glibc/sysdeps/ieee754/ldbl-128/s_erfl.c

949 lines
30 KiB
C
Raw Normal View History

/*
* ====================================================
* 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.
* ====================================================
*/
2002-08-26 Brian Youmans <3diff@gnu.org> * crypt/crypt.c: Changed copying permission notice to Lesser GPL from Library GPL. * crypt/crypt_util.c: Likewise. * crypt/ufc.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-iteratephdr.c: Likewise. * iconv/iconvconfig.h: Likewise. * linuxthreads/Examples/ex10.c: Likewise. * linuxthreads/Examples/ex11.c: Likewise. * linuxthreads/Examples/ex13.c: Likewise. * linuxthreads/Examples/ex8.c: Likewise. * linuxthreads/Examples/ex9.c: Likewise. * linuxthreads/barrier.c: Likewise. * linuxthreads/events.c: Likewise. * linuxthreads/lockfile.c: Likewise. * linuxthreads/no-tsd.c: Likewise. * linuxthreads/pt-machine.c: Likewise. * linuxthreads/ptclock_gettime.c: Likewise. * linuxthreads/ptclock_settime.c: Likewise. * linuxthreads/rwlock.c: Likewise. * linuxthreads/sysdeps/alpha/pspinlock.c: Likewise. * linuxthreads/sysdeps/alpha/pt-machine.h: Likewise. * linuxthreads/sysdeps/arm/pspinlock.c: Likewise. * linuxthreads/sysdeps/arm/pt-machine.h: Likewise. * linuxthreads/sysdeps/cris/pspinlock.c: Likewise. * linuxthreads/sysdeps/cris/pt-machine.h: Likewise. * linuxthreads/sysdeps/hppa/pspinlock.c: Likewise. * linuxthreads/sysdeps/hppa/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/i686/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/pspinlock.c: Likewise. * linuxthreads/sysdeps/i386/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/useldt.h: Likewise. * linuxthreads/sysdeps/ia64/pspinlock.c: Likewise. * linuxthreads/sysdeps/ia64/pt-machine.h: Likewise. * linuxthreads/sysdeps/m68k/pspinlock.c: Likewise. * linuxthreads/sysdeps/m68k/pt-machine.h: Likewise. * linuxthreads/sysdeps/mips/pspinlock.c: Likewise. * linuxthreads/sysdeps/mips/pt-machine.h: Likewise. * linuxthreads/sysdeps/powerpc/pspinlock.c: Likewise. * linuxthreads/sysdeps/powerpc/pt-machine.h: Likewise. * linuxthreads/sysdeps/pthread/bits/initspin.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-lock.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-tsd.h: Likewise. * linuxthreads/sysdeps/pthread/getcpuclockid.c: Likewise. * linuxthreads/sysdeps/pthread/posix-timer.h: Likewise. * linuxthreads/sysdeps/pthread/timer_create.c: Likewise. * linuxthreads/sysdeps/pthread/timer_delete.c: Likewise. * linuxthreads/sysdeps/pthread/timer_getoverr.c: Likewise. * linuxthreads/sysdeps/pthread/timer_gettime.c: Likewise. * linuxthreads/sysdeps/pthread/timer_routines.c: Likewise. * linuxthreads/sysdeps/pthread/timer_settime.c: Likewise. * linuxthreads/sysdeps/pthread/tst-timer.c: Likewise. * linuxthreads/sysdeps/s390/pspinlock.c: Likewise. * linuxthreads/sysdeps/s390/s390-32/pt-machine.h: Likewise. * linuxthreads/sysdeps/s390/s390-64/pt-machine.h: Likewise. * linuxthreads/sysdeps/sh/pspinlock.c: Likewise. * linuxthreads/sysdeps/sh/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/sparcv9/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pt-machine.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/local_lim.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/posix_opt.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/sigthread.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/hppa/bits/initspin.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/i386/bits/posix_opt.h: Likewise. * linuxthreads/tststack.c: Likewise. * linuxthreads/unload.c: Likewise. * linuxthreads/weaks.c: Likewise. * linuxthreads/wrapsyscall.c: Likewise. * malloc/arena.c: Likewise. * malloc/hooks.c: Likewise. * malloc/malloc.c: Likewise. * posix/glob/Makefile.ami: Likewise. * posix/glob/Makefile.in: Likewise. * stdlib/gmp-impl.h: Likewise. * stdlib/gmp.h: Likewise. * sysdeps/generic/dl-iteratephdr-static.c: Likewise. * sysdeps/generic/strnlen.c: Likewise. * sysdeps/mach/hurd/powerpc/bits/sigcontext.h: Likewise. * sysdeps/mach/hurd/recvmsg.c: Likewise. * sysdeps/mach/hurd/sendmsg.c: Likewise. * sysdeps/mach/hurd/spawni.c: Likewise. * sysdeps/mach/powerpc/machine-sp.h: Likewise. * sysdeps/mach/powerpc/sysdep.h: Likewise. * sysdeps/mach/powerpc/thread_state.h: Likewise. * sysdeps/unix/bsd/bsd4.4/bits/socket.h: Likewise. * sysdeps/unix/sysv/linux/ia64/dl-iteratephdr-static.c: Likewise. * sysdeps/x86_64/gmp-mparam.h: Likewise.
2002-08-26 22:40:48 +00:00
/* Modifications and expansions for 128-bit long double are
Copyright (C) 2001 Stephen L. Moshier <moshier@na-net.ornl.gov>
2013-06-05 20:44:03 +00:00
and are incorporated herein by permission of the author. The author
reserves the right to distribute this material elsewhere under different
2013-06-05 20:44:03 +00:00
copying permissions. These modifications are distributed here under
the following terms:
2002-08-26 Brian Youmans <3diff@gnu.org> * crypt/crypt.c: Changed copying permission notice to Lesser GPL from Library GPL. * crypt/crypt_util.c: Likewise. * crypt/ufc.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-iteratephdr.c: Likewise. * iconv/iconvconfig.h: Likewise. * linuxthreads/Examples/ex10.c: Likewise. * linuxthreads/Examples/ex11.c: Likewise. * linuxthreads/Examples/ex13.c: Likewise. * linuxthreads/Examples/ex8.c: Likewise. * linuxthreads/Examples/ex9.c: Likewise. * linuxthreads/barrier.c: Likewise. * linuxthreads/events.c: Likewise. * linuxthreads/lockfile.c: Likewise. * linuxthreads/no-tsd.c: Likewise. * linuxthreads/pt-machine.c: Likewise. * linuxthreads/ptclock_gettime.c: Likewise. * linuxthreads/ptclock_settime.c: Likewise. * linuxthreads/rwlock.c: Likewise. * linuxthreads/sysdeps/alpha/pspinlock.c: Likewise. * linuxthreads/sysdeps/alpha/pt-machine.h: Likewise. * linuxthreads/sysdeps/arm/pspinlock.c: Likewise. * linuxthreads/sysdeps/arm/pt-machine.h: Likewise. * linuxthreads/sysdeps/cris/pspinlock.c: Likewise. * linuxthreads/sysdeps/cris/pt-machine.h: Likewise. * linuxthreads/sysdeps/hppa/pspinlock.c: Likewise. * linuxthreads/sysdeps/hppa/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/i686/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/pspinlock.c: Likewise. * linuxthreads/sysdeps/i386/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/useldt.h: Likewise. * linuxthreads/sysdeps/ia64/pspinlock.c: Likewise. * linuxthreads/sysdeps/ia64/pt-machine.h: Likewise. * linuxthreads/sysdeps/m68k/pspinlock.c: Likewise. * linuxthreads/sysdeps/m68k/pt-machine.h: Likewise. * linuxthreads/sysdeps/mips/pspinlock.c: Likewise. * linuxthreads/sysdeps/mips/pt-machine.h: Likewise. * linuxthreads/sysdeps/powerpc/pspinlock.c: Likewise. * linuxthreads/sysdeps/powerpc/pt-machine.h: Likewise. * linuxthreads/sysdeps/pthread/bits/initspin.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-lock.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-tsd.h: Likewise. * linuxthreads/sysdeps/pthread/getcpuclockid.c: Likewise. * linuxthreads/sysdeps/pthread/posix-timer.h: Likewise. * linuxthreads/sysdeps/pthread/timer_create.c: Likewise. * linuxthreads/sysdeps/pthread/timer_delete.c: Likewise. * linuxthreads/sysdeps/pthread/timer_getoverr.c: Likewise. * linuxthreads/sysdeps/pthread/timer_gettime.c: Likewise. * linuxthreads/sysdeps/pthread/timer_routines.c: Likewise. * linuxthreads/sysdeps/pthread/timer_settime.c: Likewise. * linuxthreads/sysdeps/pthread/tst-timer.c: Likewise. * linuxthreads/sysdeps/s390/pspinlock.c: Likewise. * linuxthreads/sysdeps/s390/s390-32/pt-machine.h: Likewise. * linuxthreads/sysdeps/s390/s390-64/pt-machine.h: Likewise. * linuxthreads/sysdeps/sh/pspinlock.c: Likewise. * linuxthreads/sysdeps/sh/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/sparcv9/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pt-machine.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/local_lim.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/posix_opt.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/sigthread.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/hppa/bits/initspin.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/i386/bits/posix_opt.h: Likewise. * linuxthreads/tststack.c: Likewise. * linuxthreads/unload.c: Likewise. * linuxthreads/weaks.c: Likewise. * linuxthreads/wrapsyscall.c: Likewise. * malloc/arena.c: Likewise. * malloc/hooks.c: Likewise. * malloc/malloc.c: Likewise. * posix/glob/Makefile.ami: Likewise. * posix/glob/Makefile.in: Likewise. * stdlib/gmp-impl.h: Likewise. * stdlib/gmp.h: Likewise. * sysdeps/generic/dl-iteratephdr-static.c: Likewise. * sysdeps/generic/strnlen.c: Likewise. * sysdeps/mach/hurd/powerpc/bits/sigcontext.h: Likewise. * sysdeps/mach/hurd/recvmsg.c: Likewise. * sysdeps/mach/hurd/sendmsg.c: Likewise. * sysdeps/mach/hurd/spawni.c: Likewise. * sysdeps/mach/powerpc/machine-sp.h: Likewise. * sysdeps/mach/powerpc/sysdep.h: Likewise. * sysdeps/mach/powerpc/thread_state.h: Likewise. * sysdeps/unix/bsd/bsd4.4/bits/socket.h: Likewise. * sysdeps/unix/sysv/linux/ia64/dl-iteratephdr-static.c: Likewise. * sysdeps/x86_64/gmp-mparam.h: Likewise.
