This attempts to repair problems introduced by the bad merge
5491fd1098. The easiest way for me
to fix the badly-merged gnulib files was to run
'admin/merge-gnulib', so I did that, which also imported the
following changes:
* build-aux/update-copyright, m4/gnulib.m4:
Update from gnulib, incorporating:
2015-01-15 time: port to MinGW32 3.21
2015-01-15 update-copyright: apply to self
2015-01-11 update-copyright: recognize groff's \(co marker
2014-12-20 utimens: remove unnecessary assert
2014-12-16 stdalign: port better to HP compilers
2014-12-16 stdalign: work around Apple GCC 4.0 bug
* lib/stdalign.in.h, lib/utimens.c, m4/stdalign.m4: Update from gnulib.
2014-12-12 stddef: support C11's max_align_t
2014-12-08 apply _GL_ATTRIBUTE_PURE to some inline functions
2014-12-02 support GNU format printf and scanf on mingw
* doc/misc/texinfo.tex, lib/stat-time.h, lib/stddef.in.h:
* lib/timespec.h, m4/extensions.m4, m4/stddef_h.m4, m4/stdio_h.m4:
Update from gnulib.
* lib/gnulib.mk: Regenerate.
2014-11-14 extern-inline: update commentary about GCC bugs
2014-11-06 unistd: port to iOS
2014-11-04 update from texinfo
* doc/misc/texinfo.tex, lib/unistd.in.h, m4/extern-inline.m4:
Update from gnulib.
2014-09-02 gnulib-common.m4: port to GCC 4.2.1 and Sun Studio 12 C++
2014-09-01 manywarnings: add GCC 4.9 warnings
* m4/gnulib-common.m4, m4/manywarnings.m4: Update from gnulib.
It's not safe to call qsort or qsort_r, since they have undefined
behavior if the user-specified predicate is not a total order.
Also, watch out for garbage-collection while sorting vectors.
* admin/merge-gnulib (GNULIB_MODULES): Add vla.
* configure.ac (qsort_r): Remove, as we no longer use qsort-like
functions.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/vla.h, m4/vararrays.m4: New files, copied from gnulib.
* lib/stdlib.in.h, m4/stdlib_h.m4: Sync from gnulib, incorporating:
2014-08-29 qsort_r: new module, for GNU-style qsort_r
The previous two files' changes are boilerplate generated by
admin/merge-gnulib, and should not affect Emacs.
* src/fns.c: Include <vla.h>.
(sort_vector_predicate) [!HAVE_QSORT_R]: Remove.
(sort_vector_compare): Remove, replacing with ....
(inorder, merge_vectors, sort_vector_inplace, sort_vector_copy):
... these new functions.
(sort_vector): Rewrite to use the new functions.
GCPRO locals, since the predicate can invoke the GC.
Since it's in-place return void; caller changed.
(merge): Use 'inorder', for clarity.
Fixes: debbugs:18361
* configure.ac (HAVE_TIMERFD): Define only if TFD_CLOEXEC works,
since the code leaked file descriptors to children when !TFD_CLOEXEC.
(HAVE_TIMERFD_CLOEXEC): Remove; no longer used.
* m4/clock_time.m4 (gl_CLOCK_TIME): Don't check for clock_getres.
This reverts the previous change to this file, so it matches
gnulib again.
* src/atimer.c (TIMERFD_CREATE_FLAGS): Remove; we now assume TFD_CLOEXEC.
(alarm_timer, alarm_timer_ok, set_alarm, init_atimer):
Fall back on timer_create if timerfd_create fails at runtime.
(resolution) [HAVE_CLOCK_GETRES]: Remove; we now rely on the
kernel primitives to do resolution. All uses removed.
(timerfd) [!HAVE_TIMERFD]: Define to be -1, for convenience.
(turn_on_atimers): Clear timer_create-based timers too,
for consistency.
* configure.ac (toplevel): Check whether GNU/Linux-specific
timerfd functions and macros are available.
* m4/clock_time.m4 (gl_CLOCK_TIME): Check for clock_getres as well.
* src/atimer.c (toplevel) [HAVE_TIMERFD]: Include sys/timerfd.h.
(toplevel): Rename alarm_timer_ok to special_timer_available.
