2017-01-01 08:14:16 +08:00
|
|
|
/* Copyright (C) 2011-2017 Free Software Foundation, Inc.
|
2011-10-12 23:27:51 +08:00
|
|
|
This file is part of the GNU C Library.
|
|
|
|
Contributed by Ulrich Drepper <drepper@gmail.com>, 2011.
|
1996-03-06 05:41:30 +08:00
|
|
|
|
2011-10-12 23:27:51 +08:00
|
|
|
The GNU C 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.
|
1996-03-06 05:41:30 +08:00
|
|
|
|
2011-10-12 23:27:51 +08:00
|
|
|
The GNU C 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.
|
1996-03-06 05:41:30 +08:00
|
|
|
|
2011-10-12 23:27:51 +08:00
|
|
|
You should have received a copy of the GNU Lesser General Public
|
2012-02-10 07:18:22 +08:00
|
|
|
License along with the GNU C Library; if not, see
|
|
|
|
<http://www.gnu.org/licenses/>. */
|
1996-03-06 05:41:30 +08:00
|
|
|
|
2005-12-14 23:06:39 +08:00
|
|
|
#include <math.h>
|
2011-10-08 17:16:04 +08:00
|
|
|
#include <math_private.h>
|
Obsolete matherr, _LIB_VERSION, libieee.a.
This patch obsoletes support for SVID libm error handling (the system
where a user-defined function matherr is called on a libm function
error; only enabled if you also set _LIB_VERSION = _SVID_ or
_LIB_VERSION = _XOPEN_) and the use of the _LIB_VERSION global
variable to control libm error handling. matherr and _LIB_VERSION are
made into compat symbols, not supported for new ports or for static
linking. The libieee.a object file (which sets _LIB_VERSION = _IEEE_,
so disabling errno setting for some functions) is also removed, and
all the related definitions are removed from math.h.
The manual already recommends against using matherr, and it's already
not supported for _Float128 functions (those use new wrappers that
don't support matherr, only errno) - this patch means that it becomes
possible to e.g. add sinf32 as an alias to sinf without that resulting
in undesired matherr support in sinf32 for existing glibc ports.
matherr support is not part of any standard supported by glibc (it was
removed in XPG4).
Because matherr is a function to be defined by the user, of course
user programs defining such a function will still continue to link; it
just quietly won't be used. If they try to write to the library's
copy of _LIB_VERSION to enable SVID error handling, however, they will
get a link error (but if they define their own _LIB_VERSION variable,
they won't).
I expect the most likely case of build failures from this patch to be
programs with unconditional cargo-culted uses of -lieee (based on a
notion of "I want IEEE floating point", not any actual requirement for
that library).
Ideally, the new-port-or-static-linking case would use the new
wrappers used for _Float128. This is not implemented in this patch,
because of the complication of architecture-specific (powerpc32 and
sparc) sqrt wrappers that use _LIB_VERSION and __kernel_standard
directly. Thus, the old wrappers and __kernel_standard are still
built unconditionally, and _LIB_VERSION still exists in static libm.
But when the old wrappers and __kernel_standard are built in the
non-compat case, _LIB_VERSION and matherr are defined as macros so
code to support those features isn't actually built into static libm
or new ports' shared libm after this patch.
I intend to move to the new wrappers for static libm and new ports in
followup patches. I believe the sqrt wrappers for powerpc32 and sparc
can reasonably be removed. GCC already optimizes the normal case of
sqrt by generating code that uses a hardware instruction and only
calls the sqrt function if the argument was negative (if
-fno-math-errno, of course, it just uses the hardware instruction
without any check for negative argument being needed). Thus those
wrappers will only actually get called in the case of negative
arguments, which is not a case it makes sense to optimize for. But
even without removing the powerpc32 and sparc wrappers it should still
be possible to move to the new wrappers for static libm and new ports,
just without having those dubious architecture-specific optimizations
in static libm.
Everything said about matherr equally applies to matherrf and matherrl
(IA64-specific, undocumented), except that the structure of IA64 libm
means it won't be converted to using the new wrappers (it doesn't use
the old ones either, but its own error-handling code instead).
As with other tests of compat symbols, I expect test-matherr and
test-matherr-2 to need to become appropriately conditional once we
have a system for disabling such tests for ports too new to have the
relevant symbols.
Tested for x86_64 and x86, and with build-many-glibcs.py.
* math/math.h [__USE_MISC] (_LIB_VERSION_TYPE): Remove.
[__USE_MISC] (_LIB_VERSION): Likewise.
[__USE_MISC] (struct exception): Likewise.
[__USE_MISC] (matherr): Likewise.
[__USE_MISC] (DOMAIN): Likewise.
[__USE_MISC] (SING): Likewise.
[__USE_MISC] (OVERFLOW): Likewise.
[__USE_MISC] (UNDERFLOW): Likewise.