2002-08-26 22:40:48 +00:00
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/>. */
/* double erf(double x)
* double erfc(double x)
* x
* 2 |\
* erf(x) = --------- | exp(-t*t)dt
* sqrt(pi) \|
* 0
*
* erfc(x) = 1-erf(x)
* Note that
* erf(-x) = -erf(x)
* erfc(-x) = 2 - erfc(x)
*
* Method:
* 1. erf(x) = x + x*R(x^2) for |x| in [0, 7/8]
* Remark. The formula is derived by noting
* erf(x) = (2/sqrt(pi))*(x - x^3/3 + x^5/10 - x^7/42 + ....)
* and that
* 2/sqrt(pi) = 1.128379167095512573896158903121545171688
* is close to one.
*
* 1a. erf(x) = 1 - erfc(x), for |x| > 1.0
* erfc(x) = 1 - erf(x) if |x| < 1/4
*
* 2. For |x| in [7/8, 1], let s = |x| - 1, and
* c = 0.84506291151 rounded to single (24 bits)
* erf(s + c) = sign(x) * (c + P1(s)/Q1(s))
* Remark: here we use the taylor series expansion at x=1.
* erf(1+s) = erf(1) + s*Poly(s)
* = 0.845.. + P1(s)/Q1(s)
* Note that |P1/Q1|< 0.078 for x in [0.84375,1.25]
*
* 3. For x in [1/4, 5/4],
* erfc(s + const) = erfc(const) + s P1(s)/Q1(s)
* for const = 1/4, 3/8, ..., 9/8
* and 0 <= s <= 1/8 .
*
* 4. For x in [5/4, 107],
* erfc(x) = (1/x)*exp(-x*x-0.5625 + R(z))
* z=1/x^2
* The interval is partitioned into several segments
* of width 1/8 in 1/x.
*
* Note1:
* To compute exp(-x*x-0.5625+R/S), let s be a single
* precision number and s := x; then
* -x*x = -s*s + (s-x)*(s+x)
* exp(-x*x-0.5626+R/S) =
* exp(-s*s-0.5625)*exp((s-x)*(s+x)+R/S);
* Note2:
* Here 4 and 5 make use of the asymptotic series
* exp(-x*x)
* erfc(x) ~ ---------- * ( 1 + Poly(1/x^2) )
* x*sqrt(pi)
*
* 5. For inf > x >= 107
* erf(x) = sign(x) *(1 - tiny) (raise inexact)
* erfc(x) = tiny*tiny (raise underflow) if x > 0
* = 2 - tiny if x<0
*
* 7. Special case:
* erf(0) = 0, erf(inf) = 1, erf(-inf) = -1,
* erfc(0) = 1, erfc(inf) = 0, erfc(-inf) = 2,
* erfc/erf(NaN) is NaN
*/
#include <errno.h>
Fix erf underflow handling near 0 (bug 16516). Bug 16516 reports spurious underflows from erf (for all floating-point types), when the result is close to underflowing but does not actually underflow. erf (x) is about (2/sqrt(pi))*x for x close to 0, so there are subnormal arguments for which it does not underflow. The various implementations do (x + efx*x) (for efx = 2/sqrt(pi) - 1), for greater accuracy than if just using a single multiplication by an approximation to 2/sqrt(pi) (effectively, this way there are a few more bits in the approximation to 2/sqrt(pi)). This can introduce underflows when efx*x underflows even though the final result does not, so a scaled calculation with 8*efx is done in these cases - but 8 is not a big enough scale factor to avoid all such underflows. 16 is (any underflows with a scale factor of 16 would only occur when the final result underflows), so this patch changes the code to use that factor. Rather than recomputing all the values of the efx8 variable, it is removed, leaving it to the compiler's constant folding to compute 16*efx. As such scaling can also lose underflows when the final scaling down happens to be exact, appropriate checks are added to ensure underflow exceptions occur when required in such cases. Tested x86_64 and x86; no ulps updates needed. Also spot-checked for powerpc32 and mips64 to verify the changes to the ldbl-128ibm and ldbl-128 implementations. [BZ #16516] * sysdeps/ieee754/dbl-64/s_erf.c (efx8): Remove variable. (__erf): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/flt-32/s_erff.c (efx8): Remove variable. (__erff): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128ibm/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-96/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * math/auto-libm-test-in: Add more tests of erf. * math/auto-libm-test-out: Regenerated.
2014-05-14 12:34:03 +00:00
#include <float.h>
#include <math.h>
#include <math_private.h>
/* Evaluate P[n] x^n + P[n-1] x^(n-1) + ... + P[0] */
static _Float128
neval (_Float128 x, const _Float128 *p, int n)
{
_Float128 y;
p += n;
y = *p--;
do
{
y = y * x + *p--;
}
while (--n > 0);
return y;
}
/* Evaluate x^n+1 + P[n] x^(n) + P[n-1] x^(n-1) + ... + P[0] */
static _Float128
deval (_Float128 x, const _Float128 *p, int n)
{
_Float128 y;
p += n;
y = x + *p--;
do
{
y = y * x + *p--;
}
while (--n > 0);
return y;
}
static const _Float128
tiny = L(1e-4931),
one = 1,
two = 2,
/* 2/sqrt(pi) - 1 */
efx = L(1.2837916709551257389615890312154517168810E-1);
/* erf(x) = x + x R(x^2)
0 <= x <= 7/8
Peak relative error 1.8e-35 */
#define NTN1 8
static const _Float128 TN1[NTN1 + 1] =
{
L(-3.858252324254637124543172907442106422373E10),
L(9.580319248590464682316366876952214879858E10),
L(1.302170519734879977595901236693040544854E10),
L(2.922956950426397417800321486727032845006E9),
L(1.764317520783319397868923218385468729799E8),
L(1.573436014601118630105796794840834145120E7),
L(4.028077380105721388745632295157816229289E5),
L(1.644056806467289066852135096352853491530E4),
L(3.390868480059991640235675479463287886081E1)
};
#define NTD1 8
static const _Float128 TD1[NTD1 + 1] =
{
L(-3.005357030696532927149885530689529032152E11),
L(-1.342602283126282827411658673839982164042E11),
L(-2.777153893355340961288511024443668743399E10),
L(-3.483826391033531996955620074072768276974E9),
L(-2.906321047071299585682722511260895227921E8),
L(-1.653347985722154162439387878512427542691E7),
L(-6.245520581562848778466500301865173123136E5),
L(-1.402124304177498828590239373389110545142E4),
L(-1.209368072473510674493129989468348633579E2)
/* 1.0E0 */
};
/* erf(z+1) = erf_const + P(z)/Q(z)
-.125 <= z <= 0
Peak relative error 7.3e-36 */
static const _Float128 erf_const = L(0.845062911510467529296875);
#define NTN2 8
static const _Float128 TN2[NTN2 + 1] =
{
L(-4.088889697077485301010486931817357000235E1),
L(7.157046430681808553842307502826960051036E3),
L(-2.191561912574409865550015485451373731780E3),
L(2.180174916555316874988981177654057337219E3),
L(2.848578658049670668231333682379720943455E2),
L(1.630362490952512836762810462174798925274E2),
L(6.317712353961866974143739396865293596895E0),
L(2.450441034183492434655586496522857578066E1),
L(5.127662277706787664956025545897050896203E-1)
};
#define NTD2 8
static const _Float128 TD2[NTD2 + 1] =
{
L(1.731026445926834008273768924015161048885E4),
L(1.209682239007990370796112604286048173750E4),
L(1.160950290217993641320602282462976163857E4),
L(5.394294645127126577825507169061355698157E3),
L(2.791239340533632669442158497532521776093E3),
L(8.989365571337319032943005387378993827684E2),
L(2.974016493766349409725385710897298069677E2),
L(6.148192754590376378740261072533527271947E1),
L(1.178502892490738445655468927408440847480E1)
/* 1.0E0 */
};
/* erfc(x + 0.25) = erfc(0.25) + x R(x)
0 <= x < 0.125
Peak relative error 1.4e-35 */
#define NRNr13 8
static const _Float128 RNr13[NRNr13 + 1] =
{
L(-2.353707097641280550282633036456457014829E3),
L(3.871159656228743599994116143079870279866E2),
L(-3.888105134258266192210485617504098426679E2),
L(-2.129998539120061668038806696199343094971E1),
L(-8.125462263594034672468446317145384108734E1),
L(8.151549093983505810118308635926270319660E0),
L(-5.033362032729207310462422357772568553670E0),
L(-4.253956621135136090295893547735851168471E-2),
L(-8.098602878463854789780108161581050357814E-2)
};
#define NRDr13 7
static const _Float128 RDr13[NRDr13 + 1] =
{
L(2.220448796306693503549505450626652881752E3),
L(1.899133258779578688791041599040951431383E2),
L(1.061906712284961110196427571557149268454E3),
L(7.497086072306967965180978101974566760042E1),
L(2.146796115662672795876463568170441327274E2),
L(1.120156008362573736664338015952284925592E1),
L(2.211014952075052616409845051695042741074E1),
L(6.469655675326150785692908453094054988938E-1)
/* 1.0E0 */
};
/* erfc(0.25) = C13a + C13b to extra precision. */
static const _Float128 C13a = L(0.723663330078125);
static const _Float128 C13b = L(1.0279753638067014931732235184287934646022E-5);
/* erfc(x + 0.375) = erfc(0.375) + x R(x)
