Commit Graph

34159 Commits

Author SHA1 Message Date
Albert ARIBAUD (3ADEV)
97d91fccc8 Fix date typo in ChangeLog 2018-10-24 21:48:54 +02:00
Adhemerval Zanella
2a973ab7f1 posix: Add internal symbols for posix_spawn interface
This patch adds internal hidden definition for mostly of the posix_spawn
function so it can be used internally on both popen and system
implementations.

Checked on x86_64-linux-gnu.

	* include/spawn.h (__posix_spawn, posix_spawn_file_actions_addclose,
	__posix_spawn_file_actions_adddup2, __posix_spawn_file_actions_destroy,
	__posix_spawn_file_actions_init, __posix_spawnattr_init,
	__posix_spawnattr_destroy, __posix_spawnattr_setflags,
	__posix_spawnattr_setsigdefault, __posix_spawnattr_setsigmask): New
	prototype.
	* posix/spawn.c (__posix_spawn): Add libc_hidden_def.
	* posix/spawn_faction_addclose.c
	(__posix_spawn_file_actions_addclose): Add hidden definition.
	* posix/spawn_faction_adddup2.c
	(__posix_spawn_file_actions_adddup2): Likewise.
	* posix/spawn_faction_destroy.c
	(__posix_spawn_file_actions_destroy): Likewise.
	* posix/spawn_faction_init.c (__posix_spawn_file_actions_init):
	Likewise.
	* posix/spawnattr_destroy.c (__posix_spawnattr_destroy): Likewise.
	* posix/spawnattr_init.c (__posix_spawnattr_init): Likewise.
	* posix/spawnattr_setdefault.c (__posix_spawnattr_setsigdefault):
	Likewise.
	* posix/spawnattr_setflags.c (__posix_spawnattr_setflags): Likewise.
	* posix/spawnattr_setsigmask.c (__posix_spawnattr_setsigmask):
	Likewise.
2018-10-24 15:22:27 -03:00
Andreas Schwab
2954daf00b Add more checks for valid ld.so.cache file (bug 18093) 2018-10-24 16:34:02 +02:00
Albert ARIBAUD (3ADEV)
a27a4f4721 Y2038: provide size of default time_t for target architecture
To determine whether the default time_t interfaces are 32-bit
    and so need conversions, or are 64-bit and so are compatible
    with the internal 64-bit type without conversions, a macro
    giving the size of the  default time_t is also required.
    This macro is called __TIMESIZE.

    This macro can then be used instead of __WORDSIZE in msq-pad.h
    and shm-pad.h files, which in turn allows removing their x86
    variants, and in sem-pad.h files but keeping the x86 variant.

    This patch was tested by running 'make check' on branch master
    then applying this patch and running 'make check' again, and
    checking that both 'make check' yield identical results.
    This was done on x86_64-linux-gnu and i686-linux-gnu.

	* bits/timesize.h: New file.
	* stdlib/Makefile (headers): Add bits/timesize.h.
	* sysdeps/unix/sysv/linux/bits/msq-pad.h
	(__MSQ_PAD_AFTER_TIME): Use __TIMESIZE instead of __WORDSIZE.
	* sysdeps/unix/sysv/linux/bits/sem-pad.h
	(__SEM_PAD_AFTER_TIME): Likewise.
	* sysdeps/unix/sysv/linux/bits/shm-pad.h
	(__SHM_PAD_AFTER_TIME): Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/msq-pad.h
	(__MSQ_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/sem-pad.h
	(__SEM_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/shm-pad.h
	(__SHM_PAD_BEFORE_TIME, __SHM_PAD_BETWEEN_TIME_AND_SEGSZ): Likewise.
	* sysdeps/unix/sysv/linux/mips/bits/msq-pad.h
        (__MSQ_PAD_AFTER_TIME, __MSQ_PAD_BEFORE_TIME): Likewise.
        * sysdeps/unix/sysv/linux/powerpc/bits/msq-pad.h
        (__MSQ_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/sem-pad.h
	(__SEM_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/shm-pad.h
	(__SHM_PAD_BEFORE_TIME, __SHM_PAD_BETWEEN_TIME_AND_SEGSZ): Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/msq-pad.h
	(__MSQ_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/sem-pad.h
	(__SEM_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/shm-pad.h
	(__SHM_PAD_BEFORE_TIME): Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/msq-pad.h: Delete file.
	* sysdeps/unix/sysv/linux/x86/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/timesize.h: New file.
2018-10-24 11:37:09 +02:00
H.J. Lu
7cc65773f0 x86: Support RDTSCP for benchtests
RDTSCP waits until all previous instructions have executed and all
previous loads are globally visible before reading the counter.  RDTSC
doesn't wait until all previous instructions have been executed before
reading the counter.  All x86 processors since 2010 support RDTSCP
instruction.  This patch adds RDTSCP support to benchtests.

	* benchtests/Makefile (CPPFLAGS-nonlib): Add -DUSE_RDTSCP if
	USE_RDTSCP is defined.
	* sysdeps/x86/hp-timing.h (HP_TIMING_NOW): Use RDTSCP if
	USE_RDTSCP is defined.
2018-10-24 02:19:34 -07:00
Adhemerval Zanella
18ad0de651 Fix tst-preadvwritev2 build failure on HURD
Commit 7a16bdbb9f uses IOV_MAX, which is not defined on hurd.

Checked on a build for i686-gnu.

	* misc/tst-preadvwritev2-common.c (IOV_MAX): Define if not
	defined.
2018-10-23 14:57:02 -03:00
Adhemerval Zanella
c3d8dc45c9 x86: Fix Haswell strong flags (BZ#23709)
Th commit 'Disable TSX on some Haswell processors.' (2702856bf4) changed the
default flags for Haswell models.  Previously, new models were handled by the
default switch path, which assumed a Core i3/i5/i7 if AVX is available. After
the patch, Haswell models (0x3f, 0x3c, 0x45, 0x46) do not set the flags
Fast_Rep_String, Fast_Unaligned_Load, Fast_Unaligned_Copy, and
Prefer_PMINUB_for_stringop (only the TSX one).

This patch fixes it by disentangle the TSX flag handling from the memory
optimization ones.  The strstr case cited on patch now selects the
__strstr_sse2_unaligned as expected for the Haswell cpu.

Checked on x86_64-linux-gnu.

