* alloc.c (valid_pointer_p) [!WINDOWSNT]: Remove now-unnecessary cast.
* sysdep.c (emacs_read, emacs_write, emacs_write_sig):
Buffer arg is now void *, not char *. This matches plain
'read' and 'write' better, and avoids a constraint violation
on Solaris 9 with Oracle Studio.
Like EMACS_TIME, this portability layer is no longer needed, since
Emacs has been using fd_set as a portability layer for some time.
* sysselect.h (FD_SETSIZE): Rename from MAXDESC. All uses changed.
(SELECT_TYPE): Remove. All uses changed to fd_set.
(fd_set) [!FD_SET]: New typedef.
This portability layer is no longer needed, since Emacs has been
using struct timespec as a portability layer for some time.
Merge from gnulib, incorporating:
2013-08-27 timespec: new convenience constants and function
* src/atimer.h, src/buffer.h, src/dispextern.h, src/xgselect.h:
Include <time.h> rather than "systime.h"; that's all that's needed now.
* src/dispnew.c: Include <timespec.h> rather than "systime.h";
that's all that's needed now.
* src/systime.h (EMACS_TIME): Remove. All uses changed to struct timespec.
(EMACS_TIME_RESOLUTION): Remove. All uses changed to
TIMESPEC_RESOLUTION.
(LOG10_EMACS_TIME_RESOLUTION): Remove. All uses changed to
LOG10_TIMESPEC_RESOLUTION.
(EMACS_SECS, emacs_secs_addr): Remove. All uses changed to tv_sec.
(EMACS_NSECS): Remove. All uses changed to tv_nsec.
(make_emacs_time): Remove. All used changed to make_timespec.
(invalid_timespec): Rename from invalid_emacs_time. All uses changed.
(current_timespec): Rename from current_emacs_time. All uses changed.
(add_emacs_time): Remove. All uses changed to timespec_add.
(sub_emacs_time): Remove. All uses change dot timespec_sub.
(EMACS_TIME_SIGN): Remove. All uses changed to timespec_sign.
(timespec_valid_p): Rename from EMACS_TIME_VALID_P. All uses changed.
(EMACS_TIME_FROM_DOUBLE): Remove. All uses changed to dtotimespec.
(EMACS_TIME_TO_DOUBLE): Remove. All uses changed to timespectod.
(current_timespec): Rename from current_emacs_time. All uses changed.
(EMACS_TIME_EQ, EMACS_TIME_LT, EMACS_TIME_LE): Remove. All uses
changed to timespec_cmp.
* src/xgselect.c: Include <timespec.h>, since our .h files don't.
* process.c (Fset_process_window_size): Signal an error if
the window size is outside the range supported by the lower level.
* sysdep.c (set_window_size): Return negative on error,
nonnegative on success, rather than -1, 0, 1 on not in system,
failure, success. This is simpler. Caller changed.
(serial_configure): Remove unnecessary initialization of local.
(procfs_get_total_memory) [GNU_LINUX]: Don't assume system memory
size fits in unsigned long; this isn't true on some 32-bit hosts.
Avoid buffer overrun if some future version of /proc/meminfo has a
variable name longer than 20 bytes.
(system_process_attributes) [__FreeBSD__]:
Don't assume hw.availpages fits in 'int'.
Fix some related minor problems involving "//", vfork.
* callproc.c (encode_current_directory): New function.
(call_process): Don't append "/"; not needed.
* fileio.c (file_name_as_directory_slop): New constant.
(file_name_as_directory): Allow SRC to be longer than SRCLEN;
this can save the caller having to alloca.
(Ffile_name_as_directory, Fdirectory_file_name, Fexpand_file_name):
Use SAFE_ALLOCA, not alloca.
(directory_file_name, Fexpand_file_name): Leave leading "//"
alone, since it can be special even on POSIX platforms.
* callproc.c (call_process):
* process.c (Fformat_network_address):
* sysdep.c (sys_subshell):
Use encode_current_directory rather than rolling our own.
(create_process): No need to encode directory; caller does that now.
* process.h (encode_current_directory): New decl.
* sysdep.c (sys_subshell): Work even if vfork trashes saved_handlers.