[__USE_MISC] (TLOSS): Likewise.
[__USE_MISC] (PLOSS): Likewise.
[__USE_MISC] (HUGE): Likewise.
[__USE_XOPEN] (MAXFLOAT): Define even if [__USE_MISC].
* math/math-svid-compat.h: New file.
* conform/linknamespace.pl (@whitelist): Remove matherr, matherrf
and matherrl.
* include/math.h [!_ISOMAC] (__matherr): Remove.
* manual/arith.texi (FP Exceptions): Do not document matherr.
* math/Makefile (tests): Change test-matherr to test-matherr-3.
(tests-internal): New variable.
(install-lib): Do not add libieee.a.
(non-lib.a): Likewise.
(extra-objs): Do not add libieee.a and ieee-math.o.
(CPPFLAGS-s_lib_version.c): Remove variable.
($(objpfx)libieee.a): Remove rule.
($(addprefix $(objpfx), $(tests-internal)): Depend on $(libm).
* math/ieee-math.c: Remove.
* math/libm-test-support.c (matherr): Remove.
* math/test-matherr.c: Use <support/test-driver.c>. Add copyright
and license notices. Include <math-svid-compat.h> and
<shlib-compat.h>.
(matherr): Undefine as macro. Use compat_symbol_reference.
(_LIB_VERSION): Likewise.
* math/test-matherr-2.c: New file.
* math/test-matherr-3.c: Likewise.
* sysdeps/generic/math_private.h (__kernel_standard): Remove
declaration.
(__kernel_standard_f): Likewise.
(__kernel_standard_l): Likewise.
* sysdeps/ieee754/s_lib_version.c: Do not include <math.h> or
<math_private.h>. Include <math-svid-compat.h>.
(_LIB_VERSION): Undefine as macro.
(_LIB_VERSION_INTERNAL): Always initialize to _POSIX_. Define
only if [LIBM_SVID_COMPAT || !defined SHARED]. If
[LIBM_SVID_COMPAT], use compat_symbol.
* sysdeps/ieee754/s_matherr.c: Do not include <math.h> or
<math_private.h>. Include <math-svid-compat.h>.
(matherr): Undefine as macro.
(__matherr): Define only if [LIBM_SVID_COMPAT]. Use
compat_symbol.
* sysdeps/ia64/fpu/libm_error.c: Include <math-svid-compat.h>.
[_LIBC && LIBM_SVID_COMPAT] (matherrf): Use
compat_symbol_reference.
[_LIBC && LIBM_SVID_COMPAT] (matherrl): Likewise.
[_LIBC && !LIBM_SVID_COMPAT] (matherrf): Define as macro.
[_LIBC && !LIBM_SVID_COMPAT] (matherrl): Likewise.
* sysdeps/ia64/fpu/libm_support.h: Include <math-svid-compat.h>.
(MATHERR_D): Remove declaration.
[!_LIBC] (_LIB_VERSION_TYPE): Likewise
[!LIBM_BUILD] (_LIB_VERSIONIMF): Likewise.
[LIBM_BUILD] (pmatherrf): Likewise.
[LIBM_BUILD] (pmatherr): Likewise.
[LIBM_BUILD] (pmatherrl): Likewise.
(DOMAIN): Likewise.
(SING): Likewise.
(OVERFLOW): Likewise.
(UNDERFLOW): Likewise.
(TLOSS): Likewise.
(PLOSS): Likewise.
* sysdeps/ia64/fpu/s_matherrf.c: Include <math-svid-compat.h>.
(__matherrf): Define only if [LIBM_SVID_COMPAT]. Use
compat_symbol.
* sysdeps/ia64/fpu/s_matherrl.c: Include <math-svid-compat.h>.
(__matherrl): Define only if [LIBM_SVID_COMPAT]. Use
compat_symbol.
* math/lgamma-compat.h: Include <math-svid-compat.h>.
* math/w_acos_compat.c: Likewise.
* math/w_acosf_compat.c: Likewise.
* math/w_acosh_compat.c: Likewise.
* math/w_acoshf_compat.c: Likewise.
* math/w_acoshl_compat.c: Likewise.
* math/w_acosl_compat.c: Likewise.
* math/w_asin_compat.c: Likewise.
* math/w_asinf_compat.c: Likewise.
* math/w_asinl_compat.c: Likewise.
* math/w_atan2_compat.c: Likewise.
* math/w_atan2f_compat.c: Likewise.
* math/w_atan2l_compat.c: Likewise.
* math/w_atanh_compat.c: Likewise.
* math/w_atanhf_compat.c: Likewise.
* math/w_atanhl_compat.c: Likewise.
* math/w_cosh_compat.c: Likewise.
* math/w_coshf_compat.c: Likewise.
* math/w_coshl_compat.c: Likewise.
