This makes it easier to visualize quantities on a number line.
This patch doesn't apply to all such range checks,
only to the range checks affected by the 2013-03-24 change.
This patch reverts most of the 2013-03-24 change.
* alloc.c (xpalloc, Fgarbage_collect):
* ccl.c (ccl_driver, resolve_symbol_ccl_program):
* character.c (string_escape_byte8):
* charset.c (read_hex):
* data.c (cons_to_unsigned):
* dispnew.c (update_frame_1):
* doc.c (Fsubstitute_command_keys):
* doprnt.c (doprnt):
* editfns.c (hi_time, decode_time_components):
* fileio.c (file_offset):
* fns.c (larger_vector, make_hash_table, Fmake_hash_table):
* font.c (font_intern_prop):
* frame.c (x_set_alpha):
* gtkutil.c (get_utf8_string):
* indent.c (check_display_width):
* keymap.c (Fkey_description):
* lisp.h (FIXNUM_OVERFLOW_P, vcopy):
* lread.c (read1):
* minibuf.c (read_minibuf_noninteractive):
* process.c (wait_reading_process_output):
* search.c (Freplace_match):
* window.c (get_phys_cursor_glyph):
* xdisp.c (redisplay_internal):
* xsmfns.c (smc_save_yourself_CB):
Prefer < to > for range checks.
* dispnew.c (sit_for): Don't mishandle NaNs.
This fixes a bug introduced in the 2013-03-24 change.
* editfns.c (decode_time_components): Don't hoist comparison.
This fixes another bug introduced in the 2013-03-24 change.
* image.c (gif_load): Assume pass < 3 to pacify GCC.
* process.c (Fset_process_datagram_address)
(Fmake_network_process): Check get_lisp_to_sockaddr_size return value.
* xdisp.c (get_char_face_and_encoding):
(get_glyph_face_and_encoding): Ensure that *CHAR2B is initialized.
(get_glyph_face_and_encoding): Prepare face before possibly using it.
(get_per_char_metric): Don't use CHAR2B if it might not be initialized.
With the 2012-12-03 fix for Bug#12980 in place, an old workaround
for some of that bug's symptoms can now cause Emacs to abort.
Remove the workaround.
* process.c (wait_reading_process_output): Don't reraise SIGCHLD.
The bug that caused SIGCHLD to get lost has been fixed, and the
workaround for it can now cause Emacs to abort.
Fixes: debbugs:13192
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
* callproc.c, process.h (record_kill_process):
New function, containing part of the old call_process_kill.
(call_process_kill): Use it.
This does not change call_process_kill's behavior.
* process.c (Fdelete_process): Use record_kill_process to fix a
race condition that could cause Emacs to lose track of a child.
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
* callproc.c (Fcall_process):
* process.c (create_process):
Don't save and restore environ; no longer needed.
* callproc.c (child_setup):
Use execve, not execvp, to preserve environ.
Fixes: debbugs:13054
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 (O_RDONLY, O_RDWR, HAVE_FCNTL_H): Remove.
* admin/merge-gnulib (GNULIB_MODULES): Add fcntl-h.
* configure.ac: Do not check for fcntl.h.
* lib/gnulib.mk: Regenerate.
* lib-src/movemail.c, lib-src/update-game-score.c: Assume <fcntl.h> exists.
* nt/inc/sys/socket.h (O_NONBLOCK): Rename from O_NDELAY, since the
POSIX name for this flag is O_NONBLOCK. All uses changed.
* nt/inc/unistd.h (O_RDWR, O_NOCTTY): New macros. Like AT_FDCWD etc.
these really should be moved to a replacement <fcntl.h> if and
when that gets implemented. In the meantime, include <fcntl.h>
to make sure we don't override its definitions.
* src/callproc.c (relocate_fd): Assume F_DUPFD.
* src/emacs.c, src/term.c (O_RDWR): Remove.
* src/keyboard.c (tty_read_avail_input): Use O_NONBLOCK rather than
O_NDELAY, since O_NONBLOCK is the standard name for this flag.
* src/nsterm.m: Assume <fcntl.h> exists.
* src/process.c (NON_BLOCKING_CONNECT, allocate_pty, create_process)
(create_pty, Fmake_network_process, server_accept_connection)
(wait_reading_process_output, init_process_emacs):
Assume O_NONBLOCK.
(wait_reading_process_output): Put in a special case for WINDOWSNT
to mimick the older behavior where it had O_NDELAY but not O_NONBLOCK.
It's not clear this is needed, but it's a more-conservative change.
(create_process): Assume FD_CLOEXEC.
(create_process, create_pty): Assume O_NOCTTY.
* src/sysdep.c (init_sys_modes, reset_sys_modes): Assume F_SETFL.
(reset_sys_modes): Use O_NONBLOCK rather than O_NDELAY.
Omit if not DOS_NT, since F_GETFL is not defined there.
(serial_open): Assume O_NONBLOCK and O_NOCTTY.
* src/term.c: Include <fcntl.h>, for flags like O_NOCTTY.
(O_NOCTTY): Remove.
(init_tty): Assume O_IGNORE_CTTY is defined to 0 on platforms that
lack it, since gnulib guarantees this.
