glibc/sysdeps/ieee754/ldbl-128/e_acosl.c
Joseph Myers 1d9ab20c14 Fix ldbl-128/ldbl-128ibm acosl inaccuracy (bug 18038, bug 18039).
The ldbl-128 and ldbl-128ibm implementations of acosl have similar
bugs, using a threshold of 0x1p-57L to determine when they just return
pi/2.  Since the result pi/2 - asinl (x) is roughly pi/2 - x for small
x, the relevant cut-off is actually x being < 0.5ulp of 1.  This patch
fixes the implementations to use that cut-off and adds tests of small
acos arguments.

Tested for powerpc and mips64.  Also tested for x86_64 and x86; no
ulps updates needed.

	[BZ #18038]
	[BZ #18039]
	* sysdeps/ieee754/ldbl-128/e_acosl.c (__ieee754_acosl): Only
	return pi/2 for arguments below 0x1p-113L.
	* sysdeps/ieee754/ldbl-128ibm/e_acosl.c (__ieee754_acosl): Only
	return pi/2 for arguments below 0x1p-106L.
	* math/auto-libm-test-in: Add more tests of acos.
	* math/auto-libm-test-out: Regenerated.
2015-02-26 21:06:34 +00:00

320 lines
9.7 KiB
C

/*
* ====================================================
* Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
*
* Developed at SunPro, a Sun Microsystems, Inc. business.
* Permission to use, copy, modify, and distribute this
* software is freely granted, provided that this notice
* is preserved.
* ====================================================
*/
/*
Long double expansions are
Copyright (C) 2001 Stephen L. Moshier <moshier@na-net.ornl.gov>
and are incorporated herein by permission of the author. The author
reserves the right to distribute this material elsewhere under different
copying permissions. These modifications are distributed here under
the following terms:
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, see
<http://www.gnu.org/licenses/>. */
/* __ieee754_acosl(x)
* Method :
* acos(x) = pi/2 - asin(x)
* acos(-x) = pi/2 + asin(x)
* For |x| <= 0.375
* acos(x) = pi/2 - asin(x)
* Between .375 and .5 the approximation is
* acos(0.4375 + x) = acos(0.4375) + x P(x) / Q(x)
* Between .5 and .625 the approximation is
* acos(0.5625 + x) = acos(0.5625) + x rS(x) / sS(x)
* For x > 0.625,
* acos(x) = 2 asin(sqrt((1-x)/2))
* computed with an extended precision square root in the leading term.
* For x < -0.625
* acos(x) = pi - 2 asin(sqrt((1-|x|)/2))
*
* Special cases:
* if x is NaN, return x itself;
* if |x|>1, return NaN with invalid signal.
*
* Functions needed: __ieee754_sqrtl.
*/
#include <math.h>
#include <math_private.h>
static const long double
one = 1.0L,
pio2_hi = 1.5707963267948966192313216916397514420986L,
pio2_lo = 4.3359050650618905123985220130216759843812E-35L,
/* acos(0.5625 + x) = acos(0.5625) + x rS(x) / sS(x)
-0.0625 <= x <= 0.0625
peak relative error 3.3e-35 */
rS0 = 5.619049346208901520945464704848780243887E0L,
rS1 = -4.460504162777731472539175700169871920352E1L,
rS2 = 1.317669505315409261479577040530751477488E2L,
rS3 = -1.626532582423661989632442410808596009227E2L,
rS4 = 3.144806644195158614904369445440583873264E1L,
rS5 = 9.806674443470740708765165604769099559553E1L,
rS6 = -5.708468492052010816555762842394927806920E1L,
rS7 = -1.396540499232262112248553357962639431922E1L,
