old-cross-binutils/gdb/event-top.c

1105 lines
33 KiB
C
Raw Normal View History

1999-05-11 20:29:07 +00:00
/* Top level stuff for GDB, the GNU debugger.
Copyright (C) 1999-2015 Free Software Foundation, Inc.
1999-05-11 20:29:07 +00:00
Written by Elena Zannoni <ezannoni@cygnus.com> of Cygnus Solutions.
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
1999-05-11 20:29:07 +00:00
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>. */
1999-05-11 20:29:07 +00:00
#include "defs.h"
1999-06-14 18:08:47 +00:00
#include "top.h"
1999-05-11 20:29:07 +00:00
#include "inferior.h"
Add new infrun.h header. Move infrun.c declarations out of inferior.h to a new infrun.h file. Tested by building on: i686-w64-mingw32, enable-targets=all x86_64-linux, enable-targets=all i586-pc-msdosdjgpp And also grepped the whole tree for each symbol moved to find where infrun.h might be necessary. gdb/ 2014-05-22 Pedro Alves <palves@redhat.com> * inferior.h (debug_infrun, debug_displaced, stop_on_solib_events) (sync_execution, sched_multi, step_stop_if_no_debug, non_stop) (disable_randomization, enum exec_direction_kind) (execution_direction, stop_registers, start_remote) (clear_proceed_status, proceed, resume, user_visible_resume_ptid) (wait_for_inferior, normal_stop, get_last_target_status) (prepare_for_detach, fetch_inferior_event, init_wait_for_inferior) (insert_step_resume_breakpoint_at_sal) (follow_inferior_reset_breakpoints, stepping_past_instruction_at) (set_step_info, print_stop_event, signal_stop_state) (signal_print_state, signal_pass_state, signal_stop_update) (signal_print_update, signal_pass_update) (update_signals_program_target, clear_exit_convenience_vars) (displaced_step_dump_bytes, update_observer_mode) (signal_catch_update, gdb_signal_from_command): Move declarations ... * infrun.h: ... to this new file. * amd64-tdep.c: Include infrun.h. * annotate.c: Include infrun.h. * arch-utils.c: Include infrun.h. * arm-linux-tdep.c: Include infrun.h. * arm-tdep.c: Include infrun.h. * break-catch-sig.c: Include infrun.h. * breakpoint.c: Include infrun.h. * common/agent.c: Include infrun.h instead of inferior.h. * corelow.c: Include infrun.h. * event-top.c: Include infrun.h. * go32-nat.c: Include infrun.h. * i386-tdep.c: Include infrun.h. * inf-loop.c: Include infrun.h. * infcall.c: Include infrun.h. * infcmd.c: Include infrun.h. * infrun.c: Include infrun.h. * linux-fork.c: Include infrun.h. * linux-nat.c: Include infrun.h. * linux-thread-db.c: Include infrun.h. * monitor.c: Include infrun.h. * nto-tdep.c: Include infrun.h. * procfs.c: Include infrun.h. * record-btrace.c: Include infrun.h. * record-full.c: Include infrun.h. * remote-m32r-sdi.c: Include infrun.h. * remote-mips.c: Include infrun.h. * remote-notif.c: Include infrun.h. * remote-sim.c: Include infrun.h. * remote.c: Include infrun.h. * reverse.c: Include infrun.h. * rs6000-tdep.c: Include infrun.h. * s390-linux-tdep.c: Include infrun.h. * solib-irix.c: Include infrun.h. * solib-osf.c: Include infrun.h. * solib-svr4.c: Include infrun.h. * target.c: Include infrun.h. * top.c: Include infrun.h. * windows-nat.c: Include infrun.h. * mi/mi-interp.c: Include infrun.h. * mi/mi-main.c: Include infrun.h. * python/py-threadevent.c: Include infrun.h.
2014-05-22 11:29:11 +00:00
#include "infrun.h"
1999-10-26 03:43:48 +00:00
#include "target.h"
1999-07-07 20:19:36 +00:00
#include "terminal.h" /* for job_control */
1999-06-21 13:27:42 +00:00
#include "event-loop.h"
1999-09-22 03:28:34 +00:00
#include "event-top.h"
#include "interps.h"
#include <signal.h>
2006-07-21 14:46:56 +00:00
#include "cli/cli-script.h" /* for reset_command_nest_depth */
#include "main.h"
Add "executing" property to threads. * inferior.h (target_executing): Delete. * gdbthread.h (struct thread_info): Add executing_ field. (set_executing, is_executing): New. * thread.c (main_thread_executing): New. (init_thread_list): Clear it and also main_thread_running. (is_running): Return false if target has no execution. (any_running, is_executing, set_executing): New. * top.c: Include "gdbthread.h". (target_executing): Delete. (execute_command): Replace target_executing check by any_running. * event-top.c: Include "gdbthread.h". (display_gdb_prompt, command_handler): Replace target_executing by is_running. * inf-loop.c: Include "gdbthread.h". Don't mark as not executing here. Replace target_executing by is_running. * infrun.c (handle_inferior_event): Mark all threads as not-executing. * linux-nat.c (linux_nat_resume): Don't mark thread as executing here. * stack.c (get_selected_block): Return null if inferior is executing. * target.c (target_resume): Mark resumed ptid as executing. * breakpoint.c (until_break_command): Replace target_executing check by is_executing. * remote.c (remote_async_resume): Don't mark inferior as executing here. * mi/mi-interp.c (mi_cmd_interpreter_exec): Replace target_executing by any_running. * mi/mi-main.c (mi_cmd_exec_interrupt, mi_cmd_execute) (mi_execute_async_cli_command): Replace target_executing by is_running. * frame.c (get_current_frame): Error out if the current thread is executing. (has_stack_frames): New. (get_selected_frame, deprecated_safe_get_selected_frame): Check has_stack_frames. * Makefile.in (event-top.o, frame.o, inf-loop.o, top.o): Depend on $(gdbthread_h).
2008-07-09 22:16:15 +00:00
#include "gdbthread.h"
#include "observer.h"
#include "continuations.h"
#include "gdbcmd.h" /* for dont_repeat() */
With some changes to how software single-step (SSS) breakpoints are handled, one of those being to place SSS breakpoints on the breakpoint chain as all other breakpoints, annota1.exp times out with lots and lots of breakpoint-invalid and frame-changed annotations. All those extra annotations are actually unnecessary. For one, SSS breakpoints are internal breakpoints, so the frontend shouldn't care if they were added, removed or changed. Then, there's really no point in emitting "breakpoints-invalid" or "frames-invalid" more than once between times the frontend/user can actually issues GDB commands; the frontend will have to wait for the GDB prompt to refresh its state, so emitting those annotations at most once between prompts is enough. Non-stop or async would complicate this, but no frontend will be using annotations in those modes (one of goes of emacs switching to MI was non-stop mode support, AFAIK). The previous patch reveals there has been an intention in the past to suppress multiple breakpoints-invalid annotations caused by ignore count changes. As the previous patch shows, that's always been broken, but in any case, this patch actually makes it work. The next patch will remove several annotation-specific calls in breakpoint.c in favor of always using the breakpoint modified & friends observers, and that causes yet more of these annotations, because several calls to the corresponding annotate_* functions in breakpoint.c are missing, particularly in newer code. So all in all, here's a simple mechanism that avoids sending the same annotation to the frontend more than once until gdb is ready to accept further commands. Tested on x86_64 Fedora 17. 2013-01-22 Pedro Alves <palves@redhat.com> * annotate.c: Include "inferior.h". (frames_invalid_emitted) (breakpoints_invalid_emitted): New globals. (async_background_execution_p): New function. (annotate_breakpoints_changed, annotate_frames_invalid): Skip emitting the annotation if it has already been emitted. (annotate_display_prompt): New function. * annotate.h (annotate_display_prompt): New declaration. * event-top.c: Include annotate.h. (display_gdb_prompt): Call annotate_display_prompt.
2013-01-22 20:17:10 +00:00
#include "annotate.h"
New commands "mt set per-command {space,time,symtab} {on,off}". * NEWS: Add entry. * event-top.c: #include "maint.h". * main.c: #include "maint.h". * maint.c: #include <sys/time.h>, <time.h>, block.h, top.h, timeval-utils.h, maint.h, cli/cli-setshow.h. (per_command_time, per_command_space): New static globals. (per_command_symtab): New static global. (per_command_setlist, per_command_showlist): New static globals. (struct cmd_stats): Move here from utils.c. (set_per_command_time): Renamed from set_display_time in utils.c and moved here. All callers updated. (set_per_command_space): Renamed from set_display_space in utils.c and moved here. All callers updated. (count_symtabs_and_blocks): New function. (report_command_stats): Moved here from utils.c. Add support for printing symtab stats. Only print data if enabled before command executed. (make_command_stats_cleanup): Ditto. (sert_per_command_cmd, show_per_command_cmd): New functions. (_initialize_maint_cmds): Add new commands mt set per-command {space,time,symtab} {on,off}. * maint.h: New file. * top.c: #include "maint.h". * utils.c (reset_prompt_for_continue_wait_time): New function. (get_prompt_for_continue_wait_time): New function. * utils.h (reset_prompt_for_continue_wait_time): Declare (get_prompt_for_continue_wait_time): Declare. (make_command_stats_cleanup): Moved to maint.h. (set_display_time, set_display_space): Moved to maint.h and renamed to set_per_command_time, set_per_command_space. * cli/cli-setshow.c (parse_cli_boolean_value): Renamed from parse_binary_operation and made non-static. Don't call error, just return an error marker. All callers updated. * cli/cli-setshow.h (parse_cli_boolean_value): Declare. doc/ * gdb.texinfo (Maintenance Commands): Add docs for "mt set per-command {space,time,symtab} {on,off}". testsuite/ * gdb.base/maint.exp: Update tests for per-command stats.
2013-03-21 17:37:30 +00:00
#include "maint.h"
1999-08-31 01:14:27 +00:00
/* readline include files. */
#include "readline/readline.h"
#include "readline/history.h"
1999-05-11 20:29:07 +00:00
/* readline defines this. */
#undef savestring
1999-09-22 03:28:34 +00:00
static void rl_callback_read_char_wrapper (gdb_client_data client_data);
static void command_line_handler (char *rl);
static void change_line_handler (void);
static void command_handler (char *command);
static char *top_level_prompt (void);
1999-05-11 20:29:07 +00:00
/* Signal handlers. */
#ifdef SIGQUIT
1999-09-22 03:28:34 +00:00
static void handle_sigquit (int sig);
#endif
#ifdef SIGHUP
1999-09-22 03:28:34 +00:00
static void handle_sighup (int sig);
#endif
1999-09-22 03:28:34 +00:00
static void handle_sigfpe (int sig);
1999-05-11 20:29:07 +00:00
/* Functions to be invoked by the event loop in response to
signals. */
#if defined (SIGQUIT) || defined (SIGHUP)
1999-09-22 03:28:34 +00:00
static void async_do_nothing (gdb_client_data);
#endif
#ifdef SIGHUP
1999-09-22 03:28:34 +00:00
static void async_disconnect (gdb_client_data);
#endif
1999-09-22 03:28:34 +00:00
static void async_float_handler (gdb_client_data);
#ifdef STOP_SIGNAL
1999-09-22 03:28:34 +00:00
static void async_stop_sig (gdb_client_data);
#endif
static void async_sigterm_handler (gdb_client_data arg);
1999-05-11 20:29:07 +00:00
/* Readline offers an alternate interface, via callback
functions. These are all included in the file callback.c in the
1999-05-11 20:29:07 +00:00
readline distribution. This file provides (mainly) a function, which
the event loop uses as callback (i.e. event handler) whenever an event
is detected on the standard input file descriptor.
readline_callback_read_char is called (by the GDB event loop) whenever
there is a new character ready on the input stream. This function
1999-05-11 20:29:07 +00:00
incrementally builds a buffer internal to readline where it
accumulates the line read up to the point of invocation. In the
special case in which the character read is newline, the function
invokes a GDB supplied callback routine, which does the processing of
a full command line. This latter routine is the asynchronous analog
of the old command_line_input in gdb. Instead of invoking (and waiting
1999-05-11 20:29:07 +00:00
for) readline to read the command line and pass it back to
command_loop for processing, the new command_line_handler function has
the command line already available as its parameter. INPUT_HANDLER is
to be set to the function that readline will invoke when a complete
line of input is ready. CALL_READLINE is to be set to the function
that readline offers as callback to the event_loop. */
1999-05-11 20:29:07 +00:00
1999-09-22 03:28:34 +00:00
void (*input_handler) (char *);
void (*call_readline) (gdb_client_data);
1999-05-11 20:29:07 +00:00
/* Important variables for the event loop. */
1999-05-11 20:29:07 +00:00
/* This is used to determine if GDB is using the readline library or
its own simplified form of readline. It is used by the asynchronous
1999-06-14 18:08:47 +00:00
form of the set editing command.