* src/w32.c (fcntl): Test for O_NONBLOCK rather than O_NDELAY.
Fixes: debbugs:12881
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.
* configure.ac (fpathconf): Remove unnecessary check.
* admin/CPP-DEFINES (HAVE_FPATHCONF): Remove.
* src/process.c (pty_max_bytes): Remove; unused.
(send_process): Do not set it.
This fixes a bug that has been present in Emacs since its creation.
It was reported by Chris Torek in 1983 even before GNU Emacs existed,
which must set some sort of record. (Torek's bug report was against
a predecessor of GNU Emacs, but GNU Emacs happened to have the
same common flaw.) See Torek's Usenet posting
"setuid/setgid programs & Emacs" Article-I.D.: sri-arpa.858
Posted: Fri Apr 8 14:18:56 1983.
* .bzrignore: Add lib/fcntl.h.
* configure.ac (euidaccess): Remove check; gnulib does this for us now.
(gl_FCNTL_O_FLAGS): Define a dummy version.
* lib/at-func.c, lib/euidaccess.c, lib/faccessat.c, lib/fcntl.in.h:
* lib/getgroups.c, lib/group-member.c, lib/root-uid.h:
* lib/xalloc-oversized.h, m4/euidaccess.m4, m4/faccessat.m4:
* m4/fcntl_h.m4, m4/getgroups.m4, m4/group-member.m4:
New files, from gnulib.
* lib/gnulib.mk, m4/gnulib-comp.m4: Regenerate.
* admin/merge-gnulib (GNULIB_MODULES): Add faccessat.
(GNULIB_TOOL_FLAGS): Avoid at-internal, fchdir, malloc-posix,
openat-die, openat-h, save-cwd. Do not avoid fcntl-h.
Omit gnulib's m4/fcntl-o.m4.
* nt/inc/ms-w32.h (AT_FDCWD, AT_EACCESS): New symbols.
(access): Remove.
(faccessat): New macro.
* src/Makefile.in (LIB_EACCESS): New macro.
(LIBES): Use it.
* src/callproc.c (init_callproc):
* src/charset.c (init_charset):
* src/fileio.c (check_existing, check_executable, check_writable)
(Ffile_readable_p):
* src/lread.c (openp, load_path_check):
* src/process.c (allocate_pty):
* src/xrdb.c (file_p):
Use effective UID when checking permissions, not real UID.
* src/callproc.c (init_callproc):
* src/charset.c (init_charset):
* src/lread.c (load_path_check, init_lread):
Test whether directories are accessible, not merely whether they exist.
* src/conf_post.h (GNULIB_SUPPORT_ONLY_AT_FDCWD): New macro.
* src/fileio.c (check_existing, check_executable, check_writable)
(Ffile_readable_p):
Use symbolic names instead of integers for the flags, as they're
portable now.
(check_writable): New arg AMODE. All uses changed.
Set errno on failure.
(Ffile_readable_p): Use faccessat, not stat + open + close.
(Ffile_writable_p): No need to call check_existing + check_writable.
Just call check_writable and then look at errno. This saves a syscall.
dir should never be nil; replace an unnecessary runtime check
with an eassert. When checking the parent directory of a nonexistent
file, check that the directory is searchable as well as writable, as
we can't create files in unsearchable directories.
(file_directory_p): New function, which uses 'stat' on most platforms
but faccessat with D_OK (for efficiency) if WINDOWSNT.
(Ffile_directory_p, Fset_file_times): Use it.
(file_accessible_directory_p): New function, which uses a single
syscall for efficiency.
(Ffile_accessible_directory_p): Use it.
* src/xrdb.c (file_p): Use file_directory_p.
* src/lisp.h (file_directory_p, file_accessible_directory_p): New decls.
* src/lread.c (openp): When opening a file, use fstat rather than
stat, as that avoids a permissions race. When not opening a file,
use file_directory_p rather than stat.
(dir_warning): First arg is now a usage string, not a format.
Use errno. All uses changed.
* src/nsterm.m (ns_term_init): Remove unnecessary call to file-readable
that merely introduced a race.
* src/process.c, src/sysdep.c, src/term.c: All uses of '#ifdef O_NONBLOCK'
changed to '#if O_NONBLOCK', to accommodate gnulib O_* style,
and similarly for the other O_* flags.
* src/w32.c (sys_faccessat): Rename from sys_access and switch to
faccessat's API. All uses changed.
* src/xrdb.c: Do not include <sys/stat.h>; no longer needed.
(magic_db): Rename from magic_file_p.
(magic_db, search_magic_path): Return an XrmDatabase rather than a
char *, so that we don't have to test for file existence
separately from opening the file for reading. This removes a race
fixes a permission-checking problem, and simplifies the code.
All uses changed.
(file_p): Remove; no longer needed.
Fixes: debbugs:12632
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.
* 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.
* process.h (Lisp_Process): New boolean member 'alive'.
Fixes: debbugs:8855
* callproc.c (setpgrp): Remove macro, as we now use setpgid
and it is configured in conf_post.h.
(Fcall_process): Don't invoke both setsid and setpgid; the former
is enough, if it exists.