Rework to avoid 'goto xyzzy;'.
(procfs_ttyname): Don't use uninitialized storage if emacs_fopen
or fscanf fails.
(system_process_attributes): Prefer plain char to unsigned char
when either will do. Clean up properly if interrupted or if
memory allocations fail. Don't assume sscanf succeeds. Remove
no-longer-needed workaround to stop GCC from whining. Read
command-line once, instead of multiple times. Check read status a
bit more carefully.
* fileio.c (Fcopy_file, Finsert_file_contents, Fwrite_region):
Say "Read error" and "Write error", rather than "I/O error", or
"IO error reading", or "IO error writing", when a read or write
error occurs.
* process.c (Fmake_network_process, wait_reading_process_output)
(send_process, Fprocess_send_eof, wait_reading_process_output):
Capitalize diagnostics consistently. Put "failed foo" at the
start of the diagnostic, so that we don't capitalize the
function name "foo". Consistently say "failed" for such
diagnostics.
* sysdep.c, w32.c (serial_open): Now accepts Lisp string, not C string.
All callers changed. This is so it can use report_file_error.
* sysdep.c (serial_open, serial_configure): Capitalize I/O
diagnostics consistently as above.
* sysdep.c (emacs_pipe): New function, that implements
pipe2 (fd, O_CLOEXEC) even on hosts that lack O_CLOEXEC.
This should port better to CentOS 5 and to Mac OS X 10.6.
All calls to pipe2 changed.
Fixes: debbugs:14862
* process.c (create_process, wait_reading_process_output) [AIX]:
Remove obsolete SIGHUP-related code, as Emacs no longer disables
SIGHUP, so EBADF is no longer acceptable here (it wouldn't work in
a multithreaded environment anyway).
* sysdep.c (emacs_close): It's not dangerous to invoke emacs_close (-1).
* configure.ac (PTY_TTY_NAME_SPRINTF): Use emacs_close, not close.
* src/callproc.c (Fcall_process_region):
* src/dired.c (open_directory):
* src/emacs.c (main, Fdaemon_initialized):
* src/image.c (x_find_image_file):
* src/inotify.c (Finotify_rm_watch):
* src/lread.c (Flocate_file_internal):
* src/process.c (Fnetwork_interface_list, Fnetwork_interface_info):
* src/term.c (term_mouse_moveto, init_tty):
* src/termcap.c (tgetent):
* src/unexaix.c, src/unexcoff.c (report_error, report_error_1, adjust_lnnoptrs)
* src/unexaix.c, src/unexcoff.c, src/unexcw.c, src/unexelf.c (unexec):
* src/unexhp9k800.c, src/unexmacosx.c (unexec):
* src/callproc.c (Fcall_process_region):
Use emacs_close, not close.
* src/sysdep.c (POSIX_CLOSE_RESTART, posix_close) [!POSIX_CLOSE_RESTART]:
New macro and function, which emulates the POSIX_CLOSE_RESTART macro
and posix_close function on current platforms (which all lack them).
(emacs_close): Use it. This should fix the races on GNU/Linux and
on AIX and on future platforms that support POSIX_CLOSE_RESTART,
and it should avoid closing random victim file descriptors on
other platforms.
* lib-src/ebrowse.c (USAGE): Remove macro with too-long string literal ...
(usage_message): ... and replace it with this new static constant
containing multiple literals. All uses changed.
* lib-src/emacsclient.c (print_help_and_exit):
Rewrite to avoid string literals longer than the C89 limits.
(start_daemon_and_retry_set_socket):
Rewrite to avoid non-constant array initializer.
* lib-src/make-docfile.c (enum global_type): Omit trailing comma.
* src/bytecode.c (BYTE_CODE_THREADED): Do not define if __STRICT_ANSI__.
(B__dummy__): New dummy symbol, to pacify C89.
* src/dbusbind.c (XD_DEBUG_MESSAGE): Omit debugging on C89 hosts, since
they can't grok varargs macros.
* src/dispnew.c (add_window_display_history)
(add_frame_display_history):
* src/print.c (print_object):
* src/xdisp.c (debug_method_add):
Use %p printf format only for void pointers.
