gen-auto-libm-tests presently allows but does not require underflow
exceptions for results with magnitude in the range (greatest
subnormal, least normal].
In some cases, the magnitude of the exact result is very slightly
above the least normal, but rounding in the implementation results in
it effectively computing an infinite-precision result that is slightly
below the least normal, so raising an underflow exception. This is in
accordance with the documented accuracy goals, but results in
testsuite failures.
This patch changes the logic to allow underflows when the mathematical
result is up to 0.5ulp above the least normal (so in any case where
the round-to-nearest result is the least normal). Ideally underflows
in all these cases would be accepted only when an underflow with the
actual result is consistent with the rounding mode (in FE_TOWARDZERO
mode, a return value of the least normal implies that the
infinite-precision result did not underflow so there should be no
underflow exception, for example), so as to match the documented goals
more precisely - whereas at present the tests for exceptions are
completely independent of the tests of the returned values. (The same
applies to overflow exceptions as well - they too should be checked
for consistency with the result, as in FE_TOWARDZERO mode a result
1ulp below the largest finite value should be inconsistent with an
overflow exception and cause a failure with overflow rather than
simply being considered a 1ulp error when overflow is expected.) But
the present patch at least deals with the cases causing spurious
failures so that (a) certain existing tests no longer need to be
marked as having spurious exceptions (such markings in
auto-libm-test-in end up applying to more cases than just those they
are needed for) and (b) log1p can be tested in all rounding modes
without introducing more such failures. This patch duly moves tests
of log1p to ALL_RM_TEST.
Tested x86_64 and x86 and ulps updated accordingly.
[BZ #16357]
[BZ #16599]
* math/gen-auto-libm-tests.c (fp_format_desc): Add field
min_plus_half.
(fp_formats): Update initializers.
(init_fp_formats): Initialize new field.
(output_for_one_input_case): Allow underflow for results up to
min_plus_half.
* math/libm-test.inc (log1p_test): Use ALL_RM_TEST.
* math/auto-libm-test-in: Don't mark some underflows from asin and
atanh as spurious.
* math/auto-libm-test-out: Regenerated.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
My recent exp patch introduced warnings about implicit __isinf
declarations in exp because e_exp.c didn't include <math.h>. This
patch fixes this. Because <math.h> can't be included after
<math_private.h> (because of macro definitions of __nan*), it was
necessary to put an include in sysdeps/x86_64/fpu/multiarch/e_exp.c as
well.
Tested x86_64.
* sysdeps/ieee754/dbl-64/e_exp.c: Include <math.h>.
* sysdeps/x86_64/fpu/multiarch/e_exp.c
[HAVE_FMA4_SUPPORT || HAVE_AVX_SUPPORT]: Likewise.
An application that erroneously tries to repeatedly dlopen("a.out", ...)
may hit assertion failure:
Inconsistency detected by ld.so: dl-tls.c: 474: _dl_allocate_tls_init:
Assertion `listp != ((void *)0)' failed!
dlopen() actually fails with "./a.out: cannot dynamically load executable",
but it does so after incrementing dl_tls_max_dtv_idx.
Once we run out of TLS_SLOTINFO_SURPLUS (62), we exit with above assertion
failure.
2014-03-24 Paul Pluzhnikov <ppluzhnikov@google.com>
[BZ #16634]
* elf/dl-load.c (open_verify): Add mode parameter.
Error early when ET_EXEC and mode does not have __RTLD_OPENEXEC.
(open_path): Change from boolean 'secure' to complete flag 'mode'
(_dl_map_object): Adjust.
* elf/Makefile (tests): Add tst-dlopen-aout.
* elf/tst-dlopen-aout.c: New test.
This fixes a bug in the way the results from __nscd_getai are collected:
for every returned result a new entry is first added to the
gaih_addrtuple list, but if that result doesn't match the request this
entry remains uninitialized. So for this non-matching result an extra
result with uninitialized content is returned.
To reproduce (with nscd running):
$ getent ahostsv4 localhost
127.0.0.1 STREAM localhost
127.0.0.1 DGRAM
127.0.0.1 RAW
(null) STREAM
(null) DGRAM
(null) RAW
To reproduce:
# ip li add name dummy0 type dummy
# site_id=$(head -c6 /dev/urandom | od -tx2 -An | tr ' ' ':')
# for ((i = 0; i < 65536; i++)) do
> ip ad ad $(printf fd80$site_id::%04x $i)/128 dev dummy0
> done
# (ulimit -s 900; getent ahosts localhost)
# ip li de dummy0
The dbl-64 version of exp needs round-to-nearest mode for its internal
computations, but that has the consequence of inappropriate
overflowing and underflowing results in other rounding modes. This
patch fixes this by recomputing the relevant results in cases where
the round-to-nearest result overflows to infinity or underflows to
zero (most of the diffs are actually just consequent reindentation).