* math/w_exp10_compat.c: Likewise.
* math/w_exp10f_compat.c: Likewise.
* math/w_exp10l_compat.c: Likewise.
* math/w_exp2_compat.c: Likewise.
* math/w_exp2f_compat.c: Likewise.
* math/w_exp2l_compat.c: Likewise.
* math/w_fmod_compat.c: Likewise.
* math/w_fmodf_compat.c: Likewise.
* math/w_fmodl_compat.c: Likewise.
* math/w_hypot_compat.c: Likewise.
* math/w_hypotf_compat.c: Likewise.
* math/w_hypotl_compat.c: Likewise.
* math/w_j0_compat.c: Likewise.
* math/w_j0f_compat.c: Likewise.
* math/w_j0l_compat.c: Likewise.
* math/w_j1_compat.c: Likewise.
* math/w_j1f_compat.c: Likewise.
* math/w_j1l_compat.c: Likewise.
* math/w_jn_compat.c: Likewise.
* math/w_jnf_compat.c: Likewise.
* math/w_jnl_compat.c: Likewise.
* math/w_lgamma_main.c: Likewise.
* math/w_lgamma_r_compat.c: Likewise.
* math/w_lgammaf_main.c: Likewise.
* math/w_lgammaf_r_compat.c: Likewise.
* math/w_lgammal_main.c: Likewise.
* math/w_lgammal_r_compat.c: Likewise.
* math/w_log10_compat.c: Likewise.
* math/w_log10f_compat.c: Likewise.
* math/w_log10l_compat.c: Likewise.
* math/w_log2_compat.c: Likewise.
* math/w_log2f_compat.c: Likewise.
* math/w_log2l_compat.c: Likewise.
* math/w_log_compat.c: Likewise.
* math/w_logf_compat.c: Likewise.
* math/w_logl_compat.c: Likewise.
* math/w_pow_compat.c: Likewise.
* math/w_powf_compat.c: Likewise.
* math/w_powl_compat.c: Likewise.
* math/w_remainder_compat.c: Likewise.
* math/w_remainderf_compat.c: Likewise.
* math/w_remainderl_compat.c: Likewise.
* math/w_scalb_compat.c: Likewise.
* math/w_scalbf_compat.c: Likewise.
* math/w_scalbl_compat.c: Likewise.
* math/w_sinh_compat.c: Likewise.
* math/w_sinhf_compat.c: Likewise.
* math/w_sinhl_compat.c: Likewise.
* math/w_sqrt_compat.c: Likewise.
* math/w_sqrtf_compat.c: Likewise.
* math/w_sqrtl_compat.c: Likewise.
* math/w_tgamma_compat.c: Likewise.
* math/w_tgammaf_compat.c: Likewise.
* math/w_tgammal_compat.c: Likewise.
* sysdeps/ieee754/dbl-64/w_exp_compat.c: Likewise.
* sysdeps/ieee754/flt-32/w_expf_compat.c: Likewise.
* sysdeps/ieee754/k_standard.c: Likewise.
* sysdeps/ieee754/k_standardf.c: Likewise.
* sysdeps/ieee754/k_standardl.c: Likewise.
* sysdeps/ieee754/ldbl-128/w_expl_compat.c: Likewise.
* sysdeps/ieee754/ldbl-128ibm/w_expl_compat.c: Likewise.
* sysdeps/ieee754/ldbl-96/w_expl_compat.c: Likewise.
* sysdeps/powerpc/powerpc32/power4/fpu/w_sqrt_compat.S: Likewise.
* sysdeps/powerpc/powerpc32/power4/fpu/w_sqrtf_compat.S: Likewise.
* sysdeps/powerpc/powerpc32/power5/fpu/w_sqrt_compat.S: Likewise.
* sysdeps/powerpc/powerpc32/power5/fpu/w_sqrtf_compat.S: Likewise.
* sysdeps/sparc/sparc32/fpu/w_sqrt_compat.S: Likewise.
* sysdeps/sparc/sparc32/fpu/w_sqrtf_compat.S: Likewise.
* sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/w_sqrt_compat-vis3.S:
Likewise.
* sysdeps/sparc/sparc32/sparcv9/fpu/multiarch/w_sqrtf_compat-vis3.S:
Likewise.
* sysdeps/sparc/sparc32/sparcv9/fpu/w_sqrt_compat.S: Likewise.
* sysdeps/sparc/sparc32/sparcv9/fpu/w_sqrtf_compat.S: Likewise.
* sysdeps/sparc/sparc64/fpu/w_sqrt_compat.S: Likewise.
* sysdeps/sparc/sparc64/fpu/w_sqrtf_compat.S: Likewise.
