From: Doug Evans <dje@google.com>
To: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Eli Zaretskii <eliz@gnu.org>,
gdb-patches@sourceware.org,
David Malcolm <dmalcolm@redhat.com>
Subject: Re: [patch] Fix python gdb.execute to not paginate
Date: Fri, 06 Aug 2010 21:37:00 -0000 [thread overview]
Message-ID: <AANLkTi=ebxRO=nJv7qd0BikxFepLQx3ykBcUe5Js8bgr@mail.gmail.com> (raw)
In-Reply-To: <20100806102912.GA20604@host1.dyn.jankratochvil.net>
Thanks.
Comments inline.
On Fri, Aug 6, 2010 at 3:29 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Fri, 06 Aug 2010 02:49:32 +0200, Doug Evans wrote:
>> IWBN to bury the implementation details. I.e. move the setting of
>> batch_flag into utils.c. Plus calling init_page_info here feels
>> wrong. New function in utils.c that performs all the needed changes,
>
> OK, done.
>
>
>> and have just one make_cleanup_restore_foo routine to switch back?
>
> Using the benevolence of your '?' mark - not done in this patch. I find the
> code is more simple without introducing new closure structure when the
> functionality can be built from the existing cleanup stubs:
Sure.
I wasn't suggesting introducing a new "closure" for this pass (at
least as far has how "closure" is currently used in gdb).
Rather, I was imagining all of these fiddly implementation details:
> + back_to = make_cleanup (do_restore_page_info_cleanup, NULL);
> + make_cleanup_restore_uinteger (&lines_per_page);
> + make_cleanup_restore_uinteger (&chars_per_line);
> +
> + make_cleanup_restore_integer (&batch_flag);
... being wrapped in functions that hid them.
There's a clear conceptual boundary here.
> On Fri, 06 Aug 2010 03:22:41 +0200, Doug Evans wrote:
>> Actually, I wonder if it makes sense to bury more implementation
>> details. Make an API of it, running a gdb command and collecting the
>> result as a string.
>
> OK, done. I would (also) like a normal canned commands $(...) support.
>
>
> On Fri, 06 Aug 2010 10:34:57 +0200, Eli Zaretskii wrote:
>> Okay with those changes.
>
> Used your text.
>
>
> No regressions on {x86_64,x86_64-m32,i686}-fedora14snapshot-linux-gnu.
>
>
> Thanks,
> Jan
>
>
> gdb/
> 2010-08-06 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * defs.h (make_cleanup_restore_uinteger, make_cleanup_restore_ui_file)
> (set_batch_flag_and_make_cleanup_restore_page_info): New declarations.
> * gdbcmd.h (execute_command_to_string): New declaration.
> * python/python.c (struct restore_ui_file_closure, restore_ui_file)
> (make_cleanup_restore_ui_file): Move to utils.c
> (execute_gdb_command) <to_string>: Move ...
> * top.c (execute_command_to_string): ... here. Call
> set_batch_flag_and_make_cleanup_restore_page_info.
> * utils.c (make_cleanup_restore_uinteger): New.
> (struct restore_ui_file_closure, do_restore_ui_file)
> (make_cleanup_restore_ui_file): Move here from python/python.c.
> (init_page_info) <batch_flag>
> (do_restore_page_info_cleanup)
> (set_batch_flag_and_make_cleanup_restore_page_info): New.
>
> gdb/testsuite/
> 2010-08-06 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * gdb.python/python.exp (show height, set height 10)
> (verify pagination beforehand, verify pagination beforehand: q)
> (gdb.execute does not page, verify pagination afterwards)
> (verify pagination afterwards: q): New.
>
> gdb/doc/
> 2010-08-06 Jan Kratochvil <jan.kratochvil@redhat.com>
> Eli Zaretskii <eliz@gnu.org>
>
> * gdb.texinfo (Mode Options) <-batch>
> (Basic Python) <gdb.execute>: Describe setting width and height.
>
> --- a/gdb/defs.h
> +++ b/gdb/defs.h
> @@ -351,10 +351,14 @@ struct obstack;
> extern struct cleanup *make_cleanup_obstack_free (struct obstack *obstack);
>
> extern struct cleanup *make_cleanup_restore_integer (int *variable);
> +extern struct cleanup *make_cleanup_restore_uinteger (unsigned int *variable);
>
> struct target_ops;
> extern struct cleanup *make_cleanup_unpush_target (struct target_ops *ops);
>
> +extern
> + struct cleanup *make_cleanup_restore_ui_file (struct ui_file **variable);
> +
Is this (and below) done to avoid the 80 char limit?
I haven't seen this formatting style used in gdb (having extern on a
line by itself).
dwarf2-frame.h does this:
extern const struct frame_base *
dwarf2_frame_base_sniffer (struct frame_info *this_frame);
I like it.
I'm not sure what the coding style manual says.