	[BZ #23709]
	* sysdeps/x86/cpu-features.c (init_cpu_features): Set TSX bits
	independently of other flags.
2018-10-23 14:57:02 -03:00
Florian Weimer
f1034472e2 time/tst-mktime2: Improve test error reporting 2018-10-23 11:25:05 +02:00
Andreas Schwab
367d7cc2cb Don't use PSEUDO_END for non-PSEUDO function 2018-10-23 09:41:43 +02:00
Joseph Myers
029ad711b8 Update kernel version in syscall-names.list to 4.19.
Linux 4.19 does not add any new syscalls (some existing ones are added
to more architectures); this patch updates the version number in
syscall-names.list to reflect that it's still current for 4.19.

Tested with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/syscall-names.list: Update kernel
	version to 4.19.
2018-10-22 23:26:37 +00:00
Joseph Myers
8190520f2a Use Linux 4.19 in build-many-glibcs.py.
* scripts/build-many-glibcs.py (Context.checkout): Default Linux
	version to 4.19.
2018-10-22 16:49:37 +00:00
Joseph Myers
be8ff03f92 Stop c32rtomb and mbrtoc32 aliasing wcrtomb and mbrtowc (bug 23793).
glibc does:

/* There should be no difference between the UTF-32 handling required
   by c32rtomb and the wchar_t handling which has long since been
   implemented in wcrtomb.  */
weak_alias (__wcrtomb, c32rtomb)

/* There should be no difference between the UTF-32 handling required
   by mbrtoc32 and the wchar_t handling which has long since been
   implemented in mbrtowc.  */
weak_alias (__mbrtowc, mbrtoc32)

The reasoning in those comments to justify those aliases is incorrect:
ISO C requires that, for the case of a NULL mbstate_t* being passed,
each function has its *own* internal static mbstate_t.  Thus a program
must be able to use both wcrtomb and c32rtomb at the same time with
each keeping its own separate state, and likewise for mbrtowc and
mbrtoc32.

This patch duly sets up separarate char32_t function that wrap the
wchar_t ones.  Note that the included test only covers the mbrtoc32 /
mbrtowc pair.  While I think the change made is logically correct for
c32rtomb / wcrtomb as well, I'm not sure we have a locale with a
suitable state-dependent multibyte encoding for testing that part of
the change.

Tested for x86_64.

	[BZ #23793]
	* wcsmbs/c32rtomb.c: New file.
	* wcsmbs/mbrtoc32.c: Likewise.
	* wcsmbs/tst-c32-state.c: Likewise.
	* wcsmbs/mbrtowc.c (mbrtoc32): Do not define as alias.
	* wcsmbs/wcrtomb.c (c32rtomb): Likewise.
	* wcsmbs/Makefile (routines): Add mbrtoc32 and c32rtomb.
	(tests): Add tst-c32-state.
	[$(run-built-tests) = yes] ($(objpfx)tst-c32-state.out): Depend on
	$(gen-locales).
2018-10-22 14:52:14 +00:00
H.J. Lu
2dd8e58cc5 x86: Don't include <x86intrin.h>
Use __builtin_ia32_rdtsc directly since including <x86intrin.h> makes
building glibc very slow.  On Intel Core i5-6260U, this patch reduces
x86-64 build time from 8 minutes 33 seconds to 3 minutes 48 seconds
with "make -j4" and GCC 8.2.1.

	* sysdeps/x86/hp-timing.h: Don't include <x86intrin.h>.
	(HP_TIMING_NOW): Replace _rdtsc with __builtin_ia32_rdtsc.
2018-10-21 00:37:29 -07:00
Joseph Myers
d0a7415979 Handle surrogate pairs in c16rtomb (bug 23794, DR#488, C2X).
The c16rtomb implementation has:

  // XXX The ISO C 11 spec I have does not say anything about handling
  // XXX surrogates in this interface.

The DR#488 resolution, as applied to C2X, requires surrogate pairs to
be handled here (so the first call returns 0 and stores the high
surrogate in the mbstate_t, while the second call combines the
surrogates, produces a multibyte character and returns the number of
bytes written).  This patch implements that.  (mbrtoc16 already
handled producing surrogates as output.)

Tested for x86_64.

	[BZ #23794]
	* wcsmbs/c16rtomb.c (c16rtomb): Save first character of surrogate
	pair and return 0 in that case, and use saved character to
	interpret following character.
	* wcsmbs/tst-c16-surrogate.c: New file.
	* wcsmbs/Makefile (tests): Add tst-c16-surrogate.c.
	[$(run-built-tests) = yes] ($(objpfx)tst-c16-surrogate.out):
	Depend on $(gen-locales)
2018-10-19 16:31:29 +00:00
Ilya Yu. Malakhov
f997b4be18 signal: Use correct type for si_band in siginfo_t [BZ #23562] 2018-10-19 18:06:32 +02:00
Florian Weimer
53b2bb8747 resource: Update struct rusage comments [BZ #23689] 2018-10-19 18:03:21 +02:00
David S. Miller
94a3a27f5d Add VDSO support to sparc.
* sysdeps/unix/sysv/linux/sparc/init-first.c: New file.
	* sysdeps/unix/sysv/linux/sparc/libc-vdso.h: New file.
	* sysdeps/unix/sysv/linux/sparc/Makefile: Add dl-vdso to
	sysdep_routines in subdir elf.
	* sysdeps/unix/sysv/linux/sparc/Versions: Add GLIBC_PRIVATE
	version for __vdso_clock_gettime.
	* sysdeps/unix/sysv/linux/sparc/sysdep.h (INTERNAL_VSYSCALL_CALL):
	Define.
	(HAVE_CLOCK_GETTIME_VSYSCALL): Define.
	(HAVE_GETTIMEOFDAY_VSYSCALL): Define.
2018-10-18 11:44:32 -07:00
David S. Miller
0a271c8f25 Regenerate sparc ulps.
* sysdeps/sparc/fpu/libm-test-ulps: Regenerated.
2018-10-18 11:43:56 -07:00
H.J. Lu
72771e5375 x86: Use _rdtsc intrinsic for HP_TIMING_NOW
Since _rdtsc intrinsic is supported in GCC 4.9, we can use it for
HP_TIMING_NOW.  This patch

1. Create x86 hp-timing.h to replace i686 and x86_64 hp-timing.h.
2. Move MINIMUM_ISA from init-arch.h to isa.h so that x86 hp-timing.h
can check minimum x86 ISA to decide if _rdtsc can be used.

NB: Checking if __i686__ isn't sufficient since __i686__ may not be
defined when building for i686 class processors.