1999-05-25 18:09:09 +00:00
ezannoni: as of 1999-04-29 I expect that this
1999-05-11 20:29:07 +00:00
variable will not be used after gdb is changed to use the event
loop as default engine, and event-top.c is merged into top.c. */
1999-05-11 20:29:07 +00:00
int async_command_editing_p;
/* This is the annotation suffix that will be used when the
annotation_level is 2. */
1999-05-11 20:29:07 +00:00
char *async_annotation_suffix;
1999-08-31 01:14:27 +00:00
/* This is used to display the notification of the completion of an
asynchronous execution command. */
1999-08-31 01:14:27 +00:00
int exec_done_display_p = 0;
1999-05-11 20:29:07 +00:00
/* This is the file descriptor for the input stream that GDB uses to
read commands from. */
1999-05-11 20:29:07 +00:00
int input_fd;
Fix the processing of Meta-key commands in TUI This patch fixes the annoying bug where key sequences such as Alt_F or Alt_B (go forward or backwards by a word) do not behave promptly in TUI. You have to press a third key in order for the key sequence to register. This is mostly ncurses' fault. Calling wgetch() normally causes ncurses to read only a single key from stdin. However if the key read is the start-sequence key (^[ a.k.a. ESC) then wgetch() reads TWO keys from stdin, storing the 2nd key into an internal FIFO buffer and returning the start-sequence key. The extraneous read of the 2nd key makes us miss its corresponding stdin event, so the event loop blocks until a third key is pressed. This explains why such key sequences do not behave promptly in TUI. To fix this issue, we must somehow compensate for the missed stdin event corresponding to the 2nd byte of a key sequence. This patch achieves this by hacking up the stdin event handler to conditionally execute the readline callback multiple times in a row. This is done via a new global variable, call_stdin_event_handler_again_p, which is set from tui_getc() when we receive a start-sequence key and notice extra pending input in the ncurses buffer. Tested on x86_64-unknown-linux-gnu. gdb/ChangeLog: * event-top.h (call_stdin_event_handler_again_p): Declare. * event-top.c (call_stdin_event_handler_again_p): Define. (stdin_event_handler): Use it. * tui/tui-io.c (tui_getc): Prepare to call the stdin event handler again if there is pending input following a start sequence.
2014-11-23 10:03:39 +00:00
/* Used by the stdin event handler to compensate for missed stdin events.
Setting this to a non-zero value inside an stdin callback makes the callback
run again. */
int call_stdin_event_handler_again_p;
/* Signal handling variables. */
1999-05-11 20:29:07 +00:00
/* Each of these is a pointer to a function that the event loop will
invoke if the corresponding signal has received. The real signal
1999-05-11 20:29:07 +00:00
handlers mark these functions as ready to be executed and the event
loop, in a later iteration, calls them. See the function
invoke_async_signal_handler. */
static struct async_signal_handler *sigint_token;
1999-05-11 20:29:07 +00:00
#ifdef SIGHUP
static struct async_signal_handler *sighup_token;
1999-05-11 20:29:07 +00:00
#endif
#ifdef SIGQUIT
static struct async_signal_handler *sigquit_token;
#endif
static struct async_signal_handler *sigfpe_token;
1999-06-14 18:08:47 +00:00
#ifdef STOP_SIGNAL
static struct async_signal_handler *sigtstp_token;
1999-06-14 18:08:47 +00:00
#endif
static struct async_signal_handler *async_sigterm_token;
1999-06-14 18:08:47 +00:00
1999-05-11 20:29:07 +00:00
/* Structure to save a partially entered command. This is used when
the user types '\' at the end of a command line. This is necessary
1999-05-11 20:29:07 +00:00
because each line of input is handled by a different call to
command_line_handler, and normally there is no state retained
between different calls. */
static int more_to_come = 0;
1999-05-11 20:29:07 +00:00
struct readline_input_state
{
char *linebuffer;
char *linebuffer_ptr;
}
readline_input_state;
/* This hook is called by rl_callback_read_char_wrapper after each
character is processed. */
void (*after_char_processing_hook) (void);
1999-05-11 20:29:07 +00:00
/* Wrapper function for calling into the readline library. The event
loop expects the callback function to have a paramter, while
readline expects none. */
1999-09-22 03:28:34 +00:00
static void
rl_callback_read_char_wrapper (gdb_client_data client_data)
{
rl_callback_read_char ();
if (after_char_processing_hook)
(*after_char_processing_hook) ();
1999-09-22 03:28:34 +00:00
}
1999-05-11 20:29:07 +00:00
/* Initialize all the necessary variables, start the event loop,
register readline, and stdin, start the loop. The DATA is the
interpreter data cookie, ignored for now. */
1999-05-11 20:29:07 +00:00
void
cli_command_loop (void *data)
1999-05-11 20:29:07 +00:00
{
display_gdb_prompt (0);
1999-05-11 20:29:07 +00:00
/* Now it's time to start the event loop. */
1999-06-28 16:06:02 +00:00
start_event_loop ();
1999-05-11 20:29:07 +00:00
}
/* Change the function to be invoked every time there is a character
ready on stdin. This is used when the user sets the editing off,
1999-05-11 20:29:07 +00:00
therefore bypassing readline, and letting gdb handle the input
itself, via gdb_readline2. Also it is used in the opposite case in
1999-05-11 20:29:07 +00:00
which the user sets editing on again, by restoring readline
handling of the input. */
1999-05-25 18:09:09 +00:00
static void
1999-09-22 03:28:34 +00:00
change_line_handler (void)
1999-05-11 20:29:07 +00:00
{
/* NOTE: this operates on input_fd, not instream. If we are reading
commands from a file, instream will point to the file. However in
1999-09-22 03:28:34 +00:00
async mode, we always read commands from a file with editing
off. This means that the 'set editing on/off' will have effect
only on the interactive session. */
1999-09-22 03:28:34 +00:00
1999-05-11 20:29:07 +00:00
if (async_command_editing_p)
{
/* Turn on editing by using readline. */
1999-09-22 03:28:34 +00:00
call_readline = rl_callback_read_char_wrapper;
1999-06-14 18:08:47 +00:00
input_handler = command_line_handler;
1999-05-11 20:29:07 +00:00
}
else
{
/* Turn off editing by using gdb_readline2. */
PR python/17372 - Python hangs when displaying help() This is more of a readline/terminal issue than a Python one. PR17372 is a regression in 7.8 caused by the fix for PR17072: commit 0017922d0292d8c374584f6100874580659c9973 Author: Pedro Alves <palves@redhat.com> Date: Mon Jul 14 19:55:32 2014 +0100 Background execution + pagination aborts readline/gdb gdb_readline_wrapper_line removes the handler after a line is processed. Usually, we'll end up re-displaying the prompt, and that reinstalls the handler. But if the output is coming out of handling a stop event, we don't re-display the prompt, and nothing restores the handler. So the next input wakes up the event loop and calls into readline, which aborts. ... gdb/ 2014-07-14 Pedro Alves <palves@redhat.com> PR gdb/17072 * top.c (gdb_readline_wrapper_line): Tweak comment. (gdb_readline_wrapper_cleanup): If readline is enabled, reinstall the input handler callback. The problem is that installing the input handler callback also preps the terminal, putting it in raw mode and with echo disabled, which is bad if we're going to call a command that assumes cooked/canonical mode, and echo enabled, like in the case of the PR, Python's interactive shell. Another example I came up with that doesn't depend on Python is starting a subshell with "(gdb) shell /bin/sh" from a multi-line command. Tests covering both these examples are added. The fix is to revert the original fix for PR gdb/17072, and instead restore the callback handler after processing an asynchronous target event. Furthermore, calling rl_callback_handler_install when we already have some input in readline's line buffer discards that input, which is obviously a bad thing to do while the user is typing. No specific test is added for that, because I first tried calling it even if the callback handler was still installed and that resulted in hundreds of failures in the testsuite. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * event-top.c (change_line_handler): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (callback_handler_installed): New global. (gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New functions. (display_gdb_prompt): Call gdb_rl_callback_handler_remove and gdb_rl_callback_handler_install instead of rl_callback_handler_remove and rl_callback_handler_install. (gdb_disable_readline): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. * event-top.h (gdb_rl_callback_handler_remove) (gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New declarations. * infrun.c (reinstall_readline_callback_handler_cleanup): New cleanup function. (fetch_inferior_event): Install it. * top.c (gdb_readline_wrapper_line) Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (gdb_readline_wrapper_cleanup): Don't call rl_callback_handler_install. gdb/testsuite/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * gdb.python/python.exp: Test a multi-line command that spawns interactive Python. * gdb.base/multi-line-starts-subshell.exp: New file.
2014-10-23 16:13:35 +00:00
gdb_rl_callback_handler_remove ();
1999-05-11 20:29:07 +00:00
call_readline = gdb_readline2;
1999-06-14 18:08:47 +00:00
/* Set up the command handler as well, in case we are called as
first thing from .gdbinit. */
1999-06-14 18:08:47 +00:00
input_handler = command_line_handler;
1999-05-11 20:29:07 +00:00
}
}
PR python/17372 - Python hangs when displaying help() This is more of a readline/terminal issue than a Python one. PR17372 is a regression in 7.8 caused by the fix for PR17072: commit 0017922d0292d8c374584f6100874580659c9973 Author: Pedro Alves <palves@redhat.com> Date: Mon Jul 14 19:55:32 2014 +0100 Background execution + pagination aborts readline/gdb gdb_readline_wrapper_line removes the handler after a line is processed. Usually, we'll end up re-displaying the prompt, and that reinstalls the handler. But if the output is coming out of handling a stop event, we don't re-display the prompt, and nothing restores the handler. So the next input wakes up the event loop and calls into readline, which aborts. ... gdb/ 2014-07-14 Pedro Alves <palves@redhat.com> PR gdb/17072 * top.c (gdb_readline_wrapper_line): Tweak comment. (gdb_readline_wrapper_cleanup): If readline is enabled, reinstall the input handler callback. The problem is that installing the input handler callback also preps the terminal, putting it in raw mode and with echo disabled, which is bad if we're going to call a command that assumes cooked/canonical mode, and echo enabled, like in the case of the PR, Python's interactive shell. Another example I came up with that doesn't depend on Python is starting a subshell with "(gdb) shell /bin/sh" from a multi-line command. Tests covering both these examples are added. The fix is to revert the original fix for PR gdb/17072, and instead restore the callback handler after processing an asynchronous target event. Furthermore, calling rl_callback_handler_install when we already have some input in readline's line buffer discards that input, which is obviously a bad thing to do while the user is typing. No specific test is added for that, because I first tried calling it even if the callback handler was still installed and that resulted in hundreds of failures in the testsuite. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * event-top.c (change_line_handler): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (callback_handler_installed): New global. (gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New functions. (display_gdb_prompt): Call gdb_rl_callback_handler_remove and gdb_rl_callback_handler_install instead of rl_callback_handler_remove and rl_callback_handler_install. (gdb_disable_readline): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. * event-top.h (gdb_rl_callback_handler_remove) (gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New declarations. * infrun.c (reinstall_readline_callback_handler_cleanup): New cleanup function. (fetch_inferior_event): Install it. * top.c (gdb_readline_wrapper_line) Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (gdb_readline_wrapper_cleanup): Don't call rl_callback_handler_install. gdb/testsuite/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * gdb.python/python.exp: Test a multi-line command that spawns interactive Python. * gdb.base/multi-line-starts-subshell.exp: New file.