2017-08-22 01:45:10 +08:00
|
|
|
#include <math-svid-compat.h>
|
1996-03-06 05:41:30 +08:00
|
|
|
|
|
|
|
|
2011-10-12 23:27:51 +08:00
|
|
|
/* wrapper powf */
|
|
|
|
float
|
|
|
|
__powf (float x, float y)
|
1996-03-06 05:41:30 +08:00
|
|
|
{
|
2011-10-12 23:27:51 +08:00
|
|
|
float z = __ieee754_powf (x, y);
|
2015-06-03 22:36:34 +08:00
|
|
|
if (__glibc_unlikely (!isfinite (z)))
|
2011-10-12 23:27:51 +08:00
|
|
|
{
|
|
|
|
if (_LIB_VERSION != _IEEE_)
|
|
|
|
{
|
Fix pow (qNaN, 0) result with -lieee (bug 20919), remove dead parts of wrappers.
The dbl-64 implementation of __ieee754_pow returns a NaN for pow
(qNaN, 0) when it should return 1. Normally this is covered up by the
wrappers ending up calling __kernel_standard which fixes up the result
for this case, but for -lieee the wrappers are bypassed and the bad
result gets through as a return value.
Now, the wrappers fixing this are dealing with variant error handling
that wants a result of NaN for pow (qNaN, 0), and only ever call
__kernel_standard for this case if NaN resulted from __ieee754_pow.
This leads to a question of whether the dbl-64 code might be
deliberately returning NaN in order to use those code paths. However,
I can find no sign that this is deliberate. If it were deliberate one
would expect other implementations to do the same, and would expect
the return of NaN to be very old, but it appears it came in by
accident when the present e_pow.c implementation replaced an fdlibm
implementation in 2001. So it appears to be unintended that this path
through the pow wrapper could be used at all.
So this patch fixes the implementation to return 1 in this case as
expected. This is consistent with all the other implementations. The
relevant path through the wrappers is now unreachable, so is removed
(which is the main motivation of this patch: to avoid that path
becoming accidentally reachable when implementing TS 18661-1 semantics
that pow (sNaN, 0) should return qNaN with "invalid" raised). Another
path that would require __ieee754_pow (0, 0) to return 0 is also
unreachable (as all implementations return 1, in accordance with C99
semantics), so is removed as well.
Note: we don't have anything set up to test -lieee, which in any case
is obsolescent (at some point we should remove the ability for new
programs to access _LIB_VERSION or define matherr and have it called
by glibc). So testing will be implicit through sNaN tests added when
making sNaN inputs work correctly for pow functions.
Tested for x86_64 and x86.
[BZ #20919]
* sysdeps/ieee754/dbl-64/e_pow.c (__ieee754_pow): Do not return
NaN first argument when raised to power 0.
* math/w_pow.c (__pow): Do not check for NaN or zero results from
raising to power zero.
* math/w_powf.c (__powf): Likewise.
* math/w_powl.c (__powl): Likewise.
* sysdeps/ieee754/k_standard.c (__kernel_standard): Do not handle
pow (0, 0) or pow (NaN, 0).
2016-12-03 06:50:46 +08:00
|
|
|
if (isfinite (x) && isfinite (y))
|
2011-10-12 23:27:51 +08:00
|
|
|
{
|
2015-06-03 22:36:34 +08:00
|
|
|
if (isnan (z))
|
2011-10-12 23:27:51 +08:00
|
|
|
/* pow neg**non-int */
|
|
|
|
return __kernel_standard_f (x, y, 124);
|
|
|
|
else if (x == 0.0f && y < 0.0f)
|
|
|
|
{
|
|
|
|
if (signbit (x) && signbit (z))
|
|
|
|
/* pow(-0.0,negative) */
|
|
|
|
return __kernel_standard_f (x, y, 123);
|
|
|
|
else
|
|
|
|
/* pow(+0.0,negative) */
|
|
|
|
return __kernel_standard_f (x, y, 143);
|
|
|
|
}
|
Update.
1997-03-16 18:43 Ulrich Drepper <drepper@cygnus.com>
* manual/filesys.texi: Add documentation for scandir and alphasort.
* math/math.c (fpclassify): Correct stupid typos.
* math/libm-test.c: New file. libm test suite by Andreas Jaeger.
* nss/nss_files/files-hosts.c: Add gethostbyname2 imlementation.
* posix/Makefile (routines): Add bsd-getpgrp.
* posix/bsd-getpgrp.c: New file.
* posix/unistd.h [__FAVOR_BSD]: Define macro getpgrp which maps
calls to __bsd_getpgrp.
* sysdeps/generic/getpgrp.c: De-ANSI-declfy.
* sysdeps/i386/huge_val.h: New file. ix87 specific infinity values.
* sysdeps/m68k/huge_val.h: New file. m68k specific infinity values.
* sysdeps/generic/huge_val.h: Remove definition of long double
definition. Make it the same as the double definition.
* sysdeps/libm-i387/e_acos.S: Fix bug in FPU stack handling.
