When executing commands on a secondary UI running the MI interpreter,
some commands that should be synchronous are not. MI incorrectly
continues processing input right after the synchronous command is
sent, before the target stops.
The problem happens when we emit MI async events (=library-loaded,
etc.), and we go about restoring the previous terminal state, we end
up calling target_terminal_ours, which incorrectly always installs the
current UI's input_fd in the event loop... That is, code like this:
old_chain = make_cleanup_restore_target_terminal ();
target_terminal_ours_for_output ();
fprintf_unfiltered (mi->event_channel, "library-loaded");
...
do_cleanups (old_chain);
The fix is to move the add_file_handler/delete_file_handler calls out
of target_terminal_$foo, making these completely no-ops unless called
with the main UI as current UI.
gdb/ChangeLog:
2016-08-09 Pedro Alves <palves@redhat.com>
PR gdb/20418
* event-top.c (ui_register_input_event_handler)
(ui_unregister_input_event_handler): New functions.
(async_enable_stdin): Register input in the event loop.
(async_disable_stdin): Unregister input from the event loop.
(gdb_setup_readline): Register input in the event loop.
* infrun.c (check_curr_ui_sync_execution_done): Register input in
the event loop.
* target.c (target_terminal_inferior): Don't unregister input from
the event loop.
(target_terminal_ours): Don't register input in the event loop.
* target.h (target_terminal_inferior)
(target_terminal_ours_for_output, target_terminal_ours): Update
comments.
* top.h (ui_register_input_event_handler)
(ui_unregister_input_event_handler): New declarations.
* utils.c (ui_unregister_input_event_handler_cleanup)
(prepare_to_handle_input): New functions.
(defaulted_query, prompt_for_continue): Use
prepare_to_handle_input.
gdb/testsuite/ChangeLog:
2016-08-09 Pedro Alves <palves@redhat.com>
Simon Marchi <simon.marchi@ericsson.com>
PR gdb/20418
* gdb.mi/new-ui-mi-sync.c, gdb.mi/new-ui-mi-sync.exp: New files.
* lib/mi-support.exp (mi_expect_interrupt): Remove anchors.
This commit makes each UI have its own "stdin" stream pointer. This
is used to determine whether the "from_tty" argument to
execute_command, etc. should be true.
Related, this commit makes input_from_terminal_p take an UI parameter,
and then avoids the gdb_has_a_terminal in it. gdb_has_a_terminal only
returns info on gdb's own main/primary terminal (the real stdin).
However, the places that call input_from_terminal_p really want to
know is whether the command came from an interactive tty. This patch
thus renames input_from_terminal_p to input_interactive_p for clarity,
and then makes input_interactive_p check for "set interactive" itself,
along with ISATTY, instead of calling gdb_has_a_terminal. Actually,
quit_force wants to call input_interactive_p _after_ stdin is closed,
we can't call ISATTY that late. So instead we save the result of
ISATTY in a field of the UI.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* cli/cli-script.c (read_next_line): Adjust to per-UI stdin.
(read_command_lines): Use input_interactive_p instead of
input_from_terminal_p.
* defs.h (struct ui): Forward declare.
(input_from_terminal_p): Rename to ...
(input_interactive_p): ... this.
* event-top.c (stdin_event_handler): Pass 0 as from_tty argument
to quit_command.
(command_handler): Adjust to per-UI stdin.
(handle_line_of_input): Adjust to per-UI stdin and use
input_interactive_p instead of ISATTY and input_from_terminal_p.
(gdb_readline_no_editing_callback): Adjust to per-UI stdin.
(command_line_handler): Always pass true as "from_tty" parameter
of handle_line_of_input and execute_command.
(async_sigterm_handler): Pass 0 as from_tty argument to
quit_command.
* inflow.c (interactive_mode, show_interactive_mode): Moved to ...
(gdb_has_a_terminal): Don't check interactive_mode here.
(_initialize_inflow): Don't install "set interactive-mode" here.
* main.c (captured_command_loop): Adjust to per-UI stdin.
* mi/mi-interp.c (mi_execute_command_wrapper): Adjust to per-UI
stdin.
* top.c (new_ui): Save the stdin stream and whether it's a tty.
(dont_repeat): Adjust to per-UI stdin.
(command_line_input): Adjust to per-UI stdin and to use
input_interactive_p.
(quit_force): Write history if any UI supports interactive input.
(interactive_mode, show_interactive_mode): Move here, from
inflow.c.
(input_from_terminal_p): Rename to ...
(input_interactive_p): ... this, and check the "interactive_mode"
global instead of calling gdb_has_a_terminal.
(_initialize_top): Install "set interactive-mode" here.
* top.h (struct ui) <stdin_stream, input_interactive_p>: New
fields.
* utils.c (quit): Pass 0 as from_tty argument to quit_force.
(defaulted_query): Adjust to per-UI stdin and to use
input_interactive_p.
Without this, GDB exits if a secondary UIs terminal/input stream is
closed:
$ ./gdb -ex "new-ui mi /dev/pts/6"
New UI allocated
<<< close /dev/pts/6
(gdb) Error detected on fd 9
$
We want that for the main UI, but not secondary UIs.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c (stdin_event_handler): Don't quit gdb if it was a
secondary UI's input stream that closed. Instead, just delete the
UI.
This is preparation for being able to create more than one UI object.
The change to gdb_main to stop using catch_errors is necessary because
catch_errors references current_uiout, which expands to
current_ui->m_current_ui, which would crash because current_ui is not
initialized yet at that point. It didn't trigger earlier in the
series because before this patch, main_ui/current_ui always start out
non-NULL.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c (main_ui_): Delete.
(main_ui, current_ui, ui_list): No longer initialize here.
* main.c (captured_main): UI initialization code factored out to
new new_ui function.
(gdb_main): Wrap captured_main with TRY/CATCH instead of
catch_errors.
* top.c (highest_ui_num): New global.
(new_ui): New function.
* top.h (struct ui) <num>: New field.
(new_ui): New declaration.
When sync_execution (a boolean) is true, it means we're running a
foreground command -- we hide the prompt stop listening to input, give
the inferior the terminal, then go to the event loop waiting for the
target to stop.
With multiple independent UIs, we need to track whether each UI is
synchronously blocked waiting for the target. IOW, if you do
"continue" in one console, that console stops accepting commands, but
you should still be free to type other commands in the others
consoles.