2014-10-23 16:13:35 +00:00
/* The functions below are wrappers for rl_callback_handler_remove and
rl_callback_handler_install that keep track of whether the callback
handler is installed in readline. This is necessary because after
handling a target event of a background execution command, we may
need to reinstall the callback handler if it was removed due to a
secondary prompt. See gdb_readline_wrapper_line. We don't
unconditionally install the handler for every target event because
that also clears the line buffer, thus installing it while the user
is typing would lose input. */
/* Whether we've registered a callback handler with readline. */
static int callback_handler_installed;
/* See event-top.h, and above. */
void
gdb_rl_callback_handler_remove (void)
{
rl_callback_handler_remove ();
callback_handler_installed = 0;
}
/* See event-top.h, and above. Note this wrapper doesn't have an
actual callback parameter because we always install
INPUT_HANDLER. */
void
gdb_rl_callback_handler_install (const char *prompt)
{
/* Calling rl_callback_handler_install resets readline's input
buffer. Calling this when we were already processing input
therefore loses input. */
gdb_assert (!callback_handler_installed);
rl_callback_handler_install (prompt, input_handler);
callback_handler_installed = 1;
}
/* See event-top.h, and above. */
void
gdb_rl_callback_handler_reinstall (void)
{
if (!callback_handler_installed)
{
/* Passing NULL as prompt argument tells readline to not display
a prompt. */
gdb_rl_callback_handler_install (NULL);
}
}
/* Displays the prompt. If the argument NEW_PROMPT is NULL, the
prompt that is displayed is the current top level prompt.
Otherwise, it displays whatever NEW_PROMPT is as a local/secondary
prompt.
This is used after each gdb command has completed, and in the
following cases:
1. When the user enters a command line which is ended by '\'
indicating that the command will continue on the next line. In
that case the prompt that is displayed is the empty string.
1999-06-14 18:08:47 +00:00
2. When the user is entering 'commands' for a breakpoint, or
actions for a tracepoint. In this case the prompt will be '>'
3. On prompting for pagination. */
1999-05-11 20:29:07 +00:00
void
Fix build with Python 3.4 (PR python/16784) The type of the function pointer PyOS_ReadlineFunctionPointer (part of the Python C API), which we use, slightly changed starting with Python 3.4. The signature went from PyAPI_DATA(char) *(*PyOS_ReadlineFunctionPointer)(FILE *, FILE *, char *); to PyAPI_DATA(char) *(*PyOS_ReadlineFunctionPointer)(FILE *, FILE *, const char *); The parameter that changed is the prompt text. This commits adjust gdb accordingly by making the prompt_arg parameter const, as well as the fallouts of that. I needed to rework how annotations are added to the prompt, since the it is now const. If annotations are enabled, it will make a copy of the prompt overwrite the prompt variable that is used throughout the function. Otherwise, no copy is done and the original prompt_arg value is passed. I changed the signature of deprecated_readline_hook. I would've changed any user of it, but it seems like nothing is using it, Built-tested with python 2.7.x, 3.3.y and 3.4.z. gdb/ChangeLog: * defs.h (gdb_readline): Constify argument. (gdb_readline_wrapper): Same. (command_line_input): Same. (deprecated_readline_hook): Same. * top.c (deprecated_readline_hook): Same. (gdb_readline): Same. (gdb_readline_wrapper): Same. (command_line_input): Constify argument. Pass prompt to called functions instead of local_prompt, overwriting prompt if using annotations. * event-top.h (display_gdb_prompt): Constify argument. * event-top.c (display_gdb_prompt): Same. * python/py-gdb-readline.c (gdbpy_readline_wrapper): Constify argument if building with Python 3.4 and up. Signed-off-by: Simon Marchi <simon.marchi@ericsson.com>
2014-12-15 16:38:03 +00:00
display_gdb_prompt (const char *new_prompt)
1999-05-11 20:29:07 +00:00
{
char *actual_gdb_prompt = NULL;
struct cleanup *old_chain;
1999-05-11 20:29:07 +00:00
With some changes to how software single-step (SSS) breakpoints are handled, one of those being to place SSS breakpoints on the breakpoint chain as all other breakpoints, annota1.exp times out with lots and lots of breakpoint-invalid and frame-changed annotations. All those extra annotations are actually unnecessary. For one, SSS breakpoints are internal breakpoints, so the frontend shouldn't care if they were added, removed or changed. Then, there's really no point in emitting "breakpoints-invalid" or "frames-invalid" more than once between times the frontend/user can actually issues GDB commands; the frontend will have to wait for the GDB prompt to refresh its state, so emitting those annotations at most once between prompts is enough. Non-stop or async would complicate this, but no frontend will be using annotations in those modes (one of goes of emacs switching to MI was non-stop mode support, AFAIK). The previous patch reveals there has been an intention in the past to suppress multiple breakpoints-invalid annotations caused by ignore count changes. As the previous patch shows, that's always been broken, but in any case, this patch actually makes it work. The next patch will remove several annotation-specific calls in breakpoint.c in favor of always using the breakpoint modified & friends observers, and that causes yet more of these annotations, because several calls to the corresponding annotate_* functions in breakpoint.c are missing, particularly in newer code. So all in all, here's a simple mechanism that avoids sending the same annotation to the frontend more than once until gdb is ready to accept further commands. Tested on x86_64 Fedora 17. 2013-01-22 Pedro Alves <palves@redhat.com> * annotate.c: Include "inferior.h". (frames_invalid_emitted) (breakpoints_invalid_emitted): New globals. (async_background_execution_p): New function. (annotate_breakpoints_changed, annotate_frames_invalid): Skip emitting the annotation if it has already been emitted. (annotate_display_prompt): New function. * annotate.h (annotate_display_prompt): New declaration. * event-top.c: Include annotate.h. (display_gdb_prompt): Call annotate_display_prompt.
2013-01-22 20:17:10 +00:00
annotate_display_prompt ();
2006-07-21 14:46:56 +00:00
/* Reset the nesting depth used when trace-commands is set. */
reset_command_nest_depth ();
old_chain = make_cleanup (free_current_contents, &actual_gdb_prompt);
/* Do not call the python hook on an explicit prompt change as
passed to this function, as this forms a secondary/local prompt,
IE, displayed but not set. */
if (! new_prompt)
1999-07-19 23:30:11 +00:00
{
if (sync_execution)
{
/* This is to trick readline into not trying to display the
prompt. Even though we display the prompt using this
function, readline still tries to do its own display if
we don't call rl_callback_handler_install and
rl_callback_handler_remove (which readline detects
because a global variable is not set). If readline did
that, it could mess up gdb signal handlers for SIGINT.
Readline assumes that between calls to rl_set_signals and
rl_clear_signals gdb doesn't do anything with the signal
handlers. Well, that's not the case, because when the
target executes we change the SIGINT signal handler. If
we allowed readline to display the prompt, the signal
handler change would happen exactly between the calls to
the above two functions. Calling
rl_callback_handler_remove(), does the job. */
PR python/17372 - Python hangs when displaying help() This is more of a readline/terminal issue than a Python one. PR17372 is a regression in 7.8 caused by the fix for PR17072: commit 0017922d0292d8c374584f6100874580659c9973 Author: Pedro Alves <palves@redhat.com> Date: Mon Jul 14 19:55:32 2014 +0100 Background execution + pagination aborts readline/gdb gdb_readline_wrapper_line removes the handler after a line is processed. Usually, we'll end up re-displaying the prompt, and that reinstalls the handler. But if the output is coming out of handling a stop event, we don't re-display the prompt, and nothing restores the handler. So the next input wakes up the event loop and calls into readline, which aborts. ... gdb/ 2014-07-14 Pedro Alves <palves@redhat.com> PR gdb/17072 * top.c (gdb_readline_wrapper_line): Tweak comment. (gdb_readline_wrapper_cleanup): If readline is enabled, reinstall the input handler callback. The problem is that installing the input handler callback also preps the terminal, putting it in raw mode and with echo disabled, which is bad if we're going to call a command that assumes cooked/canonical mode, and echo enabled, like in the case of the PR, Python's interactive shell. Another example I came up with that doesn't depend on Python is starting a subshell with "(gdb) shell /bin/sh" from a multi-line command. Tests covering both these examples are added. The fix is to revert the original fix for PR gdb/17072, and instead restore the callback handler after processing an asynchronous target event. Furthermore, calling rl_callback_handler_install when we already have some input in readline's line buffer discards that input, which is obviously a bad thing to do while the user is typing. No specific test is added for that, because I first tried calling it even if the callback handler was still installed and that resulted in hundreds of failures in the testsuite. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * event-top.c (change_line_handler): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (callback_handler_installed): New global. (gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New functions. (display_gdb_prompt): Call gdb_rl_callback_handler_remove and gdb_rl_callback_handler_install instead of rl_callback_handler_remove and rl_callback_handler_install. (gdb_disable_readline): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. * event-top.h (gdb_rl_callback_handler_remove) (gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New declarations. * infrun.c (reinstall_readline_callback_handler_cleanup): New cleanup function. (fetch_inferior_event): Install it. * top.c (gdb_readline_wrapper_line) Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (gdb_readline_wrapper_cleanup): Don't call rl_callback_handler_install. gdb/testsuite/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * gdb.python/python.exp: Test a multi-line command that spawns interactive Python. * gdb.base/multi-line-starts-subshell.exp: New file.
2014-10-23 16:13:35 +00:00
gdb_rl_callback_handler_remove ();
do_cleanups (old_chain);
return;
}
else
{
/* Display the top level prompt. */
actual_gdb_prompt = top_level_prompt ();
}
1999-05-11 20:29:07 +00:00
}
else
actual_gdb_prompt = xstrdup (new_prompt);
1999-05-11 20:29:07 +00:00
if (async_command_editing_p)
{
PR python/17372 - Python hangs when displaying help() This is more of a readline/terminal issue than a Python one. PR17372 is a regression in 7.8 caused by the fix for PR17072: commit 0017922d0292d8c374584f6100874580659c9973 Author: Pedro Alves <palves@redhat.com> Date: Mon Jul 14 19:55:32 2014 +0100 Background execution + pagination aborts readline/gdb gdb_readline_wrapper_line removes the handler after a line is processed. Usually, we'll end up re-displaying the prompt, and that reinstalls the handler. But if the output is coming out of handling a stop event, we don't re-display the prompt, and nothing restores the handler. So the next input wakes up the event loop and calls into readline, which aborts. ... gdb/ 2014-07-14 Pedro Alves <palves@redhat.com> PR gdb/17072 * top.c (gdb_readline_wrapper_line): Tweak comment. (gdb_readline_wrapper_cleanup): If readline is enabled, reinstall the input handler callback. The problem is that installing the input handler callback also preps the terminal, putting it in raw mode and with echo disabled, which is bad if we're going to call a command that assumes cooked/canonical mode, and echo enabled, like in the case of the PR, Python's interactive shell. Another example I came up with that doesn't depend on Python is starting a subshell with "(gdb) shell /bin/sh" from a multi-line command. Tests covering both these examples are added. The fix is to revert the original fix for PR gdb/17072, and instead restore the callback handler after processing an asynchronous target event. Furthermore, calling rl_callback_handler_install when we already have some input in readline's line buffer discards that input, which is obviously a bad thing to do while the user is typing. No specific test is added for that, because I first tried calling it even if the callback handler was still installed and that resulted in hundreds of failures in the testsuite. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * event-top.c (change_line_handler): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (callback_handler_installed): New global. (gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New functions. (display_gdb_prompt): Call gdb_rl_callback_handler_remove and gdb_rl_callback_handler_install instead of rl_callback_handler_remove and rl_callback_handler_install. (gdb_disable_readline): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. * event-top.h (gdb_rl_callback_handler_remove) (gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New declarations. * infrun.c (reinstall_readline_callback_handler_cleanup): New cleanup function. (fetch_inferior_event): Install it. * top.c (gdb_readline_wrapper_line) Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (gdb_readline_wrapper_cleanup): Don't call rl_callback_handler_install. gdb/testsuite/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * gdb.python/python.exp: Test a multi-line command that spawns interactive Python. * gdb.base/multi-line-starts-subshell.exp: New file.