* callproc.c (Fcall_process, child_setup):
* process.c (create_process): Use setpgid.
* conf_post.h (setpgid) [!HAVE_SETPGID]: New macro, which substitutes
for the real thing.
* dispnew.c (init_display): Initialize the foreground group
if we are running a tty display.
* emacs.c (main): Do not worry about setpgrp; init_display does it now.
* lisp.h (init_foreground_group): New decl.
* sysdep.c (inherited_pgroup): New static var.
(init_foreground_group, tcsetpgrp_without_stopping)
(narrow_foreground_group, widen_foreground_group): New functions.
(init_sys_modes): Narrow foreground group.
(reset_sys_modes): Widen foreground group.
Fixes: debbugs:12697
* fileio.c (Fwrite_region): Avoid race condition if a file is
removed or renamed by some other process immediately after Emacs
writes it but before Emacs stats it. Do not assume that stat (or
fstat) succeeds.
* image.c (slurp_file): Resolve the file name with fopen + fstat
rather than stat + fopen.
(pbm_read_file) [0]: Remove unused code with stat race.
* process.c (allocate_pty) [HAVE_PTYS && !PTY_ITERATION && !PTY_OPEN]:
Remove ineffective code with stat race.
* dispnew.c (sit_for): Distinguish between 3-way display_option
and boolean do_display.
* keyboard.c (single_kboard, this_command_key_count_reset)
(waiting_for_input, echoing, immediate_quit, input_pending)
(interrupt_input, interrupts_deferred, pop_kboard)
(temporarily_switch_to_single_kboard, ignore_mouse_drag_p)
(command_loop_1, adjust_point_for_property)
(safe_run_hooks_error, input_polling_used, read_char):
(help_char_p, readable_events, kbd_buffer_events_waiting)
(kbd_buffer_get_event, timer_check_2, make_lispy_event)
(lucid_event_type_list_p, get_input_pending):
(gobble_input, menu_separator_name_p, menu_bar_item)
(parse_menu_item, parse_tool_bar_item, read_char_x_menu_prompt)
(read_char_minibuf_menu_prompt, access_keymap_keyremap)
(keyremap_step, test_undefined, read_key_sequence)
(detect_input_pending, detect_input_pending_ignore_squeezables)
(detect_input_pending_run_timers, requeued_events_pending_p)
(quit_throw_to_read_char, Fset_input_interrupt_mode):
* keymap.c (get_keymap, keymap_parent, keymap_memberp)
(access_keymap_1, access_keymap, map_keymap, get_keyelt)
(Fdefine_key, Flookup_key, struct accessible_keymaps_data)
(accessible_keymaps_1, Fkey_description, push_key_description):
(shadow_lookup, struct where_is_internal_data)
(where_is_internal, Fwhere_is_internal, where_is_internal_1)
(Fdescribe_buffer_bindings, describe_map_tree, struct describe_map_elt)
(describe_map, describe_vector):
* menu.c (single_menu_item):
* nsmenu.m (ns_update_menubar):
* process.c (wait_reading_process_output):
* search.c (scan_buffer, scan_newline):
Use bool for boolean.
* keyboard.c (timers_run, swallow_events)
(detect_input_pending_run_timers):
* process.c (wait_reading_process_output):
Use unsigned for counter where wraparound-on-overflow is desired,
since unsigned is guaranteed to have that behavior and signed is not.
(read_char): Use ptrdiff_t for string length.
(get_input_pending): Remove first argument, since it was always
the same pointer-to-int (now pointer-to-boolean) &input_pending,
and behave as if it had that value. Return new value of
input_pending. All callers changed.
* keyboard.h (struct kboard): Use unsigned : 1 for boolean member
immediate_echo. Use ptrdiff_t for echo_after_prompt, since it's
a string length.
* keymap.c (push_key_description): Omit last arg, which was always 1.
All callers changed.
A signal can arrive in the middle of a malloc, and Emacs's signal
handler can invoke strsignal, which can invoke malloc, which is
not portable. This race condition bug makes Emacs hang on GNU/Linux.
Fix it by altering the signal handler so that it does not invoke
strsignal.
* emacs.c (shut_down_emacs): Use safe_strsignal, not strsignal.
* process.c (status_message): Use const pointer, in case strsignal
is #defined to safe_strsignal.
* sysdep.c (sys_siglist, init_signals): Always define and
initialize a substitute sys_siglist if the system does not define
one, even if HAVE_STRSIGNAL.
(safe_strsignal): Rename from strsignal. Always define,
using sys_siglist. Return a const pointer.
* syssignal.h (safe_strsignal): New decl.
(strsignal) [!HAVE_STRSIGNAL]: Define in terms of safe_strsignal.
* nt/inc/ms-w32.h (emacs_raise): New macro.
* src/alloc.c (die):
* src/sysdep.c (emacs_abort) [HAVE_NTGUI]:
Avoid recursive loop if there's a fatal error in the function itself.
* src/atimer.c (pending_atimers):
* src/blockinput.h: Don't include "atimer.h"; no longer needed.
(interrupt_input_pending): Remove. All uses removed.
pending_signals now counts both atimers and ordinary interrupts.