Just simply making sync_execution be per-UI alone not sufficient,
because of this in fetch_inferior_event:
/* If the inferior was in sync execution mode, and now isn't,
restore the prompt (a synchronous execution command has finished,
and we're ready for input). */
if (current_ui->async && was_sync && !sync_execution)
observer_notify_sync_execution_done ();
We'd have to record at entry the "was_sync" state for each UI, not
just of the current UI.
This patch instead replaces the sync_execution flag by a per-UI
tristate flag indicating the command line prompt state:
enum prompt_state
{
/* The command line is blocked simulating synchronous execution.
This is used to implement the foreground execution commands
('run', 'continue', etc.). We won't display the prompt and
accept further commands until the execution is actually over. */
PROMPT_BLOCKED,
/* The command finished; display the prompt before returning back to
the top level. */
PROMPT_NEEDED,
/* We've displayed the prompt already, ready for input. */
PROMPTED,
;
I think the end result is _much_ clearer than the current code, and,
it addresses the original motivation too.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* annotate.c: Include top.h.
(async_background_execution_p): Delete.
(print_value_flags): Check the UI's prompt state rather then
async_background_execution_p.
* event-loop.c (start_event_loop): Set the prompt state to
PROMPT_NEEDED.
* event-top.c (display_gdb_prompt, async_enable_stdin)
(async_disable_stdin): Check the current UI's prompt state instead
of the sync_execution global.
(command_line_handler): Set the prompt state to PROMPT_NEEDED
before running a command, and display the prompt if still needed
afterwards.
* infcall.c (struct call_thread_fsm) <waiting_ui>: New field.
(new_call_thread_fsm): New parameter 'waiting_ui'. Store it.
(call_thread_fsm_should_stop): Set the prompt state to
PROMPT_NEEDED.
(run_inferior_call): Adjust to temporarily set the prompt state to
PROMPT_BLOCKED instead of using the sync_execution global.
(call_function_by_hand_dummy): Pass the current UI to
new_call_thread_fsm.
* infcmd.c: Include top.h.
(continue_1): Check the current UI's prompt state instead of the
sync_execution global.
(continue_command): Validate global execution state before calling
prepare_execution_command.
(step_1): Call all_uis_check_sync_execution_done.
(attach_post_wait): Don't call async_enable_stdin here. Remove
reference to sync_execution.
* infrun.c (sync_execution): Delete global.
(follow_fork_inferior)
(reinstall_readline_callback_handler_cleanup): Check the current
UI's prompt state instead of the sync_execution global.
(check_curr_ui_sync_execution_done)
(all_uis_check_sync_execution_done): New functions.
(fetch_inferior_event): Call all_uis_check_sync_execution_done
instead of trying to determine whether the global sync execution
changed.
(handle_no_resumed): Check the prompt state of all UIs.
(normal_stop): Emit the no unwait-for even to all PROMPT_BLOCKED
UIs. Emit the "Switching to" notification to all UIs. Enable
stdin in all UIs.
* infrun.h (sync_execution): Delete.
(all_uis_check_sync_execution_done): Declare.
* main.c (captured_command_loop): Don't call
interp_pre_command_loop if the prompt is blocked.
(catch_command_errors, catch_command_errors_const): Adjust.
(captured_main): Set the initial prompt state to PROMPT_NEEDED.
* mi/mi-interp.c (display_mi_prompt): Set the prompt state to
PROMPTED.
(mi_interpreter_resume): Don't clear sync_execution. Remove hack
comment.
(mi_execute_command_input_handler): Set the prompt state to
PROMPT_NEEDED before executing the command, and only display the
prompt if the prompt state is PROMPT_NEEDED afterwards.
(mi_on_resume_1): Adjust to check the prompt state.
* target.c (target_terminal_inferior): Adjust to check the prompt
state.
* top.c (wait_sync_command_done, maybe_wait_sync_command_done)
(execute_command): Check the current UI's prompt state instead of
sync_execution.
* top.h (enum prompt_state): New.
(struct ui) <prompt_state>: New field.
(ALL_UIS): New macro.
All interpreter types (CLI/TUI/MI) print the prompt, and then call
start_event_loop.
Because we'll need an interpreter hook to display the
interpreter-specific prompt before going back to the event loop,
without actually starting an event loop, this patch moves the
start_event_loop call to common code, and replaces the command_loop
hook with a pre_command_look hook, that now just prints the prompt.
Turns out to be a cleanup on its own right anyway.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* cli/cli-interp.c (cli_interpreter_pre_command_loop): New
function.
(cli_interp_procs): Install it instead of cli_command_loop.
* cli/cli-interp.h (cli_interpreter_pre_command_loop): Declare.
* event-top.c (cli_command_loop): Delete.
* interps.c (interp_new): Remove reference to command_loop_proc.
(current_interp_command_loop): Delete.
(interp_pre_command_loop): New function.
(interp_command_loop_ftype): Delete.
* interps.h (interp_pre_command_loop_ftype): New typedef.
(struct interp_procs) <command_loop_proc>: Delele field.
<pre_command_loop_proc>: New field.
(current_interp_command_loop): Delete declaration.
(interp_pre_command_loop): New declaration.
* main.c (captured_command_loop): Call interp_pre_command_loop
instead of current_interp_command_loop and start an event loop.
* mi/mi-interp.c (mi_command_loop): Delete.
(mi_interpreter_pre_command_loop): New.
(mi_interp_procs): Update.
* tui/tui-interp.c (tui_interp_procs): Install
cli_interpreter_pre_command_loop instead of cli_command_loop.
This makes target events always be always processed with the main UI
as current UI. This way, warnings, debug output, etc. are always
consistently sent to the main console.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c (restore_ui_cleanup): Make extern.
* infrun.c (fetch_inferior_event): Always switch to the main UI.
* top.h (restore_ui_cleanup): Declare.
Due to the way that readline's API works (based on globals), we can
only have one instance of readline in a process. So the goal of this
patch is to only allow editing in the main UI, and make sure that only
one UI calls into readline. Some MI paths touch readline variables
currently, which is bad as that is changing variables that matter for
the main console UI. This patch fixes those.
This actually fixes a nasty bug -- starting gdb in MI mode ("gdb
-i=mi"), and then doing "set editing on" crashes GDB, because MI is
not prepared to use readline:
set editing on
&"set editing on\n"
=cmd-param-changed,param="editing",value="on"
^done
(gdb)
p 1
readline: readline_callback_read_char() called with no handler!