* src/emacs.c (usage_message): New constant, replacing ...
(USAGE1, USAGE2, USAGE3): Remove; they were too long for C89.
(main): Adjust to usage reorg.
* src/fns.c (syms_of_fns):
* src/profiler.c (syms_of_profiler):
Don't use non-constant struct initializers.
* src/gnutls.h (gnutls_initstage_t):
* src/lisp.h (enum Lisp_Fwd_Type):
* src/lread.c (lisp_file_lexically_bound_p):
* src/xsettings.c (anonymous enum):
Remove trailing comma.
* src/xsettings.c (apply_xft_settings): Use %f, not %lf; %lf is a C99ism.
* src/lisp.h (ENUM_BF): Use unsigned if pedantic.
(DEFUN_FUNCTION_INIT): New macro, that falls back on a cast if pre-C99.
(DEFUN): Use it.
* src/regex.c (const_re_char): New type, to pacify strict C89.
All uses of 'const re_char' replaced to use it.
* src/regex.h (_Restrict_): Rename from __restrict, to avoid clash
with glibc when strict C89. This change is imported from gnulib.
All uses changed.
(_Restrict_arr_): Rename from __restrict_arr, similarly.
* src/sysdep.c (time_from_jiffies) [!HAVE_LONG_LONG_INT]:
Omit GNU_LINUX implementation, since it requires long long.
* src/xterm.c (x_draw_underwave):
Do not assume the traditional order of struct's members.
(x_term_init): Rewrite to avoid the need for non-constant structure
initializers.
* 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.
This simplifies Emacs a bit, since it no longer needs to worry
about closing file descriptors by hand in some cases.
It also fixes some unlikely races. Not all such races, as
libraries often open files internally without setting
close-on-exec, but it's an improvement.
* admin/merge-gnulib (GNULIB_MODULES): Add fcntl, pipe2.
(GNULIB_TOOL_FLAGS): Avoid binary-io, close. Do not avoid fcntl.
* configure.ac (mkostemp): New function to check for.
(PTY_OPEN): Pass O_CLOEXEC to posix_openpt.
* lib/fcntl.c, lib/getdtablesize.c, lib/pipe2.c, m4/fcntl.m4:
* m4/getdtablesize.m4, m4/pipe2.m4: New files, taken from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* nt/gnulib.mk: Remove empty gl_GNULIB_ENABLED_verify section;
otherwise, gnulib-tool complains given close-on-exec changes.
* nt/inc/ms-w32.h (pipe): Remove.
* nt/mingw-cfg.site (ac_cv_func_fcntl, gl_cv_func_fcntl_f_dupfd_cloexec)
(gl_cv_func_fcntl_f_dupfd_works, ac_cv_func_pipe2): New vars.
* src/alloc.c (valid_pointer_p) [!WINDOWSNT]:
* src/callproc.c (Fcall_process) [!MSDOS]:
* src/emacs.c (main) [!DOS_NT]:
* src/nsterm.m (ns_term_init):
* src/process.c (create_process):
Use 'pipe2' with O_CLOEXEC instead of 'pipe'.
* src/emacs.c (Fcall_process_region) [HAVE_MKOSTEMP]:
* src/filelock.c (create_lock_file) [HAVE_MKOSTEMP]:
Prefer mkostemp with O_CLOEXEC to mkstemp.
* src/callproc.c (relocate_fd) [!WINDOWSNT]:
* src/emacs.c (main): Use F_DUPFD_CLOEXEC, not plain F_DUPFD.
No need to use fcntl (..., F_SETFD, FD_CLOEXEC), since we're
now using pipe2.
* src/filelock.c (create_lock_file) [! HAVE_MKOSTEMP]:
Make the resulting file descriptor close-on-exec.
* src/lisp.h, src/lread.c, src/process.c (close_load_descs, close_process_descs):
* src/lread.c (load_descriptor_list, load_descriptor_unwind):
Remove; no longer needed. All uses removed.
* src/process.c (SOCK_CLOEXEC): Define to 0 if not supplied by system.
(close_on_exec, accept4, process_socket) [!SOCK_CLOEXEC]:
New functions.