This is less racy than having three separate pending-signal flags.
(block_input, unblock_input, totally_unblock_input, unblock_input_to)
(input_blocked_p):
Rename from their upper-case counterparts BLOCK_INPUT,
UNBLOCK_INPUT, TOTALLY_UNBLOCK_INPUT, UNBLOCK_INPUT_TO,
INPUT_BLOCKED_P, and turn into functions. All uses changed.
This makes it easier to access volatile variables more accurately.
(BLOCK_INPUT_RESIGNAL): Remove. All uses replaced by unblock_input ().
(input_blocked_p): Prefer this to 'interrupt_input_blocked', as
that's more reliable if the code is buggy and sets
interrupt_input_blocked to a negative value. All uses changed.
* src/atimer.c (deliver_alarm_signal):
Remove. No need to deliver this to the parent; any thread can
handle this signal now. All uses replaced by underlying handler.
* src/atimer.c (turn_on_atimers):
* src/dispnew.c (handle_window_change_signal):
* src/emacs.c (handle_danger_signal):
* src/keyboard.c (kbd_buffer_get_event):
Don't reestablish signal handler; not needed with sigaction.
* src/blockinput.h (UNBLOCK_INPUT_TO, TOTALLY_UNBLOCK_INPUT)
(UNBLOCK_INPUT_TO):
Rework to avoid unnecessary accesses to volatile variables.
(UNBLOCK_INPUT_TO): Now a function.
(totally_unblock_input, unblock_input): New decls.
* src/data.c (handle_arith_signal, deliver_arith_signal): Move to sysdep.c
(init_data): Remove. Necessary stuff now done in init_signal.
* src/emacs.c, src/xdisp.c: Include "atimer.h", since we invoke atimer functions.
* src/emacs.c (handle_fatal_signal, deliver_fatal_signal): Move to sysdep.c.
(fatal_error_code): Remove; no longer needed.
(terminate_due_to_signal): Rename from fatal_error_backtrace, since
it doesn't always backtrace. All uses changed. No need to reset
signal to default, since sigaction and/or die does that for us now.
Use emacs_raise (FOO), not kill (getpid (), FOO).
(main): Check more-accurately whether we're dumping.
Move fatal-error setup to sysdep.c
* src/floatfns.c: Do not include "syssignal.h"; no longer needed.
* src/gtkutil.c (xg_get_file_name, xg_get_font):
Remove no-longer-needed signal-mask manipulation.
* src/keyboard.c, src/process.c (POLL_FOR_INPUT):
Don't depend on USE_ASYNC_EVENTS, a symbol that is never defined.
* src/keyboard.c (read_avail_input): Remove.
All uses replaced by gobble_input.
(Ftop_level): Use TOTALLY_UNBLOCK_INPUT rather than open code.
(kbd_buffer_store_event_hold, gobble_input):
(record_asynch_buffer_change) [USABLE_SIGIO]:
(store_user_signal_events):
No need to mess with signal mask.
(gobble_input): If blocking input and there are terminals, simply
set pending_signals to 1 and return. All hooks changed to not
worry about whether input is blocked.
(process_pending_signals): Clear pending_signals before processing
them, in case a signal comes in while we're processing.
By convention callers now test pending_signals before calling us.
(UNBLOCK_INPUT_TO, unblock_input, totally_unblock_input):
New functions, to support changes to blockinput.h.
(handle_input_available_signal): Now extern.
(reinvoke_input_signal): Remove. All uses replaced by
handle_async_input.
(quit_count): Now volatile, since a signal handler uses it.
(handle_interrupt): Now takes bool IN_SIGNAL_HANDLER as arg. All
callers changed. Block SIGINT only if not already blocked.
Clear sigmask reliably, even if Fsignal returns, which it can.
Omit unnecessary accesses to volatile var.
(quit_throw_to_read_char): No need to restore sigmask.
* src/keyboard.c (gobble_input, handle_user_signal):
* src/process.c (wait_reading_process_output):
Call signal-handling code rather than killing ourselves.
* src/lisp.h: Include <float.h>, for...
(IEEE_FLOATING_POINT): New macro, moved here to avoid duplication.
(pending_signals): Now volatile.
(syms_of_data): Now const if IEEE floating point.
(handle_input_available_signal) [USABLE_SIGIO]:
(terminate_due_to_signal, record_child_status_change): New decls.
* src/process.c (create_process): Avoid disaster if memory is exhausted
while we're processing a vfork, by tightening the critical section
around the vfork.
(send_process_frame, process_sent_to, handle_pipe_signal)
(deliver_pipe_signal): Remove. No longer needed, as Emacs now
ignores SIGPIPE.
(send_process): No need for setjmp/longjmp any more, since the
SIGPIPE stuff is now gone. Instead, report an error if errno
is EPIPE.
(record_child_status_change): Now extern. PID and W are now args.
Return void, not bool. All callers changed.
* src/sysdep.c (wait_debugging) [(BSD_SYSTEM || HPUX) && !defined (__GNU__)]:
Remove. All uses removed. This bug should be fixed now in a
different way.
