For the -ffinite-math-only versions of various x86_64 and x86 log*
functions, a zero result from log* (1) is returned with incorrect sign
in round-downward mode. This patch fixes this in a similar way to the
previous fixes for the non-*_finite versions of the functions.
Tested for x86_64 and x86 (including an i586 build), together with a
patch that will be applied separately to enable the main libm-test.inc
tests for the finite-math-only functions.
[BZ #19213]
* sysdeps/i386/fpu/e_log.S (__log_finite): Ensure +0 is always
returned for argument 1.
* sysdeps/i386/fpu/e_logf.S (__logf_finite): Likewise.
* sysdeps/i386/fpu/e_logl.S (__logl_finite): Likewise.
* sysdeps/i386/i686/fpu/e_logl.S (__logl_finite): Likewise.
* sysdeps/x86_64/fpu/e_log10l.S (__log10l_finite): Likewise.
* sysdeps/x86_64/fpu/e_log2l.S (__log2l_finite): Likewise.
* sysdeps/x86_64/fpu/e_logl.S (__logl_finite): Likewise.
The lgamma inline functions in bits/math-finite.h do not set signgam
if __USE_ISOC99, even when other feature test macros mean a standard
such as XSI POSIX is selected for which it should be set. (This is
essentially the opposite issue to bug 15421, the out-of-line versions
setting signgam even when they shouldn't.)
This patch fixes those functions to use __USE_MISC || __USE_XOPEN as
the condition for when to set signgam, since it's the condition for
when math.h declares signgam. The legacy gamma* names are only
declared at all if __USE_MISC || __USE_XOPEN, so they just set signgam
unconditionally.
Tests for certain standards or not using _GNU_SOURCE cannot use
test-skeleton.c (this is a known issue noted on the wiki todo list).
Thus, the new tests that signgam remains not set in ISO C modes do not
use test-skeleton.c. They also define _ISOMAC to avoid running into
declarations in the internal include/ headers that only work in
_GNU_SOURCE mode.
Tested for x86_64 and x86.
[BZ #19211]
* math/bits/math-finite.h (lgamma): Set signgam if [__USE_MISC ||
__USE_XOPEN], not if [!__USE_ISOC99].
(lgammaf): Likewise.
(lgammal): Likewise.
(gamma): Set signgam unconditionally, not if [!__USE_ISOC99].
(gammaf): Likewise.
(gammal): Likewise.
* math/test-signgam-finite-c11.c: New file.
* math/test-signgam-finite-c99.c: Likewise.
* math/test-signgam-finite.c: Likewise.
* math/Makefile (tests): Add test-signgam-finite,
test-signgam-finite-c99 and test-signgam-finite-c11.
(CFLAGS-test-signgam-finite.c): New variable.
(CFLAGS-test-signgam-finite-c99.c): Likewise.
(CFLAGS-test-signgam-finite-c11.c): Likewise.
features.h is not clean with -Wundef (for the installed header, of
course this only appears with -Wsystem-headers). In ISO C standards
modes, you get a series of warnings / errors relating to
_POSIX_C_SOURCE and _XOPEN_SOURCE not being defined when tested in
standards mode and uses #undef _GNU_SOURCE to avoid the default
_GNU_SOURCE from libc-symbols.h. This patch changes the relevant #if
conditionals to avoid these warnings / errors.
Tested for x86_64 and x86 (testsuite, and that installed shared
libraries are unchanged by the patch).
[BZ #19212]
* include/features.h [(_XOPEN_SOURCE - 0) >= 500]: Change
conditional to [defined _XOPEN_SOURCE && (_XOPEN_SOURCE - 0) >=
500].
[_POSIX_C_SOURCE >= 1]: Change conditional to [defined
_POSIX_C_SOURCE && _POSIX_C_SOURCE >= 1].
[(_POSIX_C_SOURCE - 0) >= 199309L]: Change conditional to [defined
_POSIX_C_SOURCE && (_POSIX_C_SOURCE - 0) >= 199309L].
[(_POSIX_C_SOURCE - 0) >= 199506L]: Change conditional to [defined
_POSIX_C_SOURCE && (_POSIX_C_SOURCE - 0) >= 199506L].
[(_POSIX_C_SOURCE - 0) >= 200112L]: Change conditional to [defined
_POSIX_C_SOURCE && (_POSIX_C_SOURCE - 0) >= 200112L].
[(_POSIX_C_SOURCE - 0) >= 200809L]: Change conditional to [defined
_POSIX_C_SOURCE && (_POSIX_C_SOURCE - 0) >= 200809L].
bits/math-finite.h maps ldexp functions to corresponding scalbn
functions. This is (a) a namespace bug for C90, which has ldexp but
not scalbn, and (b) in any case useless, since the ldexp and scalbn
functions have identical semantics (for floating-point types with
radix 2), and since the fix for bug 6803 are actually aliases
(presumably the mapping was based around the old bug of scalbn not
setting errno). This patch removes the bogus redirections.
Tested for x86_64 and x86.
[BZ #19209]
* math/bits/math-finite.h (ldexp): Remove declaration.
(ldexpf): Likewise.
(ldexpl): Likewise.
bits/math-finite.h declares -ffinite-math-only variants of various
functions under conditions not matching those under which the normal
versions are declared.
* math.h only ever includes bits/mathcalls.h to declare float and long
double functions if __USE_ISOC99, but bits/math-finite.h declares
some float functions regardless (long double ones are conditioned on
__MATH_DECLARE_LDOUBLE). (For C90 functions this isn't a
conformance bug because C90 reserves the float and long double
names, but is still contrary to good glibc practice. For some other
functions in older XSI standards it *is* a conformance bug.)
* Some functions are defined as inlines using lgamma_r functions under
conditions where those lgamma_r functions are not themselves
declared.
* hypot is declared under __USE_XOPEN || __USE_ISOC99 in
bits/mathcalls.h, __USE_ISOC99 only in bits/math-finite.h.
* float and long double versions of Bessel functions should be limited
to __USE_MISC (as in bug 18977).