	* sysdeps/i386/init-arch.h: Removed.
	* sysdeps/i386/i586/init-arch.h: Likewise.
	* sysdeps/i386/i686/init-arch.h: Likewise.
	* sysdeps/i386/i686/hp-timing.h: Likewise.
	* sysdeps/x86_64/hp-timing.h: Likewise.
	* sysdeps/i386/isa.h: New file.
	* sysdeps/i386/i586/isa.h: Likewise.
	* sysdeps/i386/i686/isa.h: Likewise.
	* sysdeps/x86_64/isa.h: Likewise.
	* sysdeps/x86/hp-timing.h: New file.
	* sysdeps/x86/init-arch.h: Include <isa.h>.
2018-10-17 15:16:45 -07:00
Joseph Myers
c2b7ccad93 Do not allow divide-by-zero exception for pow(+/- 0, -Inf).
C99 wrongly specified a divide-by-zero exception for pow(+/- 0, -Inf);
C11 made it optional after this was pointed out, and the permission
for this exception has been removed in the current C2x draft.  This
patch makes the glibc pow tests reflect the stricter requirement
(which follows the normal IEEE rules that a divide-by-zero exception
is for the case of exact infinite results from *finite* operands, not
for such results when any operand is infinite).

Tested for x86_64 and x86.  (If any other pow implementation in glibc,
not exercised on those architectures, turns out to fail the stricter
test, it should be fixed to avoid the exception in this case.)

	* math/libm-test-pow.inc (pow_test_data): Do not allow
	divide-by-zero exception for pow(+/- 0, -Inf).
2018-10-17 21:23:40 +00:00
Zack Weinberg
26756e5756
[manual] Job control is no longer optional.
Job control was made mandatory in POSIX.1-2001: compare
<http://pubs.opengroup.org/onlinepubs/7990989775/xsh/unistd.h.html> with
<http://pubs.opengroup.org/onlinepubs/009695399/basedefs/unistd.h.html>.
Seventeen years later, we need not devote an entire manual @node to
warning people that this was once an optional POSIX feature.

	* manual/job.texi (Job Control is Optional): Remove node, as
	job control has not been optional in quite some time.
	(Job Control): Mention briefly that systems older than
	POSIX.1-2001 might not support job control.
	* manual/conf.texi (_POSIX_JOB_CONTROL): Will always be
	defined on systems conforming to POSIX.1-2001.
2018-10-17 14:10:51 -04:00
Arjun Shankar
c5288d378a Remove unnecessary locking when reading iconv configuration [BZ #22062]
In iconv/gconv_conf.c, __gconv_get_path unnecessarily obtains a lock when
populating the array pointed to by __gconv_path_elem. The locking is not
necessary because all calls to __gconv_read_conf (which in turn calls
__gconv_get_path) are serialized using __libc_once.

This patch:
- removes all locking in __gconv_get_path;
- replaces all explicitly serialized __gconv_read_conf calls with calls to
  __gconv_load_conf, a new wrapper that is serialized internally;
- adds a new test, iconv/tst-iconv_mt.c, to exercise iconv initialization,
  usage, and cleanup in a multi-threaded program;
- indents __gconv_get_path correctly, removing tab characters (which makes
  the patch look a little bigger than it really is).

After removing the unnecessary locking, it was confirmed that the test case
fails if the relevant __libc_once is removed. Additionally, four localedata
and iconvdata tests also fail. This gives confidence that the testsuite
sufficiently guards against some regressions relating to multi-threading
with iconv.

Tested on x86_64 and i686.
2018-10-17 17:47:29 +02:00
Joseph Myers
729f34028a Use single bits/shm.h for all architectures.
After my patch to move SHMLBA to its own header, the bits/shm.h
headers for architectures using the Linux kernel still vary in a few
ways: the use of __syscall_ulong_t; whether padding for 32-bit systems
is present before or after time fields, or missing altogether (mips,
x32); whether shm_segsz is before or after the time fields; whether,
if after time fields, there is extra padding before shm_segsz.

This patch arranges for a single header to be used.  __syscall_ulong_t
is safe to use everywhere, while bits/shm-pad.h is added with new
macros __SHM_PAD_AFTER_TIME, __SHM_PAD_BEFORE_TIME,
__SHM_SEGSZ_AFTER_TIME and __SHM_PAD_BETWEEN_TIME_AND_SEGSZ to
describe the differences.

Tested for x86_64 and x86, and with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/Makefile (sysdep_headers): Add
	bits/shm-pad.h.
	* sysdeps/unix/sysv/linux/bits/shm.h: Include <bits/shm-pad.h>.
	(shmatt_t): Define as __syscall_ulong_t.
	(__SHM_PAD_TIME): New macro, depending on [__SHM_PAD_BEFORE_TIME]
	and [__SHM_PAD_AFTER_TIME].
	(struct shmid_ds): Define time fields using __SHM_PAD_TIME.
	Define shm_segsz and associated padding based on
	[__SHM_SEGSZ_AFTER_TIME] and [__SHM_PAD_BETWEEN_TIME_AND_SEGSZ].
	Use __syscall_ulong_t instead of unsigned long int.
	[__USE_MISC] (struct shminfo): Use __syscall_ulong_t instead of
	unsigned long int.
	[__USE_MISC] (struct shm_info): Likewise.
	* sysdeps/unix/sysv/linux/bits/shm-pad.h: New file.
	* sysdeps/unix/sysv/linux/hppa/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/mips/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/shm-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/shm.h: Remove.
	* sysdeps/unix/sysv/linux/mips/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/shm.h: Likewise.
2018-10-17 11:56:28 +00:00
Joseph Myers
8c8d2a8aff Move SHMLBA to its own header.
One difference between bits/shm.h headers for architectures using the
Linux kernel is the definition of SHMLBA.  This was noted in
<https://sourceware.org/ml/libc-alpha/2018-09/msg00175.html> as a
reason why even a new architecture (C-SKY) might need its own
bits/shm.h; thus, splitting it out of bits/shm.h can allow less
duplication of headers for new architectures.

This patch moves that definition to its own header, bits/shmlba.h, to
allow more sharing of headers between architectures.  That move allows
the arm, ia64 and sh variants of bits/shm.h to be removed, as they had
no other significant differences from the generic bits/shm.h; powerpc
and x86 have their own bits/shm.h but do not need to get their own
bits/shmlba.h because they use the same SHMLBA as the generic header.
Other architectures with their own bits/shm.h get their own
bits/shmlba.h without being able to remove their own bits/shm.h until
the generic one has been adapted to be able to handle more
architectures (where, in addition to the differences seen for
bits/msq.h and bits/sem.h, the position of shm_segsz in struct
shmid_ds also depends on the architecture).

