - .5x**2 + x**3 P(x)/Q(x) */
 	if (x < SQRTH) {
 		e -= 1;
-		x = ldexpl(x, 1) - 1.0L; /*  2x - 1  */
+		x = 2.0*x - 1.0;
 	} else {
-		x = x - 1.0L;
+		x = x - 1.0;
 	}
 	z = x*x;
 	y = x * (z * __polevll(x, P, 6) / __p1evll(x, Q, 7));
-	y = y - ldexpl(z, -1);   /* -0.5x^2 + ... */
+	y = y - 0.5*z;
 
 done:
 	/* Multiply log of fraction by log2(e)
diff --git a/src/math/logl.c b/src/math/logl.c
index ee7ca64a..ffd81038 100644
--- a/src/math/logl.c
+++ b/src/math/logl.c
@@ -119,8 +119,8 @@ long double logl(long double x)
 		return x;
 	if (x == INFINITY)
 		return x;
-	if (x <= 0.0L) {
-		if (x == 0.0L)
+	if (x <= 0.0) {
+		if (x == 0.0)
 			return -INFINITY;
 		return NAN;
 	}
@@ -137,12 +137,12 @@ long double logl(long double x)
 	if (e > 2 || e < -2) {
 		if (x < SQRTH) {  /* 2(2x-1)/(2x+1) */
 			e -= 1;
-			z = x - 0.5L;
-			y = 0.5L * z + 0.5L;
+			z = x - 0.5;
+			y = 0.5 * z + 0.5;
 		} else {  /*  2 (x-1)/(x+1)   */
-			z = x - 0.5L;
-			z -= 0.5L;
-			y = 0.5L * x  + 0.5L;
+			z = x - 0.5;
+			z -= 0.5;
+			y = 0.5 * x  + 0.5;
 		}
 		x = z / y;
 		z = x*x;
@@ -156,14 +156,14 @@ long double logl(long double x)
 	/* logarithm using log(1+x) = x - .5x**2 + x**3 P(x)/Q(x) */
 	if (x < SQRTH) {
 		e -= 1;
-		x = ldexpl(x, 1) - 1.0L; /*  2x - 1  */
+		x = 2.0*x - 1.0;
 	} else {
-		x = x - 1.0L;
+		x = x - 1.0;
 	}
 	z = x*x;
 	y = x * (z * __polevll(x, P, 6) / __p1evll(x, Q, 6));
 	y = y + e * C2;
-	z = y - ldexpl(z, -1);   /*  y - 0.5 * z  */
+	z = y - 0.5*z;
 	/* Note, the sum of above terms does not exceed x/4,
 	 * so it contributes at most about 1/4 lsb to the error.
 	 */
diff --git a/src/math/powl.c b/src/math/powl.c
index a6ee275f..a1d2f076 100644
--- a/src/math/powl.c
+++ b/src/math/powl.c
@@ -203,44 +203,44 @@ long double powl(long double x, long double y)
 	volatile long double z=0;
 	long double w=0, W=0, Wa=0, Wb=0, ya=0, yb=0, u=0;
 
-	if (y == 0.0L)
-		return 1.0L;
+	if (y == 0.0)
+		return 1.0;
 	if (isnan(x))
 		return x;
 	if (isnan(y))
 		return y;
-	if (y == 1.0L)
+	if (y == 1.0)
 		return x;
 