* gamma should not be declared for __USE_XOPEN2K (as in bug 18967).
* remainder should be restricted to __USE_XOPEN_EXTENDED ||
__USE_ISOC99, not unconditional.
* scalb should not be declared for __USE_XOPEN2K8, and scalbf and
scalbl are non-POSIX (as in bug 18967).
This patch fixes all these issues (it doesn't seem worth splitting
them into separate patches or bugs). I put __USE_ISOC99 conditionals,
where needed, around both float and long double declarations, even
though formally redundant around the long double declarations because
__MATH_DECLARE_LDOUBLE isn't defined without __USE_ISOC99; it seemed
clearer that way. The missing declarations of lgamma_r functions are
dealt with by directly using declarations of __lgamma*_r_finite, in
the implementation namespace, rather than having the inlines rely on
asm redirection of lgamma*_r.
After this patch, there are some apparently redundant nested
__USE_ISOC99 conditionals in lgamma / gamma definitions. These
actually reflect a separate bug (the correct condition for the lgamma
inline functions to set signgam is __USE_MISC || __USE_XOPEN, the
condition under which signgam is declared, rather than disabling
setting it if __USE_ISOC99, which includes XSI POSIX versions for
which signgam *should* be set). They'll be fixed as part of a fix for
that bug, which will also add tests for these inlines. I've put a
note about more general conform/ test coverage for -ffinite-math-only
on
<https://sourceware.org/glibc/wiki/Development_Todo/Master#conformtest_improvements>,
alongside other options for which this is also relevant (some of which
have also had such bugs in the past relating to mismatched
conditionals).
I also intend to enable the main libm-test.inc tests for the
math-finite.h functions, but some other bugs in __*_finite need fixing
first.
[BZ #19205]
* math/bits/math-finite.h (acosf): Condition declaration on
[__USE_ISOC99].
(acosl): Likewise.
(acoshf): Likewise.
(acoshl): Likewise.
(asinf): Likewise.
(asinl): Likewise.
(atan2f): Likewise.
(atan2l): Likewise.
(atanhf): Likewise.
(atanhl): Likewise.
(coshf): Likewise.
(coshl): Likewise.
(expf): Likewise.
(expl): Likewise.
(fmodf): Likewise.
(fmodl): Likewise.
(hypot): Change condition to [__USE_XOPEN || __USE_ISOC99].
(j0f): Change condition to [__USE_MISC && __USE_ISOC99].
(j0l): Likewise.
(y0f): Likewise.
(y0l): Likewise.
(j1f): Likewise.
(j1l): Likewise.
(y1f): Likewise.
(y1l): Likewise.
(jnf): Likewise.
(jnl): Likewise.
(ynf): Likewise.
(ynl): Likewise.
(lgammaf_r): Condition declaration on [__USE_ISOC99].
(lgammal_r): Likewise.
(__lgamma_r_finite): New declaration.
(__lgammaf_r_finite): Likewise.
(__lgammal_r_finite): Likewise.
(lgamma): Use __lgamma_r_finite.
(lgammaf): Condition definition on [__USE_ISOC99]. Use
__lgammaf_r_finite.
(lgammal): Condition definition on [__USE_ISOC99]. Use
__lgammal_r_finite.
(gamma): Do not define for [!__USE_MISC && __USE_XOPEN2K]. Use
__lgamma_r_finite.
(gammaf): Condition definition on [__USE_ISOC99]. Use
__lgammaf_r_finite.
(gammal): Condition definition on [__USE_ISOC99]. Use
__lgammal_r_finite.
(logf): Condition declaration on [__USE_ISOC99].
(logl): Likewise.
(log10f): Likewise.
(log10l): Likewise.
(ldexpf): Likewise.
(ldexpl): Likewise.
(powf): Likewise.
(powl): Likewise.
(remainder): Condition declaration on [__USE_XOPEN_EXTENDED ||
__USE_ISOC99].
(remainderf): Condition declaration on [__USE_ISOC99].
(remainderl): Likewise.
(scalb): Do not declare for [!__USE_MISC && __USE_XOPEN2K8].
(scalbf): Change condition to [__USE_MISC && __USE_ISOC99].
(scalbl): Likewise.
(sinhf): Condition declaration on [__USE_ISOC99].
(sinhl): Likewise.
(sqrtf): Likewise.
(sqrtl): Likewise.
For some large arguments, the dbl-64 implementation of remainder gives
zero results with the wrong sign, resulting from a subtraction that is
mathematically correct but does not guarantee that a zero result has
the sign of the first argument to remainder. This patch adds an
appropriate check for this case, similar to other implementations of
remainder in the case of equality, and adds tests of remainder on
inputs already used to test remquo.
Tested for x86_64 and x86.
[BZ #19201]
* sysdeps/ieee754/dbl-64/e_remainder.c (__ieee754_remainder):
Check for zero remainder in case of large exponents and ensure
correct sign of result in that case.
* math/libm-test.inc (remainder_test_data): Add more tests.
nextafter and nexttoward fail to set errno on overflow and underflow.
This patch makes them do so in cases that should include all the cases
where such errno setting is required by glibc's goals for when to set
errno (but not all cases of underflow where the result is nonzero and
so glibc's goals do not require errno setting).
Tested for x86_64, x86, mips64 and powerpc.
[BZ #6799]
* math/s_nextafter.c: Include <errno.h>.
(__nextafter): Set errno on overflow and underflow.
* math/s_nexttowardf.c: Include <errno.h>.
(__nexttowardf): Set errno on overflow and underflow.
* sysdeps/i386/fpu/s_nextafterl.c: Include <errno.h>.
(__nextafterl): Set errno on overflow and underflow.
* sysdeps/i386/fpu/s_nexttoward.c: Include <errno.h>.
(__nexttoward): Set errno on overflow and underflow.
* sysdeps/i386/fpu/s_nexttowardf.c: Include <errno.h>.
(__nexttowardf): Set errno on overflow and underflow.
* sysdeps/ieee754/flt-32/s_nextafterf.c: Include <errno.h>.
(__nextafterf): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128/s_nextafterl.c: Include <errno.h>.