* sysdeps/libm-i387/e_acosf.S: Likewise.
* sysdeps/libm-i387/e_acosl.S: Likewise.
* sysdeps/libm-i387/e_asin.S: Likewise.
* sysdeps/libm-i387/e_asinf.S: Likewise.
* sysdeps/libm-i387/e_asinl.S: Likewise.
* sysdeps/libm-i387/e_exp.S: Likewise.
* sysdeps/libm-i387/e_expf.S: Likewise.
* sysdeps/libm-i387/e_expl.S: Likewise.
* sysdeps/libm-i387/e_scalbn.S: Likewise.
* sysdeps/libm-i387/e_scalbnf.S: Likewise.
* sysdeps/libm-i387/e_scalbnl.S: Likewise.
* sysdeps/libm-i387/e_log.S: Optimize branch code.
* sysdeps/libm-i387/e_logf.S: Likewise.
* sysdeps/libm-i387/e_logl.S: Likewise.
* sysdeps/libm-i387/e_log10.S: Likewise.
* sysdeps/libm-i387/e_log10f.S: Likewise.
* sysdeps/libm-i387/e_log10l.S: Likewise.
* sysdeps/libm-i387/e_pow.S: Major rewrite to handle special cases.
* sysdeps/libm-i387/e_powf.S: Likewise.
* sysdeps/libm-i387/e_powl.S: Likewise.
* sysdeps/libm-i387/e_expm1.S: Change return value for -inf
argument to -1.0.
* sysdeps/libm-i387/e_expm1f.S: Likewise.
* sysdeps/libm-i387/e_expm1l.S: Likewise.
* sysdeps/libm-i387/e_isinfl.c: Return -1 for -inf.
* sysdeps/libm-i387/e_logbl.S: Correct return value. Discard first
stack element after fxtract.
* sysdeps/libm-ieee754/e_atan2l.c: New file. `long double'
implementation for atan2 function.
* sysdeps/libm-ieee754/k_standard.c: Return NAN for libm not in
_SVID_ mode when acos, asin, atan2, log, log10 is called with
argument out of range.
Add new error case for pow(+0,neg).
* sysdeps/libm-ieee754/s_fpclassifyf.c: Correct recognition of
NaN and +-inf.
* sysdeps/libm-ieee754/s_fpclassifyl.c: Mask out explicit leading
digit in stupid 80 bit formats.
* sysdeps/libm-ieee754/s_isinf.c: Rewrite to return -1 for -inf.
* sysdeps/libm-ieee754/s_isinff.c: Likewise.
* sysdeps/libm-ieee754/s_isinfl.c: Likewise.
* sysdeps/libm-ieee754/s_scalbnl.c (huge, tiny): Adapt values for
long double type.
* sysdeps/libm-ieee754/w_atan2.c: Do not raise exception expect when
in SVID mode.
* sysdeps/libm-ieee754/w_atan2f.c: Likewise.
* sysdeps/libm-ieee754/w_atan2l.c: Likewise.
* sysdeps/libm-ieee754/w_pow.c: Distinguish error cases for x is +0
or -0.
* sysdeps/posix/isfdtype.c: Add cast to prevent warning.
* sysdeps/stub/fcntlbits.h: Update copyright.
* sysdeps/unix/bsd/fcntlbits.h: Likewise.
* sysdeps/unix/bsd/bsd4.4/fcntlbits.h: Likewise.
* sysdeps/unix/bsd/sun/sunos4/fcntlbits.h: Likewise.
* sysdeps/unix/bsd/ultrix4/fcntlbits.h: Likewise.
* sysdeps/unix/common/fcntlbits.h: Likewise.
* sysdeps/unix/sysv/fcntlbits.h: Likewise. Define O_FSYNC as alias
of O_SYNC. Add BSD compatibility macros FAPPEND, FFSYNC, FNONBLOCK,
and FNDELAY.
* sysdeps/unix/sysv/irix4/fcntlbits.h: Likewise.
* sysdeps/unix/readdir_r.c: Don't copy whole `struct dirent' record,
only reclen bytes.
* sysdeps/unix/sysv/linux/fcntlbits.h [__USE_GNU]: Add O_READ, O_WRITE
and O_NORW.
* sysdeps/unix/sysv/linux/alpha/fcntlbits.h: Likewise.
* sysdeps/unix/sysv/linux/init-first.h: Add copyright.
* sysdeps/unix/sysv/linux/fxstat.c: New file. Rewrite kernel-level
struct stat to user-level form.
* sysdeps/unix/sysv/linux/lxstat: New file.
* sysdeps/unix/sysv/linux/xstat: New file.
* sysdeps/unix/sysv/linux/kernel_stat.h: Define struct stat used in
kernel.
* sysdeps/unix/sysv/linux/statbuf.h (struct stat): Change definition
to use prescribed types for elements.