 	// FIXME: this is wrong, see pow special cases in c99 F.9.4.4
-	if (!isfinite(y) && (x == -1.0L || x == 1.0L) )
+	if (!isfinite(y) && (x == -1.0 || x == 1.0) )
 		return y - y;   /* +-1**inf is NaN */
-	if (x == 1.0L)
-		return 1.0L;
+	if (x == 1.0)
+		return 1.0;
 	if (y >= LDBL_MAX) {
-		if (x > 1.0L)
+		if (x > 1.0)
 			return INFINITY;
-		if (x > 0.0L && x < 1.0L)
-			return 0.0L;
-		if (x < -1.0L)
+		if (x > 0.0 && x < 1.0)
+			return 0.0;
+		if (x < -1.0)
 			return INFINITY;
-		if (x > -1.0L && x < 0.0L)
-			return 0.0L;
+		if (x > -1.0 && x < 0.0)
+			return 0.0;
 	}
 	if (y <= -LDBL_MAX) {
-		if (x > 1.0L)
-			return 0.0L;
-		if (x > 0.0L && x < 1.0L)
+		if (x > 1.0)
+			return 0.0;
+		if (x > 0.0 && x < 1.0)
 			return INFINITY;
-		if (x < -1.0L)
-			return 0.0L;
-		if (x > -1.0L && x < 0.0L)
+		if (x < -1.0)
+			return 0.0;
+		if (x > -1.0 && x < 0.0)
 			return INFINITY;
 	}
 	if (x >= LDBL_MAX) {
-		if (y > 0.0L)
+		if (y > 0.0)
 			return INFINITY;
-		return 0.0L;
+		return 0.0;
 	}
 
 	w = floorl(y);
@@ -253,29 +253,29 @@ long double powl(long double x, long double y)
 	yoddint = 0;
 	if (iyflg) {
 		ya = fabsl(y);
-		ya = floorl(0.5L * ya);
-		yb = 0.5L * fabsl(w);
+		ya = floorl(0.5 * ya);
+		yb = 0.5 * fabsl(w);
 		if( ya != yb )
 			yoddint = 1;
 	}
 
 	if (x <= -LDBL_MAX) {
-		if (y > 0.0L) {
+		if (y > 0.0) {
 			if (yoddint)
 				return -INFINITY;
 			return INFINITY;
 		}
-		if (y < 0.0L) {
+		if (y < 0.0) {
 			if (yoddint)
-				return -0.0L;
+				return -0.0;
 			return 0.0;
 		}
 	}
 
 
 	nflg = 0;       /* flag = 1 if x<0 raised to integer power */
-	if (x <= 0.0L) {
-		if (x == 0.0L) {
+	if (x <= 0.0) {
+		if (x == 0.0) {
 			if (y < 0.0) {
 				if (signbit(x) && yoddint)
 					return -INFINITY;
@@ -283,12 +283,12 @@ long double powl(long double x, long double y)
 			}
 			if (y > 0.0) {
 				if (signbit(x) && yoddint)
-					return -0.0L;
+					return -0.0;
 				return 0.0;
 			}
-			if (y == 0.0L)
-				return 1.0L;  /*   0**0   */
-			return 0.0L;  /*   0**y   */
+			if (y == 0.0)
+				return 1.0;  /*   0**0   */
+			return 0.0;  /*   0**y   */
 		}
 		if (iyflg == 0)
 			return (x - x) / (x - x); /* (x<0)**(non-int) is NaN */
@@ -343,7 +343,7 @@ long double powl(long double x, long double y)
 	 */
 	z = x*x;
 	w = x * (z * __polevll(x, P, 3) / __p1evll(x, Q, 3));
-	w = w - ldexpl(z, -1);  /*  w - 0.5 * z  */
+	w = w - 0.5*z;
 
 	/* Convert to base 2 logarithm:
 	 * multiply by log2(e) = 1 + LOG2EA
@@ -355,7 +355,8 @@ long double powl(long double x, long double y)
 
 	/* Compute exponent term of the base 2 logarithm. */
 	w = -i;
-	w = ldexpl(w, -LNXT); /* divide by NXT */
+	// TODO: use w * 0x1p-5;
+	w = scalbnl(w, -LNXT); /* divide by NXT */
 	w += e;
 	/* Now base 2 log of x is w + z. */
 
@@ -380,7 +381,7 @@ long double powl(long double x, long double y)
 