Tests are enabled in all rounding modes for complex functions using
exp - but not for cexp because it turns out there are bugs causing
spurious underflows for cexp for some tests, which will need to be
fixed separately (I suspect ccos ccosh csin csinh ctan ctanh have
similar bugs, just not shown by the present set of test inputs).
Tested x86_64 and x86 and ulps updated accordingly.
[BZ #16284]
* sysdeps/ieee754/dbl-64/e_exp.c (__ieee754_exp): Use original
rounding mode to recompute results that overflow to infinity or
underflow to zero.
* math/auto-libm-test-in: Don't mark tests as expected to fail for
bug 16284.
* math/auto-libm-test-out: Regenerated.
* math/libm-test.inc (ccos_test): Use ALL_RM_TEST.
(ccosh_test): Likewise.
(csin_test_data): Use plus_oflow.
(csin_test): Use ALL_RM_TEST.
(csinh_test_data): Use plus_oflow.
(csinh_test): Use ALL_RM_TEST.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
This patch fixes -Wundef warnings related to the _ABI* macros on MIPS.
GCC predefines only the _ABI* macro related to the ABI actually in
use, meaning that a conditional such as "#if _MIPS_SIM == _ABI64" is
true only for the ABI in question (all the macros are nonzero), but
produces a -Wundef warning for the other ABIs. The normal approach to
using these macros is to include <sgidefs.h>, which ensures that all
three _ABI* macros are defined rather than just one; this patch does
so in the places that caused warnings (the bulk of the warnings
arising from <bits/wordsize.h>). Tested that the warnings are fixed.
* sysdeps/mips/bits/wordsize.h: Include <sgidefs.h>.
* sysdeps/unix/sysv/linux/mips/getrlimit64.c: Likewise.
* sysdeps/unix/sysv/linux/mips/setrlimit64.c: Likewise.
According to ISO C Annex F, log (1) should be +0 in all rounding
modes, but some implementations in glibc wrongly return -0 in
round-downward mode (mapping to log1p (x - 1) is problematic because 1
- 1 is -0 in round-downward mode, and log1p (-0) is -0). This patch
fixes this. (It helps with some implementations of other functions
such as acosh, log2 and log10 that call out to log, but not enough to
enable all-rounding-modes testing for those functions without further
fixes to other implementations of them.)
Tested x86_64 and x86 and ulps updated accordingly, and did spot tests
for mips64 for the ldbl-128 fix, and i586 for the sysdeps/i386/fpu
implementations shadowed by those in sysdeps/i386/i686/fpu.
[BZ #16731]
* sysdeps/i386/fpu/e_log.S (__ieee754_log): Take absolute value
when x - 1 is zero.
* sysdeps/i386/fpu/e_logf.S (__ieee754_logf): Likewise.
* sysdeps/i386/fpu/e_logl.S (__ieee754_logl): Likewise.
* sysdeps/i386/i686/fpu/e_logl.S (__ieee754_logl): Likewise.
* sysdeps/ieee754/dbl-64/e_log.c (__ieee754_log): Return +0 when
argument is 1.
* sysdeps/ieee754/ldbl-128/e_logl.c (__ieee754_logl): Likewise.
* sysdeps/x86_64/fpu/e_logl.S: Take absolute value when x - 1 is
zero.
* math/libm-test.inc (log_test): Use ALL_RM_TEST.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
This patch add an optimized strpbrk for POWER7 by using a different
algorithm than default implementation: it constructs a table based on
the 'accept' argument and use this table to check for any occurance on
the input string. The idea is similar as x86_64 uses.
For PowerPC some tunings were added, such as unroll loops and memory
clear using VSX instructions.
This patch makes libm-test.inc tests of most functions use ALL_RM_TEST
unless there was some reason to defer that change for a particular
function.
I started out planning to defer the change for pow (bug 16315), cexp /
ccos / ccosh / csin / csinh (likely fallout from exp, bug 16284) and
cpow (exact expectations for signs of exact zero results not wanted).
Testing on x86_64 and x86 showed additional failures for acosh, cacos,
catan, catanh, clog, clog10, jn, log, log10, log1p, log2, tgamma, yn,
so making the change for those functions was deferred as well, pending
investigation to show which of these represent distinct bugs (some
such bugs may already be filed) and appropriate fixing / XFAILing.