(socket) [!SOCK_CLOEXEC]: Supply a substitute.
(Fmake_network_process, Fnetwork_interface_list):
(Fnetwork_interface_info, server_accept_connection):
Make newly-created socket close-on-exec.
* src/sysdep.c (emacs_open, emacs_fopen):
Make new-created descriptor close-on-exec.
* src/w32.c (fcntl): Support F_DUPFD_CLOEXEC well enough for Emacs.
* src/w32.c, src/w32.h (pipe2): Rename from 'pipe', with new flags arg.
Fixes: debbugs:14803
* callproc.c, process.h (block_child_signal, unblock_child_signal):
Now extern.
* emacs.c (main): Catch SIGCHLD just before initializing gfilenotify.
* process.c (catch_child_signal): Block SIGCHLD while futzing with
the SIGCHLD handler, since the code is not atomic and (due to glib)
signals may be arriving now.
* sysdep.c (init_signals): Do not catch child signals here;
'main' now does that later, at a safer time.
Fixes: debbugs:14569
* process.c, process.h (catch_child_signal):
Now always extern, even if !NS_IMPL_GNUSTEP.
* process.c (catch_child_signal): Move glib tickler here from
init_process_emacs, so that it's done earlier in Emacs
initialization. Also move the noninteractive && !initialized
check here from init_process_emacs. This is all a bit cleaner for
GNUish platforms, and I hope it works around the Cygwin bug.
* sysdep.c (init_signals): Invoke catch_child_signal here, so
that glib signal handling is tickled before glib creates threads.
Fixes: debbugs:14569
* admin/CPP-DEFINES (BSD_SYSTEM, HAVE_FSYNC): Remove.
* admin/merge-gnulib (GNULIB_MODULES): Add fsync, fdatasync.
* configure.ac (BSD_SYSTEM, BSD_SYSTEM_AHB): Remove; no longer needed.
(fsync): Remove check; now done by gnulib.
* lib/fdatasync.c, lib/fsync.c, m4/fdatasync.m4, m4/fsync.m4:
New files, from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib-src/Makefile.in (LIB_FDATASYNC): New macro.
(emacsclient${EXEEXT}): Use it.
* lib-src/emacsclient.c (main): Use fdatasync, not fsync, since we don't
care about metadata. Keep trying if interrupted.
* lib-src/movemail.c (main, popmail): Don't worry about BSD_SYSTEM, since
fsync is available everywhere (or there is a substitute). Don't
report an error if fsync returns EINVAL.
* nt/inc/ms-w32.h (fdatasync): New macro, suggested by Eli Zaretskii.
* src/Makefile.in (LIB_FDATASYNC): New macro.
(LIBES): Use it.
* src/conf_post.h (BSD_SYSTEM, BSD_SYSTEM_AHB): Remove; no longer needed.
* src/fileio.c (Fwrite_region, write_region_inhibit_fsync):
Don't worry about HAVE_FSYNC, since a substitute fsync is
available if the system lacks one.
(Fwrite_regin): Retry fsync if interrupted.
Fixes: debbugs:13944
Include <sys/param.h> unconditionally, as that works elsewhere and
is simpler here. Include <sys/sysctl.h> if DARWIN_OS ||
__FreeBSD__, not if BSD_SYSTEM, since it's needed only for Darwin
and FreeBSD now.
* sysdep.c (list_system_processes): Make it a stub on all BSD_SYSTEM
hosts, except for DARWIN_OS and FreeBSD where it's been tested.
Fixes: debbugs:13881
* sysdep.c (KERN_PROC, kinfo_proc) [BSD_SYSTEM && !KERN_PROC]:
Define to KERN_PROC2 and kinfo_proc2, for OpenBSD 4.9.
list-system-processes still returns nil, but at least it doesn't crash.
Problem reported by Nelson H. F. Beebe in
<http://lists.gnu.org/archive/html/emacs-devel/2013-03/msg00021.html>.
On my host, this speeds up directory-files-and-attributes by a
factor of 3, when applied to Emacs's src directory.
These functions are standardized by POSIX and are common these
days; fall back on a (slower) gnulib implementation if the host
is too old to supply them.