(wait_for_termination_1): Use waitpid rather than sigsuspend,
and record the child status change directly. This avoids the
need to futz with the signal mask.
(process_fatal_action): Move here from emacs.c.
(emacs_sigaction_flags): New function, containing
much of what used to be in emacs_sigaction_init.
(emacs_sigaction_init): Use it. Block nonfatal system signals that are
caught by emacs, to make races less likely.
(deliver_process_signal): Rename from handle_on_main_thread.
All uses changed.
(BACKTRACE_LIMIT_MAX): Now at top level.
(thread_backtrace_buffer, threadback_backtrace_pointers):
New static vars.
(deliver_thread_signal, deliver_fatal_thread_signal):
New functions, for more-accurate delivery of thread-specific signals.
(handle_fatal_signal, deliver_fatal_signal): Move here from emacs.c.
(deliver_arith_signal): Handle in this thread, not
in the main thread, since it's triggered by this thread.
(maybe_fatal_sig): New function.
(init_signals): New arg DUMPING so that we can be more accurate
about whether we're dumping. Caller changed.
Treat thread-specific signals differently from process-general signals.
Block all signals while handling fatal error; that's safer.
xsignal from SIGFPE only on non-IEEE hosts, treating it as fatal
on IEEE hosts.
When batch, ignore SIGHUP, SIGINT, SIGTERM if they were already ignored.
Ignore SIGPIPE unless batch.
(emacs_backtrace): Output backtrace for the appropriate thread,
which is not necessarily the main thread.
* src/syssignal.h: Include <stdbool.h>.
(emacs_raise): New macro.
* src/xterm.c (x_connection_signal): Remove; no longer needed
now that we use sigaction.
(x_connection_closed): No need to mess with sigmask now.
(x_initialize): No need to reset SIGPIPE handler here, since
init_signals does this for us now.
Fixes: debbugs:12471
Here, we use the generic window-system configuration system we just
implemented to support the w32 window-system in the mainline build
under Cygwin. (Previously, the w32 window system could only be
compiled as part of the NT-native Emacs build process.)
The changes in this patch need to be applied atomically in order to
avoid breaking Emacs. The changes include:
- Changes throughout the Lisp and C code to not assume that
NT Emacs and the w32 window system are synonymous.
- Wiring up the regular select(2) event loop to Windows messages
- Cleaning up the w32 drag-and-drop receiving code.
- Exposing Cygwin path conversion functions to elisp.
- Unicode file dialog support when compiling for Cygwin.
- Splitting the w32 term lisp initialization code into code
applicable to any w32 window-system and code specific to
system-type windows-nt.
- Integrating the old and new w32 code into the build system.
The workaround was for improving performance on Solaris 2.4, but
is getting in the way now. Emacs will still work if someone is
still running Solaris 2.4 in a museum somewhere; Sun dropped
support for Solaris 2.4 in 2003.
* configure.ac (ac_cv_func_vfork_works): Default to 'no' on
Solaris 2.4, so that AC_FUNC_VFORK doesn't think vfork works.
* src/callproc.c (Fcall_process) [HAVE_WORKING_VFORK]:
* src/process.c (create_process) [HAVE_WORKING_VFORK]:
Omit now-unnecessary workaround for the Solaris 2.4 vfork bug,
since Emacs no longer uses vfork on that platform.
When auditing signal-handling in preparation for cleaning it up,
I found that SYNC_INPUT has race conditions and would be a real
pain to fix. Since it's an undocumented and deprecated
configure-time option, now seems like a good time to remove it.
Also see <http://bugs.gnu.org/11080#16>.
* configure.ac (SYNC_INPUT, BROKEN_SA_RESTART): Remove.
* admin/CPP-DEFINES (BROKEN_SA_RESTART, SA_RESTART): Remove.
* etc/TODO (Make SYNC_INPUT the default): Remove, as the code now
behaves as if SYNC_INPUT is always true.
* src/alloc.c (_bytes_used, __malloc_extra_blocks, _malloc_internal)
(_free_internal) [!DOUG_LEA_MALLOC]: Remove decls.
(alloc_mutex) [!SYSTEM_MALLOC && !SYNC_INPUT && HAVE_PTHREAD]:
(malloc_hysteresis):
(check_depth) [XMALLOC_OVERRUN_CHECK]:
(MALLOC_BLOCK_INPUT, MALLOC_UNBLOCK_INPUT):
(__malloc_hook, __realloc_hook, __free_hook, BYTES_USED)
(dont_register_blocks, bytes_used_when_reconsidered)
(bytes_used_when_full, emacs_blocked_free, emacs_blocked_malloc)
(emacs_blocked_realloc, reset_malloc_hooks, uninterrupt_malloc):
[!SYSTEM_MALLOC && !SYNC_INPUT]:
Remove. All uses removed.
(MALLOC_BLOCK_INPUT, MALLOC_UNBLOCK_INPUT): Use a different
implementation, one that depends on whether the new macro
XMALLOC_BLOCK_INPUT_CHECK is defined, not on whether SYNC_INPUT
is defined.