Tested for x86_64 and x86, and with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/Makefile (sysdep_headers): Add
	bits/shmlba.h.
	* sysdeps/unix/sysv/linux/bits/shm.h: Include <bits/shmlba.h>.
	(SHMLBA): Remove macro.
	(__getpagesize): Remove function declaration.
	* sysdeps/unix/sysv/linux/hppa/bits/shm.h: Include
	<bits/shmlba.h>.
	(SHMLBA): Remove macro.
	* sysdeps/unix/sysv/linux/mips/bits/shm.h: Include
	<bits/shmlba.h>.
	(SHMLBA): Remove macro.
	* sysdeps/unix/sysv/linux/powerpc/bits/shm.h: Include
	<bits/shmlba.h>.
	(SHMLBA): Remove macro.
	(__getpagesize): Remove function declaration.
	* sysdeps/unix/sysv/linux/sparc/bits/shm.h: Include
	<bits/shmlba.h>.
	(SHMLBA): Remove macro.
	(__getshmlba): Remove function declaration.
	* sysdeps/unix/sysv/linux/x86/bits/shm.h: Include <bits/shmlba.h>.
	(SHMLBA): Remove macro.
	(__getpagesize): Remove function declaration.
	* sysdeps/unix/sysv/linux/arm/bits/shm.h: Remove file.
	* sysdeps/unix/sysv/linux/ia64/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/sh/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/bits/shmlba.h: New file.
	* sysdeps/unix/sysv/linux/arm/bits/shmlba.h: Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/shmlba.h: Likewise.
	* sysdeps/unix/sysv/linux/ia64/bits/shmlba.h: Likewise.
	* sysdeps/unix/sysv/linux/mips/bits/shmlba.h: Likewise.
	* sysdeps/unix/sysv/linux/sh/bits/shmlba.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/shmlba.h: Likewise.
2018-10-17 11:55:10 +00:00
Stefan Liebler
403b4feb22 Fix race in pthread_mutex_lock while promoting to PTHREAD_MUTEX_ELISION_NP [BZ #23275]
The race leads either to pthread_mutex_destroy returning EBUSY
or triggering an assertion (See description in bugzilla).

This patch is fixing the race by ensuring that the elision path is
used in all cases if elision is enabled by the GLIBC_TUNABLES framework.

The __kind variable in struct __pthread_mutex_s is accessed concurrently.
Therefore we are now using the atomic macros.

The new testcase tst-mutex10 is triggering the race on s390x and intel.
Presumably also on power, but I don't have access to a power machine
with lock-elision. At least the code for power is the same as on the other
two architectures.

ChangeLog:

	[BZ #23275]
	* nptl/tst-mutex10.c: New File.
	* nptl/Makefile (tests): Add tst-mutex10.
	(tst-mutex10-ENV): New variable.
	* sysdeps/unix/sysv/linux/s390/force-elision.h: (FORCE_ELISION):
	Ensure that elision path is used if elision is available.
	* sysdeps/unix/sysv/linux/powerpc/force-elision.h (FORCE_ELISION):
	Likewise.
	* sysdeps/unix/sysv/linux/x86/force-elision.h: (FORCE_ELISION):
	Likewise.
	* nptl/pthreadP.h (PTHREAD_MUTEX_TYPE, PTHREAD_MUTEX_TYPE_ELISION)
	(PTHREAD_MUTEX_PSHARED): Use atomic_load_relaxed.
	* nptl/pthread_mutex_consistent.c (pthread_mutex_consistent): Likewise.
	* nptl/pthread_mutex_getprioceiling.c (pthread_mutex_getprioceiling):
	Likewise.
	* nptl/pthread_mutex_lock.c (__pthread_mutex_lock_full)
	(__pthread_mutex_cond_lock_adjust): Likewise.
	* nptl/pthread_mutex_setprioceiling.c (pthread_mutex_setprioceiling):
	Likewise.
	* nptl/pthread_mutex_timedlock.c (__pthread_mutex_timedlock): Likewise.
	* nptl/pthread_mutex_trylock.c (__pthread_mutex_trylock): Likewise.
	* nptl/pthread_mutex_unlock.c (__pthread_mutex_unlock_full): Likewise.
	* sysdeps/nptl/bits/thread-shared-types.h (struct __pthread_mutex_s):
	Add comments.
	* nptl/pthread_mutex_destroy.c (__pthread_mutex_destroy):
	Use atomic_load_relaxed and atomic_store_relaxed.
	* nptl/pthread_mutex_init.c (__pthread_mutex_init):
	Use atomic_store_relaxed.
2018-10-17 12:23:04 +02:00
Andreas Schwab
ce5a7de6cd Don't reduce test timeout to less than default
This removes all overrides of TIMEOUT that are less than or equal to the
default timeout.
2018-10-17 09:34:13 +02:00
Steve Ellcey
f0da0bcf8b Remove extra space at end of line. 2018-10-16 11:02:03 -07:00
Anton Youdkevitch
75c1aee500 aarch64: optimized memcpy implementation for thunderx2
Since aligned loads and stores are huge performance
advantage the implementation always tries to do aligned
access. Among the cases when src and dst addresses are
aligned or unaligned evenly there are cases of not evenly
unaligned src and dst. For such cases (if the length is
big enough) ext instruction is used to merge-and-shift
two memory chunks loaded from two adjacent aligned
locations and then the adjusted chunk gets stored to
aligned address.

Performance gain against the current T2 implementation:
     memcpy-large: 65K-32M: +40% - +10%
     memcpy-walk:  128-32M: +20% - +2%
2018-10-16 11:00:27 -07:00
Joseph Myers
bcdb1bfa0c Use single bits/sem.h for all architectures.
The bits/sem.h headers for architectures using the Linux kernel vary
in a few ways:

* x32 uses __syscall_ulong_t instead of unsigned long int.

* The x86 header uses padding after time fields unconditionally
  (including for both x86_64 ABIs), not just for 32-bit time (unlike
  in msqid_ds where there is only padding for 32-bit time).  Because
  this padding is present for x32, and is __syscall_ulong_t there, it
  does have to be __syscall_ulong_t, not unsigned long int.

* The MIPS header never uses padding around time fields, even when
  32-bit (unlike in msqid_ds where it has endian-dependent padding for
  32-bit time).

* Some older 32-bit big-endian architectures have padding before
  rather than after time fields, although the preferred generic
  approach is padding after the time fields independent of endianness.

(There are also insubstantial differences such as use of unsigned int
for padding instead of unsigned long int, which makes no difference to
layout since the padding fields using unsigned int are only present on
32-bit architectures.)