* .bzrignore: Add lib/dirent.h.
* lib/Makefile.am (libgnu_a_SOURCES): Add openat-die.c, save-cwd.c.
* lib/careadlinkat.c, lib/careadlinkat.h: Merge from gnulib,
incorporating: 2013-01-29 careadlinkat: do not provide careadlinkatcwd.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/dirent.in.h, lib/fdopendir.c, lib/fstatat.c, lib/openat-priv.h:
* lib/openat-proc.c, lib/openat.h, m4/dirent_h.m4, m4/fdopendir.m4:
* m4/fstatat.m4: New files, from gnulib.
* lib/openat-die.c, lib/save-cwd.c, lib/save-cwd.h: New files.
These last three are specific to Emacs and are not copied from gnulib.
They are simpler than the gnulib versions and are tuned for Emacs.
* admin/merge-gnulib (GNULIB_MODULES): Add fdopendir, fstatat, readlinkat.
(GNULIB_TOOL_FLAGS): Do not avoid at-internal, openat-h.
Avoid dup, open, opendir.
* nt/inc/sys/stat.h (fstatat):
* nt/inc/unistd.h (readlinkat): New decls.
* src/conf_post.h (GNULIB_SUPPORT_ONLY_AT_FDCWD): Remove.
* src/dired.c: Include <fcntl.h>.
(open_directory): New function, which uses open and fdopendir
rather than opendir. DOS_NT platforms still use opendir, though.
(directory_files_internal, file_name_completion): Use it.
(file_attributes): New function, with most of the old Ffile_attributes.
(directory_files_internal, Ffile_attributes): Use it.
(file_attributes, file_name_completion_stat): First arg is now fd,
not dir name. All uses changed. Use fstatat rather than lstat +
stat.
(file_attributes): Use emacs_readlinkat rather than Ffile_symlink_p.
* src/fileio.c: Include <allocator.h>, <careadlinkat.h>.
(emacs_readlinkat): New function, with much of the old
Ffile_symlink_p, but with an fd argument for speed.
It uses readlinkat rather than careadlinkatcwd, so that it
need not assume the working directory.
(Ffile_symlink_p): Use it.
* src/filelock.c (current_lock_owner): Use emacs_readlinkat
rather than emacs_readlink.
* src/lisp.h (emacs_readlinkat): New decl.
(READLINK_BUFSIZE, emacs_readlink): Remove.
* src/sysdep.c: Do not include <allocator.h>, <careadlinkat.h>.
(emacs_norealloc_allocator, emacs_readlink): Remove.
This stuff is moved to fileio.c.
* src/w32.c (fstatat, readlinkat): New functions.
(careadlinkat): Don't check that fd == AT_FDCWD.
(careadlinkatcwd): Remove; no longer needed.
Fixes: debbugs:13539
* lisp.h, sysdep.c (block_tty_out_signal, unblock_tty_out_signal):
New functions.
* term.c (init_tty): Use them instead of rolling our own code.
* sysdep.c (tcsetpgrp_without_stopping): Likewise. Here, this
switches from 'signal' to 'pthread_sigmask', which is safer in
multithreaded applications.
* term.c (Fresume_tty): Don't bother dissociating if O_IGNORE_CTTY,
which has already arranged for that.
(dissociate_if_controlling_tty): If setsid fails, fall back on TIOCNOTTY.
This is the main part of the bug fix.
Fixes: debbugs:13387
Exceptions: do not assume SIGCONT, SIGSTOP, SIGTSTP, SIGTTIN,
SIGTTOU, SIGUSR1, SIGUSR2, as Microsoft platforms lack these.
* admin/CPP-DEFINES (SIGALRM, SIGCHLD, SIGHUP, SIGKILL, SIGPIPE, SIGQUIT):
Remove.
(SIGTRAP): Remove this one too, as config.h no longer defines it.
* admin/merge-gnulib (GNULIB_MODULES): Add sig2str.
* configure.ac (PTY_OPEN, PTY_TTY_NAME_SPRINTF):
Use SIGCHLD rather than SIGCLD.