(__nextafterl): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128/s_nexttoward.c: Include <errno.h>.
(__nexttoward): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128/s_nexttowardf.c: Include <errno.h>.
(__nexttowardf): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128ibm/s_nextafterl.c: Include <errno.h>.
(__nextafterl): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128ibm/s_nexttoward.c: Include <errno.h>.
(__nexttoward): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-128ibm/s_nexttowardf.c: Include <errno.h>.
(__nexttowardf): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-96/s_nexttoward.c: Include <errno.h>.
(__nexttoward): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-96/s_nexttowardf.c: Include <errno.h>.
(__nexttowardf): Set errno on overflow and underflow.
* sysdeps/ieee754/ldbl-opt/s_nexttowardfd.c: Include <errno.h>.
(__nldbl_nexttowardf): Set errno on overflow and underflow.
* sysdeps/m68k/m680x0/fpu/s_nextafterl.c: Include <errno.h>.
(__nextafterl): Set errno on overflow and underflow.
* math/libm-test.inc (nextafter_test_data): Do not allow errno
setting to be missing on overflow. Add more tests.
(nexttoward_test_data): Likewise.
The ldbl-128 version of log1pl raises a spurious "invalid" exception
for a -qNaN argument. This patch fixes this by making the initial
check for infinities and NaNs handle arguments of both signs in such a
way that NaNs result in a NaN being returned (quietly if the input NaN
was quiet) while +Inf results in +Inf being returned and -Inf results
in a qNaN being returned with "invalid" exception raised.
Tested for mips64.
[BZ #19189]
* sysdeps/ieee754/ldbl-128/s_log1pl.c (__log1pl): Make check for
non-finite argument handle arguments with negative sign.
The libm drem functions just call the corresponding __remainder
functions. This patch removes the unnecessary wrappers by making them
into weak aliases at the ELF level.
Tested for x86_64, x86, mips64 and powerpc.
[BZ #16171]
* math/w_remainder.c (drem): Define as weak alias of __remainder.
[NO_LONG_DOUBLE] (dreml): Define as weak alias of __remainder.
* math/w_remainderf.c (dremf): Define as weak alias of
__remainderf.
* math/w_remainderl.c (dreml): Define as weak alias of
__remainderl.
* sysdeps/ia64/fpu/e_remainder.S (drem): Define as weak alias of
__remainder.
* sysdeps/ia64/fpu/e_remainderf.S (dremf): Define as weak alias of
__remainderf.
* sysdeps/ia64/fpu/e_remainderl.S (dreml): Define as weak alias of
__remainderl.
* sysdeps/ieee754/ldbl-opt/nldbl-remainder.c (dreml): Define as
weak alias of remainderl.
* sysdeps/ieee754/ldbl-opt/w_remainder.c
[LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)] (__drem): Define as strong
alias of __remainder.
[LONG_DOUBLE_COMPAT(libm, GLIBC_2_0)] (dreml): Use compat_symbol.
* sysdeps/ieee754/ldbl-opt/w_remainderl.c (__dreml): Define as
strong alias of __remainderl.
(dreml): Use long_double_symbol.
* math/Makefile (libm-calls): Remove w_drem.
* sysdeps/ieee754/ldbl-opt/Makefile (libnldbl-calls): Remove drem.
(CFLAGS-nldbl-drem.c): Remove variable.
(CFLAGS-nldbl-remainder.c): Add -fno-builtin-dreml.
* math/w_drem.c: Remove file.
* math/w_dremf.c: Likewise.
* math/w_dreml.c: Likewise.
* sysdeps/ieee754/ldbl-opt/nldbl-drem.c: Likewise.
* sysdeps/ieee754/ldbl-opt/w_drem.c: Likewise.
* sysdeps/ieee754/ldbl-opt/w_dreml.c: Likewise.
fenv_t should include architecture-specific floating-point modes and
status flags. i386 and x86_64 fesetenv limit which bits they use from
the x87 status and control words, when using saved state, and limit
which parts of the state they set to fixed values, when using
FE_DFL_ENV / FE_NOMASK_ENV. The following should be included but are
excluded in at least some cases: status and masking for the "denormal
operand" exception (which isn't part of FE_ALL_EXCEPT); precision
control (explicitly mentioned in Annex F as something that counts as
part of the floating-point environment); MXCSR FZ and DAZ bits (for
FE_DFL_ENV and FE_NOMASK_ENV). This patch arranges for this extra
state to be handled by fesetenv (and thereby by feupdateenv, which
calls fesetenv).
(Note that glibc functions using floating point are not generally
expected to work correctly with non-default values of this state,
especially precision control, but it is still logically part of the
floating-point environment and should be handled as such by fesetenv.
Changes to the state relating to subnormals ought generally to work
with libm functions when the arguments aren't subnormal and neither
are the expected results; that's a consequence of functions avoiding
spurious internal underflows.)
A question arising from this is whether FE_NOMASK_ENV should or should
not mask the "denormal operand" exception. I decided it should mask
that exception. This is the status quo - previously that exception
could only be unmasked by direct manipulation of control registers
(possibly via <fpu_control.h>). In addition, it means that use of
FE_NOMASK_ENV leaves a floating-point environment the same as could be
obtained by fesetenv (FE_DFL_ENV); feenableexcept (FE_ALL_EXCEPT);,
rather than an environment in which an exception is unmasked that
could only be masked again by using fesetenv with FE_DFL_ENV (or a
previously saved environment) - this exception not being usable with
other <fenv.h> functions because it's outside FE_ALL_EXCEPT.
Tested for x86_64 and x86.
[BZ #16068]
* sysdeps/i386/fpu/fesetenv.c: Include <fpu_control.h>.
(FE_ALL_EXCEPT_X86): New macro.
(__fesetenv): Use FE_ALL_EXCEPT_X86 in most places instead of
FE_ALL_EXCEPT. Ensure precision control is included in
floating-point state. Ensure that FE_DFL_ENV and FE_NOMASK_ENV
handle "denormal operand exception" and clear FZ and DAZ bits.