For the first, __syscall_ulong_t can be used in the generic version as
it's the same as unsigned long int everywhere except x32.  For the
other differences, this patch adds macros __SEM_PAD_BEFORE_TIME and
__SEM_PAD_AFTER_TIME in a new bits/sem-pad.h header, so that header is
the only one needing to be provided on architectures with differences
in this area, and everything else can go in a single common bits/sem.h
header.

Tested for x86_64 and x86, and with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/Makefile (sysdep_headers): Add
	bits/sem-pad.h.
	* sysdeps/unix/sysv/linux/bits/sem.h: Include <bits/sem-pad.h>
	instead of <bits/wordsize.h>.
	(__SEM_PAD_TIME): New macro, depending on [__SEM_PAD_BEFORE_TIME]
	and [__SEM_PAD_AFTER_TIME].
	(struct semid_ds): Define time fields using __SEM_PAD_TIME.  Use
	__syscall_ulong_t instead of unsigned long int.
	* sysdeps/unix/sysv/linux/bits/sem-pad.h: New file.
	* sysdeps/unix/sysv/linux/hppa/bits/sem-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/mips/bits/sem-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/sem-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/sem-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/sem-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/sem.h: Remove.
	* sysdeps/unix/sysv/linux/mips/bits/sem.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/sem.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/sem.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/sem.h: Likewise.
2018-10-15 19:28:04 +00:00
Paul Eggert
620a5d4cb1 regex: simplify by using intprops.h
[BZ#23744]
* posix/regex_internal.h [_LIBC]: Include intprops.h.
(TYPE_SIGNED, INT_ADD_WRAPV) [_LIBC]: Remove.
intprops.h defines them.
2018-10-14 23:36:55 -05:00
Paul Eggert
f4efbdfb44 regex: __builtin_expect → __glibc_unlikely
[BZ#23744]
This refactoring was prompted by a problem when the regex code is
used as part of Gnulib and when the builder’s compiler does not grok
__builtin_expect.  Problem reported for Gawk by Nelson H.F. Beebe in:
https://lists.gnu.org/r/bug-gnulib/2018-09/msg00137.html
Although this refactoring does not fix the problem directly,
we might as well have Gawk use the now-preferred glibc style for when
__builtin_expect is unavailable.
* posix/regex_internal.h (BE): Remove.
All uses replaced by __glibc_unlikely or __glibc_likely.
2018-10-14 23:36:55 -05:00
Joseph Myers
9f9feb6d5d Use single bits/msq.h for all architectures.
The bits/msq.h headers for architectures using the Linux kernel vary
in a few ways:

* x32 uses __syscall_ulong_t instead of unsigned long int.

* x32 has 64-bit time_t, so no padding around time fields despite
  __WORDSIZE == 32.

* Some older 32-bit big-endian architectures have padding before
  rather than after time fields, although the preferred generic
  approach is padding after the time fields independent of endianness.

(There are also insubstantial differences such as use of unsigned int
for padding instead of unsigned long int, which makes no difference to
layout since the padding fields using unsigned int are only present on
32-bit architectures.)

For the first, __syscall_ulong_t can be used in the generic version as
it's the same as unsigned long int everywhere except x32.  For the
other two differences, this patch adds macros __MSQ_PAD_BEFORE_TIME
and __MSQ_PAD_AFTER_TIME in a new bits/msq-pad.h header, so that
header is the only one needing to be provided on architectures with
differences in this area, and everything else can go in a single
common bits/msq.h header.  Once we have __TIMESIZE, the generic
bits/msq-pad.h can change to use that instead of __WORDSIZE, at which
point the x86 version of bits/msq-pad.h won't be needed either.

Tested for x86_64 and x86, and with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/Makefile (sysdep_headers): Add
	bits/msq-pad.h.
	* sysdeps/unix/sysv/linux/bits/msq.h: Include <bits/msq-pad.h>
	instead of <bits/wordsize.h>.
	(msgqnum_t): Define as __syscall_ulong_t.
	(msglen_t): Likewise.
	(__MSQ_PAD_TIME): New macro, depending on [__MSQ_PAD_BEFORE_TIME]
	and [__MSQ_PAD_AFTER_TIME].
	(struct msqid_ds): Define time fields using __MSQ_PAD_TIME.  Use
	__syscall_ulong_t instead of unsigned long int.
	* sysdeps/unix/sysv/linux/bits/msq-pad.h: New file.
	* sysdeps/unix/sysv/linux/hppa/bits/msq-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/mips/bits/msq-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/msq-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/msq-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/msq-pad.h: Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/msq.h: Remove.
	* sysdeps/unix/sysv/linux/mips/bits/msq.h: Likewise.
	* sysdeps/unix/sysv/linux/powerpc/bits/msq.h: Likewise.
	* sysdeps/unix/sysv/linux/sparc/bits/msq.h: Likewise.
	* sysdeps/unix/sysv/linux/x86/bits/msq.h: Likewise.
2018-10-11 12:07:27 +00:00
Joseph Myers
2a4b25fad8 Use common bits/shm.h for more architectures.
sysdeps/unix/sysv/linux/bits/shm.h has padding after time fields in
struct shmid_ds unconditionally, and thus is only suitable for 32-bit
architectures (no 64-bit configurations use this file);
sysdeps/unix/sysv/linux/generic/bits/shm.h is substantively the same,
except that the padding is conditioned on __WORDSIZE == 32, and so it
can be used for 64-bit architectures as well.

This patch adds the conditionals to
sysdeps/unix/sysv/linux/bits/shm.h.  The linux/generic/ version is
then no longer needed and so is removed, as are the alpha and s390
versions which are also no longer needed.  The other
architecture-specific versions have different padding, layout, types
or SHMLBA definitions and so are still needed after this change.

This is essentially the same change for bits/shm.h as the bits/msq.h
patch and the bits/sem.h patch.  However, the details of the padding
variations for the architectures that aren't changed are not all the
same between msqid_ds, shmid_ds and semid_ds.

Tested with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/bits/shm.h: Include <bits/wordsize.h>.
	(struct shmid_ds): Condition padding after time fields on
	[__WORDSIZE == 32].
	* sysdeps/unix/sysv/linux/alpha/bits/shm.h: Remove file.
	* sysdeps/unix/sysv/linux/generic/bits/shm.h: Likewise.
	* sysdeps/unix/sysv/linux/s390/bits/shm.h: Likewise.
2018-10-10 00:56:17 +00:00
Joseph Myers
19e62f2690 Use common bits/sem.h for more architectures.
sysdeps/unix/sysv/linux/bits/sem.h has padding after time fields in
struct semid_ds unconditionally, and thus is only suitable for 32-bit
architectures (no 64-bit configurations use this file);
sysdeps/unix/sysv/linux/generic/bits/sem.h is substantively the same,
except that the padding is conditioned on __WORDSIZE == 32, and so it
can be used for 64-bit architectures as well.