* lib/sig2str.c, lib/sig2str.h, m4/sig2str.m4: New files, from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* lib/makefile.w32-in (GNULIBOBJS): Add $(BUILD)/sig2str.$(O).
* src/process.c [subprocesses]: Include <c-ctype.h>, <sig2str.h>.
(deleted_pid_list, Fdelete_process, create_process)
(record_child_status_change, handle_child_signal, deliver_child_signal)
(init_process_emacs, syms_of_process):
Assume SIGCHLD is defined.
(parse_signal): Remove. All uses removed.
(abbr_to_signal): New static function.
(Fsignal_process): Use it to convert signal names to ints.
* src/sysdep.c (sys_suspend) [!DOS_NT]: Use kill (0, ...) rather than
kill (getpgrp (), ...).
(emacs_sigaction_init): Assume SIGCHLD is defined.
(init_signals): Assume SIGALRM, SIGCHLD, SIGHUP, SIGKILL,
SIGPIPE, and SIGQUIT are defined. Do not worry about SIGCLD any more.
* src/syssignal.h (EMACS_KILLPG): Remove.
All uses replaced by 'kill' with a negative pid.
(SIGCHLD): Remove definition, as we now assume SIGCHLD.
* src/w32proc.c (sys_kill): Support negative pids compatibly with POSIX.
Fixes: debbugs:13026
Fixing this bug required some cleanup of the signal-handling code.
As a side effect, this change also fixes a longstanding rare race
condition whereby Emacs could mistakenly kill unrelated processes,
and it fixes a bug where a second C-g does not kill a recalcitrant
synchronous process in GNU/Linux and similar platforms.
The patch should also fix the last vestiges of Bug#9488,
a bug which has mostly been fixed on the trunk by other changes.
* callproc.c, process.h (synch_process_alive, synch_process_death)
(synch_process_termsig, sync_process_retcode):
Remove. All uses removed, to simplify analysis and so that
less consing is done inside critical sections.
* callproc.c (call_process_exited): Remove. All uses replaced
with !synch_process_pid.
* callproc.c (synch_process_pid, synch_process_fd): New static vars.
These take the role of what used to be in unwind-protect arg.
All uses changed.
(block_child_signal, unblock_child_signal):
New functions, to avoid races that could kill innocent-victim processes.
(call_process_kill, call_process_cleanup, Fcall_process): Use them.
(call_process_kill): Record killed processes as deleted, so that
zombies do not clutter up the system. Do this inside a critical
section, to avoid a race that would allow the clutter.
(call_process_cleanup): Fix code so that the second C-g works again
on common platforms such as GNU/Linux.
(Fcall_process): Create the child process in a critical section,
to fix a race condition. If creating an asynchronous process,
record it as deleted so that zombies do not clutter up the system.
Do unwind-protect for WINDOWSNT too, as that's simpler in the
light of these changes. Omit unnecessary call to emacs_close
before failure, as the unwind-protect code does that.
* callproc.c (call_process_cleanup):
* w32proc.c (waitpid): Simplify now that synch_process_alive is gone.
* process.c (record_deleted_pid): New function, containing
code refactored out of Fdelete_process.
(Fdelete_process): Use it.
(process_status_retrieved): Remove. All callers changed to use
child_status_change.
(record_child_status_change): Remove, folding its contents into ...
(handle_child_signal): ... this signal handler. Now, this
function is purely a handler for SIGCHLD, and is not called after
a synchronous waitpid returns; the synchronous code is moved to
wait_for_termination. There is no need to worry about reaping
more than one child now.
* sysdep.c (get_child_status, child_status_changed): New functions.
(wait_for_termination): Now takes int * status and bool
interruptible arguments, too. Do not record child status change;
that's now the caller's responsibility. All callers changed.
Reimplement in terms of get_child_status.
(wait_for_termination_1, interruptible_wait_for_termination):
Remove. All callers changed to use wait_for_termination.
* syswait.h: Include <stdbool.h>, for bool.
(record_child_status_change, interruptible_wait_for_termination):
Remove decls.
(record_deleted_pid, child_status_changed): New decls.
(wait_for_termination): Adjust to API changes noted above.