* src/atimer.c (run_timers, handle_alarm_signal):
* src/keyboard.c (pending_signal, poll_for_input_1, poll_for_input)
(handle_async_input, process_pending_signals)
(handle_input_available_signal, init_keyboard):
* src/nsterm.m (ns_read_socket):
* src/process.c (wait_reading_process_output):
* src/regex.c (immediate_quit, IMMEDIATE_QUIT_CHECK):
* src/sysdep.c (emacs_sigaction_init) [SA_RESTART]:
(emacs_write):
* src/xterm.c (XTread_socket):
Assume SYNC_INPUT.
* src/conf_post.h (SA_RESTART) [IRIX6_5]: Do not #undef.
* src/eval.c (handling_signal): Remove. All uses removed.
* src/lisp.h (ELSE_PENDING_SIGNALS): Remove.
All uses replaced with the SYNC_INPUT version.
(reset_malloc_hooks, uninterrupt_malloc, handling_signal):
Remove decls.
* src/sysdep.c, src/syssignal.h (main_thread) [FORWARD_SIGNAL_TO_MAIN_THREAD]:
Now static.
Fixes: debbugs:12450
* configure.ac (HAVE__SETJMP, HAVE_SIGSETJMP): New symbols.
(_setjmp, _longjmp): Remove.
* src/lisp.h: Include <setjmp.h> here, since we use its symbols here.
All instances of '#include <setjmp.h>' removed, if the
only reason for the instance was because "lisp.h" was included.
(sys_jmp_buf, sys_setjmp, sys_longjmp): New symbols.
Unless otherwise specified, replace all uses of jmp_buf, _setjmp,
and _longjmp with the new symbols. Emacs already uses _setjmp if
available, so this change affects only POSIXish hosts that have
sigsetjmp but not _setjmp, such as some versions of Solaris and
Unixware. (Also, POSIX-2008 marks _setjmp as obsolescent.)
* src/image.c (_setjmp, _longjmp) [HAVE_PNG && !HAVE__SETJMP]: New macros.
(png_load_body) [HAVE_PNG]:
(PNG_LONGJMP) [HAVE_PNG && PNG_LIBPNG_VER < 10500]:
(PNG_JMPBUF) [HAVE_PNG && PNG_LIBPNG_VER >= 10500]:
Use _setjmp and _longjmp rather than sys_setjmp and sys_longjmp,
since PNG requires jmp_buf. This is the only exception to the
general rule that we now use sys_setjmp and sys_longjmp.
This exception is OK since this code does not change the signal
mask or longjmp out of a signal handler.
Fixes: debbugs:12446
The code that dealt with SIGIO was crufty and confusing, e.g., it
played tricks like "#undef SIGIO" but these tricks were not used
consistently. Simplify mostly by not #undeffing standard symbols,
e.g., use "defined USABLE_SIGIO" (our symbol, which we can define
or not as we please) rather than "defined SIGIO" (standard symbol
that we probably shouldn't #undef).
* configure.ac (NO_TERMIO, BROKEN_FIONREAD, BROKEN_SIGAIO)
(BROKEN_SIGPOLL, BROKEN_SIGPTY): Remove.
(USABLE_FIONREAD, USABLE_SIGIO): New symbols. All uses of
'defined SIGIO' replaced with 'defined USABLE_SIGIO', with no need
to #undef SIGIO now (which was error-prone). Likewise, all uses
of 'defined FIONREAD' replaced with 'defined USABLE_FIONREAD'.
* src/admin/CPP_DEFINES (BROKEN_SIGAIO, BROKEN_SIGIO, BROKEN_SIGPOLL)
(BROKEN_SIGPTY, NO_TERMIO): Remove.
* src/conf_post.h [USG5_4]: Do not include <sys/wait.h> here.
Modules that need it can include it.
[USG5_4 && emacs]: Likewise, do not include the streams stuff here.
* src/dispextern.h (ignore_sigio): New decl.
* src/emacs.c (shut_down_emacs): Invoke unrequest_sigio
unconditionally, since it's now a no-op if !USABLE_SIGIO.
* src/emacs.c (shut_down_emacs):
* src/keyboard.c (kbd_buffer_store_event_hold):
Use ignore_sigio rather than invoking 'signal' directly.
* src/keyboard.c (USABLE_FIONREAD && USG5_4): Include <sys/filio.h>,
for FIONREAD.
(FIONREAD, SIGIO): Do not #undef.
(tty_read_avail_input): Use #error rather than a syntax error.
* src/process.c [USG5_4]: Include <sys/stream.h> and <sys/stropts.h>,
for I_PIPE, used by SETUP_SLAVE_PTY.
(DATAGRAM_SOCKETS): Simplify defn, based on USABLE_FIONREAD.
* src/sysdep.c (croak): Remove; no longer needed. This bit of
temporary code, with Fred N. Fish's comment that it's temporary,
has been in Emacs since at least 1992!
(init_sigio, reset_sigio, request_sigio, unrequest_sigio):
Arrange for them to be no-ops in all cases when ! USABLE_SIGIO.
* src/syssignal.h (croak): Remove decl.
(SIGIO, SIGPOO, SIGAIO, SIGPTY): Do not #undef; that's too fragile.