rS8 = 1.126243289311910363001762058295832610344E1L,
rS9 = 4.956179821329901954211277873774472383512E-1L,
rS10 = -3.313227657082367169241333738391762525780E-1L,
sS0 = -4.645814742084009935700221277307007679325E0L,
sS1 = 3.879074822457694323970438316317961918430E1L,
sS2 = -1.221986588013474694623973554726201001066E2L,
sS3 = 1.658821150347718105012079876756201905822E2L,
sS4 = -4.804379630977558197953176474426239748977E1L,
sS5 = -1.004296417397316948114344573811562952793E2L,
sS6 = 7.530281592861320234941101403870010111138E1L,
sS7 = 1.270735595411673647119592092304357226607E1L,
sS8 = -1.815144839646376500705105967064792930282E1L,
sS9 = -7.821597334910963922204235247786840828217E-2L,
/* 1.000000000000000000000000000000000000000E0 */
acosr5625 = 9.7338991014954640492751132535550279812151E-1L,
pimacosr5625 = 2.1682027434402468335351320579240000860757E0L,
/* acos(0.4375 + x) = acos(0.4375) + x rS(x) / sS(x)
-0.0625 <= x <= 0.0625
peak relative error 2.1e-35 */
P0 = 2.177690192235413635229046633751390484892E0L,
P1 = -2.848698225706605746657192566166142909573E1L,
P2 = 1.040076477655245590871244795403659880304E2L,
P3 = -1.400087608918906358323551402881238180553E2L,
P4 = 2.221047917671449176051896400503615543757E1L,
P5 = 9.643714856395587663736110523917499638702E1L,
P6 = -5.158406639829833829027457284942389079196E1L,
P7 = -1.578651828337585944715290382181219741813E1L,
P8 = 1.093632715903802870546857764647931045906E1L,
P9 = 5.448925479898460003048760932274085300103E-1L,
P10 = -3.315886001095605268470690485170092986337E-1L,
Q0 = -1.958219113487162405143608843774587557016E0L,
Q1 = 2.614577866876185080678907676023269360520E1L,
Q2 = -9.990858606464150981009763389881793660938E1L,
Q3 = 1.443958741356995763628660823395334281596E2L,
Q4 = -3.206441012484232867657763518369723873129E1L,
Q5 = -1.048560885341833443564920145642588991492E2L,
Q6 = 6.745883931909770880159915641984874746358E1L,
Q7 = 1.806809656342804436118449982647641392951E1L,
Q8 = -1.770150690652438294290020775359580915464E1L,
Q9 = -5.659156469628629327045433069052560211164E-1L,
/* 1.000000000000000000000000000000000000000E0 */
acosr4375 = 1.1179797320499710475919903296900511518755E0L,
pimacosr4375 = 2.0236129215398221908706530535894517323217E0L,
/* asin(x) = x + x^3 pS(x^2) / qS(x^2)
0 <= x <= 0.5
peak relative error 1.9e-35 */
pS0 = -8.358099012470680544198472400254596543711E2L,
pS1 = 3.674973957689619490312782828051860366493E3L,
pS2 = -6.730729094812979665807581609853656623219E3L,
pS3 = 6.643843795209060298375552684423454077633E3L,
pS4 = -3.817341990928606692235481812252049415993E3L,
pS5 = 1.284635388402653715636722822195716476156E3L,
pS6 = -2.410736125231549204856567737329112037867E2L,
pS7 = 2.219191969382402856557594215833622156220E1L,
pS8 = -7.249056260830627156600112195061001036533E-1L,
pS9 = 1.055923570937755300061509030361395604448E-3L,
qS0 = -5.014859407482408326519083440151745519205E3L,
qS1 = 2.430653047950480068881028451580393430537E4L,
qS2 = -4.997904737193653607449250593976069726962E4L,
qS3 = 5.675712336110456923807959930107347511086E4L,