Aborted (core dumped)
The fix for that was to add an interp_proc method to query the
interpreter whether it actually supports editing. New test included.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
PR mi/20034
* cli/cli-interp.c: Include cli-interp.h and event-top.h.
(cli_interpreter_resume): Pass 1 to gdb_setup_readline. Set the
UI's input_handler here.
(cli_interpreter_supports_command_editing): New function.
(cli_interp_procs): Install it.
* cli/cli-interp.h: New file.
* event-top.c (async_command_editing_p): Rename to ...
(set_editing_cmd_var): ... this.
(change_line_handler): Add parameter 'editing', and use it. Bail
early if the interpreter doesn't support editing. Don't touch
readline state if editing is off.
(gdb_rl_callback_handler_remove, gdb_rl_callback_handler_install)
(gdb_rl_callback_handler_reinstall): Assert the current UI is the
main UI.
(display_gdb_prompt): Don't call gdb_rl_callback_handler_remove if
not using readline. Check whether the current UI is using command
editing instead of checking the async_command_editing_p global.
(set_async_editing_command): Delete.
(gdb_setup_readline): Add 'editing' parameter. Only allow editing
on the main UI. Don't touch readline state if editing is off.
(gdb_disable_readline): Don't touch readline state if editing is
off.
* event-top.h (gdb_setup_readline): Add 'int' parameter.
(set_async_editing_command): Delete declaration.
(change_line_handler, command_line_handler): Declare.
(async_command_editing_p): Rename to ...
(set_editing_cmd_var): ... this.
* infrun.c (reinstall_readline_callback_handler_cleanup): Check
whether the current UI has editing enabled rather than checking
the async_command_editing_p global.
* interps.c (interp_supports_command_editing): New function.
* interps.h (interp_supports_command_editing_ftype): New typedef.
(struct interp_procs) <supports_command_editing_proc>: New field.
(interp_supports_command_editing): Declare.
* mi/mi-interp.c (mi_interpreter_resume): Pass 0 to
gdb_setup_readline. Don't clear the async_command_editing_p
global. Update comments.
* top.c (gdb_readline_wrapper_line, gdb_readline_wrapper): Check
whether the current UI has editing enabled rather than checking
the async_command_editing_p global. Don't touch readline state if
editing is off.
(undo_terminal_modifications_before_exit): Switch to the main UI.
Unconditionally call gdb_disable_readline.
(set_editing): New function.
(show_async_command_editing_p): Rename to ...
(show_editing): ... this. Show the state of the current UI.
(_initialize_top): Adjust.
* top.h (struct ui) <command_editing>: New field.
* tui/tui-interp.c: Include cli/cli-interp.h.
(tui_resume): Pass 1 to gdb_setup_readline. Set the UI's
input_handler.
(tui_interp_procs): Install
cli_interpreter_supports_command_editing.
* tui/tui-io.c (tui_getc): Check whether the current UI has
editing enabled rather than checking the async_command_editing_p
global.
gdb/testsuite/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
PR mi/20034
* gdb.mi/mi-editing.exp: New file.
stderr_fileopen () references stderr directly, which doesn't work when
we have a separate UI with its own stderr-like stream. So this also
adds a "errstream" to "struct ui", and plumbs stderr_fileopen to take
a stream parameter.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c (gdb_setup_readline): Pass the UI's outstream and
errstream to stdout_fileopen and stderr_fileopen.
* exceptions.c: Include top.h.
(print_flush): Open the current UI's outstream file descriptor,
instead of hardcoding file descriptor 1.
* main.c (captured_main): Save the main UI's out and error
streams. Adjust stderr_fileopen call.
* top.h (struct ui) <outstream, errstream>: New fields.
* ui-file.c (stderr_fileopen): Add stream parameter. Use it
instead of stderr.
* ui-file.h (stderr_fileopen): Add stream parameter and update
comment.
And with that, we can switch the current UI to the UI whose input
descriptor woke up the event loop. IOW, if the user types in UI 2,
the event loop wakes up, switches to UI 2, and processes the input.
Next the user types in UI 3, the event loop wakes up and switches to
UI 3, etc.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c (input_fd): Delete.
(stdin_event_handler): Switch to the UI whose input descriptor got
the event. Adjust to per-UI input_fd.
(gdb_setup_readline): Don't set the input_fd global. Adjust to
per-UI input_fd.
(gdb_disable_readline): Adjust to per-UI input_fd.
* event-top.h (input_fd): Delete declaration.
* linux-nat.c (linux_nat_terminal_inferior): Don't remove input_fd
from the event-loop here.
(linux_nat_terminal_ours): Don't register input_fd in the
event-loop here.
* main.c (captured_main): Adjust to per-UI input_fd.
* remote.c (remote_terminal_inferior): Don't remove input_fd from
the event-loop here.
(remote_terminal_ours): Don't register input_fd in the event-loop
here.
* target.c: Include top.h and event-top.h.
(target_terminal_inferior): Remove input_fd from the event-loop
here.
(target_terminal_ours): Register input_fd in the event-loop.
* top.h (struct ui) <input_fd>: New field.
Async signal handlers have no connection to whichever was the current
UI, and thus always run on the main one.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-loop.c: Include top.h.
(invoke_async_signal_handlers): Switch to the main UI.
* event-top.c (main_ui_): Update comment.
(main_ui): New global.
* top.h (main_ui): Declare.
When we have multiple consoles, MI channels, etc., then we need to
broadcast breakpoint hits, etc. to all UIs. In the past, I've
adjusted most of the run control to communicate events to the
interpreters through observer notifications, so events would be
properly sent to console and MI streams, in sync and async modes.
This patch does the next logical step -- have each interpreter's
observers output interpreter-specific info to _all_ UIs.
Note that when we have multiple instances of active cli/tui
interpreters, then the cli_interp and tui_interp globals no longer
work. This is addressed by this patch.
Also, the interpreters currently register some observers when resumed
and remove them when suspended. If we have multiple instances of the
interpreters, and they can be suspended/resumed at different,
independent times, that no longer works. What we instead do is always
install the observers, and then have the observers themselves know
when to do nothing.
An earlier prototype of this series did the looping over struct UIs in
common code, and then dispatched events to the interpreters through a
matching interp_on_foo method for each observer. That turned out a
lot more complicated than the present solution, as we'd end up with
having to create a new interp method every time some interpreter
wanted to listen to some observer notification, resulting in a lot of
duplicated make-work and more coupling than desirable.