* sysdeps/x86_64/fpu/fesetenv.c: Include <fpu_control.h>.
(FE_ALL_EXCEPT_X86): New macro.
(__fesetenv): Use FE_ALL_EXCEPT_X86 in most places instead of
FE_ALL_EXCEPT. Ensure precision control is included in
floating-point state. Ensure that FE_DFL_ENV and FE_NOMASK_ENV
handle "denormal operand exception" and clear FZ and DAZ bits.
* sysdeps/x86/fpu/test-fenv-sse-2.c: New file.
* sysdeps/x86/fpu/test-fenv-x87.c: Likewise.
* sysdeps/x86/fpu/Makefile [$(subdir) = math] (tests): Add
test-fenv-x87 and test-fenv-sse-2.
[$(subdir) = math] (CFLAGS-test-fenv-sse-2.c): New variable.
The i386 and x86_64 versions of fesetenv, when called with FE_DFL_ENV
or FE_NOMASK_ENV as argument, do not clear SSE exceptions raised in
MXCSR. These arguments should, like other fenv_t values, represent
the whole of the floating-point state, so such exceptions should be
cleared; this patch adds the required clearing. (Discovered while
working on bug 16068.)
Tested for x86_64 and x86.
[BZ #19181]
* sysdeps/i386/fpu/fesetenv.c (__fesetenv): Clear already-raised
SSE exceptions when argument is FE_DFL_ENV or FE_NOMASK_ENV.
* sysdeps/x86_64/fpu/fesetenv.c (__fesetenv): Likewise.
* math/test-fenv-clear-main.c: New file.
* math/test-fenv-clear.c: Likewise.
* math/Makefile (tests): Add test-fenv-clear.
* sysdeps/x86/fpu/test-fenv-clear-sse.c: New file.
* sysdeps/x86/fpu/Makefile [$(subdir) = math] (tests): Add
test-fenv-clear-sse.
[$(subdir) = math] (CFLAGS-test-fenv-clear-sse.c): New variable.
MIPS16 atomics used __sync_* with GCC before 4.7, which as noted in
bug 17404 is missing the required barrier semantics for
atomic_exchange_rel. This patch removes the code in question as dead
now GCC before 4.7 is no longer supported for building glibc.
Sanity tested with builds for MIPS.
[BZ #17404]
* sysdeps/mips/atomic-machine.h
[__GNUC_PREREQ (4, 8) || (__mips16 && __GNUC_PREREQ (4, 7))]:
Change conditional to [__GNUC_PREREQ (4, 8) || __mips16].
[__mips16 && !__GNUC_PREREQ (4, 7)]: Remove conditional code.
The skip_lock_out_of_tbegin_retries adaptive parameter was
not being used correctly, nor as described. This prevents
a fallback for all users of the lock if a transient abort
occurs within the accepted number of retries.
[BZ #19174]
* sysdeps/powerpc/nptl/elide.h (__elide_lock): Fix usage of
.skip_lock_out_of_tbegin_retries.
* sysdeps/unix/sysv/linux/powerpc/elision-lock.c
(__lll_lock_elision): Likewise, and respect a value of
try_tbegin <= 0.
This patch implements a requirement of GCC 4.7 or later to build
glibc.
This was discussed in the thread starting at
<https://sourceware.org/ml/libc-alpha/2015-08/msg00851.html>.
Concerns were expressed by Mike and David. At
<https://sourceware.org/ml/libc-alpha/2015-10/msg00453.html> I have
provided a 14-patch series showing in outline the cleanups facilitated
by this version requirement, as requested by Mike (this patch is the
first in that series, with the addition of a NEWS entry). Given the
absence of further concerns or alternative proposals for criteria for
updates to this version requirement as requested in
<https://sourceware.org/ml/libc-alpha/2015-10/msg00065.html>, I am
interpreting this as "absence of sustained opposition" under Carlos's
definition at <https://sourceware.org/glibc/wiki/Consensus> and
proposing this patch for inclusion in glibc. I'd like to remind
people testing with 4.6 that if they move to testing with GCC 5 then
it will probably be about four years before they need to update the
compiler they use to test glibc again.
Although on the principles of time-based updates I think a move to
requiring binutils 2.23 would be reasonable, I'm not currently aware
of any cleanups that would facilitate so am not proposing that at this
time (but would expect to propose a move to requiring binutils 2.24 in
a year's time, as that brings features such as AVX512 support that
should allow some conditionals to be cleaned up). If someone thinks a
move to requiring 2.23 would help clean things up for their
architecture, please speak up. (And in general, I suspect there are
lots of architecture-specific configure tests that could be removed on
the basis of current GCC and binutils version requirements, given how
I've found architecture-independent tests obsolete on the basis of
version requirements going back 20 years.)
Tested for x86_64 and x86 (testsuite, and that installed shared
libraries are unchanged by the patch).
* configure.ac (libc_cv_compiler_ok): Require GCC 4.7 or later.
* configure: Regenerated.
* manual/install.texi (Tools for Compilation): Document
requirement for GCC 4.7 or later.
* INSTALL: Regenerated.
j1 and jn can underflow for small arguments, but fail to set errno
when underflowing to 0. This patch fixes them to set errno in that
case.
Tested for x86_64, x86, mips64 and powerpc.
[BZ #18611]
* sysdeps/ieee754/dbl-64/e_j1.c (__ieee754_j1): Set errno and
avoid excess range and precision on underflow.
* sysdeps/ieee754/dbl-64/e_jn.c (__ieee754_jn): Likewise.
* sysdeps/ieee754/flt-32/e_j1f.c (__ieee754_j1f): Likewise.
* sysdeps/ieee754/flt-32/e_jnf.c (__ieee754_jnf): Likewise.
* sysdeps/ieee754/ldbl-128/e_j1l.c (__ieee754_j1l): Set errno on
underflow.
* sysdeps/ieee754/ldbl-128/e_jnl.c (__ieee754_jnl): Likewise.
* sysdeps/ieee754/ldbl-128ibm/e_jnl.c (__ieee754_jnl): Likewise.