0 <= x < 0.125
Peak relative error 1.2e-35 */
#define NRNr14 8
static const _Float128 RNr14[NRNr14 + 1] =
{
L(-2.446164016404426277577283038988918202456E3),
L(6.718753324496563913392217011618096698140E2),
L(-4.581631138049836157425391886957389240794E2),
L(-2.382844088987092233033215402335026078208E1),
L(-7.119237852400600507927038680970936336458E1),
L(1.313609646108420136332418282286454287146E1),
L(-6.188608702082264389155862490056401365834E0),
L(-2.787116601106678287277373011101132659279E-2),
L(-2.230395570574153963203348263549700967918E-2)
};
#define NRDr14 7
static const _Float128 RDr14[NRDr14 + 1] =
{
L(2.495187439241869732696223349840963702875E3),
L(2.503549449872925580011284635695738412162E2),
L(1.159033560988895481698051531263861842461E3),
L(9.493751466542304491261487998684383688622E1),
L(2.276214929562354328261422263078480321204E2),
L(1.367697521219069280358984081407807931847E1),
L(2.276988395995528495055594829206582732682E1),
L(7.647745753648996559837591812375456641163E-1)
/* 1.0E0 */
};
/* erfc(0.375) = C14a + C14b to extra precision. */
static const _Float128 C14a = L(0.5958709716796875);
static const _Float128 C14b = L(1.2118885490201676174914080878232469565953E-5);
/* erfc(x + 0.5) = erfc(0.5) + x R(x)
0 <= x < 0.125
Peak relative error 4.7e-36 */
#define NRNr15 8
static const _Float128 RNr15[NRNr15 + 1] =
{
L(-2.624212418011181487924855581955853461925E3),
L(8.473828904647825181073831556439301342756E2),
L(-5.286207458628380765099405359607331669027E2),
L(-3.895781234155315729088407259045269652318E1),
L(-6.200857908065163618041240848728398496256E1),
L(1.469324610346924001393137895116129204737E1),
L(-6.961356525370658572800674953305625578903E0),
L(5.145724386641163809595512876629030548495E-3),
L(1.990253655948179713415957791776180406812E-2)
};
#define NRDr15 7
static const _Float128 RDr15[NRDr15 + 1] =
{
L(2.986190760847974943034021764693341524962E3),
L(5.288262758961073066335410218650047725985E2),
L(1.363649178071006978355113026427856008978E3),
L(1.921707975649915894241864988942255320833E2),
L(2.588651100651029023069013885900085533226E2),
L(2.628752920321455606558942309396855629459E1),
L(2.455649035885114308978333741080991380610E1),
L(1.378826653595128464383127836412100939126E0)
/* 1.0E0 */
};
/* erfc(0.5) = C15a + C15b to extra precision. */
static const _Float128 C15a = L(0.4794921875);
static const _Float128 C15b = L(7.9346869534623172533461080354712635484242E-6);
/* erfc(x + 0.625) = erfc(0.625) + x R(x)
0 <= x < 0.125
Peak relative error 5.1e-36 */
#define NRNr16 8
static const _Float128 RNr16[NRNr16 + 1] =
{
L(-2.347887943200680563784690094002722906820E3),
L(8.008590660692105004780722726421020136482E2),
L(-5.257363310384119728760181252132311447963E2),
L(-4.471737717857801230450290232600243795637E1),
L(-4.849540386452573306708795324759300320304E1),
L(1.140885264677134679275986782978655952843E1),
L(-6.731591085460269447926746876983786152300E0),
L(1.370831653033047440345050025876085121231E-1),
L(2.022958279982138755020825717073966576670E-2),
};
#define NRDr16 7
static const _Float128 RDr16[NRDr16 + 1] =
{
L(3.075166170024837215399323264868308087281E3),
L(8.730468942160798031608053127270430036627E2),
L(1.458472799166340479742581949088453244767E3),
L(3.230423687568019709453130785873540386217E2),
L(2.804009872719893612081109617983169474655E2),
L(4.465334221323222943418085830026979293091E1),
L(2.612723259683205928103787842214809134746E1),
L(2.341526751185244109722204018543276124997E0),
/* 1.0E0 */
};
/* erfc(0.625) = C16a + C16b to extra precision. */
static const _Float128 C16a = L(0.3767547607421875);
static const _Float128 C16b = L(4.3570693945275513594941232097252997287766E-6);
/* erfc(x + 0.75) = erfc(0.75) + x R(x)
0 <= x < 0.125
Peak relative error 1.7e-35 */
#define NRNr17 8
static const _Float128 RNr17[NRNr17 + 1] =
{
L(-1.767068734220277728233364375724380366826E3),
L(6.693746645665242832426891888805363898707E2),
L(-4.746224241837275958126060307406616817753E2),
L(-2.274160637728782675145666064841883803196E1),
L(-3.541232266140939050094370552538987982637E1),
L(6.988950514747052676394491563585179503865E0),
L(-5.807687216836540830881352383529281215100E0),
L(3.631915988567346438830283503729569443642E-1),
L(-1.488945487149634820537348176770282391202E-2)
};
#define NRDr17 7
static const _Float128 RDr17[NRDr17 + 1] =
{
L(2.748457523498150741964464942246913394647E3),
L(1.020213390713477686776037331757871252652E3),
L(1.388857635935432621972601695296561952738E3),
L(3.903363681143817750895999579637315491087E2),
L(2.784568344378139499217928969529219886578E2),
L(5.555800830216764702779238020065345401144E1),
L(2.646215470959050279430447295801291168941E1),
L(2.984905282103517497081766758550112011265E0),
/* 1.0E0 */
};
/* erfc(0.75) = C17a + C17b to extra precision. */
static const _Float128 C17a = L(0.2888336181640625);
static const _Float128 C17b = L(1.0748182422368401062165408589222625794046E-5);
/* erfc(x + 0.875) = erfc(0.875) + x R(x)
0 <= x < 0.125
Peak relative error 2.2e-35 */
#define NRNr18 8
static const _Float128 RNr18[NRNr18 + 1] =
{
L(-1.342044899087593397419622771847219619588E3),
L(6.127221294229172997509252330961641850598E2),
L(-4.519821356522291185621206350470820610727E2),
L(1.223275177825128732497510264197915160235E1),
L(-2.730789571382971355625020710543532867692E1),
L(4.045181204921538886880171727755445395862E0),
L(-4.925146477876592723401384464691452700539E0),
L(5.933878036611279244654299924101068088582E-1),
L(-5.557645435858916025452563379795159124753E-2)
};
#define NRDr18 7
static const _Float128 RDr18[NRDr18 + 1] =
{
L(2.557518000661700588758505116291983092951E3),
L(1.070171433382888994954602511991940418588E3),
L(1.344842834423493081054489613250688918709E3),
L(4.161144478449381901208660598266288188426E2),
L(2.763670252219855198052378138756906980422E2),
L(5.998153487868943708236273854747564557632E1),
L(2.657695108438628847733050476209037025318E1),
L(3.252140524394421868923289114410336976512E0),
/* 1.0E0 */
};
/* erfc(0.875) = C18a + C18b to extra precision. */
static const _Float128 C18a = L(0.215911865234375);
static const _Float128 C18b = L(1.3073705765341685464282101150637224028267E-5);
/* erfc(x + 1.0) = erfc(1.0) + x R(x)
0 <= x < 0.125
Peak relative error 1.6e-35 */
#define NRNr19 8
static const _Float128 RNr19[NRNr19 + 1] =
{
L(-1.139180936454157193495882956565663294826E3),
L(6.134903129086899737514712477207945973616E2),
L(-4.628909024715329562325555164720732868263E2),
L(4.165702387210732352564932347500364010833E1),
L(-2.286979913515229747204101330405771801610E1),
L(1.870695256449872743066783202326943667722E0),
L(-4.177486601273105752879868187237000032364E0),
L(7.533980372789646140112424811291782526263E-1),
L(-8.629945436917752003058064731308767664446E-2)
};
#define NRDr19 7
static const _Float128 RDr19[NRDr19 + 1] =
{
L(2.744303447981132701432716278363418643778E3),
L(1.266396359526187065222528050591302171471E3),
L(1.466739461422073351497972255511919814273E3),
L(4.868710570759693955597496520298058147162E2),
L(2.993694301559756046478189634131722579643E2),
L(6.868976819510254139741559102693828237440E1),
L(2.801505816247677193480190483913753613630E1),
L(3.604439909194350263552750347742663954481E0),
/* 1.0E0 */
};
/* erfc(1.0) = C19a + C19b to extra precision. */
static const _Float128 C19a = L(0.15728759765625);
static const _Float128 C19b = L(1.1609394035130658779364917390740703933002E-5);
/* erfc(x + 1.125) = erfc(1.125) + x R(x)
0 <= x < 0.125
Peak relative error 3.6e-36 */
#define NRNr20 8
static const _Float128 RNr20[NRNr20 + 1] =
{
L(-9.652706916457973956366721379612508047640E2),
L(5.577066396050932776683469951773643880634E2),
L(-4.406335508848496713572223098693575485978E2),
L(5.202893466490242733570232680736966655434E1),
L(-1.931311847665757913322495948705563937159E1),
L(-9.364318268748287664267341457164918090611E-2),