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* cli/cli-interp.c (cli_interp): Delete.
(as_cli_interp): New function.
(cli_on_normal_stop, cli_on_signal_received)
(cli_on_end_stepping_range, cli_on_signal_exited, cli_on_exited)
(cli_on_no_history): Send output to all CLI UIs.
(cli_on_sync_execution_done, cli_on_command_error): Skip output if
the top level interpreter is not a CLI.
(cli_interpreter_init): Don't set cli_interp or install observers
here.
(_initialize_cli_interp): Install observers here.
* event-top.c (main_ui_, ui_list): New globals.
(current_ui): Point to main_ui_.
(restore_ui_cleanup, switch_thru_all_uis_init)
(switch_thru_all_uis_cond, switch_thru_all_uis_next): New
functions.
* mi/mi-interp.c (as_mi_interp): New function.
(mi_interpreter_init): Don't install observers here.
(mi_on_sync_execution_done): Skip output if the top level
interpreter is not a MI.
(mi_new_thread, mi_thread_exit, mi_record_changed)
(mi_inferior_added, mi_inferior_appeared, mi_inferior_exit)
(mi_inferior_removed): Send output to all MI UIs.
(find_mi_interpreter, mi_interp_data): Delete.
(find_mi_interp): New function.
(mi_on_signal_received, mi_on_end_stepping_range)
(mi_on_signal_exited, mi_on_exited, mi_on_no_history): Send output
to all MI UIs.
(mi_on_normal_stop): Rename to ...
(mi_on_normal_stop_1): ... this.
(mi_on_normal_stop): Reimplement, sending output to all MI UIs.
(mi_traceframe_changed, mi_tsv_created, mi_tsv_deleted)
(mi_tsv_modified, mi_breakpoint_created, mi_breakpoint_deleted)
(mi_breakpoint_modified, mi_output_running_pid): Send output to
all MI UIs.
(mi_on_resume): Rename to ...
(mi_on_resume_1): ... this. Don't handle infcalls here.
(mi_on_resume): Reimplement, sending output to all MI UIs.
(mi_solib_loaded, mi_solib_unloaded, mi_command_param_changed)
(mi_memory_changed): Send output to all MI UIs.
(report_initial_inferior): Install observers here.
* top.h (struct ui) <next>: New field.
(ui_list): Declare.
(struct switch_thru_all_uis): New.
(switch_thru_all_uis_init, switch_thru_all_uis_cond)
(switch_thru_all_uis_next): Declare.
(SWITCH_THRU_ALL_UIS): New macro.
* tui/tui-interp.c (tui_interp): Delete global.
(as_tui_interp): New function.
(tui_on_normal_stop, tui_on_signal_received)
(tui_on_end_stepping_range, tui_on_signal_exited, tui_on_exited)
(tui_on_no_history): Send output to all TUI UIs.
(tui_on_sync_execution_done, tui_on_command_error): Skip output if
the top level interpreter is not a TUI.
(tui_init): Don't set tui_interp or install observers here.
(_initialize_tui_interp): Install observers here.
This is a step towards supporting multiple consoles/MIs, each on its
own stdio streams / terminal.
See intro comment in top.h.
(I've had trouble picking a name for this object. I've started out
with "struct console" originally. But then this is about MI as well,
and there's "interpreter-exec console", which is specifically about
the CLI...
So I changed to "struct terminal", but, then we have a terminal object
that works when the input is not a terminal as well ...
Then I sort of gave up and renamed it to "struct top_level". But it
then gets horribly confusing when we talk about the "top level
interpreter that's running on the current top level".
In the end, I realized we're already sort of calling this "ui", in
struct ui_out, struct ui_file, and a few coments here and there.)
gdb/ChangeLog:
2016-06-21 Pedro Alves <palves@redhat.com>
* event-top.c: Update readline-related comments.
(input_handler, call_readline): Delete globals.
(gdb_rl_callback_handler): Call the current UI's input_handler
method.
(change_line_handler): Adjust to set current UI's properties
instead of globals.
(current_ui_, current_ui): New globals.
(get_command_line_buffer): Rewrite to refer to the current UI.
(stdin_event_handler): Adjust to call the call_readline method of
the current UI.
(gdb_readline_no_editing_callback): Adjust to call the current UI's
input_handler method.
(gdb_setup_readline): Adjust to set current UI's properties
instead of globals.
* event-top.h (call_readline, input_handler): Delete declarations.
* mi/mi-interp.c (mi_interpreter_resume): Adjust to set current
UI's properties instead of globals.
* top.c (gdb_readline_wrapper_cleanup): Adjust to set current UI's
properties instead of globals.
(gdb_readline_wrapper): Adjust to call and set current UI's
methods instead of globals.
* top.h: Include buffer.h and event-loop.h.
(struct ui): New struct.
(current_ui): New declaration.
If we map GDB'S TRY/CATCH macros to C++ try/catch, GDB breaks on
systems where readline isn't built with exceptions support. The
problem is that readline calls into GDB through the callback
interface, and if GDB's callback throws a C++ exception/error, the
system unwinder won't manage to unwind past the readline frame, and
ends up calling std::terminate(), which aborts the process:
(gdb) whatever-command-that-causes-an-error
terminate called after throwing an instance of 'gdb_exception_RETURN_MASK_ERROR'
Aborted
$
This went unnoticed for so long because:
- the x86-64 ABI requires -fasynchronous-unwind-tables, making it
possible for exceptions to cross readline with no special handling.
But e.g., on ARM or AIX, unless you build readline with
-fexceptions, you trip on the problem.
- TRY/CATCH was mapped to setjmp/longjmp, even in C++ mode, until
quite recently.
The fix is to catch and save any GDB exception that is thrown inside
the GDB readline callback, and then once the callback returns back to
the GDB code that called into readline in the first place, rethrow the
saved GDB exception.
This is similar in spirit to how we catch/map GDB exceptions at the
GDB/Python and GDB/Guile API boundaries.
The next question is then: if we intercept all exceptions within GDB's
readline callback, should we simply return normally to readline? The
callback prototype has no way to signal an error back to readline (*).