Failures include wrong signs of zero results, errors slightly above
the 9ulp bound (in such cases it may make sense for functions to set
round-to-nearest internally to reduce error accumulation), large
errors and incorrect overflow/underflow for the rounding mode (with
consequent missing errno settings in some cases). It's possible some
could be issues with test expectations, though I didn't notice any
that were obviously like that (I added NO_TEST_INLINE for cases that
were failing for ildoubl on x86 and where it seemed reasonable for
them to fail for the fast-math inlines).
There may of course be failures on other architectures for functions
that didn't fail on x86_64 or x86, in which case the usual rule
applies: file a bug (preferably identifying the underlying problem
function, in cases where function A calls function B and a problem
with function B may present in the test results for function A) if not
already in Bugzilla then fix or XFAIL.
Tested x86_64 and x86 and ulps updated accordingly.
* math/libm-test.inc (asinh_test): Use ALL_RM_TEST.
(atan_test): Likewise.
(atanh_test_data): Use NO_TEST_INLINE for two tests.
(atanh_test): Use ALL_RM_TEST.
(atan2_test_data): Likewise.
(cabs_test): Likewise.
(cacosh_test): Likewise.
(carg_test): Likewise.
(casin_test): Likewise.
(casinh_test): Likewise.
(cbrt_test): Likewise.
(csqrt_test): Likewise.
(erf_test): Likewise.
(erfc_test): Likewise.
(pow10_test): Likewise.
(exp2_test): Likewise.
(hypot_test): Likewise.
(j0_test): Likewise.
(j1_test): Likewise.
(lgamma_test): Likewise.
(gamma_test): Likewise.
(sincos_test): Likewise.
(tanh_test): Likewise.
(y0_test): Likewise.
(y1_test): Likewise.
* sysdeps/i386/fpu/libm-test-ulps: Update.
* sysdeps/x86_64/fpu/libm-test-ulps: Likewise.
This patch add a optimized strcspn for POWER7 by using a different
algorithm than default implementation: it constructs a table based on
the 'accept' argument and use this table to check for any occurance
on the input string. The idea is similar as x86_64 uses.
For PowerPC some tunings were added, such as unroll loops and align
stack memory to table to 16 bytes (so VSX clean can ran without
alignment issues).
Continuing the move to using ALL_RM_TEST for tests in libm-test.inc,
this patch converts the tests of fdim, ldexp and scalb. fdim and
scalb are cases where tests could depend on the rounding mode though
none of the present test inputs do; ldexp is such a case where the
function is equivalent to scalbn (for binary floating point) and the
tests used were a subset of those for scalbn, so this patch makes
ldexp testing use the scalbn tests, as done for other cases of libm
function aliases.
Tested x86_64 and x86.
* math/libm-test.inc (fdim_test): Use ALL_RM_TEST.
(ldexp_test_data): Remove.
(ldexp_test): Move to after scalbn_test. Use ALL_RM_TEST with
scalbn_test_data.
(scalb_test): Use ALL_RM_TEST.
Reviewing (for all architectures, with a baseline kernel version of
2.6.32) the kernel support for features for which __ASSUME_* macros
would be affected by a move to 2.6.32 as minimum kernel version showed
up that __ASSUME_PREADV and __ASSUME_PWRITEV were wrongly defined for
MicroBlaze (despite the corresponding syscall table entries not being
wired up in the kernel) and Alpha for 2.6.30 and above (although the
support on Alpha was added in 2.6.33). This patch makes the
kernel-features.h files undefine those macros for appropriate
versions.
[BZ #16649]
* sysdeps/unix/sysv/linux/alpha/kernel-features.h
[__LINUX_KERNEL_VERSION < 0x020621] (__ASSUME_PREADV): Undefine.
[__LINUX_KERNEL_VERSION < 0x020621] (__ASSUME_PWRITEV): Likewise.
* sysdeps/unix/sysv/linux/microblaze/kernel-features.h
(__ASSUME_PREADV): Undefine.
(__ASSUME_PWRITEV): Likewise.
This patch adds the -std=c99 option when preprocessing the data files
from the conform testcases. It fixes an issue where the compiler may
split the 'macro bool' defition from stdbool.h-data in two lines and
thus breaking the conform script.
Continuing the move to systematically testing libm functions in all
rounding modes with ALL_RM_TEST, this patch converts the tests of
scalbn and scalbln to use that macro.
Those tests include cases of underflow and overflow, meaning the
expected results depend on the rounding mode. For convenience in
writing such tests manually, the patch adds the notation plus_oflow,
minus_oflow, plus_uflow and minus_uflow for overflowing / underflowing
results of each sign appropriate to the rounding mode being used, and
gen-libm-test.pl is made to substitute in the appropriate values. The
tests of underflow and overflow are extended to include negative
arguments to provide better coverage (otherwise minus_oflow and
minus_uflow wouldn't have been used at all).
