about summary refs log tree commit diff
path: root/sysdeps
Commit message (Collapse)AuthorAgeFilesLines
* Add a generic scalb implementationTulio Magno Quites Machado Filho2020-02-145-3/+8
| | | | | | | | | | | | | | | | | | | | | | | This is a preparatory patch to enable building a _Float128 variant to ease reuse when building a _Float128 variant to alias this long double only symbol. Notably, stubs are added where missing to the native _Float128 sysdep dir to prevent building these newly templated variants created inside the build directories. Also noteworthy are the changes around LIBM_SVID_COMPAT. These changes are not intuitive. The templated version is only enabled when !LIBM_SVID_COMPAT, and the compat version is predicated entirely on LIBM_SVID_COMPAT. Thus, exactly one is stubbed out entirely when building. The nldbl scalb compat files are updated to account for this. Likewise, fixup the reuse of m68k's e_scalb{f,l}.c to include it's override of e_scalb.c. Otherwise, the search path finds the templated copy in the build directory. This could be futher simplified by providing an overridden template, but I lack the hardware to verify.
* Adjust thresholds in Bessel function implementations (bug 14469).Joseph Myers2020-02-144-8/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | A recent discussion in bug 14469 notes that a threshold in float Bessel function implementations, used to determine when to use a simpler implementation approach, results in substantially inaccurate results. As I discussed in <https://sourceware.org/ml/libc-alpha/2013-03/msg00345.html>, a heuristic argument suggests 2^(S+P) as the right order of magnitude for a suitable threshold, where S is the number of significand bits in the floating-point type and P is the number of significant bits in the representation of the floating-point type, and the float and ldbl-96 implementations use thresholds that are too small. Some threshold does need using, there or elsewhere in the implementation, to avoid spurious underflow and overflow for large arguments. This patch sets the thresholds in the affected implementations to more heuristically justifiable values. Results will still be inaccurate close to zeroes of the functions (thus this patch does *not* fix any of the bugs for Bessel function inaccuracy); fixing that would require a different implementation approach, likely along the lines described in <http://www.cl.cam.ac.uk/~jrh13/papers/bessel.ps.gz>. So the justification for a change such as this would be statistical rather than based on particular tests that had excessive errors and no longer do so (no doubt such tests could be found, but would probably be too fragile to add to the testsuite, as liable to give large errors again from very small implementation changes or even from compiler changes). See <https://sourceware.org/ml/libc-alpha/2020-02/msg00638.html> for such statistics of the resulting improvements for float functions. Tested (glibc testsuite) for x86_64.
* Use gcc -finput-charset=ascii for check-installed-headers.Stefan Liebler2020-02-134-4/+4
| | | | | | | | | | | | | | | | | | | | A non-ascii character in the installed headers leads now to: error: failure to convert ascii to UTF-8 Such a finding in s390 specific fenv.h leads to fails in GCC testsuite. See glibc commit 08aea89ef67c5780ae734073494df0a451bce20f. Adding this gcc option also to our tests was proposed by Florian Weimer. This change also found a hit in resource.h where now "microseconds" is used. I've adjusted all the resource.h files. I've used the following command to check for further hits in headers. LC_ALL=C find -name "*.h" -exec grep -PHn "[\x80-\xFF]" {} \; Tested on s390x and x86_64. Reviewed-by: Zack Weinberg <zackw@panix.com>
* math/test-sinl-pseudo: Use stack protector only if availableFlorian Weimer2020-02-131-0/+2
| | | | | This fixes commit 9333498794cde1d5cca518bad ("Avoid ldbl-96 stack corruption from range reduction of pseudo-zero (bug 25487).").
* alpha: Fix static gettimeofday symbolAdhemerval Zanella2020-02-131-5/+2
| | | | | | | | By undef strong_alias on alpha implementation, the default_symbol_version macro becomes an empty macro on static build. It fixes the issue introduced at c953219420. Checked on alpha-linux-gnu with a 'make check run-built-tests=no'.