(_STAT_VER): Change to value 3.
* sysdeps/unix/sysv/linux/alph/statbuf.h: Likewise.
* sysdeps/unix/sysv/linux/Dist: Add kernel_stat.h.
* sysdeps/unix/sysv/linux/alpha/Dist: Likewise.
* time/Makefile: Correct dependencies for test-tz.
1997-03-16 14:59 Philip Blundell <phil@london.uk.eu.org>
* resolv/netdb.h: Add prototypes for gai_strerror and getnameinfo
(needed for IPv6 basic sockets API).
1997-03-16 15:02 a sun <asun@zoology.washington.edu>
* sysdeps/unix/sysv/linux/net/if_ppp.h: Don't use incompatible
kernel header.
* sysdeps/unix/sysv/linux/net/ppp_defs.h: Likewise.
1997-03-14 17:15 Ulrich Drepper <drepper@cygnus.com>
* db/hash/hash_bigkey.c (__big_delete): Don't call __free_ovflpage
without testing for last_bfp to be NULL.
Reported by fabsoft@fabserver1.zarm.uni-bremen.de.
1997-03-13 11:42 Jim Meyering <meyering@asic.sc.ti.com>
* time/mktime.c (TIME_T_MIN): Work around a bug in Cray C 5.0.3.0.
1997-03-14 04:00 Kurt Garloff <garloff@kg1.ping.de>
* sysdeps/unix/sysv/linux/fcntlbits.h (O_FSYNC): Make alias for O_SYNC.
(FASYNC): Move to __USE_BSD section. Create new macro O_ASYNC.
1997-03-14 02:50 Ulrich Drepper <drepper@cygnus.com>
* nis/nss_nis/nis-hosts.c (_nss_nis_gethostbyname2_r): New
functions. Compare result for correct address type.
(_nss_nis_gethostbyname_r): Use _nss_nis_gethostbyname2_r.
Reported by Mirko Streckenbach <mirko@marian.hil.de>.
1997-02-17 01:40 Zlatko Calusic <zcalusic@srce.hr>
* time/strptime.c (recursive): Return rp to caller.
(strptime_internal): First check for long names, then abbreviated
(month & weekday).
1997-03-10 19:44 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* Makeconfig: Remove useless definitions of ASFLAGS-%.
* config.make.in (ASFLAGS-.so): Remove.
* configure.in: Don't substitute ASFLAGS_SO.
* sysdeps/sparc/configure.in: Remove file.
* sysdeps/sparc/Makefile (ASFLAGS-.so): Define.
1997-03-11 17:00 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de>
* time/strptime.c (strptime_internal) [case 'Y']: Always subtract
1900 from year, regardless of century.
1997-03-12 05:43 Ulrich Drepper <drepper@cygnus.com>
* stdlib/strtod.c (_tens_in_limb) [BITS_PER_MP_LIMB > 32]: Make
all numbers unsigned to make buggy gccs happy.
Patch by Bryan W. Headley <bheadley@interaccess.com>.
* sysdeps/unix/sysv/linux/netinet/ip.h: Add backward-compatibility
definitions. Patch by a sun <asun@zoology.washington.edu>.
Pretty print header.
* Makerules (build-shlib): Also create symlink if library is versioned.
based on a patch by H.J. Lu <hjl@gnu.ai.mit.edu>.
Remove special rule to libc.so symlink.
1997-03-11 20:16 Andreas Jaeger <aj@arthur.pfalz.de>
* manual/math.texi (Domain and Range Errors): Change descriptions
according to recent changes for ISO C 9X.
1997-03-11 22:39 Ulrich Drepper <drepper@cygnus.com>
* sysdeps/libm-ieee754/k_standard.c (__kernel_standard): Correct
return values for acos, asin, and atan2.
Reported by Andreas Jaeger <aj@arthur.pfalz.de>.
1997-03-10 18:16 Thorsten Kukuk <kukuk@vt.uni-paderborn.de>
* ypclnt.c (__yp_bind): Fix possible buffer overflow.
1997-03-10 18:06 Bernd Schmidt <crux@Pool.Informatik.RWTH-Aachen.DE>
* dirent/alphasort.c (alphasort): Interpret arguments as pointers
to pointers to directory entries so that alphasort really can be
used as argument for scandir.
1997-03-09 23:33 Andreas Jaeger <aj@arthur.pfalz.de>
* string/strdup.c: Declare memcpy if !(_LIBC || STDC_HEADERS)
instead of strcpy.
1997-03-10 03:34 Ulrich Drepper <drepper@cygnus.com>
* catgets/catgets.c (catopen): Always add NLSPATH to search path for
catalogs, not only if the envvar NLSPATH is not available.
Reported by Andries.Brouwer@cwi.nl.