(A subsequent patch will make ldexp use the scalbn tests, as those
functions are equivalent for binary floating point.)
Tested x86_64 and x86.
* math/gen-libm-test.pl (parse_args): Handle plus_oflow,
minus_oflow, plus_uflow and minus_uflow in expected results.
* math/libm-test.inc (scalbn_test_data): Add more tests of
negative arguments. Use plus_oflow, minus_oflow, plus_uflow and
minus_uflow.
(scalbn_test): Use ALL_RM_TEST.
(scalbln_test_data): Add more tests of negative arguments. Use
plus_oflow, minus_oflow, plus_uflow and minus_uflow.
(scalbln_test): Use ALL_RM_TEST.
This comment appears to have been copied from the ARM port where it
makes more sense.
2014-03-18 Will Newton <will.newton@linaro.org>
* sysdeps/unix/sysv/linux/aarch64/sysdep.h: Remove
inaccurate comment.
This patch arranges for rint and nearbyint to be tested by
libm-test.inc with the same inputs (previously each had some test
inputs the other didn't, although there was a lot of overlap as well),
and for nearbyint to be tested in all rounding modes where previously
it was only tested in round-to-nearest mode. The expected results are
the same for each function, except that rint is expected to have
"inexact" exceptions for non-integer input and nearbyint is expected
not to have those exceptions.
Tested x86_64 and x86.
* math/libm-test.inc (nearbyint_test_data): Include all tests used
for rint. Include results for all rounding modes.
(nearbyint_test): Use ALL_RM_TEST.
(rint_test_data): Include all tests used for nearbyint.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* sysdeps/generic/math_private.h: Check whether
HAVE_RM_CTX is defined with #ifdef rather
than #if.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* argp/argp-fmtstream.h: Check whether
__STRICT_ANSI__ is defined with #ifdef rather
than #if.
* argp/argp.h: Likewise.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* libio/genops.c: Check whether
_IO_JUMPS_OFFSET is defined with #ifdef rather
than #if.
* libio/libioP.h: Likewise.
* stdio-common/vfprintf.c: Likewise.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* sysdeps/generic/ldsodefs.h: Check whether
HP_SMALL_TIMING_AVAIL is defined with #ifdef rather
than #if.
ChangeLog:
2014-03-17 Will Newton <will.newton@linaro.org>
* nptl/sysdeps/pthread/pthread.h: Check
__PTHREAD_MUTEX_HAVE_ELISION is defined before testing
its value.
fdopen should only be allowed to change the offset in the file it
attaches to if it is setting O_APPEND. If O_APPEND is already set, it
should not change the state of the handle.
The ftell implementation was made conservative to ensure that
incorrectly cached offsets never affect it. However, this causes
problems for append mode when a file stream is rewound. Additionally,
the 'clever' trick of using stat to get position for append mode files
caused more problems than it solved and broke old behavior. I have
described the various problems that it caused and then finally the
solution.
For a and a+ mode files, rewinding the stream should result in ftell
returning 0 as the offset, but the stat() trick caused it to
(incorrectly) always return the end of file. Now I couldn't find
anything in POSIX that specifies the stream position after rewind()
for a file opened in 'a' mode, but for 'a+' mode it should be set to
0. For 'a' mode too, it probably makes sense to keep it set to 0 in
the interest of retaining old behavior.
The initial file position for append mode files is implementation
defined, so the implementation could either retain the current file
position or move the position to the end of file. The earlier ftell
implementation would move the offset to end of file for append-only
mode, but retain the old offset for a+ mode. It would also cache the
offset (this detail is important). My patch broke this and would set
the initial position to end of file for both append modes, thus
breaking old behavior. I was ignorant enough to write an incorrect
test case for it too.
The Change:
I have now brought back the behavior of seeking to end of file for
append-only streams, but with a slight difference. I don't cache the
offset though, since we would want ftell to query the current file
position through lseek while the stream is not active. Since the
offset is moved to the end of file, we can rely on the file position
reported by lseek and we don't need to resort to the stat() nonsense.
Finally, the cache is always reliable, except when there are unflished
writes in an append mode stream (i.e. both a and a+). In the latter
case, it is safe to just do an lseek to SEEK_END. The value can be
safely cached too, since the file handle is already active at this
point. Incidentally, this is the only state change we affect in the
file handle (apart from taking locks of course).
I have also updated the test case to correct my impression of the
initial file position for a+ streams to the initial behavior. I have
verified that this does not break any existing tests in the testsuite
and also passes with the new tests.