* sysdeps/ieee754/ldbl-96/e_j1l.c (__ieee754_j1l): Likewise.
* sysdeps/ieee754/ldbl-96/e_jnl.c (__ieee754_jnl): Likewise.
* math/auto-libm-test-in: Do not allow missing errno setting for
tests of j1 and jn.
* math/auto-libm-test-out: Regenerated.
The implementations of nearbyint functions using x87 floating point
(i386 all versions, x86_64 long double only) use the fclex
instruction, which clears any exceptions that were raised before the
function was called. These functions must not clear exceptions that
were raised before they were called.
This patch fixes these functions to save and restore the whole
floating-point environment (fnstenv / fldenv) as the way of avoiding
raising "inexact" (recall that there isn't an x87 instruction for
loading just the status word, so the whole environment has to be saved
and loaded instead - the code already saved and loaded the control
word, which is now obtained from the saved environment after this
patch, to disable traps on "inexact"). In the case of the long double
functions, any "invalid" exception from frndint (applied to a
signaling NaN) needs merging into the saved state; this issue doesn't
apply to the float and double functions because that exception would
have been raised when the argument is loaded, before the environment
is saved.
[BZ #15491]
* sysdeps/i386/fpu/s_nearbyint.S (__nearbyint): Save and restore
floating-point environment instead of clearing all exceptions.
* sysdeps/i386/fpu/s_nearbyintf.S (__nearbyintf): Likewise.
* sysdeps/i386/fpu/s_nearbyintl.S (__nearbyintl): Likewise,
merging in "invalid" exceptions from frndint.
* sysdeps/x86_64/fpu/s_nearbyintl.S (__nearbyintl): Likewise.
* math/test-nearbyint-except.c: New file.
* math/Makefile (tests): Add test-nearbyint-except.
Wilco fixed this tilegx bug in commit fe8c2b33ae ("Since we
now inline isinf, isnan and isfinite in math.h, replace uses
of __isinf_ns(l/f) with isinf, and remove the unused inlines
__isinf_ns(l/f), __isnan(f) and __finite(f).")
I verified that reverting math/s_cprojf.c back to using the
sysdeps/ieee754/flt-32/math_private.h version of __isinf_nsf()
instead of isinf() brings back the bug on tilegx.
My recent addition of more tests for j0 showed up that the ldbl-128
implementation of j0l produces spurious underflow exceptions for
arguments close to 0 (when the result is very close to 1). This patch
fixes this by just returning the argument in that case.
Tested for mips64 (where it fixes the recently-added tests that were
previously failing).
[BZ #19156]
* sysdeps/ieee754/ldbl-128/e_j0l.c (__ieee754_j0l): Return 1 for
arguments very close to 0.
The previous code used to evaluate the preprocessor token is_lock_free to
a variable before starting a transaction. This behavior can cause an
error if another thread got the lock (without using a transaction)
between the evaluation of the token and the beginning of the transaction.
This bug can be triggered with the following order of events:
1. The lock accessed by is_lock_free is free.
2. Thread T1 evaluates is_lock_free and stores into register R1 that the
lock is free.
3. Thread T2 acquires the same lock used in is_lock_free.
4. T1 begins the transaction, creating a memory barrier where is_lock_free
is false, but R1 is true.
5. T1 reads R1 and doesn't abort the transaction.
6. T1 calls ELIDE_UNLOCK, which reads false from is_lock_free and decides
to unlock a lock acquired by T2, leading to undefined behavior.
This patch delays the evaluation of is_lock_free to inside a transaction
by moving this part of the code to the macro ELIDE_LOCK.
[BZ #18743]
* sysdeps/powerpc/nptl/elide.h (__elide_lock): Move most of this
code to...
(ELIDE_LOCK): ...here.
(__get_new_count): New function with part of the code from
__elide_lock that updates the value of adapt_count after a
transaction abort.
(__elided_trylock): Moved this code to...
(ELIDE_TRYLOCK): ...here.
The previous (11th) version of the Hungarian spelling rules (released
in 1984) said that the separator had to be a dot, e.g. 10.35 meaning
10 o'clock 35 minutes. glibc correctly implements this.
The brand new (12th) version, in effect since September 1, 2015 adopts
to the common use of colon (especially in the digital world) and
allows to use either separator, without even expressing a preference.
For computer systems, using colons is way more typical and probably
easier to recognize. Dot is typically used in printed materials.
It also avoids an almost ambiguous situation where a space makes a
difference, e.g. "10.15-ig" means "until 10 o'clock 15 minutes"
whereas "10. 15-ig" means "until 15th of October". So I believe using
the colon as the separator is not only more frequent in the computer
world, but is also easier and quicker to recognize for the brain that
it's about hour:minute rather than month and day. And luckily it's now
equally correct according to the official rules.
11th edition: http://helyesiras.mta.hu/helyesiras/default/akh11
12th edition: http://helyesiras.mta.hu/helyesiras/default/akh12
In both editions it's the very last (299th and 300th, respectively) rule.
Microsoft also uses and recommends a colon since at least May 2011:
http://download.microsoft.com/download/e/6/1/e61266b2-d8b4-4fe0-a553-f01dc3976675/hun-hun-StyleGuide.pdf
The time format is different in common language and in the language of
IT. In common texts we usually do not abbreviate, so the full forms are
used: “7 óra 10 perckor csörgött a telefon”. However, the short format,
consisting of numerals only, can also be used. In this case a period
must be used between the two numbers and there must not be a space
between them: “találkozzunk 10.45-kor”.
However, in software mostly the short format is used, and the numbers
are separated by a colon. An obvious example is the clock in the bottom
right corner of your screen, thus 18:31.
Only i386 implements epoll_pwait in assembly code withot cancellation
support. All other architectures implement epoll_pwait in epoll_pwait.c
with
int epoll_pwait (int epfd, struct epoll_event *events,
int maxevents, int timeout,
const sigset_t *set)
{
return SYSCALL_CANCEL (epoll_pwait, epfd, events, maxevents,
timeout, set, _NSIG / 8);
}
Although there is no test for epoll_pwait in glibc, since SYSCALL_CANCEL
works on i386 and epoll_pwait.c works for other architectures, it is
safe to assume that epoll_pwait.c with SYSCALL_CANCEL also works on
i386.