[HAVE_TIMERFD]: Declare timerfd.
[HAVE_CLOCK_GETRES]: Declare resolution.
(start_atimer) [HAVE_CLOCK_GETRES]: Round up timestamp to
system timer resolution.
(set_alarm) [HAVE_TIMERFD]: Use timerfd_settime.
(timerfd_callback) [HAVE_TIMERFD]: New function.
(atimer_result, debug_timer_callback, Fdebug_timer_check)
[ENABLE_CHECKING]: New function for the sake of automated tests.
(init_atimer) [HAVE_TIMERFD]: Setup timerfd.
[HAVE_CLOCK_GETRES]: Likewise for system timer resolution.
[ENABLE_CHECKING]: Defsubr test function.
* src/atimer.h (timerfd_callback) [HAVE_TIMERFD]: Add prototype.
* src/lisp.h (add_timer_wait_descriptor) [HAVE_TIMERFD]: Likewise.
* src/process.c (add_timer_wait_descriptor) [HAVE_TIMERFD]: New function.
* test/automated/timer-tests.el (timer-tests-debug-timer-check): New test.
This fixes a bug with the shell freezing. See:
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=17598#185
Merge from gnulib, incorporating:
2014-05-31 dup2, fcntl, fcntl-h: port to AIX 7.1
* lib/fcntl.in.h, m4/dup2.m4, m4/fcntl.m4:
Update from gnulib.
This incorporates:
2014-06-02 acl: apply pure attribute to two functions
2014-06-01 gnulib-common.m4: add _GL_UNUSED_LABEL
2014-05-31 dup2, fcntl, fcntl-h: port to AIX 7.1
2014-05-30 ftoastr: work around compiler bug in IBM xlc 12.1
* lib/acl-internal.h, lib/fcntl.in.h, lib/ftoastr.h:
* m4/dup2.m4, m4/fcntl.m4, m4/gnulib-common.m4:
Update from gnulib.
sigprocmask isn't portable in a multithreaded process.
* configure.ac (gl_THREADLIB): Remove dummy.
Merge from gnulib, incorporating:
2014-05-28 pthread_sigmask, timer-time: use gl_THREADLIB only if needed
* m4/pthread_sigmask.m4, m4/timer_time.m4: Update from gnulib.
* src/Makefile.in (LIB_PTHREAD_SIGMASK): Remove; all uses removed.
Fixes: debbugs:17561
* INSTALL: Prefer './configure FOO=BAR' to 'FOO=BAR ./configure'.
* INSTALL.REPO: pkg-config is no longer required to build from
the repository.
* autogen.sh: Don't check for pkg-config.
(progs): Remove pkg-config.
(pkg_config_min, AUTORECONF_ENV, env_space, ACLOCAL_PATH):
Remove. All uses removed.
* m4/pkg.m4: New file, built by admin/merge-pkg-config.
* configure.ac: Remove unnecessary m4_pattern_forbid of ^PKG_ and
an AC_ARG_VAR of PKG_CONFIG_PATH. pkg.m4 does that for us.
(EMACS_CHECK_MODULES): Remove workaround for old pkg-config bug,
as we use pkg.m4 from a newer pkg-config.
* admin/merge-pkg-config: New script.
* admin/notes/copyright: Update for m4/*.m4, in particular m4/pkg.m4.
* etc/NEWS: Prefer './configure FOO=BAR' to 'FOO=BAR ./configure'.
* etc/PROBLEMS (Build-time-problems): Remove pkg-config problem
that is no longer an issue.
* nt/INSTALL: Remove no-longer-needed notes about pkg-config.
2014-02-21 timer: fix uClibc detection of threading
2014-02-21 maintainer-makefiles: provide AC_PROG_SED for older autoconf
* texinfo.tex: Update from gnulib.
2013-12-17 gettimeofday: port recent C++ fix to Emacs
2013-12-17 gettimeofday: fix C++ crosscompilation
2013-12-17 qacl: port to Windows better
* lib/file-has-acl.c, lib/time.in.h, m4/gettimeofday.m4, m4/time_h.m4:
Update from gnulib.
* lib/gnulib.mk: Regenerate.
This scorches the earth and waits for spring;
see Ted Zlatanov and Stefan Monnier in
<http://lists.gnu.org/archive/html/emacs-devel/2013-12/msg00323.html>.