The answer is no -- if we return normally, we'll be returning to a
loop inside rl_callback_read_char that continues processing pending
input, calling into GDB again, redisplaying the prompt, etc. Thus if
we want to error out of rl_callback_read_char, we need to long jump
across it, just like we always did before TRY/CATCH were ever mapped
to C++ exceptions.
My first approach built a specialized API to handle this, with a
couple macros to hide the setjmp/longjmp and the struct gdb_exception
saving/rethrowing.
However, I realized that we need to:
- Handle multiple active rl_callback_read_char invocations. If,
while processing input something triggers a secondary prompt, we
end up in a nested rl_callback_read_char call, through
gdb_readline_wrapper.
- Propagate a struct gdb_exception along with the longjmp.
... and that this is exactly what the setjmp/longjmp-based TRY/CATCH
does.
So the fix makes the setjmp/longjmp TRY/CATCH always available under
new TRY_SJLJ/CATCH_SJLJ aliases, even when TRY/CATCH is mapped to C++
try/catch, and then uses TRY_SJLJ/CATCH_SJLJ to propagate GDB
exceptions across the readline callback.
This turns out to be a much better looking fix than my bespoke API
attempt, even. We'll probably be able to simplify TRY_SJLJ/CATCH_SJLJ
when we finally get rid of TRY/CATCH all over the tree, but until
then, this reuse seems quite nice for avoiding a second parallel
setjmp/longjmp mechanism.
(*) - maybe we could propose a readline API change, but we still need
to handle current readline, anyway.
gdb/ChangeLog:
2016-04-22 Pedro Alves <palves@redhat.com>
* common/common-exceptions.c (enum catcher_state, struct catcher)
(current_catcher): Define in C++ mode too.
(exceptions_state_mc_catch): Call throw_exception_sjlj instead of
throw_exception.
(throw_exception_sjlj, throw_exception_cxx): New functions,
factored out from throw_exception.
(throw_exception): Reimplement.
* common/common-exceptions.h (exceptions_state_mc_init)
(exceptions_state_mc_action_iter)
(exceptions_state_mc_action_iter_1, exceptions_state_mc_catch):
Declare in C++ mode too.
(TRY): Rename to ...
(TRY_SJLJ): ... this.
(CATCH): Rename to ...
(CATCH_SJLJ): ... this.
(END_CATCH): Rename to ...
(END_CATCH_SJLJ): ... this.
[GDB_XCPT == GDB_XCPT_SJMP] (TRY, CATCH, END_CATCH): Map to SJLJ
equivalents.
(throw_exception): Update comments.
(throw_exception_sjlj): Declare.
* event-top.c (gdb_rl_callback_read_char_wrapper): Extend intro
comment. Wrap body in TRY_SJLJ/CATCH_SJLJ and rethrow any
intercepted exception.
(gdb_rl_callback_handler): New function.
(gdb_rl_callback_handler_install): Always install
gdb_rl_callback_handler as readline callback.
Use the "gdb_rl_" prefix like other gdb readline function wrappers to
make it clear this is a gdb function, not a readline function.
gdb/ChangeLog:
2016-04-22 Pedro Alves <palves@redhat.com>
* event-top.c (rl_callback_read_char_wrapper): Rename to ...
(gdb_rl_callback_read_char_wrapper): ... this.
(change_line_handler, gdb_setup_readline): Adjust.
This finally gets rid of immediate_quit (and surrounding
infrustruture), as nothing sets it anymore.
gdb_call_async_signal_handler was only necessary in order to handle
immediate_quit. We can just call mark_async_signal_handler directly
on all hosts now.
In turn, we can clean up mingw-hdep.c's gdb_select a bit, as
sigint_event / sigint_handler is no longer needed.
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* defs.h: Update comments on SIGINT handling.
(immediate_quit): Delete declaration.
* event-loop.c (call_async_signal_handler): Delete.
* event-loop.h (call_async_signal_handler): Delete declaration.
(mark_async_signal_handler): Update comments.
(gdb_call_async_signal_handler): Delete declaration.
* event-top.c (handle_sigint): Call mark_async_signal_handler
instead of gdb_call_async_signal_handler.
* exceptions.c (prepare_to_throw_exception): Remove reference to
immediate_quit.
(exception_fprintf): Remove comments about immediate_quit.
* mingw-hdep.c (sigint_event, sigint_handler): Delete.
(gdb_select): Don't wait on sigint_event.
(gdb_call_async_signal_handler): Delete.
(_initialize_mingw_hdep): Delete.
* posix-hdep.c (gdb_call_async_signal_handler): Delete.
* utils.c (immediate_quit): Delete.
remote.c is the last user of immediate_quit. It's relied on to
immediately break the initial remote connection sync up, if the user
does Ctrl-C, assuming that was because the target isn't responding.
At that stage, since the connection isn't synced yet, disconnecting is
the only safe thing to do. This commit reworks that, to not rely on
throwing from the SIGINT signal handler.
So, this commit:
- Introduces the concept of a "quit handler". This is used to
override what does the QUIT macro do when the quit flag is set.
- Makes the "struct serial" reachar / write code call QUIT in the
partial read/write loops, so the current quit handler is invoked
whenever a serial->read_prim / serial->write_prim returns EINTR.
- Makes the "struct serial" reachar / write code call
interruptible_select instead of gdb_select, so that QUITs are
detected in a race-free manner.
- Stops remote.c from setting immediate_quit during the initial
connection.
- Instead, we install a custom quit handler whenever we're calling
into the serial code. This custom quit handler knows to immediately
throw a quit when we're in the initial connection setup, and
otherwise defer handling the quit/Ctrl-C request to later, when
we're safely out of a packet command/response sequence. This also
is what is now responsible for handling "double Ctrl-C because
target connection is stuck/wedged."
- remote.c no longer installs a specialized SIGINT handlers, and
instead re-uses the quit flag. Since we want to rely on the QUIT
macro, the SIGINT handler must also set the quit. And the easiest
is just to not install custom SIGINT handler in remote.c. Let the
standard SIGINT handler do its job of setting the quit flag.
Centralizing SIGINT handlers seems like a good thing to me, anyway.
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* defs.h (quit_handler_ftype, quit_handler)
(make_cleanup_override_quit_handler, default_quit_handler): New.
(QUIT): Adjust comments.
* event-top.c (default_quit_handler): New function.
(quit_handler): New global.
(struct quit_handler_cleanup_data): New.
(restore_quit_handler, restore_quit_handler_dtor)
(make_cleanup_override_quit_handler): New.