* nss: Add function types and NSS_DECLARE_MODULE_FUNCTIONS macro to <nss.h>Florian Weimer2020-02-131-15/+3
| | | | | | | This macro allows to add type safety to the implementation of NSS service modules. Reviewed-by: DJ Delorie <dj@redhat.com>
* Avoid ldbl-96 stack corruption from range reduction of pseudo-zero (bug 25487).Joseph Myers2020-02-123-1/+55
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Bug 25487 reports stack corruption in ldbl-96 sinl on a pseudo-zero argument (an representation where all the significand bits, including the explicit high bit, are zero, but the exponent is not zero, which is not a valid representation for the long double type). Although this is not a valid long double representation, existing practice in this area (see bug 4586, originally marked invalid but subsequently fixed) is that we still seek to avoid invalid memory accesses as a result, in case of programs that treat arbitrary binary data as long double representations, although the invalid representations of the ldbl-96 format do not need to be consistently handled the same as any particular valid representation. This patch makes the range reduction detect pseudo-zero and unnormal representations that would otherwise go to __kernel_rem_pio2, and returns a NaN for them instead of continuing with the range reduction process. (Pseudo-zero and unnormal representations whose unbiased exponent is less than -1 have already been safely returned from the function before this point without going through the rest of range reduction.) Pseudo-zero representations would previously result in the value passed to __kernel_rem_pio2 being all-zero, which is definitely unsafe; unnormal representations would previously result in a value passed whose high bit is zero, which might well be unsafe since that is not a form of input expected by __kernel_rem_pio2. Tested for x86_64.
* mips: Fix argument passing for inlined syscalls on Linux [BZ #25523]WANG Xuerui2020-02-123-52/+104
| | | | | | | | | | | | | | According to [gcc documentation][1], temporary variables must be used for the desired content to not be call-clobbered. Fix the Linux inline syscall templates by adding temporary variables, much like what x86 did before (commit 381a0c26d73e0f074c962e0ab53b99a6c327066d). Tested with gcc 9.2.0, both cross-compiled and natively on Loongson 3A4000. [1]: https://gcc.gnu.org/onlinedocs/gcc/Local-Register-Variables.html
* mips: Use 'long int' and 'long long int' in linux syscall codeWANG Xuerui2020-02-1214-220/+230
| | | | Style fixes only, no functional change.
* alpha: Use generic gettimeofday implementationAdhemerval Zanella2020-02-122-2/+8
| | | | | | | | | It makes alpha no longer reports information about a system-wide time zone and moves the version logic on the alpha implementation. Checked on a build and check-abi for alpha-linux-gnu. Reviewed-by: Lukasz Majewski <lukma@denx.de>
* Rename RWF_WRITE_LIFE_NOT_SET to RWH_WRITE_LIFE_NOT_SET following Linux 5.5.Joseph Myers2020-02-121-1/+2
| | | | | | | | Linux 5.5 renames RWF_WRITE_LIFE_NOT_SET to RWH_WRITE_LIFE_NOT_SET, with the old name kept as an alias. This patch makes the corresponding change in glibc. Tested for x86_64.
* S390: Fix non-ascii character in fenv.h.Stefan Liebler2020-02-121-2/+2
| | | | | | | | | | | | | | | | | The comment "isn't" contained a non-ascii character which leads to an error if compiled with -finput-charset=ascii: error: failure to convert ascii to UTF-8 This is observable in GCC testsuite: FAIL: 17_intro/headers/c++1998/charset.cc (test for excess errors) FAIL: 17_intro/headers/c++2011/charset.cc (test for excess errors) FAIL: 17_intro/headers/c++2014/charset.cc (test for excess errors) FAIL: 17_intro/headers/c++2017/charset.cc (test for excess errors) FAIL: 17_intro/headers/c++2020/charset.cc (test for excess errors) Also rewrite the comment above. Reported-by: Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
* Linux: Emulate fchmodat with AT_SYMLINK_NOFOLLOW using O_PATH [BZ #14578]Florian Weimer2020-02-121-10/+47
| | | | | /proc/self/fd files are special and chmod on O_PATH descriptors in that directory operates on the symbolic link itself (like lchmod).
* io: Implement lchmod using fchmodat [BZ #14578]Florian Weimer2020-02-122-0/+2
|
* htl C11 threads: Avoid pthread_ symbols visibility in static librarySamuel Thibault2020-02-1035-38/+34
|
* hurd: Add __pthread_spin_wait and use itSamuel Thibault2020-02-1117-24/+33
| | | | | | | | | 900778283ac3 ("htl: make pthread_spin_lock really spin") made pthread_spin_lock really spin and not block, but the current users of __pthread_spin_lock were assuming that it blocks, i.e. they use it as a lightweight mutex fitting in just one int. __pthread_spin_wait provides that support back.