L(-3.306390351286352764891355375882586201069E0),
L(7.573806045289044647727613003096916516475E-1),
L(-9.611744011489092894027478899545635991213E-2)
};
#define NRDr20 7
static const _Float128 RDr20[NRDr20 + 1] =
{
L(3.032829629520142564106649167182428189014E3),
L(1.659648470721967719961167083684972196891E3),
L(1.703545128657284619402511356932569292535E3),
L(6.393465677731598872500200253155257708763E2),
L(3.489131397281030947405287112726059221934E2),
L(8.848641738570783406484348434387611713070E1),
L(3.132269062552392974833215844236160958502E1),
L(4.430131663290563523933419966185230513168E0)
/* 1.0E0 */
};
/* erfc(1.125) = C20a + C20b to extra precision. */
static const _Float128 C20a = L(0.111602783203125);
static const _Float128 C20b = L(8.9850951672359304215530728365232161564636E-6);
/* erfc(1/x) = 1/x exp (-1/x^2 - 0.5625 + R(1/x^2))
7/8 <= 1/x < 1
Peak relative error 1.4e-35 */
#define NRNr8 9
static const _Float128 RNr8[NRNr8 + 1] =
{
L(3.587451489255356250759834295199296936784E1),
L(5.406249749087340431871378009874875889602E2),
L(2.931301290625250886238822286506381194157E3),
L(7.359254185241795584113047248898753470923E3),
L(9.201031849810636104112101947312492532314E3),
L(5.749697096193191467751650366613289284777E3),
L(1.710415234419860825710780802678697889231E3),
L(2.150753982543378580859546706243022719599E2),
L(8.740953582272147335100537849981160931197E0),
L(4.876422978828717219629814794707963640913E-2)
};
#define NRDr8 8
static const _Float128 RDr8[NRDr8 + 1] =
{
L(6.358593134096908350929496535931630140282E1),
L(9.900253816552450073757174323424051765523E2),
L(5.642928777856801020545245437089490805186E3),
L(1.524195375199570868195152698617273739609E4),
L(2.113829644500006749947332935305800887345E4),
L(1.526438562626465706267943737310282977138E4),
L(5.561370922149241457131421914140039411782E3),
L(9.394035530179705051609070428036834496942E2),
L(6.147019596150394577984175188032707343615E1)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp (-1/x^2 - 0.5625 + R(1/x^2))
0.75 <= 1/x <= 0.875
Peak relative error 2.0e-36 */
#define NRNr7 9
static const _Float128 RNr7[NRNr7 + 1] =
{
L(1.686222193385987690785945787708644476545E1),
L(1.178224543567604215602418571310612066594E3),
L(1.764550584290149466653899886088166091093E4),
L(1.073758321890334822002849369898232811561E5),
L(3.132840749205943137619839114451290324371E5),
L(4.607864939974100224615527007793867585915E5),
L(3.389781820105852303125270837910972384510E5),
L(1.174042187110565202875011358512564753399E5),
L(1.660013606011167144046604892622504338313E4),
L(6.700393957480661937695573729183733234400E2)
};
#define NRDr7 9
static const _Float128 RDr7[NRDr7 + 1] =
{
L(-1.709305024718358874701575813642933561169E3),
L(-3.280033887481333199580464617020514788369E4),
L(-2.345284228022521885093072363418750835214E5),
L(-8.086758123097763971926711729242327554917E5),
L(-1.456900414510108718402423999575992450138E6),
L(-1.391654264881255068392389037292702041855E6),
L(-6.842360801869939983674527468509852583855E5),
L(-1.597430214446573566179675395199807533371E5),
L(-1.488876130609876681421645314851760773480E4),
L(-3.511762950935060301403599443436465645703E2)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
5/8 <= 1/x < 3/4
Peak relative error 1.9e-35 */
#define NRNr6 9
static const _Float128 RNr6[NRNr6 + 1] =
{
L(1.642076876176834390623842732352935761108E0),
L(1.207150003611117689000664385596211076662E2),
L(2.119260779316389904742873816462800103939E3),
L(1.562942227734663441801452930916044224174E4),
L(5.656779189549710079988084081145693580479E4),
L(1.052166241021481691922831746350942786299E5),
L(9.949798524786000595621602790068349165758E4),
L(4.491790734080265043407035220188849562856E4),
L(8.377074098301530326270432059434791287601E3),
L(4.506934806567986810091824791963991057083E2)
};
#define NRDr6 9
static const _Float128 RDr6[NRDr6 + 1] =
{
L(-1.664557643928263091879301304019826629067E2),
L(-3.800035902507656624590531122291160668452E3),
L(-3.277028191591734928360050685359277076056E4),
L(-1.381359471502885446400589109566587443987E5),
L(-3.082204287382581873532528989283748656546E5),
L(-3.691071488256738343008271448234631037095E5),
L(-2.300482443038349815750714219117566715043E5),
L(-6.873955300927636236692803579555752171530E4),
L(-8.262158817978334142081581542749986845399E3),
L(-2.517122254384430859629423488157361983661E2)
/* 1.00 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
1/2 <= 1/x < 5/8
Peak relative error 4.6e-36 */
#define NRNr5 10
static const _Float128 RNr5[NRNr5 + 1] =
{
L(-3.332258927455285458355550878136506961608E-3),
L(-2.697100758900280402659586595884478660721E-1),
L(-6.083328551139621521416618424949137195536E0),
L(-6.119863528983308012970821226810162441263E1),
L(-3.176535282475593173248810678636522589861E2),
L(-8.933395175080560925809992467187963260693E2),
L(-1.360019508488475978060917477620199499560E3),
L(-1.075075579828188621541398761300910213280E3),
L(-4.017346561586014822824459436695197089916E2),
L(-5.857581368145266249509589726077645791341E1),
L(-2.077715925587834606379119585995758954399E0)
};
#define NRDr5 9
static const _Float128 RDr5[NRDr5 + 1] =
{
L(3.377879570417399341550710467744693125385E-1),
L(1.021963322742390735430008860602594456187E1),
L(1.200847646592942095192766255154827011939E2),
L(7.118915528142927104078182863387116942836E2),
L(2.318159380062066469386544552429625026238E3),
L(4.238729853534009221025582008928765281620E3),
L(4.279114907284825886266493994833515580782E3),
L(2.257277186663261531053293222591851737504E3),
L(5.570475501285054293371908382916063822957E2),
L(5.142189243856288981145786492585432443560E1)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
3/8 <= 1/x < 1/2
Peak relative error 2.0e-36 */
#define NRNr4 10
static const _Float128 RNr4[NRNr4 + 1] =
{
L(3.258530712024527835089319075288494524465E-3),
L(2.987056016877277929720231688689431056567E-1),
L(8.738729089340199750734409156830371528862E0),
L(1.207211160148647782396337792426311125923E2),
L(8.997558632489032902250523945248208224445E2),
L(3.798025197699757225978410230530640879762E3),
L(9.113203668683080975637043118209210146846E3),
L(1.203285891339933238608683715194034900149E4),
L(8.100647057919140328536743641735339740855E3),
L(2.383888249907144945837976899822927411769E3),
L(2.127493573166454249221983582495245662319E2)
};
#define NRDr4 10
static const _Float128 RDr4[NRDr4 + 1] =
{
L(-3.303141981514540274165450687270180479586E-1),
L(-1.353768629363605300707949368917687066724E1),
L(-2.206127630303621521950193783894598987033E2),
L(-1.861800338758066696514480386180875607204E3),
L(-8.889048775872605708249140016201753255599E3),
L(-2.465888106627948210478692168261494857089E4),
L(-3.934642211710774494879042116768390014289E4),
L(-3.455077258242252974937480623730228841003E4),
L(-1.524083977439690284820586063729912653196E4),
L(-2.810541887397984804237552337349093953857E3),
L(-1.343929553541159933824901621702567066156E2)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
1/4 <= 1/x < 3/8
Peak relative error 8.4e-37 */
#define NRNr3 11
static const _Float128 RNr3[NRNr3 + 1] =
{
L(-1.952401126551202208698629992497306292987E-6),
L(-2.130881743066372952515162564941682716125E-4),
L(-8.376493958090190943737529486107282224387E-3),
L(-1.650592646560987700661598877522831234791E-1),
L(-1.839290818933317338111364667708678163199E0),
L(-1.216278715570882422410442318517814388470E1),
L(-4.818759344462360427612133632533779091386E1),
L(-1.120994661297476876804405329172164436784E2),
L(-1.452850765662319264191141091859300126931E2),
L(-9.485207851128957108648038238656777241333E1),
L(-2.563663855025796641216191848818620020073E1),
L(-1.787995944187565676837847610706317833247E0)
};
#define NRDr3 10
static const _Float128 RDr3[NRDr3 + 1] =
{
L(1.979130686770349481460559711878399476903E-4),
L(1.156941716128488266238105813374635099057E-2),
L(2.752657634309886336431266395637285974292E-1),