[BZ #19137]
* sysdeps/unix/sysv/linux/i386/Makefile (CFLAGS-epoll_pwait.c):
Add -fomit-frame-pointer.
* sysdeps/unix/sysv/linux/i386/epoll_pwait.S: Remove file.
Honoring the LD_POINTER_GUARD environment variable in AT_SECURE mode
has security implications. This commit enables pointer guard
unconditionally, and the environment variable is now ignored.
[BZ #18928]
* sysdeps/generic/ldsodefs.h (struct rtld_global_ro): Remove
_dl_pointer_guard member.
* elf/rtld.c (_rtld_global_ro): Remove _dl_pointer_guard
initializer.
(security_init): Always set up pointer guard.
(process_envvars): Do not process LD_POINTER_GUARD.
The powerpc32 implementation of lround and lroundf can produce
spurious exceptions from adding 0.5 then converting to integer. This
includes "inexact" from the conversion to integer (not allowed for
integer arguments to these functions), and, for larger integer
arguments, "inexact", and "overflow" when rounding upward, from the
addition. In addition, "inexact" is not allowed together with
"invalid" and so inexact addition must be avoided when the integer
will be out of range of 32-bit long, whether or not the argument is an
integer.
This patch fixes these problems. As in the powerpc64 llround
implementation, a check is added for too-large arguments; in the
powerpc64 case that means arguments at least 2^52 in magnitude (so
that 0.5 cannot be added exactly), while in this case it means
arguments for which the result would overflow "long". In those cases
a suitable overflowing value is used for the integer conversion
without adding 0.5, while for smaller arguments it's tested whether
the argument is an integer (by adding and subtracting 2^52 to the
absolute value and comparing with the original absolute value) to
avoid adding 0.5 to integers and generating spurious "inexact".
This code is not used when the power5+ sysdeps directories are used,
as there's a separate power5+ version of these functions..
Tested for powerpc. This gets test-float (for a default powerpc32
hard-float build without any --with-cpu) back to the point where it
should pass once powerpc ulps are regenerated; test-double still needs
another problem with exceptions fixed to get back to that point (and I
haven't looked lately at what default powerpc64 results are like).
[BZ #19134]
* sysdeps/powerpc/powerpc32/fpu/s_lround.S (.LC1): New object.
(.LC2): Likewise.
(.LC3): Likewise.
(__lround): Do not add 0.5 to integer or out-of-range arguments.
_dl_tlsdesc_resolve_hold calls into a C function that clobbers r0,
but it assumes the original argument is still in r0 after the call.
This can cause crash in case of concurrent TLS access when TLSDESC
is in use (-mtls-dialect=gnu2).
Run into this while fixing BZ 18572.
Both r0 and r1 are saved/restored so the stack remains 8 byte aligned.
[BZ #19129]
* sysdeps/arm/dl-tlsdesc.S (_dl_tlsdesc_resolve_hold): Save and restore
r0 and r1.
Linker in binutils 2.26 and newer generate GOT references instead
PLT references when -z now is passed to linker. We need to extend
scripts/localplt.awk to allow PLT or GOT references.
[BZ #19007]
* scripts/localplt.awk: Also allow GOT references.
* sysdeps/unix/sysv/linux/i386/localplt.data: Mark
_Unwind_Find_FDE, calloc, memalign, realloc and __libc_memalign
with "+ REL R_386_GLOB_DAT".
* sysdeps/x86_64/localplt.data: Mark calloc, memalign, realloc
and __libc_memalign with "+ RELA R_X86_64_GLOB_DAT".
The powerpc32 implementations of llroundf and llround produce spurious
and missing exceptions (some arising from such exceptions from
conversions to long long, some present even when fctidz is used).
This patch fixes those problems in a similar way to the llrint /
llrintf fixes. The spurious exceptions in the fctidz case for large
arguments arise from a converted value that saturated as LLONG_MAX
being converted back to float or double (the conversion back being
inexact, but "inexact" must not be raised together with "invalid"),
and from the subtraction x - xrf also being inexact for sufficiently
large arguments (whether the saturation was to LLONG_MAX or
LLONG_MIN); those are fixed by returning early if the argument is
large enough that no rounding is needed.
This code is not used for --with-cpu=power4 builds (I suspect the code
used in that case may also produce spurious "inexact" exceptions, but
that's something to investigate later).
Tested for powerpc.
[BZ #19125]
* sysdeps/powerpc/powerpc32/fpu/s_llround.c: Include <limits.h>,
<math_private.h> and <stdint.h>.
(__llround): Avoid conversions to and from long long int, and
subtractions, where those might raise spurious exceptions.
* sysdeps/powerpc/powerpc32/fpu/s_llroundf.c: Include
<math_private.h> and <stdint.h>.
(__llroundf): Avoid conversions to and from long long int, and
subtractions, where those might raise spurious exceptions.
When x86-64 assmebler doesn't support AVX512, we should make
_dl_runtime_resolve_avx512/_dl_runtime_profile_avx512 as aliases of
_dl_runtime_resolve_avx/_dl_runtime_profile_avx. Tested on x86-64
using GCC 5.2 with binutils 20151008 and GCC 4.8 with binutils 20130219.
There are no differences in ld.so with binutils 20151008. There are no
unexpected failures with binutils 20130219 and 20151008.
[BZ #19124]
* sysdeps/x86_64/dl-trampoline.S [!HAVE_AVX512_ASM_SUPPORT]
(_dl_runtime_resolve_avx512): Make it a hidden alias of
_dl_runtime_resolve_avx.
(_dl_runtime_profile_avx512): Make it a hidden alias of
_dl_runtime_profile_avx.
The versions of llrint and llrintf for older powerpc32 processors
convert the results of __rint / __rintf to long long int, resulting in
spurious exceptions from such casts in certain cases. This patch
makes glibc work around the problems with the libgcc conversions when
the compiler used to build glibc doesn't use the fctidz instruction
for them.