2014-10-23 16:13:35 +00:00
gdb_rl_callback_handler_remove ();
gdb_rl_callback_handler_install (actual_gdb_prompt);
1999-05-11 20:29:07 +00:00
}
/* new_prompt at this point can be the top of the stack or the one
passed in. It can't be NULL. */
else
1999-05-11 20:29:07 +00:00
{
/* Don't use a _filtered function here. It causes the assumed
character position to be off, since the newline we read from
the user is not accounted for. */
fputs_unfiltered (actual_gdb_prompt, gdb_stdout);
1999-05-11 20:29:07 +00:00
gdb_flush (gdb_stdout);
}
do_cleanups (old_chain);
1999-05-11 20:29:07 +00:00
}
/* Return the top level prompt, as specified by "set prompt", possibly
overriden by the python gdb.prompt_hook hook, and then composed
with the prompt prefix and suffix (annotations). The caller is
responsible for freeing the returned string. */
static char *
top_level_prompt (void)
1999-05-11 20:29:07 +00:00
{
char *prefix;
char *prompt = NULL;
char *suffix;
char *composed_prompt;
size_t prompt_length;
1999-05-11 20:29:07 +00:00
/* Give observers a chance of changing the prompt. E.g., the python
`gdb.prompt_hook' is installed as an observer. */
observer_notify_before_prompt (get_prompt ());
prompt = xstrdup (get_prompt ());
1999-05-11 20:29:07 +00:00
if (annotation_level >= 2)
1999-05-11 20:29:07 +00:00
{
/* Prefix needs to have new line at end. */
prefix = (char *) alloca (strlen (async_annotation_suffix) + 10);
strcpy (prefix, "\n\032\032pre-");
strcat (prefix, async_annotation_suffix);
strcat (prefix, "\n");
/* Suffix needs to have a new line at end and \032 \032 at
beginning. */
suffix = (char *) alloca (strlen (async_annotation_suffix) + 6);
strcpy (suffix, "\n\032\032");
strcat (suffix, async_annotation_suffix);
strcat (suffix, "\n");
1999-05-11 20:29:07 +00:00
}
else
{
prefix = "";
suffix = "";
1999-05-11 20:29:07 +00:00
}
prompt_length = strlen (prefix) + strlen (prompt) + strlen (suffix);
composed_prompt = xmalloc (prompt_length + 1);
1999-05-11 20:29:07 +00:00
strcpy (composed_prompt, prefix);
strcat (composed_prompt, prompt);
strcat (composed_prompt, suffix);
1999-05-11 20:29:07 +00:00
xfree (prompt);
return composed_prompt;
1999-05-11 20:29:07 +00:00
}
1999-09-22 03:28:34 +00:00
/* When there is an event ready on the stdin file desriptor, instead
of calling readline directly throught the callback function, or
instead of calling gdb_readline2, give gdb a chance to detect
errors and do something. */
1999-09-22 03:28:34 +00:00
void
1999-10-05 23:13:56 +00:00
stdin_event_handler (int error, gdb_client_data client_data)
1999-09-22 03:28:34 +00:00
{
if (error)
{
printf_unfiltered (_("error detected on stdin\n"));
1999-10-05 23:13:56 +00:00
delete_file_handler (input_fd);
/* If stdin died, we may as well kill gdb. */
2000-01-18 00:55:13 +00:00
quit_command ((char *) 0, stdin == instream);
1999-09-22 03:28:34 +00:00
}
else
Fix the processing of Meta-key commands in TUI This patch fixes the annoying bug where key sequences such as Alt_F or Alt_B (go forward or backwards by a word) do not behave promptly in TUI. You have to press a third key in order for the key sequence to register. This is mostly ncurses' fault. Calling wgetch() normally causes ncurses to read only a single key from stdin. However if the key read is the start-sequence key (^[ a.k.a. ESC) then wgetch() reads TWO keys from stdin, storing the 2nd key into an internal FIFO buffer and returning the start-sequence key. The extraneous read of the 2nd key makes us miss its corresponding stdin event, so the event loop blocks until a third key is pressed. This explains why such key sequences do not behave promptly in TUI. To fix this issue, we must somehow compensate for the missed stdin event corresponding to the 2nd byte of a key sequence. This patch achieves this by hacking up the stdin event handler to conditionally execute the readline callback multiple times in a row. This is done via a new global variable, call_stdin_event_handler_again_p, which is set from tui_getc() when we receive a start-sequence key and notice extra pending input in the ncurses buffer. Tested on x86_64-unknown-linux-gnu. gdb/ChangeLog: * event-top.h (call_stdin_event_handler_again_p): Declare. * event-top.c (call_stdin_event_handler_again_p): Define. (stdin_event_handler): Use it. * tui/tui-io.c (tui_getc): Prepare to call the stdin event handler again if there is pending input following a start sequence.
2014-11-23 10:03:39 +00:00
{
do
{
call_stdin_event_handler_again_p = 0;
(*call_readline) (client_data);
} while (call_stdin_event_handler_again_p != 0);
}
1999-09-22 03:28:34 +00:00
}
1999-09-28 21:55:21 +00:00
/* Re-enable stdin after the end of an execution command in
synchronous mode, or after an error from the target, and we aborted
the exec operation. */
1999-09-28 21:55:21 +00:00
void
async_enable_stdin (void)
1999-09-28 21:55:21 +00:00
{
if (sync_execution)
{
/* See NOTE in async_disable_stdin(). */
/* FIXME: cagney/1999-09-27: Call this before clearing
sync_execution. Current target_terminal_ours() implementations
check for sync_execution before switching the terminal. */
target_terminal_ours ();
sync_execution = 0;
}
1999-09-28 21:55:21 +00:00
}
/* Disable reads from stdin (the console) marking the command as
synchronous. */
1999-09-28 21:55:21 +00:00
void
async_disable_stdin (void)
{
sync_execution = 1;
1999-09-28 21:55:21 +00:00
}
1999-05-11 20:29:07 +00:00
1999-09-28 21:55:21 +00:00
/* Handles a gdb command. This function is called by
1999-05-11 20:29:07 +00:00
command_line_handler, which has processed one or more input lines
into COMMAND. */
1999-05-25 18:09:09 +00:00
/* NOTE: 1999-04-30 This is the asynchronous version of the command_loop
1999-05-11 20:29:07 +00:00
function. The command_loop function will be obsolete when we
switch to use the event loop at every execution of gdb. */
1999-05-25 18:09:09 +00:00
static void
1999-09-22 03:28:34 +00:00
command_handler (char *command)
1999-05-11 20:29:07 +00:00
{
int stdin_is_tty = ISATTY (stdin);
struct cleanup *stat_chain;
1999-05-11 20:29:07 +00:00
clear_quit_flag ();
1999-05-11 20:29:07 +00:00
if (instream == stdin && stdin_is_tty)
reinitialize_more_filter ();
/* If readline returned a NULL command, it means that the connection
with the terminal is gone. This happens at the end of a
testsuite run, after Expect has hung up but GDB is still alive.
In such a case, we just quit gdb killing the inferior program
too. */
1999-05-11 20:29:07 +00:00
if (command == 0)
{
printf_unfiltered ("quit\n");
execute_command ("quit", stdin == instream);
}
1999-05-11 20:29:07 +00:00
stat_chain = make_command_stats_cleanup (1);
1999-05-11 20:29:07 +00:00
execute_command (command, instream == stdin);
1999-07-07 20:19:36 +00:00
/* Do any commands attached to breakpoint we stopped at. */
bpstat_do_actions ();
1999-07-07 20:19:36 +00:00
do_cleanups (stat_chain);
1999-07-05 17:58:44 +00:00
}
/* Handle a complete line of input. This is called by the callback
mechanism within the readline library. Deal with incomplete
commands as well, by saving the partial input in a global
buffer. */
1999-05-11 20:29:07 +00:00
1999-05-25 18:09:09 +00:00
/* NOTE: 1999-04-30 This is the asynchronous version of the
command_line_input function; command_line_input will become
1999-05-11 20:29:07 +00:00
obsolete once we use the event loop as the default mechanism in
GDB. */
1999-05-11 20:29:07 +00:00
static void
1999-09-22 03:28:34 +00:00
command_line_handler (char *rl)
1999-05-11 20:29:07 +00:00
{
static char *linebuffer = 0;
static unsigned linelength = 0;
char *p;
1999-05-11 20:29:07 +00:00
char *p1;
char *nline;
int repeat = (instream == stdin);
if (annotation_level > 1 && instream == stdin)
{
printf_unfiltered (("\n\032\032post-"));
puts_unfiltered (async_annotation_suffix);
printf_unfiltered (("\n"));
1999-05-11 20:29:07 +00:00
}
if (linebuffer == 0)
{
linelength = 80;
linebuffer = (char *) xmalloc (linelength);
Fix uninitialized value access when very first GDB command entered is <RET> While running GDB under Valgrind, I noticed that if the very first command entered is just <RET>, GDB accesses an uninitialized value: $ valgrind ./gdb -q -nx ==26790== Memcheck, a memory error detector ==26790== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al. ==26790== Using Valgrind-3.9.0 and LibVEX; rerun with -h for copyright info ==26790== Command: ./gdb -q -nx ==26790== (gdb) ==26790== Conditional jump or move depends on uninitialised value(s) ==26790== at 0x619DFC: command_line_handler (event-top.c:588) ==26790== by 0x7813D5: rl_callback_read_char (callback.c:220) ==26790== by 0x6194B4: rl_callback_read_char_wrapper (event-top.c:166) ==26790== by 0x61988A: stdin_event_handler (event-top.c:372) ==26790== by 0x61847D: handle_file_event (event-loop.c:762) ==26790== by 0x617964: process_event (event-loop.c:339) ==26790== by 0x617A2B: gdb_do_one_event (event-loop.c:403) ==26790== by 0x617A7B: start_event_loop (event-loop.c:428) ==26790== by 0x6194E6: cli_command_loop (event-top.c:181) ==26790== by 0x60F86B: current_interp_command_loop (interps.c:317) ==26790== by 0x610A34: captured_command_loop (main.c:321) ==26790== by 0x60C728: catch_errors (exceptions.c:237) ==26790== (gdb) It's this check here: /* If we just got an empty line, and that is supposed to repeat the previous command, return the value in the global buffer. */ if (repeat && p == linebuffer && *p != '\\') { The problem is that linebuffer's contents were never initialized at this point. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> * event-top.c (command_line_handler): Clear the first byte of linebuffer, when it is first allocated.