qS4 = -3.881523118339661268482937768522572588022E4L,
qS5 = 1.634202194895541569749717032234510811216E4L,
qS6 = -4.151452662440709301601820849901296953752E3L,
qS7 = 5.956050864057192019085175976175695342168E2L,
qS8 = -4.175375777334867025769346564600396877176E1L;
/* 1.000000000000000000000000000000000000000E0 */
long double
__ieee754_acosl (long double x)
{
long double z, r, w, p, q, s, t, f2;
int32_t ix, sign;
ieee854_long_double_shape_type u;
u.value = x;
sign = u.parts32.w0;
ix = sign & 0x7fffffff;
u.parts32.w0 = ix; /* |x| */
if (ix >= 0x3fff0000) /* |x| >= 1 */
{
if (ix == 0x3fff0000
&& (u.parts32.w1 | u.parts32.w2 | u.parts32.w3) == 0)
{ /* |x| == 1 */
if ((sign & 0x80000000) == 0)
return 0.0; /* acos(1) = 0 */
else
return (2.0 * pio2_hi) + (2.0 * pio2_lo); /* acos(-1)= pi */
}
return (x - x) / (x - x); /* acos(|x| > 1) is NaN */
}
else if (ix < 0x3ffe0000) /* |x| < 0.5 */
{
if (ix < 0x3f8e0000) /* |x| < 2**-113 */
return pio2_hi + pio2_lo;
if (ix < 0x3ffde000) /* |x| < .4375 */
{
/* Arcsine of x. */
z = x * x;
p = (((((((((pS9 * z
+ pS8) * z
+ pS7) * z
+ pS6) * z
+ pS5) * z
+ pS4) * z
+ pS3) * z
+ pS2) * z
+ pS1) * z
+ pS0) * z;
q = (((((((( z
+ qS8) * z
+ qS7) * z
+ qS6) * z
+ qS5) * z
+ qS4) * z
+ qS3) * z
+ qS2) * z
+ qS1) * z
+ qS0;
r = x + x * p / q;
z = pio2_hi - (r - pio2_lo);
return z;
}
/* .4375 <= |x| < .5 */
t = u.value - 0.4375L;
p = ((((((((((P10 * t
+ P9) * t
+ P8) * t
+ P7) * t
+ P6) * t
+ P5) * t
+ P4) * t
+ P3) * t
+ P2) * t
+ P1) * t
+ P0) * t;
q = (((((((((t
+ Q9) * t
+ Q8) * t
+ Q7) * t
+ Q6) * t
+ Q5) * t
+ Q4) * t
+ Q3) * t
+ Q2) * t
+ Q1) * t
+ Q0;
r = p / q;
if (sign & 0x80000000)
r = pimacosr4375 - r;
else
r = acosr4375 + r;
return r;
}
else if (ix < 0x3ffe4000) /* |x| < 0.625 */
{
t = u.value - 0.5625L;
p = ((((((((((rS10 * t
+ rS9) * t
+ rS8) * t
+ rS7) * t
+ rS6) * t
+ rS5) * t
+ rS4) * t
+ rS3) * t
+ rS2) * t
+ rS1) * t
+ rS0) * t;
q = (((((((((t
+ sS9) * t
+ sS8) * t
+ sS7) * t
+ sS6) * t
+ sS5) * t
+ sS4) * t
+ sS3) * t
+ sS2) * t
+ sS1) * t
+ sS0;
if (sign & 0x80000000)
r = pimacosr5625 - p / q;
else
r = acosr5625 + p / q;
return r;
}
else
{ /* |x| >= .625 */
z = (one - u.value) * 0.5;
s = __ieee754_sqrtl (z);
/* Compute an extended precision square root from
the Newton iteration s -> 0.5 * (s + z / s).
The change w from s to the improved value is
w = 0.5 * (s + z / s) - s = (s^2 + z)/2s - s = (z - s^2)/2s.
Express s = f1 + f2 where f1 * f1 is exactly representable.
w = (z - s^2)/2s = (z - f1^2 - 2 f1 f2 - f2^2)/2s .
s + w has extended precision. */
u.value = s;
u.parts32.w2 = 0;
u.parts32.w3 = 0;
f2 = s - u.value;
w = z - u.value * u.value;
w = w - 2.0 * u.value * f2;
w = w - f2 * f2;
w = w / (2.0 * s);
/* Arcsine of s. */
p = (((((((((pS9 * z
+ pS8) * z
+ pS7) * z
+ pS6) * z
+ pS5) * z
+ pS4) * z
+ pS3) * z
+ pS2) * z
+ pS1) * z
+ pS0) * z;
q = (((((((( z
+ qS8) * z
+ qS7) * z
+ qS6) * z
+ qS5) * z
+ qS4) * z
+ qS3) * z
+ qS2) * z
+ qS1) * z
+ qS0;
r = s + (w + s * p / q);
if (sign & 0x80000000)
w = pio2_hi + (pio2_lo - r);
else
w = r;
return 2.0 * w;
}
}
strong_alias (__ieee754_acosl, __acosl_finite)