* configure.ac (with_openssl_default, HAVE_LIB_CRYPTO): Remove.
Do not say whether Emacs is configured to use a crypto library,
since it's no longer an option.
(gl_CRYPTO_CHECK): Define a dummy.
* lib/gl_openssl.h, m4/gl-openssl.m4: Remove.
* admin/merge-gnulib: Remove lib/gl_openssh.h and m4/gl-openssl.m4.
* src/Makefile.in (LIB_CRYPTO): Remove.
(LIBES): Don't use it.
On commonly used platform libcrypto uses architecture-specific
assembly code, which is significantly faster than the C code we
were using. See Pádraig Brady's note in
<http://lists.gnu.org/archive/html/bug-gnulib/2013-12/msg00000.html>.
Merge from gnulib, incorporating:
2013-12-07 md5, sha1, sha256, sha512: add gl_SET_CRYPTO_CHECK_DEFAULT
2013-12-07 md5, sha1, sha256, sha512: add 'auto', and set-default method
2013-12-04 include_next: minimize code duplication
2013-12-03 md5, sha1, sha256, sha512: support mandating use of openssl
2013-12-02 md5, sha1, sha256, sha512: use openssl routines if available
* configure.ac (--without-all): Set with_openssl_default too.
Use gl_SET_CRYPTO_CHECK_DEFAULT to default to 'auto'.
(HAVE_LIB_CRYPTO): New var.
Say whether Emacs is configured to use a crypto library.
* lib/gl_openssl.h, m4/absolute-header.m4, m4/gl-openssl.m4:
New files, copied from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/md5.c, lib/md5.h, lib/sha1.c, lib/sha1.h:
* lib/sha256.c, lib/sha256.h, lib/sha512.c, lib/sha512.h:
* m4/include_next.m4, m4/md5.m4, m4/sha1.m4, m4/sha256.m4, m4/sha512.m4:
Update from gnulib.
* src/Makefile.in (LIB_CRYPTO): New macro.
(LIBES): Use it.
This incorporates:
2013-11-08 extern-inline: port better to OS X 10.9
2013-11-08 fpending: fix regression on DragonFly BSD
* lib/fpending.h, m4/extern-inline.m4, m4/fpending.m4:
Update from gnulib.
This incorporates:
2013-10-14 acl: allow cross-compilation to Gentoo
2013-10-18 extern-inline: make safe for -Wundef usage
2013-09-30 fpending: use pure+const function attrs
* lib/fpending.h, m4/acl.m4, m4/extern-inline.m4: Update from gnulib.
This incorporates:
2013-10-10 strtoumax: port to Solaris 8
2013-10-09 strtoimax, strtoumax: port to HP-UX 11.11
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/inttypes.in.h, lib/strtoimax.c, m4/inttypes.m4, m4/strtoimax.m4:
* m4/strtoumax.m4:
Update from gnulib.
Do this by using the Gnulib modules for this.
This should generate faster code on non-GCC, non-MSC platforms,
and make the code a bit more portable, at least in theory.
* admin/merge-gnulib (GNULIB_MODULES): Add count-one-bits
and count-trailing-zeros.
* lib/count-one-bits.c, lib/count-one-bits.h:
* lib/count-trailing-zeros.c, lib/count-trailing-zeros.h:
* m4/count-one-bits.m4, m4/count-trailing-zeros.m4:
New files, copied from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* nt/gnulib.mk: Merge changes from lib/gnulib.mk.
* src/data.c: Include <count-one-bits.h>, <count-trailing-zeros.h>.
(USE_MSC_POPCOUNT, POPCOUNT_STATIC_INLINE)
(NEED_GENERIC_POPCOUNT, popcount_size_t_generic)
(popcount_size_t_msc, popcount_size_t_gcc):
Remove; now done by Gnulib.
(popcount_size_t): Now a macro that defers to Gnulib.
(count_trailing_zero_bits): Return int, for consistency with
Gnulib and because Emacs prefers signed to unsigned int.
Don't assume that size_t is either unsigned int or unsigned long
or unsigned long long.
(size_t_to_host_endian): Do not assume that size_t is either
exactly 32 or exactly 64 bits wide.