2014-10-29 11:57:03 +00:00
linebuffer[0] = '\0';
1999-05-11 20:29:07 +00:00
}
p = linebuffer;
if (more_to_come)
{
strcpy (linebuffer, readline_input_state.linebuffer);
p = readline_input_state.linebuffer_ptr;
2000-12-15 01:01:51 +00:00
xfree (readline_input_state.linebuffer);
1999-05-11 20:29:07 +00:00
more_to_come = 0;
}
#ifdef STOP_SIGNAL
if (job_control)
1999-06-14 18:08:47 +00:00
signal (STOP_SIGNAL, handle_stop_sig);
1999-05-11 20:29:07 +00:00
#endif
/* Make sure that all output has been output. Some machines may let
you get away with leaving out some of the gdb_flush, but not
all. */
1999-05-11 20:29:07 +00:00
wrap_here ("");
gdb_flush (gdb_stdout);
gdb_flush (gdb_stderr);
if (source_file_name != NULL)
++source_line_number;
1999-05-11 20:29:07 +00:00
/* If we are in this case, then command_handler will call quit
and exit from gdb. */
1999-05-11 20:29:07 +00:00
if (!rl || rl == (char *) EOF)
{
command_handler (0);
return; /* Lint. */
1999-05-11 20:29:07 +00:00
}
if (strlen (rl) + 1 + (p - linebuffer) > linelength)
{
linelength = strlen (rl) + 1 + (p - linebuffer);
nline = (char *) xrealloc (linebuffer, linelength);
p += nline - linebuffer;
linebuffer = nline;
}
p1 = rl;
/* Copy line. Don't copy null at end. (Leaves line alone
if this was just a newline). */
1999-05-11 20:29:07 +00:00
while (*p1)
*p++ = *p1++;
2000-12-15 01:01:51 +00:00
xfree (rl); /* Allocated in readline. */
1999-05-11 20:29:07 +00:00
if (p > linebuffer && *(p - 1) == '\\')
1999-05-11 20:29:07 +00:00
{
*p = '\0';
1999-05-11 20:29:07 +00:00
p--; /* Put on top of '\'. */
readline_input_state.linebuffer = xstrdup (linebuffer);
readline_input_state.linebuffer_ptr = p;
/* We will not invoke a execute_command if there is more
input expected to complete the command. So, we need to
print an empty prompt here. */
more_to_come = 1;
display_gdb_prompt ("");
return;
1999-05-11 20:29:07 +00:00
}
#ifdef STOP_SIGNAL
if (job_control)
signal (STOP_SIGNAL, SIG_DFL);
#endif
#define SERVER_COMMAND_LENGTH 7
server_command =
(p - linebuffer > SERVER_COMMAND_LENGTH)
&& strncmp (linebuffer, "server ", SERVER_COMMAND_LENGTH) == 0;
1999-05-11 20:29:07 +00:00
if (server_command)
{
/* Note that we don't set `line'. Between this and the check in
dont_repeat, this insures that repeating will still do the
right thing. */
*p = '\0';
command_handler (linebuffer + SERVER_COMMAND_LENGTH);
display_gdb_prompt (0);
return;
}
/* Do history expansion if that is wished. */
if (history_expansion_p && instream == stdin
&& ISATTY (instream))
{
char *history_value;
int expanded;
*p = '\0'; /* Insert null now. */
expanded = history_expand (linebuffer, &history_value);
if (expanded)
{
/* Print the changes. */
printf_unfiltered ("%s\n", history_value);
/* If there was an error, call this function again. */
if (expanded < 0)
{
2000-12-15 01:01:51 +00:00
xfree (history_value);
1999-05-11 20:29:07 +00:00
return;
}
if (strlen (history_value) > linelength)
{
linelength = strlen (history_value) + 1;
linebuffer = (char *) xrealloc (linebuffer, linelength);
}
strcpy (linebuffer, history_value);
p = linebuffer + strlen (linebuffer);
}
xfree (history_value);
1999-05-11 20:29:07 +00:00
}
/* If we just got an empty line, and that is supposed to repeat the
previous command, return the value in the global buffer. */
1999-05-11 20:29:07 +00:00
if (repeat && p == linebuffer && *p != '\\')
{
command_handler (saved_command_line);
1999-05-11 20:29:07 +00:00
display_gdb_prompt (0);
return;
}
for (p1 = linebuffer; *p1 == ' ' || *p1 == '\t'; p1++);
if (repeat && !*p1)
{
command_handler (saved_command_line);
1999-05-11 20:29:07 +00:00
display_gdb_prompt (0);
return;
}
*p = 0;
/* Add line to history if appropriate. */
if (*linebuffer && input_from_terminal_p ())
gdb_add_history (linebuffer);
1999-05-11 20:29:07 +00:00
/* Note: lines consisting solely of comments are added to the command
history. This is useful when you type a command, and then
realize you don't want to execute it quite yet. You can comment
out the command and then later fetch it from the value history
and remove the '#'. The kill ring is probably better, but some
people are in the habit of commenting things out. */
if (*p1 == '#')
*p1 = '\0'; /* Found a comment. */
1999-05-11 20:29:07 +00:00
/* Save into global buffer if appropriate. */
if (repeat)
{
if (linelength > saved_command_line_size)
1999-05-11 20:29:07 +00:00
{
saved_command_line = xrealloc (saved_command_line, linelength);
saved_command_line_size = linelength;
1999-05-11 20:29:07 +00:00
}
strcpy (saved_command_line, linebuffer);
1999-05-11 20:29:07 +00:00
if (!more_to_come)
{
command_handler (saved_command_line);
1999-05-11 20:29:07 +00:00
display_gdb_prompt (0);
}
return;
}
command_handler (linebuffer);
display_gdb_prompt (0);
return;
}
/* Does reading of input from terminal w/o the editing features
provided by the readline library. */
1999-05-11 20:29:07 +00:00
/* NOTE: 1999-04-30 Asynchronous version of gdb_readline; gdb_readline
1999-05-11 20:29:07 +00:00
will become obsolete when the event loop is made the default
execution for gdb. */
1999-06-28 16:06:02 +00:00
void
1999-09-22 03:28:34 +00:00
gdb_readline2 (gdb_client_data client_data)
1999-05-11 20:29:07 +00:00
{
int c;
char *result;
int input_index = 0;
int result_size = 80;
1999-08-16 19:57:19 +00:00
static int done_once = 0;
/* Unbuffer the input stream, so that, later on, the calls to fgetc
fetch only one char at the time from the stream. The fgetc's will
1999-08-16 19:57:19 +00:00
get up to the first newline, but there may be more chars in the
stream after '\n'. If we buffer the input and fgetc drains the
1999-08-16 19:57:19 +00:00
stream, getting stuff beyond the newline as well, a select, done
afterwards will not trigger. */
1999-08-16 19:57:19 +00:00
if (!done_once && !ISATTY (instream))
{
setbuf (instream, NULL);
done_once = 1;
}
1999-05-11 20:29:07 +00:00
result = (char *) xmalloc (result_size);
/* We still need the while loop here, even though it would seem
obvious to invoke gdb_readline2 at every character entered. If
not using the readline library, the terminal is in cooked mode,
which sends the characters all at once. Poll will notice that the
input fd has changed state only after enter is pressed. At this
point we still need to fetch all the chars entered. */
1999-05-11 20:29:07 +00:00
while (1)
{
/* Read from stdin if we are executing a user defined command.
This is the right thing for prompt_for_continue, at least. */
c = fgetc (instream ? instream : stdin);
if (c == EOF)
{
if (input_index > 0)
/* The last line does not end with a newline. Return it,
and if we are called again fgetc will still return EOF
and we'll return NULL then. */
1999-05-11 20:29:07 +00:00
break;
2000-12-15 01:01:51 +00:00
xfree (result);
1999-06-14 18:08:47 +00:00
(*input_handler) (0);
return;
1999-05-11 20:29:07 +00:00
}
if (c == '\n')
{
if (input_index > 0 && result[input_index - 1] == '\r')
input_index--;
break;
}
result[input_index++] = c;
while (input_index >= result_size)
{
result_size *= 2;
result = (char *) xrealloc (result, result_size);
}
}
result[input_index++] = '\0';
1999-06-14 18:08:47 +00:00
(*input_handler) (result);
1999-05-11 20:29:07 +00:00
}
/* Initialization of signal handlers and tokens. There is a function
handle_sig* for each of the signals GDB cares about. Specifically:
1999-05-11 20:29:07 +00:00
SIGINT, SIGFPE, SIGQUIT, SIGTSTP, SIGHUP, SIGWINCH. These
functions are the actual signal handlers associated to the signals
via calls to signal(). The only job for these functions is to
enqueue the appropriate event/procedure with the event loop. Such
procedures are the old signal handlers. The event loop will take
1999-05-11 20:29:07 +00:00
care of invoking the queued procedures to perform the usual tasks
associated with the reception of the signal. */
1999-05-25 18:09:09 +00:00
/* NOTE: 1999-04-30 This is the asynchronous version of init_signals.
1999-05-11 20:29:07 +00:00
init_signals will become obsolete as we move to have to event loop
as the default for gdb. */
1999-05-11 20:29:07 +00:00
void
1999-09-22 03:28:34 +00:00
async_init_signals (void)
1999-07-07 20:19:36 +00:00
{
1999-05-11 20:29:07 +00:00
signal (SIGINT, handle_sigint);
sigint_token =
1999-06-14 18:08:47 +00:00
create_async_signal_handler (async_request_quit, NULL);
signal (SIGTERM, handle_sigterm);
async_sigterm_token
= create_async_signal_handler (async_sigterm_handler, NULL);
1999-05-11 20:29:07 +00:00
/* If SIGTRAP was set to SIG_IGN, then the SIG_IGN will get passed
to the inferior and breakpoints will be ignored. */
#ifdef SIGTRAP
signal (SIGTRAP, SIG_DFL);
#endif
#ifdef SIGQUIT
1999-05-11 20:29:07 +00:00
/* If we initialize SIGQUIT to SIG_IGN, then the SIG_IGN will get
passed to the inferior, which we don't want. It would be
possible to do a "signal (SIGQUIT, SIG_DFL)" after we fork, but
on BSD4.3 systems using vfork, that can affect the
GDB process as well as the inferior (the signal handling tables
might be in memory, shared between the two). Since we establish
a handler for SIGQUIT, when we call exec it will set the signal
to SIG_DFL for us. */
signal (SIGQUIT, handle_sigquit);
sigquit_token =
1999-06-14 18:08:47 +00:00
create_async_signal_handler (async_do_nothing, NULL);
#endif
1999-05-11 20:29:07 +00:00
#ifdef SIGHUP
if (signal (SIGHUP, handle_sighup) != SIG_IGN)
sighup_token =
1999-06-14 18:08:47 +00:00
create_async_signal_handler (async_disconnect, NULL);
1999-05-11 20:29:07 +00:00
else
sighup_token =
1999-06-14 18:08:47 +00:00
create_async_signal_handler (async_do_nothing, NULL);
1999-05-11 20:29:07 +00:00
#endif
signal (SIGFPE, handle_sigfpe);
sigfpe_token =
1999-06-14 18:08:47 +00:00
create_async_signal_handler (async_float_handler, NULL);
1999-05-11 20:29:07 +00:00
1999-06-14 18:08:47 +00:00
#ifdef STOP_SIGNAL
sigtstp_token =
create_async_signal_handler (async_stop_sig, NULL);
#endif
}
/* Tell the event loop what to do if SIGINT is received.