Fixes: debbugs:12980
This is a backport from the trunk, consisting of:
2012-11-17 Eli Zaretskii <eliz@gnu.org>
* nt/inc/sys/wait.h: New file, with prototype of waitpid and
definitions of macros it needs.
* nt/inc/ms-w32.h (wait): Don't define, 'wait' is not used anymore.
(sys_wait): Remove prototype.
* nt/config.nt (HAVE_SYS_WAIT_H): Define to 1.
* src/w32proc.c (create_child): Don't clip the PID of the child
process to fit into an Emacs integer, as this is no longer a
restriction.
(waitpid): Rename from sys_wait. Emulate a Posix 'waitpid' by
reaping only the process specified by PID argument, if that is
positive. Use PID instead of dead_child to know which process to
reap. Wait for the child to die only if WNOHANG is not in
OPTIONS.
(sys_select): Don't set dead_child.
* src/sysdep.c (wait_for_termination_1): Remove the WINDOWSNT portion,
as it is no longer needed.
* src/process.c (waitpid, WUNTRACED) [!WNOHANG]: Remove definitions,
no longer needed.
(record_child_status_change): Remove the setting of
record_at_most_one_child for the !WNOHANG case.
2012-11-03 Paul Eggert <eggert@cs.ucla.edu>
Fix a race condition that causes Emacs to mess up glib (Bug#8855).
This is a backport from the trunk.
The symptom is a diagnostic "GLib-WARNING **: In call to
g_spawn_sync(), exit status of a child process was requested but
SIGCHLD action was set to SIG_IGN and ECHILD was received by
waitpid(), so exit status can't be returned." The diagnostic
is partly wrong, as the SIGCHLD action is not set to SIG_IGN.
The real bug is a race condition between Emacs and glib: Emacs
does a waitpid (-1, ...) and reaps glib's subprocess by mistake,
so that glib can't find it. Work around the bug by invoking
waitpid only on subprocesses that Emacs itself creates.
* src/process.c (create_process, record_child_status_change):
Don't use special value -1 in pid field, as the caller now must
know the pid rather than having the callee infer it. The
inference was sometimes incorrect anyway, due to another race.
(create_process): Set new 'alive' member if child is created.
(process_status_retrieved): New function.
(record_child_status_change): Use it.
Accept negative 1st argument, which means to wait for the
processes that Emacs already knows about. Move special-case code
for DOS_NT (which lacks WNOHANG) here, from caller. Keep track of
processes that have already been waited for, by testing and
clearing new 'alive' member.
(CAN_HANDLE_MULTIPLE_CHILDREN): Remove, as record_child_status_change
now does this internally.
(handle_child_signal): Let record_child_status_change do all
the work, since we do not want to reap all exited child processes,
only the child processes that Emacs itself created.
* src/process.h (Lisp_Process): New boolean member 'alive'.
* admin/CPP-DEFINES (HAVE_CLOSEDIR, HAVE_DIRENT_H): Remove.
* admin/notes/copyright: Adjust to src/ndir.h -> nt/inc/dirent.h renaming.
* configure.ac: Do not check for dirent.h or closdir.
* nt/inc/dirent.h: Rename from ../src/ndir.h, with these changes:
(struct dirent): Rename from struct direct. All uses changed.
* nt/inc/sys/dir.h: Remove.
* src/dired.c: Assume HAVE_DIRENT_H.
(NAMLEN): Remove, replacing with ...
(dirent_namelen): New function. All uses changed. Use the GNU macro
_D_EXACT_NAMELEN if available, as it's faster than strlen.
(DIRENTRY): Remove, replacing all uses with 'struct dirent'.
(DIRENTRY_NONEMPTY): Remove. All callers now assume it's nonzero.
* src/makefile.w32-in (DIR_H): Remove. All uses replaced with
$(NT_INC)/dirent.h.
($(BLD)/w32.$(O)): Do not depend on $(SRC)/ndir.h.
* src/ndir.h: Rename to ../nt/inc/dirent.h.
* src/sysdep.h (closedir) [!HAVE_CLOSEDIR]: Remove.
Do not include <dirent.h>; no longer needed.
* src/w32.c: Include <dirent.h> rather than "ndir.h".
Fixes: debbugs:12958