(async_request_quit): Call QUIT.
* remote.c (struct remote_state) <got_ctrlc_during_io>: New field.
(async_sigint_remote_twice_token, async_sigint_remote_token):
Delete.
(remote_close): Update comments.
(remote_start_remote): Don't set immediate_quit. Set starting_up
earlier.
(remote_serial_quit_handler, remote_unpush_and_throw): New
functions.
(remote_open_1): Clear got_ctrlc_during_io. Set
remote_async_terminal_ours_p unconditionally.
(async_initialize_sigint_signal_handler)
(async_handle_remote_sigint, async_handle_remote_sigint_twice)
(remote_check_pending_interrupt, async_remote_interrupt)
(async_remote_interrupt_twice)
(async_cleanup_sigint_signal_handler, ofunc)
(sync_remote_interrupt, sync_remote_interrupt_twice): Delete.
(remote_terminal_inferior, remote_terminal_ours): Remove async
checks.
(remote_wait_as): Don't install a SIGINT handler in sync mode.
(readchar, remote_serial_write): Override the quit handler with
remote_serial_quit_handler.
(getpkt_or_notif_sane_1): Don't call QUIT.
(initialize_remote_ops): Don't install
remote_check_pending_interrupt.
(_initialize_remote): Don't create async_sigint_remote_token and
async_sigint_remote_twice_token.
* ser-base.c (ser_base_wait_for): Call QUIT and use
interruptible_select.
(ser_base_write): Call QUIT.
* ser-go32.c (dos_readchar, dos_write): Call QUIT.
* ser-unix.c (wait_for): Don't use VTIME. Always take the
gdb_select path, but call QUIT and interruptible_select.
* utils.c (maybe_quit): Call the current quit handler. Don't call
target_check_pending_interrupt.
(defaulted_query, prompt_for_continue): Override the quit handler
with the default quit handler.
This just looks totally wrong to me, for completetly discarding a
user-requested Ctrl-C. I can't think of why we'd want do this here.
Actually, I digged the history, and found out that this has been here
since at least 7b4ac7e1ed (gdb-2.4, the initial revision, 1988), at
a time were we had a top level setjmp/longjmp, long before that got
wrapped in throw_exception and friends, and this code was in an
explicit loop, with the quit_flag cleared on every iteration, before
executing a command...
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* event-top.c (command_handler): Don't call clear_quit_flag.
We have places where we call a blocking gdb_select expecting that a
Ctrl-C will unblock it. However, if the Ctrl-C is pressed just before
gdb_select, the SIGINT handler runs before gdb_select, and thus
gdb_select won't return.
For example gdb_readline_no_editing:
QUIT;
/* Wait until at least one byte of data is available. Control-C
can interrupt gdb_select, but not fgetc. */
FD_ZERO (&readfds);
FD_SET (fd, &readfds);
if (gdb_select (fd + 1, &readfds, NULL, NULL, NULL) == -1)
and stdio_file_read:
/* For the benefit of Windows, call gdb_select before reading from
the file. Wait until at least one byte of data is available.
Control-C can interrupt gdb_select, but not read. */
{
fd_set readfds;
FD_ZERO (&readfds);
FD_SET (stdio->fd, &readfds);
if (gdb_select (stdio->fd + 1, &readfds, NULL, NULL, NULL) == -1)
return -1;
}
return read (stdio->fd, buf, length_buf);
This is a race classically fixed with either the self-pipe trick, or
by blocking SIGINT and then using pselect instead of select.
Blocking SIGINT most of the time would mean that check_quit_flag (and
thus QUIT) would need to do a syscall every time it is called, which
sounds best avoided, since QUIT is called in many loops. Thus we take
the self-pipe trick route (wrapped in a serial event).
Instead of having all places that need this manually add an extra file
descriptor to the set of gdb_select's watched file descriptors, we
introduce a wrapper, interruptible_select, that does that.
The Windows version of gdb_select actually does not suffer from this,
because mingw-hdep.c:gdb_call_async_signal_handler sets a Windows
event that gdb_select always waits on. So this patch can be seen as
generalization of that technique. We can't remove that extra event
from mingw-hdep.c until we get rid of immediate_quit though.
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* defs.h: Extend QUIT-related comments to mention
interruptible_select.
(quit_serial_event_set, quit_serial_event_clear): Declare.
* event-top.c: Include "ser-event.h" and "gdb_select.h".
(quit_serial_event): New global.
(async_init_signals): Make quit_serial_event.
(quit_serial_event_set, quit_serial_event_clear)
(quit_serial_event_fd, interruptible_select): New functions.
* extension.c (set_quit_flag): Set the quit serial event.
(check_quit_flag): Clear the quit serial event.
* gdb_select.h (interruptible_select): New declaration.
* guile/scm-ports.c (ioscm_input_waiting): Use
interruptible_select instead of gdb_select.
* top.c (gdb_readline_no_editing): Likewise.
* ui-file.c (stdio_file_read): Likewise.
GDB's core signal handling suffers from a classical signal handler /
mainline code race:
int
gdb_do_one_event (void)
{
...
/* First let's see if there are any asynchronous signal handlers
that are ready. These would be the result of invoking any of the
signal handlers. */
if (invoke_async_signal_handlers ())
return 1;
...
/* Block waiting for a new event. (...). */
if (gdb_wait_for_event (1) < 0)
return -1;
...
}
If a signal is delivered while gdb is blocked in the poll/select
inside gdb_wait_for_event, then the select/poll breaks with EINTR,
we'll loop back around and call invoke_async_signal_handlers.
However, if the signal handler runs between
invoke_async_signal_handlers and gdb_wait_for_event,
gdb_wait_for_event will block, until the next unrelated event...
The fix is to a struct serial_event, and register it in the set of
files that select/poll in gdb_wait_for_event waits on. The signal
handlers that defer work to invoke_async_signal_handlers call
mark_async_signal_handler, which is adjusted to also set the new
serial event in addition to setting a flag, and is thus now is
garanteed to immediately unblock the next gdb_select/poll call, up
until invoke_async_signal_handlers is called and the event is cleared.
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* event-loop.c: Include "ser-event.h".
(async_signal_handlers_serial_event): New global.
(async_signals_handler, initialize_async_signal_handlers): New
functions.
(mark_async_signal_handler): Set
async_signal_handlers_serial_event.