This patch adds the conditionals to
sysdeps/unix/sysv/linux/bits/sem.h.  The linux/generic/ version is
then no longer needed and so is removed, as are the alpha, ia64 and
s390 versions which are also no longer needed.  The other
architecture-specific versions have different padding or types and so
are still needed after this change.

This is essentially the same change for bits/sem.h as the bits/msq.h
patch.  However, the details of the padding variations for the
architectures that aren't changed are not all the same between
msqid_ds and semid_ds.

Tested with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/bits/sem.h: Include <bits/wordsize.h>.
	(struct semid_ds): Condition padding after time fields on
	[__WORDSIZE == 32].
	* sysdeps/unix/sysv/linux/alpha/bits/sem.h: Remove file.
	* sysdeps/unix/sysv/linux/generic/bits/sem.h: Likewise.
	* sysdeps/unix/sysv/linux/ia64/bits/sem.h: Likewise.
	* sysdeps/unix/sysv/linux/s390/bits/sem.h: Likewise.
2018-10-10 00:54:27 +00:00
Joseph Myers
b9c0f6c11a Use common bits/msq.h for more architectures.
sysdeps/unix/sysv/linux/bits/msq.h has padding after time fields in
struct msqid_ds unconditionally, and thus is only suitable for 32-bit
architectures (no 64-bit configurations use this file);
sysdeps/unix/sysv/linux/generic/bits/msq.h is substantively the same,
except that the padding is conditioned on __WORDSIZE == 32, and so it
can be used for 64-bit architectures as well.

This patch adds the conditionals to
sysdeps/unix/sysv/linux/bits/msq.h.  The linux/generic/ version is
then no longer needed and so is removed, as are the alpha, ia64 and
s390 versions which are also no longer needed.  The other
architecture-specific versions have different padding or types and so
are still needed after this change.

Tested with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/bits/msq.h: Include <bits/wordsize.h>.
	(struct msqid_ds): Condition padding after time fields on
	[__WORDSIZE == 32].
	* sysdeps/unix/sysv/linux/alpha/bits/msq.h: Remove file.
	* sysdeps/unix/sysv/linux/generic/bits/msq.h: Likewise.
	* sysdeps/unix/sysv/linux/ia64/bits/msq.h: Likewise.
	* sysdeps/unix/sysv/linux/s390/bits/msq.h: Likewise.
2018-10-10 00:52:47 +00:00
Szabolcs Nagy
97785a61af Increase timeout of nss/tst-nss-files-hosts-multi
Increase timeout from the default 20s to 40s. This test makes close to
2 million syscalls with distribution:

1180249 connect
 297952 getsockname
 144040 lseek
 143734 read
  14466 close
...

connect can be slow, so the default timeout was not enough on slow
systems.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>

	* nss/tst-nss-files-hosts-multi.c (TIMEOUT): Define.
2018-10-09 15:57:01 +01:00
Szabolcs Nagy
ed643089cd Increase timeout of libio/tst-readline
Increase timeout from the default 20s to 100s. This test makes close to
20 million syscalls with distribution:

12327675 read
 4143204 lseek
  929475 close
  929471 openat
   92817 fstat
    1431 write
...

The default timeout assumes each can finish in 1us on average which
is not true on slow machines.

Reviewed-by: Carlos O'Donell <carlos@redhat.com>

	* libio/tst-readline.c (TIMEOUT): Define.
2018-10-09 15:56:12 +01:00
Paul Eggert
5a58064311 mktime fix for Gnulib + coreutils
[BZ#23745]
This fix affects only Gnulib.  Problem discovered when
mktime.c was used as part of Gnulib in bleeding-edge Coreutils.
* time/mktime.c:
(my_tzset) [!_LIBC && !NEED_MKTIME_WORKING && !NEED_MKTIME_WINDOWS]:
Do not define since it is not used.  Defining an unused static
function prompts a warning from GCC when Coreutils is configured
with --enable-gcc-warnings.
2018-10-08 11:28:25 -07:00
Leonardo Sandoval
c892ae04f4 benchtests: Set float type on --threshold argument
Otherwise, we see the following runtime error when using the parameter:

  File "./glibc/benchtests/scripts/compare_bench.py", line 46, in do_compare
    if d > threshold:
TypeError: '>' not supported between instances of 'float' and 'str'

	* benchtests/scripts/compare_bench.py (main): set float type on
	threshold argument.
2018-10-08 09:11:30 -05:00
Rafal Luzynski
a68ec8eac2 kl_GL: Update the month names and date formats (bug 23740).
Month names as provided by Oqaasileriffik, the official Greenlandic
language regulator.  They have recently reached the consensus regarding
the orthography of the month names.

Date formats updated to match the correct Greenlandic order which is MDY.

	[BZ #23740]
	* localedata/locales/kl_GL (mon): Update, the relative case.
	(alt_mon): Add, fill with month names in the nominative case.
	(d_t_fmt): Set to "%a %b %d %Y %T %Z".
	(d_fmt): Set to "%b %d %Y".
2018-10-08 12:28:02 +02:00
Joseph Myers
47c49251d5 Use bits/mman-linux.h for hppa.
hppa currently has a bits/mman.h that does not include
bits/mman-linux.h, unlike all other architectures using the Linux
kernel.  This sort of variation between architectures is generally
unhelpful when making global changes for new constants added to new
Linux kernel releases.

This patch changes hppa to use bits/mman-linux.h, overriding constants
with different values as necessary (including with #undef after
bits/mman.h inclusion when needed, as already done for alpha).  While
there could possibly be further improvements through e.g. splitting
more sets of definitions into separate bits/ headers, I think this is
still an improvement on the current state.  diffstat shows 27 lines
added, 51 deleted (and some of that is actually existing lines moving
to a different place in the file).

Tested with build-many-glibcs.py for hppa-linux-gnu.