See event-signal.c. */
1999-07-07 20:19:36 +00:00
void
1999-09-22 03:28:34 +00:00
handle_sigint (int sig)
1999-05-11 20:29:07 +00:00
{
signal (sig, handle_sigint);
/* We could be running in a loop reading in symfiles or something so
it may be quite a while before we get back to the event loop. So
set quit_flag to 1 here. Then if QUIT is called before we get to
the event loop, we will unwind as expected. */
set_quit_flag ();
1999-05-11 20:29:07 +00:00
/* If immediate_quit is set, we go ahead and process the SIGINT right
away, even if we usually would defer this to the event loop. The
1999-05-11 20:29:07 +00:00
assumption here is that it is safe to process ^C immediately if
immediate_quit is set. If we didn't, SIGINT would be really
1999-05-11 20:29:07 +00:00
processed only the next time through the event loop. To get to
that point, though, the command that we want to interrupt needs to
finish first, which is unacceptable. If immediate quit is not set,
we process SIGINT the next time through the loop, which is fine. */
gdb_call_async_signal_handler (sigint_token, immediate_quit);
1999-05-11 20:29:07 +00:00
}
/* Handle GDB exit upon receiving SIGTERM if target_can_async_p (). */
static void
async_sigterm_handler (gdb_client_data arg)
{
quit_force (NULL, stdin == instream);
}
/* See defs.h. */
volatile int sync_quit_force_run;
/* Quit GDB if SIGTERM is received.
GDB would quit anyway, but this way it will clean up properly. */
void
handle_sigterm (int sig)
{
signal (sig, handle_sigterm);
sync_quit_force_run = 1;
set_quit_flag ();
mark_async_signal_handler (async_sigterm_token);
}
/* Do the quit. All the checks have been done by the caller. */
1999-07-07 20:19:36 +00:00
void
1999-09-22 03:28:34 +00:00
async_request_quit (gdb_client_data arg)
1999-05-11 20:29:07 +00:00
{
/* If the quit_flag has gotten reset back to 0 by the time we get
back here, that means that an exception was thrown to unwind the
current command before we got back to the event loop. So there
is no reason to call quit again here. */
if (check_quit_flag ())
quit ();
1999-05-11 20:29:07 +00:00
}
#ifdef SIGQUIT
/* Tell the event loop what to do if SIGQUIT is received.
See event-signal.c. */
1999-07-07 20:19:36 +00:00
static void
1999-09-22 03:28:34 +00:00
handle_sigquit (int sig)
1999-05-11 20:29:07 +00:00
{
mark_async_signal_handler (sigquit_token);
1999-05-11 20:29:07 +00:00
signal (sig, handle_sigquit);
}
#endif
1999-05-11 20:29:07 +00:00
#if defined (SIGQUIT) || defined (SIGHUP)
/* Called by the event loop in response to a SIGQUIT or an
ignored SIGHUP. */
1999-07-07 20:19:36 +00:00
static void
1999-09-22 03:28:34 +00:00
async_do_nothing (gdb_client_data arg)
1999-05-11 20:29:07 +00:00
{
/* Empty function body. */
1999-05-11 20:29:07 +00:00
}
#endif
1999-05-11 20:29:07 +00:00
#ifdef SIGHUP
/* Tell the event loop what to do if SIGHUP is received.
See event-signal.c. */
1999-07-07 20:19:36 +00:00
static void
2000-07-30 01:48:28 +00:00
handle_sighup (int sig)
1999-05-11 20:29:07 +00:00
{
mark_async_signal_handler (sighup_token);
1999-05-11 20:29:07 +00:00
signal (sig, handle_sighup);
}
/* Called by the event loop to process a SIGHUP. */
1999-07-07 20:19:36 +00:00
static void
1999-09-22 03:28:34 +00:00
async_disconnect (gdb_client_data arg)
1999-05-11 20:29:07 +00:00
{
Split TRY_CATCH into TRY + CATCH This patch splits the TRY_CATCH macro into three, so that we go from this: ~~~ volatile gdb_exception ex; TRY_CATCH (ex, RETURN_MASK_ERROR) { } if (ex.reason < 0) { } ~~~ to this: ~~~ TRY { } CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH ~~~ Thus, we'll be getting rid of the local volatile exception object, and declaring the caught exception in the catch block. This allows reimplementing TRY/CATCH in terms of C++ exceptions when building in C++ mode, while still allowing to build GDB in C mode (using setjmp/longjmp), as a transition step. TBC, after this patch, is it _not_ valid to have code between the TRY and the CATCH blocks, like: TRY { } // some code here. CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH Just like it isn't valid to do that with C++'s native try/catch. By switching to creating the exception object inside the CATCH block scope, we can get rid of all the explicitly allocated volatile exception objects all over the tree, and map the CATCH block more directly to C++'s catch blocks. The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was done with a script, rerun from scratch at every rebase, no manual editing involved. After the mechanical conversion, a few places needed manual intervention, to fix preexisting cases where we were using the exception object outside of the TRY_CATCH block, and cases where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH after this patch]. The result was folded into this patch so that GDB still builds at each incremental step. END_CATCH is necessary for two reasons: First, because we name the exception object in the CATCH block, which requires creating a scope, which in turn must be closed somewhere. Declaring the exception variable in the initializer field of a for block, like: #define CATCH(EXCEPTION, mask) \ for (struct gdb_exception EXCEPTION; \ exceptions_state_mc_catch (&EXCEPTION, MASK); \ EXCEPTION = exception_none) would avoid needing END_CATCH, but alas, in C mode, we build with C90, which doesn't allow mixed declarations and code. Second, because when TRY/CATCH are wired to real C++ try/catch, as long as we need to handle cleanup chains, even if there's no CATCH block that wants to catch the exception, we need for stop at every frame in the unwind chain and run cleanups, then rethrow. That will be done in END_CATCH. After we require C++, we'll still need TRY/CATCH/END_CATCH until cleanups are completely phased out -- TRY/CATCH in C++ mode will save/restore the current cleanup chain, like in C mode, and END_CATCH catches otherwise uncaugh exceptions, runs cleanups and rethrows, so that C++ cleanups and exceptions can coexist. IMO, this still makes the TRY/CATCH code look a bit more like a newcomer would expect, so IMO worth it even if we weren't considering C++. gdb/ChangeLog. 2015-03-07 Pedro Alves <palves@redhat.com> * common/common-exceptions.c (struct catcher) <exception>: No longer a pointer to volatile exception. Now an exception value. <mask>: Delete field. (exceptions_state_mc_init): Remove all parameters. Adjust. (exceptions_state_mc): No longer pop the catcher here. (exceptions_state_mc_catch): New function. (throw_exception): Adjust. * common/common-exceptions.h (exceptions_state_mc_init): Remove all parameters. (exceptions_state_mc_catch): Declare. (TRY_CATCH): Rename to ... (TRY): ... this. Remove EXCEPTION and MASK parameters. (CATCH, END_CATCH): New. All callers adjusted. gdb/gdbserver/ChangeLog: 2015-03-07 Pedro Alves <palves@redhat.com> Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH instead.
2015-03-07 15:14:14 +00:00
TRY
{
quit_cover ();
}
Split TRY_CATCH into TRY + CATCH This patch splits the TRY_CATCH macro into three, so that we go from this: ~~~ volatile gdb_exception ex; TRY_CATCH (ex, RETURN_MASK_ERROR) { } if (ex.reason < 0) { } ~~~ to this: ~~~ TRY { } CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH ~~~ Thus, we'll be getting rid of the local volatile exception object, and declaring the caught exception in the catch block. This allows reimplementing TRY/CATCH in terms of C++ exceptions when building in C++ mode, while still allowing to build GDB in C mode (using setjmp/longjmp), as a transition step. TBC, after this patch, is it _not_ valid to have code between the TRY and the CATCH blocks, like: TRY { } // some code here. CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH Just like it isn't valid to do that with C++'s native try/catch. By switching to creating the exception object inside the CATCH block scope, we can get rid of all the explicitly allocated volatile exception objects all over the tree, and map the CATCH block more directly to C++'s catch blocks. The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was done with a script, rerun from scratch at every rebase, no manual editing involved. After the mechanical conversion, a few places needed manual intervention, to fix preexisting cases where we were using the exception object outside of the TRY_CATCH block, and cases where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH after this patch]. The result was folded into this patch so that GDB still builds at each incremental step. END_CATCH is necessary for two reasons: First, because we name the exception object in the CATCH block, which requires creating a scope, which in turn must be closed somewhere. Declaring the exception variable in the initializer field of a for block, like: #define CATCH(EXCEPTION, mask) \ for (struct gdb_exception EXCEPTION; \ exceptions_state_mc_catch (&EXCEPTION, MASK); \ EXCEPTION = exception_none) would avoid needing END_CATCH, but alas, in C mode, we build with C90, which doesn't allow mixed declarations and code. Second, because when TRY/CATCH are wired to real C++ try/catch, as long as we need to handle cleanup chains, even if there's no CATCH block that wants to catch the exception, we need for stop at every frame in the unwind chain and run cleanups, then rethrow. That will be done in END_CATCH. After we require C++, we'll still need TRY/CATCH/END_CATCH until cleanups are completely phased out -- TRY/CATCH in C++ mode will save/restore the current cleanup chain, like in C mode, and END_CATCH catches otherwise uncaugh exceptions, runs cleanups and rethrows, so that C++ cleanups and exceptions can coexist. IMO, this still makes the TRY/CATCH code look a bit more like a newcomer would expect, so IMO worth it even if we weren't considering C++. gdb/ChangeLog. 2015-03-07 Pedro Alves <palves@redhat.com> * common/common-exceptions.c (struct catcher) <exception>: No longer a pointer to volatile exception. Now an exception value. <mask>: Delete field. (exceptions_state_mc_init): Remove all parameters. Adjust. (exceptions_state_mc): No longer pop the catcher here. (exceptions_state_mc_catch): New function. (throw_exception): Adjust. * common/common-exceptions.h (exceptions_state_mc_init): Remove all parameters. (exceptions_state_mc_catch): Declare. (TRY_CATCH): Rename to ... (TRY): ... this. Remove EXCEPTION and MASK parameters. (CATCH, END_CATCH): New. All callers adjusted. gdb/gdbserver/ChangeLog: 2015-03-07 Pedro Alves <palves@redhat.com> Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH instead.
2015-03-07 15:14:14 +00:00
CATCH (exception, RETURN_MASK_ALL)
{
fputs_filtered ("Could not kill the program being debugged",
gdb_stderr);
exception_print (gdb_stderr, exception);
}
Split TRY_CATCH into TRY + CATCH This patch splits the TRY_CATCH macro into three, so that we go from this: ~~~ volatile gdb_exception ex; TRY_CATCH (ex, RETURN_MASK_ERROR) { } if (ex.reason < 0) { } ~~~ to this: ~~~ TRY { } CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH ~~~ Thus, we'll be getting rid of the local volatile exception object, and declaring the caught exception in the catch block. This allows reimplementing TRY/CATCH in terms of C++ exceptions when building in C++ mode, while still allowing to build GDB in C mode (using setjmp/longjmp), as a transition step. TBC, after this patch, is it _not_ valid to have code between the TRY and the CATCH blocks, like: TRY { } // some code here. CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH Just like it isn't valid to do that with C++'s native try/catch. By switching to creating the exception object inside the CATCH block scope, we can get rid of all the explicitly allocated volatile exception objects all over the tree, and map the CATCH block more directly to C++'s catch blocks. The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was done with a script, rerun from scratch at every rebase, no manual editing involved. After the mechanical conversion, a few places needed manual intervention, to fix preexisting cases where we were using the exception object outside of the TRY_CATCH block, and cases where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH after this patch]. The result was folded into this patch so that GDB still builds at each incremental step. END_CATCH is necessary for two reasons: First, because we name the exception object in the CATCH block, which requires creating a scope, which in turn must be closed somewhere. Declaring the exception variable in the initializer field of a for block, like: #define CATCH(EXCEPTION, mask) \ for (struct gdb_exception EXCEPTION; \ exceptions_state_mc_catch (&EXCEPTION, MASK); \ EXCEPTION = exception_none) would avoid needing END_CATCH, but alas, in C mode, we build with C90, which doesn't allow mixed declarations and code. Second, because when TRY/CATCH are wired to real C++ try/catch, as long as we need to handle cleanup chains, even if there's no CATCH block that wants to catch the exception, we need for stop at every frame in the unwind chain and run cleanups, then rethrow. That will be done in END_CATCH. After we require C++, we'll still need TRY/CATCH/END_CATCH until cleanups are completely phased out -- TRY/CATCH in C++ mode will save/restore the current cleanup chain, like in C mode, and END_CATCH catches otherwise uncaugh exceptions, runs cleanups and rethrows, so that C++ cleanups and exceptions can coexist. IMO, this still makes the TRY/CATCH code look a bit more like a newcomer would expect, so IMO worth it even if we weren't considering C++. gdb/ChangeLog. 2015-03-07 Pedro Alves <palves@redhat.com> * common/common-exceptions.c (struct catcher) <exception>: No longer a pointer to volatile exception. Now an exception value. <mask>: Delete field. (exceptions_state_mc_init): Remove all parameters. Adjust. (exceptions_state_mc): No longer pop the catcher here. (exceptions_state_mc_catch): New function. (throw_exception): Adjust. * common/common-exceptions.h (exceptions_state_mc_init): Remove all parameters. (exceptions_state_mc_catch): Declare. (TRY_CATCH): Rename to ... (TRY): ... this. Remove EXCEPTION and MASK parameters. (CATCH, END_CATCH): New. All callers adjusted. gdb/gdbserver/ChangeLog: 2015-03-07 Pedro Alves <palves@redhat.com> Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH instead.