1997-03-10 02:46 Ulrich Drepper <drepper@cygnus.com>
* Makeconfig (localtime-file): Don't define using installation
directory.
(inst_localtime-file): New variable.
* time/Makefile (installed-localtime-file): Use inst_localtime-file.
Reported by Edward Seidl <seidl@janed.com>.
1997-03-10 02:31 H.J. Lu <hjl@gnu.ai.mit.edu>
* time/Makefile: Add source files to dependencies for test data.
1997-03-09 22:53 Thorsten Kukuk <kukuk@weber.uni-paderborn.de>
* nis/nss_nis/nis-ethers.c: Don't ignore return value of yp_all.
* nis/nss_nis/nis-proto.c: Likewise.
* nis/nss_nis/nis-rpc.c: Likewise.
* nis/nss_nis/nis-service.c: Likewise.
1997-03-08 14:37 Miguel de Icaza <miguel@nuclecu.unam.mx>
* sysdeps/sparc/dl-machine.h (elf_machine_rela): Upgrade to
versioning; Added missing R_SPARC_WDISP30 handling.
(RTLD_START): Implement it.
* sysdeps/unix/sysv/linux/sparc/brk.c: Fix.
* sysdeps/unix/sysv/linux/sparc/start.c: Startup code for
Linux/SPARC.
1997-03-02 18:06 Miguel de Icaza <miguel@nuclecu.unam.mx>
* sysdeps/sparc/dl-machine.h (RTLD_START): Make arg as expected by
the dynamic linker instead of having a new conditional define.
Thanks to Richard Henderson for pointing this out.
* elf/rtld.c: Remove usage of ELF_ADJUST_ARG.
1997-03-20 20:44 Thomas Bushnell, n/BSG <thomas@gnu.ai.mit.edu>
* sysdeps/mach/hurd/euidaccess.c: Define as __euidaccess and make
euidaccess weak alias.
1997-03-07 10:30 Thomas Bushnell, n/BSG <thomas@gnu.ai.mit.edu>
* stdio-common/printf_fphex.c (MIN): New macro.
* sysdeps/generic/netinet/in.h: Include <sys/types.h>.
* sysdeps/generic/sys/mman.h (msync): Mention third arg.
* sysdeps/generic/netinet/in.h: Add definitions for IPv6 basic
API. (See change by Philip Blundell on Feb 16, 1997.)
1997-03-05 10:40 Thomas Bushnell, n/BSG <thomas@gnu.ai.mit.edu>
* hurd/hurd.h (vpprintf): Include <stdarg.h>. New declaration.
* hurd/set-host.c (_hurd_set_host_config): Cast second arg to
__file_name_split.
* mach/mach_error.c (mach_error_string_int): Give full prototype.
* mach/errstring.c (mach_error_string_int): Likewise.
* mach/error_compat.c (__mach_error_map_compat): Likewise.
* hurd/vpprintf.c (pwrite, vpprintf): Likewise.
* stdio/vasprintf.c (vasprintf): Likewise.
* mach/mach/mach_traps.h: Include <mach/kern_return.h>.
* mach/spin-solid.c: Include <mach/mach_traps.h>.
* mach/spin-solid.c (__spin_lock_solid): Provide arg to
swtch_pri.
* mach/mach_init.c: Include <mach/mig_support.h>.
* mach/mach_error.h (mach_error_string, mach_error,
mach_error_type): Always provide prototypes.
* mach/mach/error.h (mach_error_fn_t): Comment out declaration; it
appears to be entirely unused dead code.
* stdio/stdio.h (freopen): Fix spelling error.
1997-03-02 13:38 Miles Bader <miles@gnu.ai.mit.edu>
* string/argz.h (__need_error_t): New macro, before including <errno.h>
[!__const] (__const): New macro.
[!__error_t_defined] (error_t): New typedef.
* sysdeps/generic/socketbits.h: Add PF_FILE as synonym for PF_LOCAL
* sysdeps/unix/sysv/linux/socketbits.h: Likewise.
1997-03-17 04:28:07 +08:00
|
|
|
else
|
2011-10-12 23:27:51 +08:00
|
|
|
/* pow overflow */
|
|
|
|
return __kernel_standard_f (x, y, 121);
|
1998-09-14 19:43:26 +08:00
|
|
|
}
|
1996-03-06 05:41:30 +08:00
|
|
|
}
|
2011-10-12 23:27:51 +08:00
|
|
|
}
|
Fix pow (qNaN, 0) result with -lieee (bug 20919), remove dead parts of wrappers.
The dbl-64 implementation of __ieee754_pow returns a NaN for pow
(qNaN, 0) when it should return 1. Normally this is covered up by the
wrappers ending up calling __kernel_standard which fixes up the result
for this case, but for -lieee the wrappers are bypassed and the bad
result gets through as a return value.