(invoke_async_signal_handlers): Clear
async_signal_handlers_serial_event.
* event-top.c (async_init_signals): Call
initialize_async_signal_handlers.
immediate_quit used to be necessary back when prompt_for_continue used
blocking fread, but nowadays it uses gdb_readline_wrapper, which is
implemented in terms of a nested event loop, which already knows how
to react to SIGINT:
#0 throw_it (reason=RETURN_QUIT, error=GDB_NO_ERROR, fmt=0x9d6d7e "Quit", ap=0x7fffffffcb88)
at .../src/gdb/common/common-exceptions.c:324
#1 0x00000000007bab5d in throw_vquit (fmt=0x9d6d7e "Quit", ap=0x7fffffffcb88) at .../src/gdb/common/common-exceptions.c:366
#2 0x00000000007bac9f in throw_quit (fmt=0x9d6d7e "Quit") at .../src/gdb/common/common-exceptions.c:385
#3 0x0000000000773a2d in quit () at .../src/gdb/utils.c:1039
#4 0x000000000065d81b in async_request_quit (arg=0x0) at .../src/gdb/event-top.c:893
#5 0x000000000065c27b in invoke_async_signal_handlers () at .../src/gdb/event-loop.c:949
#6 0x000000000065aeef in gdb_do_one_event () at .../src/gdb/event-loop.c:280
#7 0x0000000000770838 in gdb_readline_wrapper (prompt=0x7fffffffcd40 "---Type <return> to continue, or q <return> to quit---")
at .../src/gdb/top.c:873
The need for the QUIT in stdin_event_handler is then exposed by the
gdb.base/double-prompt-target-event-error.exp test, which has:
# We're now stopped in a pagination query while handling a
# target event (printing where the program stopped). Quitting
# the pagination should result in only one prompt being
# output.
send_gdb "\003p 1\n"
Without that change we'd get:
Continuing.
---Type <return> to continue, or q <return> to quit---PASS: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: continue to pagination
^CpQuit
(gdb) 1
Undefined command: "1". Try "help".
(gdb) PASS: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: first prompt
ERROR: Undefined command "".
UNRESOLVED: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: no double prompt
Vs:
Continuing.
---Type <return> to continue, or q <return> to quit---PASS: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: continue to pagination
^CQuit
(gdb) p 1
$1 = 1
(gdb) PASS: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: first prompt
PASS: gdb.base/double-prompt-target-event-error.exp: ctrlc target event: continue: no double prompt
gdb/ChangeLog:
2016-04-12 Pedro Alves <palves@redhat.com>
* event-top.c (stdin_event_handler): Call QUIT;
(prompt_for_continue): Don't run with immediate_quit set.
I didn't manage to usefully split this further into smaller
independent pieces, so:
- Use "struct buffer" more.
- Split out the responsibility of composing a complete command line
from multiple input lines split with backslash
(
E.g.:
(gdb) print \
1 + \
2
$1 = 3
(gdb)
)
to a separate function. Note we don't need the separate
readline_input_state and more_to_come globals at all. They were
just obfuscating the logic.
- Factor out the tricky mostly duplicated code in
command_line_handler and command_line_input.
gdb/ChangeLog
2016-03-09 Pedro Alves <palves@redhat.com>
* event-top.c (more_to_come): Delete.
(struct readline_input_state): Delete.
(readline_input_state): Delete.
(get_command_line_buffer): New function.
(command_handler): Update comments. Don't handle NULL commands
here. Do not execute commented lines.
(command_line_append_input_line): New function.
(handle_line_of_input): New function, partly based on
command_line_handler and command_line_input.
(command_line_handler): Rewrite.
* event-top.h (command_handler): New declaration.
(command_loop): Defer command execution to command_handler.
(command_line_input): Update comments. Simplify, using struct
buffer and handle_line_of_input.
* top.h (struct buffer): New forward declaration.
(handle_line_of_input): New declaration.
There doesn't seem to be much point in trying to reuse this buffer.
Prefer simplicity instead.
(In case you're wondering whether this fixes an off-by-one: linelength
is misnamed; it's really a size including terminating null char.)
gdb/ChangeLog:
2016-03-09 Pedro Alves <palves@redhat.com>
* event-top.c (command_line_handler): Use xfree + xstrdup instead
of xrealloc + strcpy.
* main.c (captured_main): Use xstrdup instead of xmalloc plus
manual clear.
* top.c (saved_command_line): Rewrite comment.
(saved_command_line_size): Delete.
(command_line_input): Use xfree + xstrdup instead of xrealloc +
strcpy.
* top.h (saved_command_line_size): Delete declaration.
gdb/ChangeLog:
2016-03-09 Pedro Alves <palves@redhat.com>
* event-top.c: Include buffer.h.
(gdb_readline_no_editing_callback): Use struct buffer instead
of xrealloc.
The comments and existence of this global are a bit of misleading
obfuscation, since this is only ever used to print the prompt
annotation, and never changes. Just hardcode "prompt" where
necessary, as done for most other annotations.
gdb/ChangeLog:
2016-03-09 Pedro Alves <palves@redhat.com>
* event-top.c (async_annotation_suffix): Delete.
(top_level_prompt, command_line_handler): Don't use
'async_annotation_suffix' and simplify.
* event-top.h (async_annotation_suffix): Delete declaration.
(init_main): Remove reference to 'async_annotation_suffix'.
2015-12-18 Sandra Loosemore <sandra@codesourcery.com>
gdb/
* event-top.c (command_handler): Don't require stdin to be a tty
for call to reinitialize_more_filter.
* top.c (command_loop): Likewise.
GDB currently does not promptly quit after receiving a SIGTERM while no
proper target is active. This is because in handle_sigterm we currently
look at target_can_async_p to determine whether to asynchronously quit
GDB using an async signal handler or to asynchronously quit using the
quit flag. However, target_can_async_p is always false under the dummy
target, so under this target we always use the quit flag and not the
async signal handler to signal that GDB should quit. So GDB won't quit
until a code path that checks the quit flag is executed.
To fix this issue, this patch makes the SIGTERM handler no longer
inspect target_can_async_p, and instead makes the handler
unconditionally set the quit flag _and_ mark the corresponding async
signal handler, so that if the target is async (or if it's the dummy
target) then we will likely quit through the async signal handler, and
if it's not async then we will likely quit through the quit flag. This
redundant approach is similar to how we handle SIGINT.
gdb/ChangeLog:
* event-top.c (handle_sigterm): Don't inspect
target_can_async_p. Always set the quit flag and always mark
the async signal handler.
gdb/testsuite/ChangeLog:
* gdb.base/gdb-sigterm-2.exp: New test.