* src/lisp.h (BITS_PER_SIZE_T): Define consistently with BITS_PER_LONG
etc., so that it's now an enum constant, not a macro.
No need to assume that it's either 32 or 64.
Fixes: debbugs:15550
* src/conf_post.h (__has_builtin, assume): Remove; gnulib now does these.
* src/lisp.h: Include <verify.h>, for 'assume'.
This also incorpoprates:
2013-10-02 verify: new macro 'assume'
2013-09-26 dup2, dup3: work around another cygwin crasher
2013-09-26 getdtablesize: work around cygwin issue
This incorporates:
2013-09-24 manywarnings: enable nicer gcc warning messages
2013-09-23 warnings: port --enable-gcc-warnings to Solaris Studio 12.3
2013-09-21 timespec: use the new TIMESPEC_RESOLUTION elsewhere
* configure.ac (WERROR_CFLAGS): Omit -fdiagnostics-show-option
and -funit-at-a-time, since manywarnings does that for us now.
This incorporates the following changes:
2013-09-19 stdio: OS X port of putc_unlocked + extern inline
2013-09-19 signal: OS X port of sigaddset etc. + extern inline
2013-09-19 extern-inline: do not always suppress extern inline on OS X
2013-09-17 getgroups: statement without effect
2013-08-28 headers: check that _GL_INLINE_HEADER_BEGIN is defined
Gnulib's emulation of mkostemp doesn't have races that Emacs's does.
* configure.ac (mkostemp): Remove check for this function;
gnulib does the check now.
(mkstemp): Remove check for this no-longer-used function.
* lib/mkostemp.c, lib/secure_getenv.c, lib/tempname.c, lib/tempname.h:
* m4/mkostemp.m4, m4/secure_getenv.m4, m4/tempname.m4:
New files, copied from Gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* admin/merge-gnulib (GNULIB_MODULES): Add mkostemp.
* lib-src/movemail.c (main):
* lib-src/update-game-score.c (write_scores):
Use mkostemp (which now works on all platforms, due to changes
in the portability layer) rather than mktemp (which has a race)
or mkstemp (which we no longer bother with).
* src/callproc.c (create_temp_file):
* src/filelock.c (create_lock_file):
Assume mkostemp, since it's now provided by Gnulib.
Fixes: debbugs:15015
* lib/binary-io.c, lib/binary-io.h: New files.
Merge from gnulib, incorporating:
2013-07-09 accept4, dup3, pipe2: port to Cygwin
* lib/pipe2.c: Update from gnulib, as part of this merge.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
Fixes: debbugs:14821
* lib/ignore-value.h: Remove this gnulib-imported file.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* admin/merge-gnulib (GNULIB_MODULES): Remove ignore-value.
* src/callproc.c (child_setup) [!DOS_NT]: Don't try to stuff an error
number into an exit status. Instead, use EXIT_CANCELED.
(child_setup) [!MSDOS]: Avoid possible deadlock with vfork.
* src/callproc.c (relocate_fd):
* src/emacs.c (close_output_streams, main):
* src/process.c (create_process):
* src/sysdep.c (sys_subshell) [!DOS_NT || !WINDOWSNT]:
Use emacs_perror for simplicity.
* src/callproc.c (relocate_fd, main):
* src/sysdep.c (sys_subshell):
Exit with EXIT_CANCELED etc., not 1, when exec setup fails.
(shut_down_emacs): Use emacs_write, not write.
* src/emacs.c, src/sysdep.c: Don't include <ignore-value.h>.
* src/fileio.c (Fcopy_file, e_write):
* src/nsterm.m (ns_select):
* src/process.c (send_process):
* src/sound.c (vox_write):
Use emacs_write_sig, not emacs_write.
* src/lisp.h (emacs_write_sig, emacs_perror): New decls.
* src/process.h (EXIT_CANCELED), EXIT_CANNOT_INVOKE, EXIT_ENOENT):
New constants.
* src/sysdep.c (emacs_backtrace): Use emacs_write, not ignore_value
of write.
(emacs_full_write): New function.
(emacs_write): Rewrite to use it.
(emacswrite_sig, emacs_perror): New functions.
* src/xrdb.c (fatal): Don't invoke perror, since errno might be garbage.