Now, the wrappers fixing this are dealing with variant error handling
that wants a result of NaN for pow (qNaN, 0), and only ever call
__kernel_standard for this case if NaN resulted from __ieee754_pow.
This leads to a question of whether the dbl-64 code might be
deliberately returning NaN in order to use those code paths. However,
I can find no sign that this is deliberate. If it were deliberate one
would expect other implementations to do the same, and would expect
the return of NaN to be very old, but it appears it came in by
accident when the present e_pow.c implementation replaced an fdlibm
implementation in 2001. So it appears to be unintended that this path
through the pow wrapper could be used at all.
So this patch fixes the implementation to return 1 in this case as
expected. This is consistent with all the other implementations. The
relevant path through the wrappers is now unreachable, so is removed
(which is the main motivation of this patch: to avoid that path
becoming accidentally reachable when implementing TS 18661-1 semantics
that pow (sNaN, 0) should return qNaN with "invalid" raised). Another
path that would require __ieee754_pow (0, 0) to return 0 is also
unreachable (as all implementations return 1, in accordance with C99
semantics), so is removed as well.
Note: we don't have anything set up to test -lieee, which in any case
is obsolescent (at some point we should remove the ability for new
programs to access _LIB_VERSION or define matherr and have it called
by glibc). So testing will be implicit through sNaN tests added when
making sNaN inputs work correctly for pow functions.
Tested for x86_64 and x86.
[BZ #20919]
* sysdeps/ieee754/dbl-64/e_pow.c (__ieee754_pow): Do not return
NaN first argument when raised to power 0.
* math/w_pow.c (__pow): Do not check for NaN or zero results from
raising to power zero.
* math/w_powf.c (__powf): Likewise.
* math/w_powl.c (__powl): Likewise.
* sysdeps/ieee754/k_standard.c (__kernel_standard): Do not handle
pow (0, 0) or pow (NaN, 0).
2016-12-03 06:50:46 +08:00
|
|
|
else if (__builtin_expect (z == 0.0f, 0)
|
|
|
|
&& isfinite (x) && x != 0 && isfinite (y)
|
2011-10-12 23:27:51 +08:00
|
|
|
&& _LIB_VERSION != _IEEE_)
|
Fix pow (qNaN, 0) result with -lieee (bug 20919), remove dead parts of wrappers.
The dbl-64 implementation of __ieee754_pow returns a NaN for pow
(qNaN, 0) when it should return 1. Normally this is covered up by the
wrappers ending up calling __kernel_standard which fixes up the result
for this case, but for -lieee the wrappers are bypassed and the bad
result gets through as a return value.
Now, the wrappers fixing this are dealing with variant error handling
that wants a result of NaN for pow (qNaN, 0), and only ever call
__kernel_standard for this case if NaN resulted from __ieee754_pow.
This leads to a question of whether the dbl-64 code might be
deliberately returning NaN in order to use those code paths. However,
I can find no sign that this is deliberate. If it were deliberate one
would expect other implementations to do the same, and would expect
the return of NaN to be very old, but it appears it came in by
accident when the present e_pow.c implementation replaced an fdlibm
implementation in 2001. So it appears to be unintended that this path
through the pow wrapper could be used at all.
So this patch fixes the implementation to return 1 in this case as
expected. This is consistent with all the other implementations. The
relevant path through the wrappers is now unreachable, so is removed
(which is the main motivation of this patch: to avoid that path
becoming accidentally reachable when implementing TS 18661-1 semantics
that pow (sNaN, 0) should return qNaN with "invalid" raised). Another
path that would require __ieee754_pow (0, 0) to return 0 is also
unreachable (as all implementations return 1, in accordance with C99
semantics), so is removed as well.
Note: we don't have anything set up to test -lieee, which in any case
is obsolescent (at some point we should remove the ability for new
programs to access _LIB_VERSION or define matherr and have it called
by glibc). So testing will be implicit through sNaN tests added when
making sNaN inputs work correctly for pow functions.
Tested for x86_64 and x86.
[BZ #20919]
* sysdeps/ieee754/dbl-64/e_pow.c (__ieee754_pow): Do not return
NaN first argument when raised to power 0.
* math/w_pow.c (__pow): Do not check for NaN or zero results from
raising to power zero.
* math/w_powf.c (__powf): Likewise.
* math/w_powl.c (__powl): Likewise.
* sysdeps/ieee754/k_standard.c (__kernel_standard): Do not handle
pow (0, 0) or pow (NaN, 0).
2016-12-03 06:50:46 +08:00
|
|
|
/* pow underflow */
|
|
|
|
return __kernel_standard_f (x, y, 122);
|
2011-10-12 23:27:51 +08:00
|
|
|
|
|
|
|
return z;
|
1996-03-06 05:41:30 +08:00
|
|
|
}
|
|
|
|
weak_alias (__powf, powf)
|