* src/systty.h [!NO_TERMIO]: Do not include <termio.h>; no longer needed
now that we're termios-only.
(FIONREAD, ASYNC) [BROKEN_FIONREAD]: Do not #undef.
* src/term.c (dissociate_if_controlling_tty): Use #error rather than
a run-time error.
Fixes: debbugs:12408
* lib-src/pop.c (socket_connection) [HAVE_GETADDRINFO]:
* src/bidi.c (bidi_push_it, bidi_pop_it):
* src/fns.c (copy_hash_table):
* src/image.c (define_image_type):
* src/keyboard.c (kbd_buffer_store_event_hold):
* src/process.c (Fprocess_send_eof):
* src/xfaces.c (x_create_gc) [HAVE_NS]:
* src/xgselect.c (xg_select):
Use assignment, not memcpy, as either will do here, and assignment is
more likely to catch type errors.
Emacs's signal handlers were written in the old 4.2BSD style with
sigblock and sigmask and so forth, and this led to some
inefficiencies and confusion. Rewrite these to use
pthread_sigmask etc. without copying signal sets around. Also,
get rid of the confusing macros 'SIGNAL_THREAD_CHECK' and
'signal', and instead use functions that do not attempt to take
over the system name space. This patch causes Emacs's text
segment to shrink by 0.7% on my platform, Fedora 17 x86-64.
* configure.ac (PTY_OPEN, PTY_TTY_NAME_SPRINTF):
Adjust to syssignal.h changes.
(SIGNAL_H_AB): Remove; no longer needed.
* src/alloc.c, src/emacsgtkfixed.c, src/nsfns.m, src/widget.c, src/xmenu.c:
Do not include <signal.h> or "syssignal.h", as these
modules do not use signals.
* src/atimer.c, src/callproc.c, src/data.c, src/dispnew.c, src/emacs.c:
* src/floatfns.c, src/gtkutil.c, src/keyboard.c, src/process.c, src/sound.c:
* src/sysdep.c, src/term.c, src/xterm.c:
Do not include <signal.h>, as "syssignal.h" does that for us now.
* src/atimer.c (sigmask_atimers): New function.
(block_atimers, unblock_atimers): New functions,
replacing the old macros BLOCK_ATIMERS and UNBLOCK_ATIMERS.
All uses replaced.
* src/conf_post.h [SIGNAL_H_AHB]: Do not include <signal.h>;
no longer needed here.
* src/emacs.c (main): Inspect existing signal handler with sigaction,
so that there's no need to block and unblock SIGHUP.
* src/sysdep.c (struct save_signal): New member 'action', replacing
old member 'handler'.
(save_signal_handlers, restore_signal_handlers):
Use sigaction instead of 'signal' to save and restore.
(get_set_sighandler, set_sighandler) [!WINDOWSNT]:
New function. All users of 'signal' modified to use set_sighandler
if they're writeonly, and to use sys_signal if they're read+write.
(emacs_sigaction_init, forwarded_signal): New functions.
(sys_signal): Remove. All uses replaced by calls to sigaction
and emacs_sigaction_init, or by direct calls to 'signal'.
(sys_sigmask) [!__GNUC__]: Remove; no longer needed.
(sys_sigblock, sys_sigunblock, sys_sigsetmask): Remove;
all uses replaced by pthread_sigmask etc. calls.
* src/syssignal.h: Include <signal.h>.
(emacs_sigaction_init, forwarded_signal): New decls.
(SIGMASKTYPE): Remove. All uses replaced by its definiens, sigset_t.
(SIGEMPTYMASK): Remove; all uses replaced by its definiens, empty_mask.
(sigmask, sys_sigmask): Remove; no longer needed.
(sigpause): Remove. All uses replaced by its definiens, sigsuspend.
(sigblock, sigunblock, sigfree):
(sigsetmask) [!defined sigsetmask]:
Remove. All uses replaced by pthread_sigmask.
(signal): Remove. Its remaining uses (with SIG_DFL and SIG_IGN)
no longer need to be replaced, and its typical old uses
are now done via emacs_sigaction_init and sigaction.
(sys_sigblock, sys_sigunblock, sys_sigsetmask): Remove decls.
(sys_sigdel): Remove; unused.
(NSIG): Remove a FIXME; the code's fine. Remove an unnecessary ifdef.
Fixes: debbugs:12327
Be more systematic about preserving errno whenever a signal
handler returns, even if it's not in the main thread. Do this by
renaming signal handlers to distinguish between signal delivery
and signal handling. All uses changed.
* atimer.c (deliver_alarm_signal): Rename from alarm_signal_handler.
* data.c (deliver_arith_signal): Rename from arith_error.
* dispnew.c (deliver_window_change_signal): Rename from
window_change_signal.
* emacs.c (deliver_error_signal): Rename from fatal_error_signal.
(deliver_danger_signal) [SIGDANGER]: Rename from memory_warning_signal.
* keyboard.c (deliver_input_available_signal): Rename from
input_available_signal.
(deliver_user_signal): Rename from handle_user_signal.
(deliver_interrupt_signal): Rename from interrupt_signal.
* process.c (deliver_pipe_signal): Rename from send_process_trap.