	* sysdeps/unix/sysv/linux/hppa/bits/mman.h: Include
	<bits/mman-linux.h>.
	(PROT_READ): Don't define here.
	(PROT_WRITE): Likewise.
	(PROT_EXEC): Likewise.
	(PROT_NONE): Likewise.
	(PROT_GROWSDOWN): Likewise.
	(PROT_GROWSUP): Likewise.
	(MAP_SHARED): Likewise.
	(MAP_PRIVATE): Likewise.
	[__USE_MISC] (MAP_SHARED_VALIDATE): Likewise.
	[__USE_MISC] (MAP_FILE): Likewise.
	[__USE_MISC] (MAP_ANONYMOUS): Likewise.
	[__USE_MISC] (MAP_ANON): Likewise.
	[__USE_MISC] (MAP_HUGE_SHIFT): Likewise.
	[__USE_MISC] (MAP_HUGE_MASK): Likewise.
	(MCL_CURRENT): Likewise.
	(MCL_FUTURE): Likewise.
	(MCL_ONFAULT): Likewise.
	[__USE_MISC] (MADV_NORMAL): Likewise.
	[__USE_MISC] (MADV_RANDOM): Likewise.
	[__USE_MISC] (MADV_SEQUENTIAL): Likewise.
	[__USE_MISC] (MADV_WILLNEED): Likewise.
	[__USE_MISC] (MADV_DONTNEED): Likewise.
	[__USE_MISC] (MADV_FREE): Likewise.
	[__USE_MISC] (MADV_REMOVE): Likewise.
	[__USE_MISC] (MADV_DONTFORK): Likewise.
	[__USE_MISC] (MADV_DOFORK): Likewise.
	[__USE_MISC] (MADV_HWPOISON): Likewise.
	[__USE_XOPEN2K] (POSIX_MADV_NORMAL): Likewise.
	[__USE_XOPEN2K] (POSIX_MADV_RANDOM): Likewise.
	[__USE_XOPEN2K] (POSIX_MADV_SEQUENTIAL): Likewise.
	[__USE_XOPEN2K] (POSIX_MADV_WILLNEED): Likewise.
	[__USE_XOPEN2K] (POSIX_MADV_DONTNEED): Likewise.
	(__MAP_ANONYMOUS): New macro.
	[__USE_MISC] (MAP_TYPE): Undefine and redefine after
	<bits/mman-linux.h> inclusion.
	(MAP_FIXED): Likewise.
	(MS_SYNC): Likewise.
	(MS_ASYNC): Likewise.
	(MS_INVALIDATE): Likewise.
	[__USE_MISC] (MADV_MERGEABLE): Likewise.
	[__USE_MISC] (MADV_UNMERGEABLE): Likewise.
	[__USE_MISC] (MADV_HUGEPAGE): Likewise.
	[__USE_MISC] (MADV_NOHUGEPAGE): Likewise.
	[__USE_MISC] (MADV_DONTDUMP): Likewise.
	[__USE_MISC] (MADV_DODUMP): Likewise.
	[__USE_MISC] (MADV_WIPEONFORK): Likewise.
	[__USE_MISC] (MADV_KEEPONFORK): Likewise.
2018-10-04 19:47:23 +00:00
Joseph Myers
a19876214a Fix libnldbl_nonshared.a references to internal libm symbols (bug 23735).
The redirection of built-in functions such as sqrt in include/math.h
applies when the wrappers for those functions in libnldbl_nonshared.a
are built, resulting in references to internal names such as
__ieee754_sqrt that aren't actually exported from the shared libm.
(This applies for sqrt in 2.28, also for the round-to-integer
functions in current master because of my changes there.)  This patch
arranges for NO_MATH_REDIRECT to be used for all the affected
functions, and adds a test for those functions in
libnldbl_nonshared.a.

(We could of course choose to obsolete libnldbl_nonshared.a and
require that people building with -mlong-double-64 either include the
relevant headers and have a compiler supporting asm redirection, or
have some other means of achieving that redirection at compile time if
not including those headers.  But while we have libnldbl_nonshared.a,
it seems appropriate to fix such bugs in it.)

Tested for powerpc, and with build-many-glibcs.py.

	[BZ #23735]
	* sysdeps/ieee754/ldbl-opt/nldbl-compat.h (NO_MATH_REDIRECT):
	Define.
	* sysdeps/ieee754/ldbl-opt/test-nldbl-redirect.c: New file.
	* sysdeps/ieee754/ldbl-opt/Makefile [$(subdir) = math] (tests):
	Add test-nldbl-redirect.
	[$(subdir) = math] (CFLAGS-test-nldbl-redirect.c): New variable.
	[$(subdir) = math] ($(objpfx)test-nldbl-redirect): Depend on
	$(objpfx)libnldbl_nonshared.a.
2018-10-04 12:16:05 +00:00
Stefan Liebler
e7624d708d Adjust name of ld.so in test-container.c.
The test-container.c file assumes that ld.so is always named
something like /elf/ld-linux-*.
But e.g. on s390x it is named ld64.so.1 or ld.so.1 on s390.
There are other architectures like power or mips with similar names.

This patch introduces the new global variable support_objdir_elf_ldso
which contains the absolute path to the runtime linker used by the
testsuite, e.g. OBJDIR_PATH/elf/ld-linux-x86-64.so.2.
The check in test-container.c is now comparing against this path.
Without this patch, test-container.c is searching invalid files / directories
and fails to find glibc/nss/tst-nss-test3.root/tst-nss-test3.script.
Then the test tst-nss-test3 fails!

Reviewed-by: Carlos O'Donell <carlos@redhat.com>

ChangeLog:

	* support/support.h (support_objdir_elf_ldso): New variable.
	* support/support_paths.c (support_objdir_elf_ldso): Likewise.
	* support/Makefile (CFLAGS-support_paths.c): Add definition
	for OBJDIR_ELF_LDSO_PATH.
	* support/test-container.c (main): Search for the ld.so
	which is also used by the testsuite.
2018-10-04 13:07:29 +02:00
Rafal Luzynski
dae3ed958c kl_GL: Fix spelling of Sunday, should be "sapaat" (bug 20209).
Although CLDR says otherwise, it is confirmed by Oqaasileriffik, the
official Greenlandic language regulator, that this change is correct.

	[BZ #20209]
	* localedata/locales/kl_GL: (abday): Fix spelling of Sun (Sunday),
	should be "sap" rather than "sab".
	(day): Fix spelling of Sunday, should be "sapaat" rather than
	"sabaat".
2018-10-02 23:55:13 +02:00
Joseph Myers
f346b5b38c Add more fma tests.
In my review
<https://sourceware.org/ml/libc-alpha/2018-06/msg00375.html> of a
patch for bug 23584, I expressed concern that the proposed changes
didn't deal with certain cases similar to the ones in the bug but
where test coverage was missing.

This patch adds such tests of fma (Inf, finite, finite) and fma
(finite, Inf, finite) to libm-test-fma.inc.  It does *not* do anything
to fix the bug, simply adds test coverage to provide stronger evidence
of whether any proposed revised fix does address the cases I was
concerned with.