2015-03-07 15:14:14 +00:00
END_CATCH
Split TRY_CATCH into TRY + CATCH This patch splits the TRY_CATCH macro into three, so that we go from this: ~~~ volatile gdb_exception ex; TRY_CATCH (ex, RETURN_MASK_ERROR) { } if (ex.reason < 0) { } ~~~ to this: ~~~ TRY { } CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH ~~~ Thus, we'll be getting rid of the local volatile exception object, and declaring the caught exception in the catch block. This allows reimplementing TRY/CATCH in terms of C++ exceptions when building in C++ mode, while still allowing to build GDB in C mode (using setjmp/longjmp), as a transition step. TBC, after this patch, is it _not_ valid to have code between the TRY and the CATCH blocks, like: TRY { } // some code here. CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH Just like it isn't valid to do that with C++'s native try/catch. By switching to creating the exception object inside the CATCH block scope, we can get rid of all the explicitly allocated volatile exception objects all over the tree, and map the CATCH block more directly to C++'s catch blocks. The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was done with a script, rerun from scratch at every rebase, no manual editing involved. After the mechanical conversion, a few places needed manual intervention, to fix preexisting cases where we were using the exception object outside of the TRY_CATCH block, and cases where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH after this patch]. The result was folded into this patch so that GDB still builds at each incremental step. END_CATCH is necessary for two reasons: First, because we name the exception object in the CATCH block, which requires creating a scope, which in turn must be closed somewhere. Declaring the exception variable in the initializer field of a for block, like: #define CATCH(EXCEPTION, mask) \ for (struct gdb_exception EXCEPTION; \ exceptions_state_mc_catch (&EXCEPTION, MASK); \ EXCEPTION = exception_none) would avoid needing END_CATCH, but alas, in C mode, we build with C90, which doesn't allow mixed declarations and code. Second, because when TRY/CATCH are wired to real C++ try/catch, as long as we need to handle cleanup chains, even if there's no CATCH block that wants to catch the exception, we need for stop at every frame in the unwind chain and run cleanups, then rethrow. That will be done in END_CATCH. After we require C++, we'll still need TRY/CATCH/END_CATCH until cleanups are completely phased out -- TRY/CATCH in C++ mode will save/restore the current cleanup chain, like in C mode, and END_CATCH catches otherwise uncaugh exceptions, runs cleanups and rethrows, so that C++ cleanups and exceptions can coexist. IMO, this still makes the TRY/CATCH code look a bit more like a newcomer would expect, so IMO worth it even if we weren't considering C++. gdb/ChangeLog. 2015-03-07 Pedro Alves <palves@redhat.com> * common/common-exceptions.c (struct catcher) <exception>: No longer a pointer to volatile exception. Now an exception value. <mask>: Delete field. (exceptions_state_mc_init): Remove all parameters. Adjust. (exceptions_state_mc): No longer pop the catcher here. (exceptions_state_mc_catch): New function. (throw_exception): Adjust. * common/common-exceptions.h (exceptions_state_mc_init): Remove all parameters. (exceptions_state_mc_catch): Declare. (TRY_CATCH): Rename to ... (TRY): ... this. Remove EXCEPTION and MASK parameters. (CATCH, END_CATCH): New. All callers adjusted. gdb/gdbserver/ChangeLog: 2015-03-07 Pedro Alves <palves@redhat.com> Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH instead.
2015-03-07 15:14:14 +00:00
TRY
{
gdb/ Code cleanup. * bfd-target.c (target_bfd_xclose): Remove parameter quitting. * bsd-kvm.c (bsd_kvm_close): Likewise. * bsd-uthread.c (bsd_uthread_close): Likewise. * corelow.c (core_close): Likewise. (core_close_cleanup): Remove parameter quitting from a caller. * event-top.c (async_disconnect): Likewise. * exec.c (exec_close_1): Remove parameter quitting. * go32-nat.c (go32_close): Likewise. * linux-nat.c (linux_nat_close): Remove parameter quitting. Remove parameter quitting from a caller. * mips-linux-nat.c (super_close): Remove parameter quitting from the variable. (mips_linux_close): Remove parameter quitting. Remove parameter quitting from a caller. * monitor.c (monitor_close): Remove parameter quitting. * monitor.h (monitor_close): Likewise. * record-btrace.c (record_btrace_close): Likewise. * record-full.c (record_full_close): Likewise. * remote-m32r-sdi.c (m32r_close): Remove parameter quitting and remove it also from fprintf_unfiltered. * remote-mips.c (mips_close): Remove parameter quitting. (mips_detach): Remove parameter quitting from a caller. * remote-sim.c (gdbsim_close): Remove parameter quitting. (gdbsim_close): Remove duplicate function comment. Remove parameter quitting and remove it also from printf_filtered. * remote.c (remote_close): Remove parameter quitting. * solib-svr4.c (enable_break): Remove parameter quitting from a caller. * target.c (update_current_target): Remove parameter int from to_close de_fault. (push_target, unpush_target, pop_target): Remove parameter quitting from a caller. (pop_all_targets_above, pop_all_targets): Remove parameter quitting. Remove parameter quitting from a caller. (target_preopen): Remove parameter quitting from a caller. (target_close): Remove parameter quitting. Remove parameter quitting from a caller two times. Remove parameter quitting also from fprintf_unfiltered. * target.h (struct target_ops): Remove parameter quitting and as int from fields to_xclose and to_close. (extern struct target_ops current_target): (target_close, pop_all_targets): Remove parameter quitting. Update the comment. (pop_all_targets_above): Remove parameter quitting. * top.c (quit_target): Remove parameter quitting from a caller. * tracepoint.c (tfile_close): Remove parameter quitting. * windows-nat.c (windows_close): Remove parameter quitting.
2013-03-20 15:46:24 +00:00
pop_all_targets ();
}
Split TRY_CATCH into TRY + CATCH This patch splits the TRY_CATCH macro into three, so that we go from this: ~~~ volatile gdb_exception ex; TRY_CATCH (ex, RETURN_MASK_ERROR) { } if (ex.reason < 0) { } ~~~ to this: ~~~ TRY { } CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH ~~~ Thus, we'll be getting rid of the local volatile exception object, and declaring the caught exception in the catch block. This allows reimplementing TRY/CATCH in terms of C++ exceptions when building in C++ mode, while still allowing to build GDB in C mode (using setjmp/longjmp), as a transition step. TBC, after this patch, is it _not_ valid to have code between the TRY and the CATCH blocks, like: TRY { } // some code here. CATCH (ex, RETURN_MASK_ERROR) { } END_CATCH Just like it isn't valid to do that with C++'s native try/catch. By switching to creating the exception object inside the CATCH block scope, we can get rid of all the explicitly allocated volatile exception objects all over the tree, and map the CATCH block more directly to C++'s catch blocks. The majority of the TRY_CATCH -> TRY+CATCH+END_CATCH conversion was done with a script, rerun from scratch at every rebase, no manual editing involved. After the mechanical conversion, a few places needed manual intervention, to fix preexisting cases where we were using the exception object outside of the TRY_CATCH block, and cases where we were using "else" after a 'if (ex.reason) < 0)' [a CATCH after this patch]. The result was folded into this patch so that GDB still builds at each incremental step. END_CATCH is necessary for two reasons: First, because we name the exception object in the CATCH block, which requires creating a scope, which in turn must be closed somewhere. Declaring the exception variable in the initializer field of a for block, like: #define CATCH(EXCEPTION, mask) \ for (struct gdb_exception EXCEPTION; \ exceptions_state_mc_catch (&EXCEPTION, MASK); \ EXCEPTION = exception_none) would avoid needing END_CATCH, but alas, in C mode, we build with C90, which doesn't allow mixed declarations and code. Second, because when TRY/CATCH are wired to real C++ try/catch, as long as we need to handle cleanup chains, even if there's no CATCH block that wants to catch the exception, we need for stop at every frame in the unwind chain and run cleanups, then rethrow. That will be done in END_CATCH. After we require C++, we'll still need TRY/CATCH/END_CATCH until cleanups are completely phased out -- TRY/CATCH in C++ mode will save/restore the current cleanup chain, like in C mode, and END_CATCH catches otherwise uncaugh exceptions, runs cleanups and rethrows, so that C++ cleanups and exceptions can coexist. IMO, this still makes the TRY/CATCH code look a bit more like a newcomer would expect, so IMO worth it even if we weren't considering C++. gdb/ChangeLog. 2015-03-07 Pedro Alves <palves@redhat.com> * common/common-exceptions.c (struct catcher) <exception>: No longer a pointer to volatile exception. Now an exception value. <mask>: Delete field. (exceptions_state_mc_init): Remove all parameters. Adjust. (exceptions_state_mc): No longer pop the catcher here. (exceptions_state_mc_catch): New function. (throw_exception): Adjust. * common/common-exceptions.h (exceptions_state_mc_init): Remove all parameters. (exceptions_state_mc_catch): Declare. (TRY_CATCH): Rename to ... (TRY): ... this. Remove EXCEPTION and MASK parameters. (CATCH, END_CATCH): New. All callers adjusted. gdb/gdbserver/ChangeLog: 2015-03-07 Pedro Alves <palves@redhat.com> Adjust all callers of TRY_CATCH to use TRY/CATCH/END_CATCH instead.
2015-03-07 15:14:14 +00:00
CATCH (exception, RETURN_MASK_ALL)
{
}
END_CATCH
signal (SIGHUP, SIG_DFL); /*FIXME: ??????????? */
raise (SIGHUP);
1999-05-11 20:29:07 +00:00
}
#endif
1999-06-14 18:08:47 +00:00
#ifdef STOP_SIGNAL
1999-07-07 20:19:36 +00:00
void
1999-09-22 03:28:34 +00:00
handle_stop_sig (int sig)
1999-06-14 18:08:47 +00:00
{
mark_async_signal_handler (sigtstp_token);
1999-07-07 20:19:36 +00:00
signal (sig, handle_stop_sig);
1999-06-14 18:08:47 +00:00
}
static void
1999-09-22 03:28:34 +00:00
async_stop_sig (gdb_client_data arg)
1999-06-14 18:08:47 +00:00
{
char *prompt = get_prompt ();
1999-06-14 18:08:47 +00:00
#if STOP_SIGNAL == SIGTSTP
signal (SIGTSTP, SIG_DFL);
1999-10-05 23:13:56 +00:00
#if HAVE_SIGPROCMASK
{
sigset_t zero;
1999-10-05 23:13:56 +00:00
sigemptyset (&zero);
sigprocmask (SIG_SETMASK, &zero, 0);
}
#elif HAVE_SIGSETMASK
1999-06-14 18:08:47 +00:00
sigsetmask (0);
1999-10-05 23:13:56 +00:00
#endif
raise (SIGTSTP);
1999-06-14 18:08:47 +00:00
signal (SIGTSTP, handle_stop_sig);
#else
signal (STOP_SIGNAL, handle_stop_sig);
#endif
printf_unfiltered ("%s", prompt);
gdb_flush (gdb_stdout);
/* Forget about any previous command -- null line now will do
nothing. */
1999-06-14 18:08:47 +00:00
dont_repeat ();
}
#endif /* STOP_SIGNAL */
/* Tell the event loop what to do if SIGFPE is received.