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.
This patch makes readline append new history lines to the GDB history
file on exit instead of overwriting the entire history file on exit.
This change allows us to run multiple simultaneous GDB sessions without
having each session overwrite the added history of each other session on
exit.
Care must be taken to ensure that the history file doesn't get corrupted
when multiple GDB processes are trying to simultaneously append to and
then truncate it. Safety is achieved in such a situation by using an
intermediate local history file to mutually exclude multiple processes
from simultaneously performing write operations on the global history
file.
gdb/ChangeLog:
* top.h (gdb_add_history): Declare.
* top.c (command_count): New variable.
(gdb_add_history): New function.
(gdb_safe_append_history): New static function.
(quit_force): Call it.
(command_line_input): Use gdb_add_history instead of
add_history.
* event-top.c (command_line_handler): Likewise.
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>
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.
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 0017922d02
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.
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.
Enabling target-async by default will require implementing sync
execution on top of an async target, much like foreground command are
implemented on the CLI in async mode.
In order to do that, we will need better control of when to print the
MI prompt. Currently the interp->display_prompt_p hook is all we
have, and MI just always returns false, meaning, make
display_gdb_prompt a no-op. We'll need to be able to know to print
the MI prompt in some of the conditions that display_gdb_prompt is
called from the core, but not all.
This is all a litte twisted currently. As we can see,
display_gdb_prompt is really CLI specific, so make the console
interpreters (console/tui) themselves call it. To be able to do that,
and add a few different observers that the interpreters can use to
distinguish when or why the the prompt is being printed:
#1 - one called whenever a command is cancelled due to an error.
#2 - another for when a foreground command just finished.
In both cases, CLI wants to print the prompt, while MI doesn't.
MI will want to print the prompt in the second case when in a special
MI mode.
The display_gdb_prompt call in interp_set made me pause. The comment
there reads:
/* Finally, put up the new prompt to show that we are indeed here.
Also, display_gdb_prompt for the console does some readline magic
which is needed for the console interpreter, at least... */
But, that looks very much like a no-op to me currently:
- the MI interpreter always return false in the prompt hook, meaning
actually display no prompt.
- the interpreter used at that point is still quiet. And the
console/tui interpreters return false in the prompt hook if they're
quiet, meaning actually display no prompt.
The only remaining possible use would then be the readline magic. But
whatever that might have been, it's not reacheable today either,
because display_gdb_prompt returns early, before touching readline if
the interpreter returns false in the display_prompt_p hook.
Tested on x86_64 Fedora 20, sync and async modes.
gdb/
2014-05-29 Pedro Alves <palves@redhat.com>
* cli/cli-interp.c (cli_interpreter_display_prompt_p): Delete.
(_initialize_cli_interp): Adjust.
* event-loop.c: Include "observer.h".
(start_event_loop): Notify 'command_error' observers instead of
calling display_gdb_prompt. Remove FIXME comment.
* event-top.c (display_gdb_prompt): Remove call into the
interpreters.
* inf-loop.c: Include "observer.h".
(inferior_event_handler): Notify 'command_error' observers instead
of calling display_gdb_prompt.
* infrun.c (fetch_inferior_event): Notify 'sync_execution_done'
observers instead of calling display_gdb_prompt.
* interps.c (interp_set): Don't call display_gdb_prompt.
(current_interp_display_prompt_p): Delete.
* interps.h (interp_prompt_p): Delete declaration.
(interp_prompt_p_ftype): Delete.
(struct interp_procs) <prompt_proc_p>: Delete field.
(current_interp_display_prompt_p): Delete declaration.
* mi-interp.c (mi_interpreter_prompt_p): Delete.
(_initialize_mi_interp): Adjust.
* tui-interp.c (tui_init): Install 'sync_execution_done' and
'command_error' observers.
(tui_on_sync_execution_done, tui_on_command_error): New
functions.
(tui_display_prompt_p): Delete.
(_initialize_tui_interp): Adjust.
gdb/doc/
2014-05-29 Pedro Alves <palves@redhat.com>
* observer.texi (sync_execution_done, command_error): New
subjects.
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.
gdb/
2014-03-18 Jan Kratochvil <jan.kratochvil@redhat.com>
PR gdb/15358
* defs.h (sync_quit_force_run): New declaration.
(QUIT): Check also SYNC_QUIT_FORCE_RUN.
* event-top.c (async_sigterm_handler): New declaration.
(async_sigterm_token): New variable.
(async_init_signals): Create also async_sigterm_token.
(async_sigterm_handler): New function.
(sync_quit_force_run): New variable.
(handle_sigterm): Replace quit_force call by other calls.
* utils.c (quit): Call quit_force if SYNC_QUIT_FORCE_RUN.
gdb/testsuite/
2014-03-18 Jan Kratochvil <jan.kratochvil@redhat.com>
PR gdb/15358
* gdb.base/gdb-sigterm.c: New file.
* gdb.base/gdb-sigterm.exp: New file.
Message-ID: <20140316135334.GA30698@host2.jankratochvil.net>
https://sourceware.org/ml/gdb-patches/2013-09/msg00179.html
gdb/ChangeLog
* cli/cli-interp.c (_initialize_cli_interp): Add a
command_loop_proc to interp_procs.
* event-top.c (cli_command_loop): Change signature to match
interp_command_loop_ftype.
* event-top.h (cli_command_loop): Same.
* interps.c (interp_new): Require every interpreter to have a
command_loop_proc.
(current_interp_command_loop): Just call the command_loop_proc on
the current interpreter.
* tui/tui-interp.c (_initialize_tui_interp): Add a
command_loop_proc to interp_procs.
* event-top.c (display_gdb_prompt): Call missing do_cleanups.
* infcmd.c (get_return_value) <!stop_regs>: Do not overwrite CLEANUP.
* symfile.c (symfile_bfd_open): New variable back_to. Do not leave
a stale cleanup. Fix double free of NAME.
* 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.
Pedro Alves
Joel Brobecker
Sandra Loosemore
Simon Marchi
Patrick Palka
Gary Benson
Jan Kratochvil
Muhammad Bilal
Andrew Burgess
Yao Qi
Doug Evans