* ldbl-128ibm-compat: set PRINTF_CHK flag in {,v}sprintf_chkPaul E. Murphy2020-02-102-0/+8
| | | | | | | This should be unconditionally set to match the common implementation, and fixes multiple test failures related to sprintf. Reviewed-by: Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
* nptl: update default pthread-offsets.hAndreas Schwab2020-02-101-16/+6
| | | | | Remove unused definitions, and correct __PTHREAD_RWLOCK_FLAGS_OFFSET for __WORDSIZE == 64.
* nptl: add missing pthread-offsets.hAndreas Schwab2020-02-105-0/+25
| | | | | All architectures using their own definition of struct __pthread_rwlock_arch_t need to provide their own pthread-offsets.h.
* pthread: Move some join tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-107-0/+516
| | | | So they can be checked with htl too.
* pthread: Move most barrier tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-106-0/+531
| | | | So they can be checked with htl too.
* htl: Fix barrier_wait with one threadSamuel Thibault2020-02-101-1/+3
|
* pthread: Move most sem tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-1014-0/+1205
| | | | So they can be checked with htl too.
* htl: Make sem_wait/sem_timedwait interruptibleSamuel Thibault2020-02-105-9/+37
|
* htl: Make sem_open return ENOSYSSamuel Thibault2020-02-101-1/+1
| | | | instead of EOPNOTSUPP, which is for sockets.
* htl: Add support for semaphore maximum valueSamuel Thibault2020-02-102-0/+10
|
* pthread: Move key tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-105-0/+497
| | | | So they can be checked with htl too.
* hurd: Make nanosleep a cancellation pointSamuel Thibault2020-02-101-0/+5
|
* htl: Add support for libc cancellation pointsSamuel Thibault2020-02-102-2/+19
|
* htl: clean __pthread_get_cleanup_stack hidden protoSamuel Thibault2020-02-101-1/+1
|
* htl: XFAIL rwlock tests which need pshared supportSamuel Thibault2020-02-091-0/+2
|
* pthread: Move some rwlock tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-098-0/+1106
| | | | So they can be checked with htl too.
* pthread: Move most once tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-097-0/+525
| | | | So they can be checked with htl too.
* htl: support cancellation during pthread_onceSamuel Thibault2020-02-091-0/+10
|
* pthread: Move most cond tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-0930-0/+3789
| | | | So they can be checked with htl too.
* htl: make pthread_cond_destroy return EBUSY on waitersSamuel Thibault2020-02-091-1/+8
|
* htl: Report missing mutex lock on pthread_cond_*waitSamuel Thibault2020-02-092-0/+55
|
* pthread: Move some attr tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-092-0/+306
| | | | So they can be checked with htl too.
* htl: Fix default guard sizeSamuel Thibault2020-02-092-1/+5
| | | | | When it is not hardcoded by the architecture with PAGESIZE, we need to use the dynamic values from __vm_page_size.
* pthread: Move most mutex tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-099-1/+1468
| | | | | | So they can be checked with htl too. XFAIL tst-mutex4, for which support is still missing in htl.
* pthread: Move spin tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-095-0/+376
| | | | So they can be checked with htl too.
* htl: make pthread_spin_lock really spinSamuel Thibault2020-02-091-1/+1
| | | | | __spin_lock would actually use gsync_wait to block, which is not what pthread_spin_lock is about.
* htl: Avoid check-installed-headers looking at inlinesSamuel Thibault2020-02-091-2/+4
|
* htl: Do not put spin_lock inlines in public headersSamuel Thibault2020-02-092-36/+36
| | | | | | They were not getting used anyway. Also do not make libsupport use them, it would make tests using it have to be made to link against libmachuser for gsync_wait.
* pthread: Move basic tests from nptl to sysdeps/pthreadSamuel Thibault2020-02-098-1/+675
| | | | So they can be checked with htl too.
* htl: Fix calling pthread_exit in the child of a forkSamuel Thibault2020-02-091-0/+9
| | | | | We need to reset the threads counter, otherwise pthread_exit() would not call exit(0).
* x86: Remove <bits/select.h> and use the generic versionFlorian Weimer2020-02-091-63/+0
| | | | | | Particularly on CPUs without ERMS, the string instructions are slow, so it is unclear whether this architecture-specific implementation is in fact an optimization.
* C11 threads: Move implementation to sysdeps/pthreadSamuel Thibault2020-02-0947-64/+1633
| | | | | | | so it gets shared by nptl and htl. Also add htl versions of thrd_current and thrd_yield. Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
* htl: Add C11 threads types definitionsSamuel Thibault2020-02-091-0/+13
|
* nptl: Move nptl-specific types to separate headerSamuel Thibault2020-02-092-8/+15
| | | | Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-11-08 09:29:14 +0000