Tested for powerpc.
[BZ #16422]
* sysdeps/powerpc/powerpc32/fpu/configure.ac (libc_cv_ppc_fctidz):
New configure test.
* sysdeps/powerpc/powerpc32/fpu/configure: Regenerated.
* config.h.in [_LIBC] (HAVE_PPC_FCTIDZ): New macro.
* sysdeps/powerpc/powerpc32/fpu/s_llrint.c: Include <limits.h>,
<math_private.h> and <stdint.h>.
(__llrint): Avoid conversions to long long int where those might
raise spurious exceptions.
* sysdeps/powerpc/powerpc32/fpu/s_llrintf.c: Include
<math_private.h> and <stdint.h>.
(__llrintf): Avoid conversions to long long int where those might
raise spurious exceptions.
Similar to the recent fix for MIPS, ARM is also missing correct
exceptions on overflow from llrint and llround functions because casts
from floating-point types to long long do not result in correct
exceptions on overflow. This patch enables the fix for this for ARM.
Tested for ARM.
[BZ #15470]
* sysdeps/arm/fix-fp-int-convert-overflow.h: New file.
For 32-bit MIPS and some other systems, various of the lrint, llrint,
lround, llround functions can be missing exceptions on overflow
because casts do not (in current GCC) result in the proper
exceptions. In the MIPS case there are two problems here: MIPS I code
generation uses an assembler macro that doesn't raise exceptions,
while the libgcc conversions of floating-point values to long long
also do not raise "invalid" on all overflow cases (and can raise
spurious "inexact").
This patch adds support in the generic code (only the functions for
which this problem has actually been seen) for forcing the "invalid"
exception in the problem cases, and enables that support for the
affected MIPS cases.
Tested for MIPS; also tested for x86_64 and x86 that installed
stripped shared libraries are unchanged by this patch.
[BZ #16399]
* sysdeps/generic/fix-fp-int-convert-overflow.h: New file.
* sysdeps/ieee754/dbl-64/s_llrint.c: Include <fenv.h>, <limits.h>
and <fix-fp-int-convert-overflow.h>.
(__llrint) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/dbl-64/s_llround.c: Include <fenv.h>, <limits.h>
and <fix-fp-int-convert-overflow.h>.
(__llround) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/dbl-64/s_lrint.c: Include
<fix-fp-int-convert-overflow.h>.
(__lrint) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/dbl-64/s_lround.c: Include
<fix-fp-int-convert-overflow.h>.
(__lround) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/flt-32/s_llrintf.c: Include <fenv.h>, <limits.h>
and <fix-fp-int-convert-overflow.h>.
(__llrintf) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/flt-32/s_llroundf.c: Include <fenv.h>,
<limits.h> and <fix-fp-int-convert-overflow.h>.
(__llroundf) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/flt-32/s_lrintf.c: Include <fenv.h>, <limits.h>
and <fix-fp-int-convert-overflow.h>.
(__lrintf) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/ieee754/flt-32/s_lroundf.c: Include <fenv.h>, <limits.h>
and <fix-fp-int-convert-overflow.h>.
(__lroundf) [FE_INVALID]: Force FE_INVALID exception as needed if
FIX_DBL_LLONG_CONVERT_OVERFLOW.
* sysdeps/mips/mips32/fpu/fix-fp-int-convert-overflow.h: New file.
The dbl-64 implementation of lrint produces incorrect results for some
arguments with 64-bit long because a 32-bit (unsigned) low part of the
mantissa is shifted left, losing high bits in the process. This patch
fixes this by casting to long int before shifting, as in lround (as
this case only applies for 64-bit long, there are no issues with
sign-extension).
Tested for mips64 (n64).
[BZ #19095]
* sysdeps/ieee754/dbl-64/s_lrint.c (__lrint): Cast low part of
mantissa to long int before shifting left.
The dbl-64, ldbl-96 and ldbl-128 implementations of lrint and llrint
fail to produce "invalid" exceptions in cases where the rounded result
overflows the target type, but truncating the floating-point argument
to the next integer towards zero does not overflow it (so in
particular casts do not produce such exceptions). (This issue cannot
arise for float, or for double with 64-bit target type, or for ldbl-96
with 64-bit target type and negative arguments, because of
insufficient precision in the floating-point type for arguments with
the relevant property to exist. It also obviously cannot arise in
FE_TOWARDZERO mode.)
This patch fixes these problems by inserting checks for the special
cases that can occur in each implementation, and explicitly raising
FE_INVALID (and avoiding the cast if it might raise spurious
FE_INEXACT, while raising FE_INEXACT explicitly in the cases where it
is needed; unlike lround and llround, FE_INEXACT is required, not
optional, for these functions for a within-range inexact result).
The fixes are conditional on FE_INVALID or FE_INEXACT being defined.
If any future architecture supports one but not both of those
exceptions, the code will fail to compile and need fixing to handle
that case (this seemed better than conditioning on both macros being
defined, resulting in code that would compile but quietly miss
exceptions on such a system).
Tested for x86_64, x86 and mips64. Tested the ldbl-96 changes (only
relevant for ia64, it appears) on x86_64 by removing the x86_64
versions of lrintl / llrintl.
[BZ #19094]
* sysdeps/ieee754/dbl-64/s_lrint.c: Include <fenv.h> and
<limits.h>.
(__lrint) [FE_INVALID || FE_INEXACT]: Force FE_INVALID exception
when result overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-128/s_llrintl.c: Include <fenv.h> and
<limits.h>.
(__llrintl) [FE_INVALID || FE_INEXACT]: Force FE_INVALID exception
when result overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-128/s_lrintl.c: Include <fenv.h> and
<limits.h>.
(__lrintl) [FE_INVALID || FE_INEXACT]: Force FE_INVALID exception
when result overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-96/s_llrintl.c: Include <fenv.h> and
<limits.h>.
(__llrintl) [FE_INVALID || FE_INEXACT]: Force FE_INVALID exception
when result overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-96/s_lrintl.c: Include <fenv.h> and
<limits.h>.