 	H = Fb + Gb;
 	Ha = reducl(H);
-	w = ldexpl( Ga+Ha, LNXT );
+	w = scalbnl( Ga+Ha, LNXT );
 
 	/* Test the power of 2 for overflow */
 	if (w > MEXP)
@@ -391,9 +392,9 @@ long double powl(long double x, long double y)
 	e = w;
 	Hb = H - Ha;
 
-	if (Hb > 0.0L) {
+	if (Hb > 0.0) {
 		e += 1;
-		Hb -= 1.0L/NXT;  /*0.0625L;*/
+		Hb -= 1.0/NXT;  /*0.0625L;*/
 	}
 
 	/* Now the product y * log2(x)  =  Hb + e/NXT.
@@ -415,16 +416,16 @@ long double powl(long double x, long double y)
 	w = douba(e);
 	z = w * z;  /*  2**-e * ( 1 + (2**Hb-1) )  */
 	z = z + w;
-	z = ldexpl(z, i);  /* multiply by integer power of 2 */
+	z = scalbnl(z, i);  /* multiply by integer power of 2 */
 
 	if (nflg) {
 		/* For negative x,
 		 * find out if the integer exponent
 		 * is odd or even.
 		 */
-		w = ldexpl(y, -1);
+		w = 0.5*y;
 		w = floorl(w);
-		w = ldexpl(w, 1);
+		w = 2.0*w;
 		if (w != y)
 			z = -z;  /* odd exponent */
 	}
@@ -438,9 +439,9 @@ static long double reducl(long double x)
 {
 	long double t;
 
-	t = ldexpl(x, LNXT);
+	t = scalbnl(x, LNXT);
 	t = floorl(t);
-	t = ldexpl(t, -LNXT);
+	t = scalbnl(t, -LNXT);
 	return t;
 }
 
@@ -483,18 +484,18 @@ static long double powil(long double x, int nn)
 	long double s;
 	int n, e, sign, asign, lx;
 
-	if (x == 0.0L) {
+	if (x == 0.0) {
 		if (nn == 0)
-			return 1.0L;
+			return 1.0;
 		else if (nn < 0)
 			return LDBL_MAX;
-		return 0.0L;
+		return 0.0;
 	}
 
 	if (nn == 0)
-		return 1.0L;
+		return 1.0;
 
-	if (x < 0.0L) {
+	if (x < 0.0) {
 		asign = -1;
 		x = -x;
 	} else
@@ -516,7 +517,7 @@ static long double powil(long double x, int nn)
 	e = (lx - 1)*n;
 	if ((e == 0) || (e > 64) || (e < -64)) {
 		s = (s - 7.0710678118654752e-1L) / (s +  7.0710678118654752e-1L);
-		s = (2.9142135623730950L * s - 0.5L + lx) * nn * LOGE2L;
+		s = (2.9142135623730950L * s - 0.5 + lx) * nn * LOGE2L;
 	} else {
 		s = LOGE2L * e;
 	}
@@ -530,8 +531,8 @@ static long double powil(long double x, int nn)
 	 * since roundoff error in 1.0/x will be amplified.
 	 * The precise demarcation should be the gradual underflow threshold.
 	 */
-	if (s < -MAXLOGL+2.0L) {
-		x = 1.0L/x;
+	if (s < -MAXLOGL+2.0) {
+		x = 1.0/x;
 		sign = -sign;
 	}
 
@@ -539,7 +540,7 @@ static long double powil(long double x, int nn)
 	if (n & 1)
 		y = x;
 	else {
-		y = 1.0L;
+		y = 1.0;
 		asign = 0;
 	}
 
@@ -555,7 +556,7 @@ static long double powil(long double x, int nn)
 	if (asign)
 		y = -y;  /* odd power of negative number */
 	if (sign < 0)
-		y = 1.0L/y;
+		y = 1.0/y;
 	return y;
 }
 
-- 
2.25.1