(deliver_child_signal): Rename from sigchld_handler.
* atimer.c (handle_alarm_signal):
* data.c (handle_arith_signal):
* dispnew.c (handle_window_change_signal):
* emacs.c (handle_fatal_signal, handle_danger_signal):
* keyboard.c (handle_input_available_signal):
* keyboard.c (handle_user_signal, handle_interrupt_signal):
* process.c (handle_pipe_signal, handle_child_signal):
New functions, with the actual signal-handling code taken from the
original respective signal handlers, sans the sporadic attempts to
preserve errno, since that's now done by handle_on_main_thread.
* atimer.c (alarm_signal_handler): Remove unnecessary decl.
* emacs.c, floatfns.c, lisp.h: Remove unused FLOAT_CATCH_SIGKILL cruft.
* emacs.c (main_thread) [FORWARD_SIGNAL_TO_MAIN_THREAD]:
Move to sysdep.c.
(main) [FORWARD_SIGNAL_TO_MAIN_THREAD]:
Move initialization of main_thread to sysdep.c's init_signals.
* process.c (waitpid) [!WNOHANG]: #define to wait; that's good enough for
our usage, and simplifies the mainline code.
(record_child_status_change): New static function, as a helper
for handle_child_signal, and with most of the old child handler's
contents.
(CAN_HANDLE_MULTIPLE_CHILDREN): New constant.
(handle_child_signal): Use the above.
* sysdep.c (main_thread) [FORWARD_SIGNAL_TO_MAIN_THREAD]:
Moved here from emacs.c.
(init_signals) [FORWARD_SIGNAL_TO_MAIN_THREAD]: Initialize it;
code moved here from emacs.c's main function.
* sysdep.c, syssignal.h (handle_on_main_thread): New function,
replacing the old SIGNAL_THREAD_CHECK. All uses changed. This
lets callers save and restore errno properly.
Do not try to redefine the 'abort' function. Instead, redo
the code so that it calls 'emacs_abort' rather than 'abort'.
This removes the need for the NO_ABORT configure-time macro
and makes it easier to change the abort code to do a backtrace.
* configure.ac (NO_ABRT): Remove.
* admin/CPP-DEFINES (NO_ABORT): Remove.
* nt/inc/ms-w32.h (w32_abort) [HAVE_NTGUI]: Remove.
* src/.gdbinit: Just stop at emacs_abort, not at w32_abort or abort.
* src/emacs.c (abort) [!DOS_NT && !NO_ABORT]:
Remove; sysdep.c's emacs_abort now takes its place.
* src/lisp.h (emacs_abort): New decl. All calls from Emacs code to
'abort' changed to use 'emacs_abort'.
* src/msdos.c (dos_abort) [defined abort]: Remove; not used.
(abort) [!defined abort]: Rename to ...
(emacs_abort): ... new name.
* src/sysdep.c (emacs_abort) [!HAVE_NTGUI]: New function, taking
the place of the old 'abort' in emacs.c.
* src/w32.c, src/w32fns.c (abort): Do not #undef.
* src/w32.c (emacs_abort): Rename from w32_abort.
* alloc.c (test_setjmp, mark_stack):
* image.c (PNG_LONGJMP) [PNG_LIBPNG_VER < 10500]:
(PNG_JMPBUF) [! (PNG_LIBPNG_VER < 10500)]:
(png_load, my_error_exit, jpeg_load):
* process.c (send_process_trap, send_process):
Uniformly prefer _setjmp and _longjmp to setjmp and longjmp.
The underscored versions are up to 30x faster on some hosts.
Formerly, the code used setjmp+longjmp sometimes and
_setjmp+_longjmp at other times, with no particular reason to
prefer setjmp+longjmp.
* emacs.c (initialized, inhibit_window_system, running_asynch_code):
(malloc_using_checking) [DOUG_LEA_MALLOC]:
(display_arg) [HAVE_X_WINDOWS || HAVE_NS]:
(noninteractive, no_site_lisp, fatal_error_in_progress, argmatch)
(main, decode_env_path, Fdaemon_initialized):
* eval.c (call_debugger, Finteractive_p, interactive_p):
(unwind_to_catch, Fsignal, wants_debugger, skip_debugger)
(maybe_call_debugger, Fbacktrace):
* process.c (read_process_output, exec_sentinel):
Use bool for booleans.
* emacs.c (shut_down_emacs): Omit unused boolean argument NO_X.
All callers changed.
* eval.c (interactive_p): Omit always-true boolean argument
EXCLUDE_SUBRS_P. All callers changed.
* dispextern.h, lisp.h: Reflect above API changes.
* firstfile.c (dummy): Use the address of 'main', whose signature
won't change, instead of the address of 'initialize', whose
signature just changed from int to bool.
* lisp.h (fatal_error_in_progress): New decl of boolean, moved here ...
* msdos.c (fatal_error_in_progress): ... from here.
* xdisp.c (redisplaying_p): Now a boolean. Set it to 1 instead
of incrementing it.
(redisplay_internal, unwind_redisplay): Simply clear
REDISPLAYING_P when unwinding, instead of saving its previous,
always-false value and then restoring it.