L(3.482245457248318787349778336603569327521E0),
L(2.569347069372696358578399521203959253162E1),
L(1.142279000180457419740314694631879921561E2),
L(3.056503977190564294341422623108332700840E2),
L(4.780844020923794821656358157128719184422E2),
L(4.105972727212554277496256802312730410518E2),
L(1.724072188063746970865027817017067646246E2),
L(2.815939183464818198705278118326590370435E1)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
1/8 <= 1/x < 1/4
Peak relative error 1.5e-36 */
#define NRNr2 11
static const _Float128 RNr2[NRNr2 + 1] =
{
L(-2.638914383420287212401687401284326363787E-8),
L(-3.479198370260633977258201271399116766619E-6),
L(-1.783985295335697686382487087502222519983E-4),
L(-4.777876933122576014266349277217559356276E-3),
L(-7.450634738987325004070761301045014986520E-2),
L(-7.068318854874733315971973707247467326619E-1),
L(-4.113919921935944795764071670806867038732E0),
L(-1.440447573226906222417767283691888875082E1),
L(-2.883484031530718428417168042141288943905E1),
L(-2.990886974328476387277797361464279931446E1),
L(-1.325283914915104866248279787536128997331E1),
L(-1.572436106228070195510230310658206154374E0)
};
#define NRDr2 10
static const _Float128 RDr2[NRDr2 + 1] =
{
L(2.675042728136731923554119302571867799673E-6),
L(2.170997868451812708585443282998329996268E-4),
L(7.249969752687540289422684951196241427445E-3),
L(1.302040375859768674620410563307838448508E-1),
L(1.380202483082910888897654537144485285549E0),
L(8.926594113174165352623847870299170069350E0),
L(3.521089584782616472372909095331572607185E1),
L(8.233547427533181375185259050330809105570E1),
L(1.072971579885803033079469639073292840135E2),
L(6.943803113337964469736022094105143158033E1),
L(1.775695341031607738233608307835017282662E1)
/* 1.0E0 */
};
/* erfc(1/x) = 1/x exp(-1/x^2 - 0.5625 + R(1/x^2))
1/128 <= 1/x < 1/8
Peak relative error 2.2e-36 */
#define NRNr1 9
static const _Float128 RNr1[NRNr1 + 1] =
{
L(-4.250780883202361946697751475473042685782E-8),
L(-5.375777053288612282487696975623206383019E-6),
L(-2.573645949220896816208565944117382460452E-4),
L(-6.199032928113542080263152610799113086319E-3),
L(-8.262721198693404060380104048479916247786E-2),
L(-6.242615227257324746371284637695778043982E-1),
L(-2.609874739199595400225113299437099626386E0),
L(-5.581967563336676737146358534602770006970E0),
L(-5.124398923356022609707490956634280573882E0),
L(-1.290865243944292370661544030414667556649E0)
};
#define NRDr1 8
static const _Float128 RDr1[NRDr1 + 1] =
{
L(4.308976661749509034845251315983612976224E-6),
L(3.265390126432780184125233455960049294580E-4),
L(9.811328839187040701901866531796570418691E-3),
L(1.511222515036021033410078631914783519649E-1),
L(1.289264341917429958858379585970225092274E0),
L(6.147640356182230769548007536914983522270E0),
L(1.573966871337739784518246317003956180750E1),
L(1.955534123435095067199574045529218238263E1),
L(9.472613121363135472247929109615785855865E0)
/* 1.0E0 */
};
_Float128
__erfl (_Float128 x)
{
_Float128 a, y, z;
int32_t i, ix, sign;
ieee854_long_double_shape_type u;
u.value = x;
sign = u.parts32.w0;
ix = sign & 0x7fffffff;
if (ix >= 0x7fff0000)
{ /* erf(nan)=nan */
i = ((sign & 0xffff0000) >> 31) << 1;
return (_Float128) (1 - i) + one / x; /* erf(+-inf)=+-1 */
}
if (ix >= 0x3fff0000) /* |x| >= 1.0 */
{
if (ix >= 0x40030000 && sign > 0)
return one; /* x >= 16, avoid spurious underflow from erfc. */
y = __erfcl (x);
return (one - y);
/* return (one - __erfcl (x)); */
}
u.parts32.w0 = ix;
a = u.value;
z = x * x;
if (ix < 0x3ffec000) /* a < 0.875 */
{
if (ix < 0x3fc60000) /* |x|<2**-57 */
{
if (ix < 0x00080000)
Fix erf underflow handling near 0 (bug 16516). Bug 16516 reports spurious underflows from erf (for all floating-point types), when the result is close to underflowing but does not actually underflow. erf (x) is about (2/sqrt(pi))*x for x close to 0, so there are subnormal arguments for which it does not underflow. The various implementations do (x + efx*x) (for efx = 2/sqrt(pi) - 1), for greater accuracy than if just using a single multiplication by an approximation to 2/sqrt(pi) (effectively, this way there are a few more bits in the approximation to 2/sqrt(pi)). This can introduce underflows when efx*x underflows even though the final result does not, so a scaled calculation with 8*efx is done in these cases - but 8 is not a big enough scale factor to avoid all such underflows. 16 is (any underflows with a scale factor of 16 would only occur when the final result underflows), so this patch changes the code to use that factor. Rather than recomputing all the values of the efx8 variable, it is removed, leaving it to the compiler's constant folding to compute 16*efx. As such scaling can also lose underflows when the final scaling down happens to be exact, appropriate checks are added to ensure underflow exceptions occur when required in such cases. Tested x86_64 and x86; no ulps updates needed. Also spot-checked for powerpc32 and mips64 to verify the changes to the ldbl-128ibm and ldbl-128 implementations. [BZ #16516] * sysdeps/ieee754/dbl-64/s_erf.c (efx8): Remove variable. (__erf): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/flt-32/s_erff.c (efx8): Remove variable. (__erff): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128ibm/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-96/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * math/auto-libm-test-in: Add more tests of erf. * math/auto-libm-test-out: Regenerated.
2014-05-14 12:34:03 +00:00
{
/* Avoid spurious underflow. */
_Float128 ret = 0.0625 * (16.0 * x + (16.0 * efx) * x);
Refactor code forcing underflow exceptions. Various floating-point functions have code to force underflow exceptions if a tiny result was computed in a way that might not have resulted in such exceptions even though the result is inexact. This typically uses math_force_eval to ensure that the underflowing expression is evaluated, but sometimes uses volatile. This patch refactors such code to use three new macros math_check_force_underflow, math_check_force_underflow_nonneg and math_check_force_underflow_complex (which in turn use math_force_eval). In the limited number of cases not suited to a simple conversion to these macros, existing uses of volatile are changed to use math_force_eval instead. The converted code does not always execute exactly the same sequence of operations as the original code, but the overall effects should be the same. Tested for x86_64, x86, mips64 and powerpc. * sysdeps/generic/math_private.h (fabs_tg): New macro. (min_of_type): Likewise. (math_check_force_underflow): Likewise. (math_check_force_underflow_nonneg): Likewise. (math_check_force_underflow_complex): Likewise. * math/e_exp2l.c (__ieee754_exp2l): Use math_check_force_underflow_nonneg. * math/k_casinh.c (__kernel_casinh): Likewise. * math/k_casinhf.c (__kernel_casinhf): Likewise. * math/k_casinhl.c (__kernel_casinhl): Likewise. * math/s_catan.c (__catan): Use math_check_force_underflow_complex. * math/s_catanf.c (__catanf): Likewise. * math/s_catanh.c (__catanh): Likewise. * math/s_catanhf.c (__catanhf): Likewise. * math/s_catanhl.c (__catanhl): Likewise. * math/s_catanl.c (__catanl): Likewise. * math/s_ccosh.c (__ccosh): Likewise. * math/s_ccoshf.c (__ccoshf): Likewise. * math/s_ccoshl.c (__ccoshl): Likewise. * math/s_cexp.c (__cexp): Likewise. * math/s_cexpf.c (__cexpf): Likewise. * math/s_cexpl.c (__cexpl): Likewise. * math/s_clog.c (__clog): Use math_check_force_underflow_nonneg. * math/s_clog10.c (__clog10): Likewise. * math/s_clog10f.c (__clog10f): Likewise. * math/s_clog10l.c (__clog10l): Likewise. * math/s_clogf.c (__clogf): Likewise. * math/s_clogl.c (__clogl): Likewise. * math/s_csin.c (__csin): Use math_check_force_underflow_complex. * math/s_csinf.c (__csinf): Likewise. * math/s_csinh.c (__csinh): Likewise. * math/s_csinhf.c (__csinhf): Likewise. * math/s_csinhl.c (__csinhl): Likewise. * math/s_csinl.c (__csinl): Likewise. * math/s_csqrt.c (__csqrt): Use math_check_force_underflow. * math/s_csqrtf.c (__csqrtf): Likewise. * math/s_csqrtl.c (__csqrtl): Likewise. * math/s_ctan.c (__ctan): Use math_check_force_underflow_complex. * math/s_ctanf.c (__ctanf): Likewise. * math/s_ctanh.c (__ctanh): Likewise. * math/s_ctanhf.c (__ctanhf): Likewise. * math/s_ctanhl.c (__ctanhl): Likewise. * math/s_ctanl.c (__ctanl): Likewise. * stdlib/strtod_l.c (round_and_return): Use math_force_eval instead of volatile. * sysdeps/ieee754/dbl-64/e_asin.c (__ieee754_asin): Use math_check_force_underflow. * sysdeps/ieee754/dbl-64/e_atanh.c (__ieee754_atanh): Likewise. * sysdeps/ieee754/dbl-64/e_exp.c (__ieee754_exp): Do not use volatile when forcing underflow. * sysdeps/ieee754/dbl-64/e_exp2.c (__ieee754_exp2): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/dbl-64/e_gamma_r.c (__ieee754_gamma_r): Likewise. * sysdeps/ieee754/dbl-64/e_j1.c (__ieee754_j1): Use math_check_force_underflow. * sysdeps/ieee754/dbl-64/e_jn.c (__ieee754_jn): Likewise. * sysdeps/ieee754/dbl-64/e_sinh.c (__ieee754_sinh): Likewise. * sysdeps/ieee754/dbl-64/s_asinh.c (__asinh): Likewise. * sysdeps/ieee754/dbl-64/s_atan.c (atan): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/dbl-64/s_erf.c (__erf): Use math_check_force_underflow. * sysdeps/ieee754/dbl-64/s_expm1.c (__expm1): Likewise. * sysdeps/ieee754/dbl-64/s_fma.c (__fma): Use math_force_eval instead of volatile. * sysdeps/ieee754/dbl-64/s_log1p.c (__log1p): Use math_check_force_underflow. * sysdeps/ieee754/dbl-64/s_sin.c (__sin): Likewise. * sysdeps/ieee754/dbl-64/s_tan.c (tan): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/dbl-64/s_tanh.c (__tanh): Use math_check_force_underflow. * sysdeps/ieee754/flt-32/e_asinf.c (__ieee754_asinf): Likewise. * sysdeps/ieee754/flt-32/e_atanhf.c (__ieee754_atanhf): Likewise. * sysdeps/ieee754/flt-32/e_exp2f.c (__ieee754_exp2f): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/flt-32/e_gammaf_r.c (__ieee754_gammaf_r): Likewise. * sysdeps/ieee754/flt-32/e_j1f.c (__ieee754_j1f): Use math_check_force_underflow. * sysdeps/ieee754/flt-32/e_jnf.c (__ieee754_jnf): Likewise. * sysdeps/ieee754/flt-32/e_sinhf.c (__ieee754_sinhf): Likewise. * sysdeps/ieee754/flt-32/k_sinf.c (__kernel_sinf): Likewise. * sysdeps/ieee754/flt-32/k_tanf.c (__kernel_tanf): Likewise. * sysdeps/ieee754/flt-32/s_asinhf.c (__asinhf): Likewise. * sysdeps/ieee754/flt-32/s_atanf.c (__atanf): Likewise. * sysdeps/ieee754/flt-32/s_erff.c (__erff): Likewise. * sysdeps/ieee754/flt-32/s_expm1f.c (__expm1f): Likewise. * sysdeps/ieee754/flt-32/s_log1pf.c (__log1pf): Likewise. * sysdeps/ieee754/flt-32/s_tanhf.c (__tanhf): Likewise. * sysdeps/ieee754/ldbl-128/e_asinl.c (__ieee754_asinl): Likewise. * sysdeps/ieee754/ldbl-128/e_atanhl.c (__ieee754_atanhl): Likewise. * sysdeps/ieee754/ldbl-128/e_expl.c (__ieee754_expl): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/ldbl-128/e_gammal_r.c (__ieee754_gammal_r): Likewise. * sysdeps/ieee754/ldbl-128/e_j1l.c (__ieee754_j1l): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-128/e_jnl.c (__ieee754_jnl): Likewise. * sysdeps/ieee754/ldbl-128/e_sinhl.c (__ieee754_sinhl): Likewise. * sysdeps/ieee754/ldbl-128/k_sincosl.c (__kernel_sincosl): Likewise. * sysdeps/ieee754/ldbl-128/k_sinl.c (__kernel_sinl): Likewise. * sysdeps/ieee754/ldbl-128/k_tanl.c (__kernel_tanl): Likewise. * sysdeps/ieee754/ldbl-128/s_asinhl.c (__asinhl): Likewise. * sysdeps/ieee754/ldbl-128/s_atanl.c (__atanl): Likewise. * sysdeps/ieee754/ldbl-128/s_erfl.c (__erfl): Likewise. * sysdeps/ieee754/ldbl-128/s_expm1l.c (__expm1l): Likewise. * sysdeps/ieee754/ldbl-128/s_fmal.c (__fmal): Use math_force_eval instead of volatile. * sysdeps/ieee754/ldbl-128/s_log1pl.c (__log1pl): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-128/s_tanhl.c (__tanhl): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_asinl.c (__ieee754_asinl): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-128ibm/e_atanhl.c (__ieee754_atanhl): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_gammal_r.c (__ieee754_gammal_r): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/ldbl-128ibm/e_jnl.c (__ieee754_jnl): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-128ibm/e_sinhl.c (__ieee754_sinhl): Likewise. * sysdeps/ieee754/ldbl-128ibm/k_sincosl.c (__kernel_sincosl): Likewise. * sysdeps/ieee754/ldbl-128ibm/k_sinl.c (__kernel_sinl): Likewise. * sysdeps/ieee754/ldbl-128ibm/k_tanl.c (__kernel_tanl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_asinhl.c (__asinhl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_atanl.c (__atanl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_erfl.c (__erfl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_tanhl.c (__tanhl): Likewise. * sysdeps/ieee754/ldbl-96/e_asinl.c (__ieee754_asinl): Likewise. * sysdeps/ieee754/ldbl-96/e_atanhl.c (__ieee754_atanhl): Likewise. * sysdeps/ieee754/ldbl-96/e_gammal_r.c (__ieee754_gammal_r): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/ldbl-96/e_j1l.c (__ieee754_j1l): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-96/e_jnl.c (__ieee754_jnl): Likewise. * sysdeps/ieee754/ldbl-96/e_sinhl.c (__ieee754_sinhl): Likewise. * sysdeps/ieee754/ldbl-96/k_sinl.c (__kernel_sinl): Likewise. * sysdeps/ieee754/ldbl-96/k_tanl.c (__kernel_tanl): Use math_check_force_underflow_nonneg. * sysdeps/ieee754/ldbl-96/s_asinhl.c (__asinhl): Use math_check_force_underflow. * sysdeps/ieee754/ldbl-96/s_erfl.c (__erfl): Likewise. * sysdeps/ieee754/ldbl-96/s_fmal.c (__fmal): Use math_force_eval instead of volatile. * sysdeps/ieee754/ldbl-96/s_tanhl.c (__tanhl): Use math_check_force_underflow.
2015-09-23 22:42:30 +00:00
math_check_force_underflow (ret);
Fix erf underflow handling near 0 (bug 16516). Bug 16516 reports spurious underflows from erf (for all floating-point types), when the result is close to underflowing but does not actually underflow. erf (x) is about (2/sqrt(pi))*x for x close to 0, so there are subnormal arguments for which it does not underflow. The various implementations do (x + efx*x) (for efx = 2/sqrt(pi) - 1), for greater accuracy than if just using a single multiplication by an approximation to 2/sqrt(pi) (effectively, this way there are a few more bits in the approximation to 2/sqrt(pi)). This can introduce underflows when efx*x underflows even though the final result does not, so a scaled calculation with 8*efx is done in these cases - but 8 is not a big enough scale factor to avoid all such underflows. 16 is (any underflows with a scale factor of 16 would only occur when the final result underflows), so this patch changes the code to use that factor. Rather than recomputing all the values of the efx8 variable, it is removed, leaving it to the compiler's constant folding to compute 16*efx. As such scaling can also lose underflows when the final scaling down happens to be exact, appropriate checks are added to ensure underflow exceptions occur when required in such cases. Tested x86_64 and x86; no ulps updates needed. Also spot-checked for powerpc32 and mips64 to verify the changes to the ldbl-128ibm and ldbl-128 implementations. [BZ #16516] * sysdeps/ieee754/dbl-64/s_erf.c (efx8): Remove variable. (__erf): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/flt-32/s_erff.c (efx8): Remove variable. (__erff): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-128ibm/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * sysdeps/ieee754/ldbl-96/s_erfl.c: Include <float.h>. (efx8): Remove variable. (__erfl): Scale by 16 instead of 8 in potentially underflowing case. Ensure exception if result actually underflows. * math/auto-libm-test-in: Add more tests of erf. * math/auto-libm-test-out: Regenerated.