(__lrintl) [FE_INVALID || FE_INEXACT]: Force FE_INVALID exception
when result overflows but exception would not result from cast.
* math/libm-test.inc (lrint_test_data): Add more tests.
(llrint_test_data): Likewise.
The optimization introduced in commit
f13c2a8dff, causes regressions in
sorting for languages that have digraphs that change sort order, like
cs_CZ which sorts ch between h and i.
My analysis shows the fast-forwarding optimization in STRCOLL advances
through a digraph while possibly stopping in the middle which results
in a subsequent skipping of the digraph and incorrect sorting. The
optimization is incorrect as implemented and because of that I'm
removing it for 2.23, and I will also commit this fix for 2.22 where
it was originally introduced.
This patch reverts the optimization, introduces a new bug-strcoll2.c
regression test that tests both cs_CZ.UTF-8 and da_DK.ISO-8859-1 and
ensures they sort one digraph each correctly. The optimization can't be
applied without regressing this test.
Checked on x86_64, bug-strcoll2.c fails without this patch and passes
after. This will also get a fix on 2.22 which has the same bug.
The dbl-64, ldbl-96 and ldbl-128 implementations of lround and llround
fail to produce "invalid" exceptions in cases where the rounded result
overflows the target type, but truncating the floating-point argument
to the next integer towards zero does not overflow it (so in
particular casts do not produce such exceptions). (This issue cannot
arise for float, or for double with 64-bit target type, or for ldbl-96
with 64-bit target type and negative arguments, because of
insufficient precision in the floating-point type for arguments with
the relevant property to exist.)
This patch fixes these problems by inserting checks for the special
cases that can occur in each implementation, and explicitly raising
FE_INVALID (and avoiding the cast if it might raise spurious
FE_INEXACT).
Tested for x86_64, x86 and mips64.
[BZ #19088]
* sysdeps/ieee754/dbl-64/s_lround.c: Include <fenv.h> and
<limits.h>.
(__lround) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* sysdeps/ieee754/dbl-64/wordsize-64/s_lround.c: Include <fenv.h>
and <limits.h>.
(__lround) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-128/s_llroundl.c: Include <fenv.h> and
<limits.h>.
(__llroundl) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-128/s_lroundl.c: Include <fenv.h> and
<limits.h>.
(__lroundl) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-96/s_llroundl.c: Include <fenv.h> and
<limits.h>.
(__llroundl) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* sysdeps/ieee754/ldbl-96/s_lroundl.c: Include <fenv.h> and
<limits.h>.
(__lroundl) [FE_INVALID]: Force FE_INVALID exception when result
overflows but exception would not result from cast.
* math/libm-test.inc (lround_test_data): Add more tests.
(llround_test_data): Likewise.
The ldbl-128 implementations of lrintl and lroundl miss "invalid"
exceptions on systems with 32-bit long for arguments that overflow
long but have exponent below 48. This patch fixes this by rearranging
the sequence of tests in the code so the exponent < 48 case is only
used for exponents that don't overflow long.
Tested for mips64 (n32 and n64).
[BZ #19085]
* sysdeps/ieee754/ldbl-128/s_lrintl.c (__lrintl): Move test for
exponent below 48 inside case for non-overflowing exponent.
* sysdeps/ieee754/ldbl-128/s_lroundl.c (__lroundl): Likewise.
The implementation of lround in dbl-64/wordsize-64 as an alias or
wrapper for llround is always incorrect when long is not 64-bit,
because it misses required exceptions in overflow cases, as shown by
my recently added tests. This patch removes that alias / wrapper in
the non-LP64 case, together with the REGISTER_CAST_INT32_TO_INT64
macro, restoring the previous version of lround for dbl-64/wordsize-64
(newly conditioned on !_LP64).
Tested for x86_64, and for mips64 with use of dbl-64/wordsize-64
enabled.
[BZ #19079]
* sysdeps/ieee754/dbl-64/wordsize-64/s_lround.c: Restore previous
file, conditioned on [!_LP64].
* sysdeps/ieee754/dbl-64/wordsize-64/s_llround.c
[!_LP64] (__lround): Do not define as function or alias.
[!_LP64] (lround): Likewise.
[!_LP64] (__lroundl): Likewise.
[!_LP64] (lroundl): Likewise.
* sysdeps/tile/sysdep.h (REGISTER_CAST_INT32_TO_INT64): Remove
macro.
* sysdeps/x86_64/x32/sysdep.h (REGISTER_CAST_INT32_TO_INT64):
Likewise.
The ldbl-128ibm expl wrapper checks the argument to determine when to
call __kernel_standard_l, thereby overriding overflowing results from
__ieee754_expl that could otherwise (given appropriately patched
libgcc) be correct for the rounding mode. This patch changes it to
check the result of __ieee754_expl instead, as other versions of this
wrapper do.
Tested for powerpc.
[BZ #19078]
* sysdeps/ieee754/ldbl-128ibm/w_expl.c (o_thres): Remove variable.
(u_thres): Likewise.
(__expl): Determine whether to call __kernel_standard_l based on
value of result, not argument.
The ldbl-128ibm implementation of logl produces a zero with the wrong
sign for logl (1) in FE_DOWNWARD mode. This patch makes it explicitly
return 0.0L in that case, as in e.g. the ldbl-128 implementation.
Tested for powerpc.
[BZ #19077]
* sysdeps/ieee754/ldbl-128ibm/e_logl.c (__ieee754_logl): Return
0.0L for argument 1.0L.
The ldbl-128ibm implementation of log1pl produces an infinity with the
wrong sign for log1pl (-1) in FE_DOWNWARD mode. This patch fixes this
by changing a division (-1.0L / (x - x)) (incorrect in FE_DOWNWARD
mode) to (-1.0L / 0.0L) (correct in all rounding modes).
Tested for powerpc.
[BZ #19076]
* sysdeps/ieee754/ldbl-128ibm/s_log1pl.c (__log1pl): Divide by
constant 0.0L when computing infinite result.