Tested for x86_64 and x86.

	* math/libm-test-fma.inc (fma_test_data): Add more tests.
2018-10-02 17:17:15 +00:00
Martin Jansa
4a06ceea33 sysdeps/ieee754/soft-fp: ignore maybe-uninitialized with -O [BZ #19444]
* with -O, -O1, -Os it fails with:

In file included from ../soft-fp/soft-fp.h:318,
                 from ../sysdeps/ieee754/soft-fp/s_fdiv.c:28:
../sysdeps/ieee754/soft-fp/s_fdiv.c: In function '__fdiv':
../soft-fp/op-2.h:98:25: error: 'R_f1' may be used uninitialized in this function [-Werror=maybe-uninitialized]
        X##_f0 = (X##_f1 << (_FP_W_TYPE_SIZE - (N)) | X##_f0 >> (N) \
                         ^~
../sysdeps/ieee754/soft-fp/s_fdiv.c:38:14: note: 'R_f1' was declared here
   FP_DECL_D (R);
              ^
../soft-fp/op-2.h:37:36: note: in definition of macro '_FP_FRAC_DECL_2'
   _FP_W_TYPE X##_f0 _FP_ZERO_INIT, X##_f1 _FP_ZERO_INIT
                                    ^
../soft-fp/double.h:95:24: note: in expansion of macro '_FP_DECL'
 # define FP_DECL_D(X)  _FP_DECL (2, X)
                        ^~~~~~~~
../sysdeps/ieee754/soft-fp/s_fdiv.c:38:3: note: in expansion of macro 'FP_DECL_D'
   FP_DECL_D (R);
   ^~~~~~~~~
../soft-fp/op-2.h:101:17: error: 'R_f0' may be used uninitialized in this function [-Werror=maybe-uninitialized]
       : (X##_f0 << (_FP_W_TYPE_SIZE - (N))) != 0)); \
                 ^~
../sysdeps/ieee754/soft-fp/s_fdiv.c:38:14: note: 'R_f0' was declared here
   FP_DECL_D (R);
              ^
../soft-fp/op-2.h:37:14: note: in definition of macro '_FP_FRAC_DECL_2'
   _FP_W_TYPE X##_f0 _FP_ZERO_INIT, X##_f1 _FP_ZERO_INIT
              ^
../soft-fp/double.h:95:24: note: in expansion of macro '_FP_DECL'
 # define FP_DECL_D(X)  _FP_DECL (2, X)
                        ^~~~~~~~
../sysdeps/ieee754/soft-fp/s_fdiv.c:38:3: note: in expansion of macro 'FP_DECL_D'
   FP_DECL_D (R);
   ^~~~~~~~~

Build tested with Yocto for ARM, AARCH64, X86, X86_64, PPC, MIPS, MIPS64
with -O, -O1, -Os.
For AARCH64 it needs one more fix in locale for -Os.

	[BZ #19444]
	* sysdeps/ieee754/soft-fp/s_fdiv.c: Include <libc-diag.h> and use
	DIAG_PUSH_NEEDS_COMMENT, DIAG_IGNORE_NEEDS_COMMENT and
	DIAG_POP_NEEDS_COMMENT to disable -Wmaybe-uninitialized.
2018-10-02 15:40:57 +00:00
Adhemerval Zanella
a06fc4a533 Fix build from commit 0b727ed
* sysdeps/unix/sysv/linux/fd_to_filename.h: Add missing includes.
2018-10-02 12:29:02 -03:00
H.J. Lu
a1ccc0654b x86: Use RTM intrinsics in pthread mutex lock elision
Since RTM intrinsics are supported in GCC 4.9, we can use them in
pthread mutex lock elision.

	* sysdeps/unix/sysv/linux/x86/Makefile (CFLAGS-elision-lock.c):
	Add -mrtm.
	(CFLAGS-elision-unlock.c): Likewise.
	(CFLAGS-elision-timed.c): Likewise.
	(CFLAGS-elision-trylock.c): Likewise.
	* sysdeps/unix/sysv/linux/x86/hle.h: Rewritten.
2018-10-02 07:51:57 -07:00
Adhemerval Zanella
0b727ed4d6 libio: Flush stream at freopen (BZ#21037)
As POSIX states [1] a freopen call should first flush the stream as if by a
call fflush.  C99 (n1256) and C11 (n1570) only states the function should
first close any file associated with the specific stream.  Although current
implementation only follow C specification, current BSD and other libc
implementation (musl) are in sync with POSIX and fflush the stream.

This patch change freopen{64} to fflush the stream before actually reopening
it (or returning if the stream does not support reopen).  It also changes the
Linux implementation to avoid a dynamic allocation on 'fd_to_filename'.

Checked on x86_64-linux-gnu.

	[BZ #21037]
	* libio/Makefile (tests): Add tst-memstream4 and tst-wmemstream4.
	* libio/freopen.c (freopen): Sync stream before reopen and adjust to
	new fd_to_filename interface.
	* libio/freopen64.c (freopen64): Likewise.
	* libio/tst-memstream.h: New file.
	* libio/tst-memstream4.c: Likewise.
	* libio/tst-wmemstream4.c: Likewise.
	* sysdeps/generic/fd_to_filename.h (fd_to_filename): Change signature.
	* sysdeps/unix/sysv/linux/fd_to_filename.h (fd_to_filename): Likewise
	and remove internal dynamic allocation.

[1] http://pubs.opengroup.org/onlinepubs/9699919799/
2018-10-02 10:01:56 -03:00
Joseph Myers
d0e91e8c4e Move MREMAP_* to bits/mman-shared.h.
The MREMAP_* flags are identical between bits/mman-linux.h and the
hppa bits/mman.h; thus, they should be in bits/mman-shared.h instead
to avoid unnecessary duplication.  This patch moves them there.

Tested for x86_64, and with build-many-glibcs.py.

	* sysdeps/unix/sysv/linux/bits/mman-linux.h [__USE_GNU]
	(MREMAP_MAYMOVE): Do not define here.
	[__USE_GNU] (MREMAP_FIXED): Likewise.
	* sysdeps/unix/sysv/linux/bits/mman-shared.h [__USE_GNU]
	(MREMAP_MAYMOVE): Define here instead.
	[__USE_GNU] (MREMAP_FIXED): Likewise.
	* sysdeps/unix/sysv/linux/hppa/bits/mman.h [__USE_GNU]
	(MREMAP_MAYMOVE): Remove.
	[__USE_GNU] (MREMAP_FIXED): Likewise.
2018-10-01 20:30:57 +00:00