See event-signal.c. */
1999-07-07 20:19:36 +00:00
static void
1999-09-22 03:28:34 +00:00
handle_sigfpe (int sig)
1999-05-11 20:29:07 +00:00
{
mark_async_signal_handler (sigfpe_token);
1999-05-11 20:29:07 +00:00
signal (sig, handle_sigfpe);
}
/* Event loop will call this functin to process a SIGFPE. */
1999-07-07 20:19:36 +00:00
static void
1999-09-22 03:28:34 +00:00
async_float_handler (gdb_client_data arg)
1999-05-11 20:29:07 +00:00
{
/* This message is based on ANSI C, section 4.7. Note that integer
divide by zero causes this, so "float" is a misnomer. */
2005-02-10 Andrew Cagney <cagney@gnu.org> Mark up all error and warning messages. * ada-lang.c, amd64-tdep.c, arch-utils.c, breakpoint.c: Update. * bsd-kvm.c, bsd-uthread.c, coff-solib.h, coffread.c: Update. * core-aout.c, core-regset.c, corefile.c, corelow.c: Update. * cp-abi.c, cp-support.c, cp-valprint.c, cris-tdep.c: Update. * dbxread.c, demangle.c, doublest.c, dsrec.c: Update. * dve3900-rom.c, dwarf2expr.c, dwarf2loc.c: Update. * dwarf2read.c, dwarfread.c, elfread.c, eval.c: Update. * event-top.c, exec.c, expprint.c, f-lang.c: Update. * f-typeprint.c, f-valprint.c, fbsd-nat.c, findvar.c: Update. * frame.c, frv-linux-tdep.c, gcore.c, gdbtypes.c: Update. * gnu-nat.c, gnu-v2-abi.c, gnu-v3-abi.c, go32-nat.c: Update. * hpacc-abi.c, hppa-hpux-nat.c, hppa-hpux-tdep.c: Update. * hppa-linux-nat.c, hppa-linux-tdep.c, hppa-tdep.c: Update. * hpread.c, hpux-thread.c, i386-linux-nat.c: Update. * i386-linux-tdep.c, i386-tdep.c, i386bsd-nat.c: Update. * i386gnu-nat.c, i387-tdep.c, ia64-linux-nat.c: Update. * ia64-tdep.c, inf-child.c, inf-ptrace.c, inf-ttrace.c: Update. * infcall.c, infcmd.c, inflow.c, infptrace.c, infrun.c: Update. * inftarg.c, interps.c, irix5-nat.c, jv-lang.c: Update. * kod-cisco.c, kod.c, language.c, libunwind-frame.c: Update. * linespec.c, linux-nat.c, linux-thread-db.c, m2-lang.c: Update. * m32r-rom.c, m68hc11-tdep.c, m68k-tdep.c: Update. * m68klinux-nat.c, macrocmd.c, macroexp.c, main.c: Update. * maint.c, mdebugread.c, mem-break.c, memattr.c: Update. * mips-linux-tdep.c, mips-tdep.c, mipsread.c, monitor.c: Update. * nlmread.c, nto-procfs.c, objc-lang.c, objfiles.c: Update. * observer.c, ocd.c, p-lang.c, p-typeprint.c: Update. * p-valprint.c, pa64solib.c, parse.c, ppc-linux-tdep.c: Update. * ppcnbsd-tdep.c, printcmd.c, procfs.c, remote-e7000.c: Update. * remote-fileio.c, remote-m32r-sdi.c, remote-rdi.c: Update. * remote-rdp.c, remote-sim.c, remote-st.c: Update. * remote-utils.c, remote-utils.h, remote.c: Update. * rom68k-rom.c, rs6000-nat.c, s390-tdep.c, scm-lang.c: Update. * ser-e7kpc.c, ser-tcp.c, ser-unix.c, sh-tdep.c: Update. * sh3-rom.c, shnbsd-tdep.c, sol-thread.c, solib-aix5.c: Update. * solib-frv.c, solib-irix.c, solib-osf.c, solib-pa64.c: Update. * solib-som.c, solib-sunos.c, solib-svr4.c, solib.c: Update. * somread.c, somsolib.c, source.c, stabsread.c: Update. * stack.c, std-regs.c, symfile-mem.c, symfile.c: Update. * symmisc.c, symtab.c, target.c, thread.c, top.c: Update. * tracepoint.c, trad-frame.c, typeprint.c, utils.c: Update. * uw-thread.c, valarith.c, valops.c, valprint.c: Update. * value.c, varobj.c, version.in, win32-nat.c, wince.c: Update. * xcoffread.c, xcoffsolib.c, cli/cli-cmds.c: Update. * cli/cli-decode.c, cli/cli-dump.c, cli/cli-logging.c: Update. * cli/cli-script.c, cli/cli-setshow.c, mi/mi-cmd-break.c: Update. * mi/mi-cmd-disas.c, mi/mi-cmd-env.c, mi/mi-cmd-file.c: Update. * mi/mi-cmd-stack.c, mi/mi-cmd-var.c, mi/mi-getopt.c: Update. * mi/mi-symbol-cmds.c, tui/tui-layout.c, tui/tui-stack.c: Update. * tui/tui-win.c: Update.
2005-02-11 04:06:14 +00:00
error (_("Erroneous arithmetic operation."));
1999-05-11 20:29:07 +00:00
}
/* Called by do_setshow_command. */
void
set_async_editing_command (char *args, int from_tty,
struct cmd_list_element *c)
1999-05-11 20:29:07 +00:00
{
change_line_handler ();
}
1999-06-14 18:08:47 +00:00
/* Set things up for readline to be invoked via the alternate
interface, i.e. via a callback function (rl_callback_read_char),
and hook up instream to the event loop. */
1999-06-14 18:08:47 +00:00
void
gdb_setup_readline (void)
1999-06-14 18:08:47 +00:00
{
/* This function is a noop for the sync case. The assumption is
that the sync setup is ALL done in gdb_init, and we would only
mess it up here. The sync stuff should really go away over
time. */
if (!batch_silent)
gdb_stdout = stdio_fileopen (stdout);
gdb_stderr = stderr_fileopen ();
gdb_stdlog = gdb_stderr; /* for moment */
gdb_stdtarg = gdb_stderr; /* for moment */
gdb_stdtargerr = gdb_stderr; /* for moment */
/* If the input stream is connected to a terminal, turn on
editing. */
if (ISATTY (instream))
1999-06-21 13:27:42 +00:00
{
/* Tell gdb that we will be using the readline library. This
could be overwritten by a command in .gdbinit like 'set
editing on' or 'off'. */
async_command_editing_p = 1;
/* When a character is detected on instream by select or poll,
readline will be invoked via this callback function. */
call_readline = rl_callback_read_char_wrapper;
1999-06-21 13:27:42 +00:00
}
else
{
async_command_editing_p = 0;
call_readline = gdb_readline2;
}
/* When readline has read an end-of-line character, it passes the
complete line to gdb for processing; command_line_handler is the
function that does this. */
input_handler = command_line_handler;
/* Tell readline to use the same input stream that gdb uses. */
rl_instream = instream;
/* Get a file descriptor for the input stream, so that we can
register it with the event loop. */
input_fd = fileno (instream);
/* Now we need to create the event sources for the input file
descriptor. */
/* At this point in time, this is the only event source that we
register with the even loop. Another source is going to be the
target program (inferior), but that must be registered only when
it actually exists (I.e. after we say 'run' or after we connect
to a remote target. */
add_file_handler (input_fd, stdin_event_handler, 0);
1999-06-14 18:08:47 +00:00
}
/* Disable command input through the standard CLI channels. Used in
the suspend proc for interpreters that use the standard gdb readline
interface, like the cli & the mi. */
void
gdb_disable_readline (void)
{
/* FIXME - It is too heavyweight to delete and remake these every
time you run an interpreter that needs readline. It is probably
better to have the interpreters cache these, which in turn means
that this needs to be moved into interpreter specific code. */
#if 0
ui_file_delete (gdb_stdout);
ui_file_delete (gdb_stderr);
gdb_stdlog = NULL;
gdb_stdtarg = NULL;
gdb_stdtargerr = NULL;
#endif
PR python/17372 - Python hangs when displaying help() This is more of a readline/terminal issue than a Python one. PR17372 is a regression in 7.8 caused by the fix for PR17072: commit 0017922d0292d8c374584f6100874580659c9973 Author: Pedro Alves <palves@redhat.com> Date: Mon Jul 14 19:55:32 2014 +0100 Background execution + pagination aborts readline/gdb gdb_readline_wrapper_line removes the handler after a line is processed. Usually, we'll end up re-displaying the prompt, and that reinstalls the handler. But if the output is coming out of handling a stop event, we don't re-display the prompt, and nothing restores the handler. So the next input wakes up the event loop and calls into readline, which aborts. ... gdb/ 2014-07-14 Pedro Alves <palves@redhat.com> PR gdb/17072 * top.c (gdb_readline_wrapper_line): Tweak comment. (gdb_readline_wrapper_cleanup): If readline is enabled, reinstall the input handler callback. The problem is that installing the input handler callback also preps the terminal, putting it in raw mode and with echo disabled, which is bad if we're going to call a command that assumes cooked/canonical mode, and echo enabled, like in the case of the PR, Python's interactive shell. Another example I came up with that doesn't depend on Python is starting a subshell with "(gdb) shell /bin/sh" from a multi-line command. Tests covering both these examples are added. The fix is to revert the original fix for PR gdb/17072, and instead restore the callback handler after processing an asynchronous target event. Furthermore, calling rl_callback_handler_install when we already have some input in readline's line buffer discards that input, which is obviously a bad thing to do while the user is typing. No specific test is added for that, because I first tried calling it even if the callback handler was still installed and that resulted in hundreds of failures in the testsuite. gdb/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * event-top.c (change_line_handler): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (callback_handler_installed): New global. (gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New functions. (display_gdb_prompt): Call gdb_rl_callback_handler_remove and gdb_rl_callback_handler_install instead of rl_callback_handler_remove and rl_callback_handler_install. (gdb_disable_readline): Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. * event-top.h (gdb_rl_callback_handler_remove) (gdb_rl_callback_handler_install) (gdb_rl_callback_handler_reinstall): New declarations. * infrun.c (reinstall_readline_callback_handler_cleanup): New cleanup function. (fetch_inferior_event): Install it. * top.c (gdb_readline_wrapper_line) Call gdb_rl_callback_handler_remove instead of rl_callback_handler_remove. (gdb_readline_wrapper_cleanup): Don't call rl_callback_handler_install. gdb/testsuite/ 2014-10-29 Pedro Alves <palves@redhat.com> PR python/17372 * gdb.python/python.exp: Test a multi-line command that spawns interactive Python. * gdb.base/multi-line-starts-subshell.exp: New file.
2014-10-23 16:13:35 +00:00
gdb_rl_callback_handler_remove ();
delete_file_handler (input_fd);
}