2014-05-14 12:34:03 +00:00
return ret;
}
return x + efx * x;
}
y = a + a * neval (z, TN1, NTN1) / deval (z, TD1, NTD1);
}
else
{
a = a - one;
y = erf_const + neval (a, TN2, NTN2) / deval (a, TD2, NTD2);
}
if (sign & 0x80000000) /* x < 0 */
y = -y;
return( y );
}
weak_alias (__erfl, erfl)
_Float128
__erfcl (_Float128 x)
{
_Float128 y, z, p, r;
int32_t i, ix, sign;
ieee854_long_double_shape_type u;
u.value = x;
sign = u.parts32.w0;
ix = sign & 0x7fffffff;
u.parts32.w0 = ix;
if (ix >= 0x7fff0000)
{ /* erfc(nan)=nan */
/* erfc(+-inf)=0,2 */
Consistently use uintN_t not u_intN_t in libm. This patch changes libm code to make consistent use of C99 uintN_t types instead of sometimes using those and sometimes using the older nonstandard u_intN_t names. This makes sense as a cleanup in its own right, and also facilitates merges to GCC's libquadmath (which gets the types from stdint.h and so may not have u_intN_t available at all). Tested for x86_64, and with build-many-glibcs.py. * math/s_nextafter.c (__nextafter): Use uintN_t instead of u_intN_t. * math/s_nexttowardf.c (__nexttowardf): Likewise. * sysdeps/generic/math_private.h (ieee_double_shape_type): Likewise. (ieee_float_shape_type): Likewise. * sysdeps/i386/fpu/s_fpclassifyl.c (__fpclassifyl): Likewise. * sysdeps/i386/fpu/s_isnanl.c (__isnanl): Likewise. * sysdeps/i386/fpu/s_nextafterl.c (__nextafterl): Likewise. * sysdeps/i386/fpu/s_nexttoward.c (__nexttoward): Likewise. * sysdeps/i386/fpu/s_nexttowardf.c (__nexttowardf): Likewise. * sysdeps/ieee754/dbl-64/e_acosh.c (__ieee754_acosh): Likewise. * sysdeps/ieee754/dbl-64/e_cosh.c (__ieee754_cosh): Likewise. * sysdeps/ieee754/dbl-64/e_fmod.c (__ieee754_fmod): Likewise. * sysdeps/ieee754/dbl-64/e_gamma_r.c (__ieee754_gamma_r): Likewise. * sysdeps/ieee754/dbl-64/e_hypot.c (__ieee754_hypot): Likewise. * sysdeps/ieee754/dbl-64/e_jn.c (__ieee754_jn): Likewise. (__ieee754_yn): Likewise. * sysdeps/ieee754/dbl-64/e_log10.c (__ieee754_log10): Likewise. * sysdeps/ieee754/dbl-64/e_log2.c (__ieee754_log2): Likewise. * sysdeps/ieee754/dbl-64/e_rem_pio2.c (__ieee754_rem_pio2): Likewise. * sysdeps/ieee754/dbl-64/e_sinh.c (__ieee754_sinh): Likewise. * sysdeps/ieee754/dbl-64/s_ceil.c (__ceil): Likewise. * sysdeps/ieee754/dbl-64/s_copysign.c (__copysign): Likewise. * sysdeps/ieee754/dbl-64/s_erf.c (__erf): Likewise. (__erfc): Likewise. * sysdeps/ieee754/dbl-64/s_expm1.c (__expm1): Likewise. * sysdeps/ieee754/dbl-64/s_finite.c (FINITE): Likewise. * sysdeps/ieee754/dbl-64/s_floor.c (__floor): Likewise. * sysdeps/ieee754/dbl-64/s_fpclassify.c (__fpclassify): Likewise. * sysdeps/ieee754/dbl-64/s_isnan.c (__isnan): Likewise. * sysdeps/ieee754/dbl-64/s_issignaling.c (__issignaling): Likewise. * sysdeps/ieee754/dbl-64/s_llrint.c (__llrint): Likewise. * sysdeps/ieee754/dbl-64/s_llround.c (__llround): Likewise. * sysdeps/ieee754/dbl-64/s_lrint.c (__lrint): Likewise. * sysdeps/ieee754/dbl-64/s_lround.c (__lround): Likewise. * sysdeps/ieee754/dbl-64/s_modf.c (__modf): Likewise. * sysdeps/ieee754/dbl-64/s_nextup.c (__nextup): Likewise. * sysdeps/ieee754/dbl-64/s_remquo.c (__remquo): Likewise. * sysdeps/ieee754/dbl-64/s_round.c (__round): Likewise. * sysdeps/ieee754/dbl-64/s_trunc.c (__trunc): Likewise. * sysdeps/ieee754/dbl-64/wordsize-64/s_issignaling.c (__issignaling): Likewise. * sysdeps/ieee754/flt-32/e_atan2f.c (__ieee754_atan2f): Likewise. * sysdeps/ieee754/flt-32/e_fmodf.c (__ieee754_fmodf): Likewise. * sysdeps/ieee754/flt-32/e_gammaf_r.c (__ieee754_gammaf_r): Likewise. * sysdeps/ieee754/flt-32/e_jnf.c (__ieee754_ynf): Likewise. * sysdeps/ieee754/flt-32/e_log10f.c (__ieee754_log10f): Likewise. * sysdeps/ieee754/flt-32/e_powf.c (__ieee754_powf): Likewise. * sysdeps/ieee754/flt-32/e_rem_pio2f.c (__ieee754_rem_pio2f): Likewise. * sysdeps/ieee754/flt-32/e_remainderf.c (__ieee754_remainderf): Likewise. * sysdeps/ieee754/flt-32/e_sqrtf.c (__ieee754_sqrtf): Likewise. * sysdeps/ieee754/flt-32/s_ceilf.c (__ceilf): Likewise. * sysdeps/ieee754/flt-32/s_copysignf.c (__copysignf): Likewise. * sysdeps/ieee754/flt-32/s_erff.c (__erff): Likewise. (__erfcf): Likewise. * sysdeps/ieee754/flt-32/s_expm1f.c (__expm1f): Likewise. * sysdeps/ieee754/flt-32/s_finitef.c (FINITEF): Likewise. * sysdeps/ieee754/flt-32/s_floorf.c (__floorf): Likewise. * sysdeps/ieee754/flt-32/s_fpclassifyf.c (__fpclassifyf): Likewise. * sysdeps/ieee754/flt-32/s_isnanf.c (__isnanf): Likewise. * sysdeps/ieee754/flt-32/s_issignalingf.c (__issignalingf): Likewise. * sysdeps/ieee754/flt-32/s_llrintf.c (__llrintf): Likewise. * sysdeps/ieee754/flt-32/s_llroundf.c (__llroundf): Likewise. * sysdeps/ieee754/flt-32/s_lrintf.c (__lrintf): Likewise. * sysdeps/ieee754/flt-32/s_lroundf.c (__lroundf): Likewise. * sysdeps/ieee754/flt-32/s_modff.c (__modff): Likewise. * sysdeps/ieee754/flt-32/s_remquof.c (__remquof): Likewise. * sysdeps/ieee754/flt-32/s_roundf.c (__roundf): Likewise. * sysdeps/ieee754/ldbl-128/e_acoshl.c (__ieee754_acoshl): Likewise. * sysdeps/ieee754/ldbl-128/e_atan2l.c (__ieee754_atan2l): Likewise. * sysdeps/ieee754/ldbl-128/e_atanhl.c (__ieee754_atanhl): Likewise. * sysdeps/ieee754/ldbl-128/e_fmodl.c (__ieee754_fmodl): Likewise. * sysdeps/ieee754/ldbl-128/e_gammal_r.c (__ieee754_gammal_r): Likewise. * sysdeps/ieee754/ldbl-128/e_hypotl.c (__ieee754_hypotl): Likewise. * sysdeps/ieee754/ldbl-128/e_jnl.c (__ieee754_jnl): Likewise. (__ieee754_ynl): Likewise. * sysdeps/ieee754/ldbl-128/e_powl.c (__ieee754_powl): Likewise. * sysdeps/ieee754/ldbl-128/e_rem_pio2l.c (__ieee754_rem_pio2l): Likewise. * sysdeps/ieee754/ldbl-128/e_remainderl.c (__ieee754_remainderl): Likewise. * sysdeps/ieee754/ldbl-128/e_sinhl.c (__ieee754_sinhl): Likewise. * sysdeps/ieee754/ldbl-128/k_cosl.c (__kernel_cosl): Likewise. * sysdeps/ieee754/ldbl-128/k_sincosl.c (__kernel_sincosl): Likewise. * sysdeps/ieee754/ldbl-128/k_sinl.c (__kernel_sinl): Likewise. * sysdeps/ieee754/ldbl-128/s_ceill.c (__ceill): Likewise. * sysdeps/ieee754/ldbl-128/s_copysignl.c (__copysignl): Likewise. * sysdeps/ieee754/ldbl-128/s_erfl.c (__erfcl): Likewise. * sysdeps/ieee754/ldbl-128/s_fabsl.c (__fabsl): Likewise. * sysdeps/ieee754/ldbl-128/s_finitel.c (__finitel): Likewise. * sysdeps/ieee754/ldbl-128/s_floorl.c (__floorl): Likewise. * sysdeps/ieee754/ldbl-128/s_fpclassifyl.c (__fpclassifyl): Likewise. * sysdeps/ieee754/ldbl-128/s_frexpl.c (__frexpl): Likewise. * sysdeps/ieee754/ldbl-128/s_isnanl.c (__isnanl): Likewise. * sysdeps/ieee754/ldbl-128/s_issignalingl.c (__issignalingl): Likewise. * sysdeps/ieee754/ldbl-128/s_llrintl.c (__llrintl): Likewise. * sysdeps/ieee754/ldbl-128/s_llroundl.c (__llroundl): Likewise. * sysdeps/ieee754/ldbl-128/s_lrintl.c (__lrintl): Likewise. * sysdeps/ieee754/ldbl-128/s_lroundl.c (__lroundl): Likewise. * sysdeps/ieee754/ldbl-128/s_modfl.c (__modfl): Likewise. * sysdeps/ieee754/ldbl-128/s_nearbyintl.c (__nearbyintl): Likewise. * sysdeps/ieee754/ldbl-128/s_nextafterl.c (__nextafterl): Likewise. * sysdeps/ieee754/ldbl-128/s_nexttoward.c (__nexttoward): Likewise. * sysdeps/ieee754/ldbl-128/s_nexttowardf.c (__nexttowardf): Likewise. * sysdeps/ieee754/ldbl-128/s_nextupl.c (__nextupl): Likewise. * sysdeps/ieee754/ldbl-128/s_remquol.c (__remquol): Likewise. * sysdeps/ieee754/ldbl-128/s_rintl.c (__rintl): Likewise. * sysdeps/ieee754/ldbl-128/s_roundl.c (__roundl): Likewise. * sysdeps/ieee754/ldbl-128/s_tanhl.c (__tanhl): Likewise. * sysdeps/ieee754/ldbl-128/s_truncl.c (__truncl): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_fmodl.c (__ieee754_fmodl): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_gammal_r.c (__ieee754_gammal_r): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_powl.c (__ieee754_powl): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_rem_pio2l.c (__ieee754_rem_pio2l): Likewise. * sysdeps/ieee754/ldbl-128ibm/e_remainderl.c (__ieee754_remainderl): Likewise. * sysdeps/ieee754/ldbl-128ibm/k_cosl.c (__kernel_cosl): Likewise. * sysdeps/ieee754/ldbl-128ibm/k_sinl.c (__kernel_sinl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_fabsl.c (__fabsl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_fpclassifyl.c (___fpclassifyl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_modfl.c (__modfl): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_nexttowardf.c (__nexttowardf): Likewise. * sysdeps/ieee754/ldbl-128ibm/s_remquol.c (__remquol): Likewise. * sysdeps/ieee754/ldbl-96/e_acoshl.c (__ieee754_acoshl): Likewise. * sysdeps/ieee754/ldbl-96/e_asinl.c (__ieee754_asinl): Likewise. * sysdeps/ieee754/ldbl-96/e_atanhl.c (__ieee754_atanhl): Likewise. * sysdeps/ieee754/ldbl-96/e_coshl.c (__ieee754_coshl): Likewise. * sysdeps/ieee754/ldbl-96/e_gammal_r.c (__ieee754_gammal_r): Likewise. * sysdeps/ieee754/ldbl-96/e_hypotl.c (__ieee754_hypotl): Likewise. * sysdeps/ieee754/ldbl-96/e_j0l.c (__ieee754_j0l): Likewise. (__ieee754_y0l): Likewise. (pzero): Likewise. (qzero): Likewise. * sysdeps/ieee754/ldbl-96/e_j1l.c (__ieee754_j1l): Likewise. (__ieee754_y1l): Likewise. (pone): Likewise. (qone): Likewise. * sysdeps/ieee754/ldbl-96/e_jnl.c (__ieee754_jnl): Likewise. (__ieee754_ynl): Likewise. * sysdeps/ieee754/ldbl-96/e_lgammal_r.c (sin_pi): Likewise. (__ieee754_lgammal_r): Likewise. * sysdeps/ieee754/ldbl-96/e_rem_pio2l.c (__ieee754_rem_pio2l): Likewise. * sysdeps/ieee754/ldbl-96/e_sinhl.c (__ieee754_sinhl): Likewise. * sysdeps/ieee754/ldbl-96/s_copysignl.c (__copysignl): Likewise. * sysdeps/ieee754/ldbl-96/s_erfl.c (__erfl): Likewise. (__erfcl): Likewise. * sysdeps/ieee754/ldbl-96/s_frexpl.c (__frexpl): Likewise. * sysdeps/ieee754/ldbl-96/s_issignalingl.c (__issignalingl): Likewise. * sysdeps/ieee754/ldbl-96/s_llrintl.c (__llrintl): Likewise. * sysdeps/ieee754/ldbl-96/s_llroundl.c (__llroundl): Likewise. * sysdeps/ieee754/ldbl-96/s_lrintl.c (__lrintl): Likewise. * sysdeps/ieee754/ldbl-96/s_lroundl.c (__lroundl): Likewise. * sysdeps/ieee754/ldbl-96/s_modfl.c (__modfl): Likewise. * sysdeps/ieee754/ldbl-96/s_nexttoward.c (__nexttoward): Likewise. * sysdeps/ieee754/ldbl-96/s_nexttowardf.c (__nexttowardf): Likewise. * sysdeps/ieee754/ldbl-96/s_nextupl.c (__nextupl): Likewise. * sysdeps/ieee754/ldbl-96/s_remquol.c (__remquol): Likewise. * sysdeps/ieee754/ldbl-96/s_roundl.c (__roundl): Likewise. * sysdeps/ieee754/ldbl-96/s_tanhl.c (__tanhl): Likewise. * sysdeps/ieee754/ldbl-opt/s_nexttowardfd.c (__nldbl_nexttowardf): Likewise. * sysdeps/m68k/m680x0/fpu/e_pow.c (s(__ieee754_pow)): Likewise. * sysdeps/m68k/m680x0/fpu/s_fpclassifyl.c (__fpclassifyl): Likewise. * sysdeps/m68k/m680x0/fpu/s_llrint.c (__llrint): Likewise. * sysdeps/m68k/m680x0/fpu/s_llrintf.c (__llrintf): Likewise. * sysdeps/m68k/m680x0/fpu/s_llrintl.c (__llrintl): Likewise. * sysdeps/m68k/m680x0/fpu/s_nextafterl.c (__nextafterl): Likewise. * sysdeps/x86/fpu/powl_helper.c (__powl_helper): Likewise.
2017-08-03 19:55:04 +00:00
return (_Float128) (((uint32_t) sign >> 31) << 1) + one / x;
}
if (ix < 0x3ffd0000) /* |x| <1/4 */
{
if (ix < 0x3f8d0000) /* |x|<2**-114 */
return one - x;
return one - __erfl (x);
}
if (ix < 0x3fff4000) /* 1.25 */
{
x = u.value;
i = 8.0 * x;
switch (i)
{
case 2:
z = x - L(0.25);
y = C13b + z * neval (z, RNr13, NRNr13) / deval (z, RDr13, NRDr13);
y += C13a;
break;
case 3:
z = x - L(0.375);
y = C14b + z * neval (z, RNr14, NRNr14) / deval (z, RDr14, NRDr14);
y += C14a;
break;
case 4:
z = x - L(0.5);
y = C15b + z * neval (z, RNr15, NRNr15) / deval (z, RDr15, NRDr15);
y += C15a;
break;
case 5:
z = x - L(0.625);
y = C16b + z * neval (z, RNr16, NRNr16) / deval (z, RDr16, NRDr16);
y += C16a;
break;
case 6:
z = x - L(0.75);
y = C17b + z * neval (z, RNr17, NRNr17) / deval (z, RDr17, NRDr17);
y += C17a;
break;
case 7:
z = x - L(0.875);
y = C18b + z * neval (z, RNr18, NRNr18) / deval (z, RDr18, NRDr18);
y += C18a;
break;
case 8:
z = x - 1;
y = C19b + z * neval (z, RNr19, NRNr19) / deval (z, RDr19, NRDr19);
y += C19a;
break;
default: /* i == 9. */
z = x - L(1.125);
y = C20b + z * neval (z, RNr20, NRNr20) / deval (z, RDr20, NRDr20);
y += C20a;
break;
}
if (sign & 0x80000000)
y = 2 - y;
return y;
}
/* 1.25 < |x| < 107 */
if (ix < 0x4005ac00)
{
/* x < -9 */
if ((ix >= 0x40022000) && (sign & 0x80000000))
return two - tiny;
x = fabsl (x);
z = one / (x * x);
i = 8.0 / x;
switch (i)
{
default:
case 0:
p = neval (z, RNr1, NRNr1) / deval (z, RDr1, NRDr1);
break;
case 1:
p = neval (z, RNr2, NRNr2) / deval (z, RDr2, NRDr2);
break;
case 2:
p = neval (z, RNr3, NRNr3) / deval (z, RDr3, NRDr3);
break;
case 3:
p = neval (z, RNr4, NRNr4) / deval (z, RDr4, NRDr4);
break;
case 4:
p = neval (z, RNr5, NRNr5) / deval (z, RDr5, NRDr5);
break;
case 5:
p = neval (z, RNr6, NRNr6) / deval (z, RDr6, NRDr6);
break;
case 6:
p = neval (z, RNr7, NRNr7) / deval (z, RDr7, NRDr7);
break;
case 7:
p = neval (z, RNr8, NRNr8) / deval (z, RDr8, NRDr8);
break;
}
u.value = x;
u.parts32.w3 = 0;
u.parts32.w2 &= 0xfe000000;
z = u.value;
r = __ieee754_expl (-z * z - 0.5625) *
__ieee754_expl ((z - x) * (z + x) + p);
if ((sign & 0x80000000) == 0)
{
_Float128 ret = r / x;
if (ret == 0)
__set_errno (ERANGE);
return ret;
}
else
return two - r / x;
}
else
{
if ((sign & 0x80000000) == 0)
{
__set_errno (ERANGE);
return tiny * tiny;
}
else
return two - tiny;
}
}
weak_alias (__erfcl, erfcl)