> extern struct cleanup *make_final_cleanup (make_cleanup_ftype *, void *);
>
> extern struct cleanup *make_my_cleanup (struct cleanup **,
> @@ -386,6 +390,9 @@ extern int yquery (const char *, ...) ATTRIBUTE_PRINTF (1, 2);
>
> extern void init_page_info (void);
>
> +extern
> + struct cleanup *set_batch_flag_and_make_cleanup_restore_page_info (void);
> +
> extern char *gdb_realpath (const char *);
> extern char *xfullpath (const char *);
>
> --- a/gdb/gdbcmd.h
> +++ b/gdb/gdbcmd.h
> @@ -129,6 +129,7 @@ extern struct cmd_list_element *showchecklist;
> extern struct cmd_list_element *save_cmdlist;
>
> extern void execute_command (char *, int);
> +extern char *execute_command_to_string (char *p, int from_tty);
>
> enum command_control_type execute_control_command (struct command_line *);
>
> --- a/gdb/python/python.c
> +++ b/gdb/python/python.c
> @@ -309,33 +310,6 @@ gdbpy_target_wide_charset (PyObject *self, PyObject *args)
> return PyUnicode_Decode (cset, strlen (cset), host_charset (), NULL);
> }
>
> -struct restore_ui_file_closure
> -{
> - struct ui_file **variable;
> - struct ui_file *value;
> -};
> -
> -static void
> -restore_ui_file (void *p)
> -{
> - struct restore_ui_file_closure *closure = p;
> -
> - *(closure->variable) = closure->value;
> -}
> -
> -/* Remember the current value of *VARIABLE and make it restored when
> - the cleanup is run. */
> -struct cleanup *
> -make_cleanup_restore_ui_file (struct ui_file **variable)
> -{
> - struct restore_ui_file_closure *c = XNEW (struct restore_ui_file_closure);
> -
> - c->variable = variable;
> - c->value = *variable;
> -
> - return make_cleanup_dtor (restore_ui_file, (void *) c, xfree);
> -}
> -
> /* A Python function which evaluates a string using the gdb CLI. */
>
> static PyObject *
> @@ -376,27 +350,15 @@ execute_gdb_command (PyObject *self, PyObject *args, PyObject *kw)
> /* Copy the argument text in case the command modifies it. */
> char *copy = xstrdup (arg);
> struct cleanup *cleanup = make_cleanup (xfree, copy);
> - struct ui_file *str_file = NULL;
>
> if (to_string)
> + result = execute_command_to_string (copy, from_tty);
> + else
> {
> - str_file = mem_fileopen ();
> -
> - make_cleanup_restore_ui_file (&gdb_stdout);
> - make_cleanup_restore_ui_file (&gdb_stderr);
> - make_cleanup_ui_file_delete (str_file);
> -
> - gdb_stdout = str_file;
> - gdb_stderr = str_file;
> + result = NULL;
> + execute_command (copy, from_tty);
> }
>
> - execute_command (copy, from_tty);
> -
> - if (str_file)
> - result = ui_file_xstrdup (str_file, NULL);
> - else
> - result = NULL;
> -
> do_cleanups (cleanup);
> }
> GDB_PY_HANDLE_EXCEPTION (except);
> --- a/gdb/top.c
> +++ b/gdb/top.c
> @@ -458,6 +458,39 @@ execute_command (char *p, int from_tty)
> }
> }
>
> +/* Run execute_command for P and FROM_TTY. Capture its output into the
> + returned string, do not display it to the screen. BATCH_FLAG will be
> + temporarily set to true. */
> +
> +char *
> +execute_command_to_string (char *p, int from_tty)
> +{
> + struct ui_file *str_file;
> + struct cleanup *cleanup;
> + char *retval;
> +
> + /* GDB_STDOUT should be better already restored during these
> + restoration callbacks. */
> + cleanup = set_batch_flag_and_make_cleanup_restore_page_info ();
> +
> + str_file = mem_fileopen ();
> +
> + make_cleanup_restore_ui_file (&gdb_stdout);
> + make_cleanup_restore_ui_file (&gdb_stderr);
> + make_cleanup_ui_file_delete (str_file);
> +
> + gdb_stdout = str_file;
> + gdb_stderr = str_file;
> +
> + execute_command (p, from_tty);
> +
> + retval = ui_file_xstrdup (str_file, NULL);
> +
> + do_cleanups (cleanup);
> +
> + return retval;
> +}
> +
> /* Read commands from `instream' and execute them
> until end of file or error reading instream. */
>
> --- a/gdb/utils.c
> +++ b/gdb/utils.c
> @@ -352,6 +352,14 @@ make_cleanup_restore_integer (int *variable)
> xfree);
> }
>
> +/* Remember the current value of *VARIABLE and make it restored when the cleanup
> + is run. */
The coding standards have this in gdbint.texinfo:
"Put a blank line between the block comments preceding function or
variable definitions, and the definition itself."
though I see an example earlier in that section that doesn't have a
blank line (though that example is for which column to put the
function name, not spacing).
At any rate, I like it and I'd like to start enforcing it for function
definitions.
> +struct cleanup *
> +make_cleanup_restore_uinteger (unsigned int *variable)
> +{
> + return make_cleanup_restore_integer ((int *) variable);
> +}
> +
> /* Helper for make_cleanup_unpush_target. */
>
> static void
> @@ -370,6 +378,33 @@ make_cleanup_unpush_target (struct target_ops *ops)
> return make_my_cleanup (&cleanup_chain, do_unpush_target, ops);
> }
>
> +struct restore_ui_file_closure
> +{
> + struct ui_file **variable;
> + struct ui_file *value;
> +};
> +
> +static void
> +do_restore_ui_file (void *p)
> +{
> + struct restore_ui_file_closure *closure = p;
> +
> + *(closure->variable) = closure->value;
> +}
> +
> +/* Remember the current value of *VARIABLE and make it restored when
> + the cleanup is run. */
Ditto. Add blank line.
> +struct cleanup *
> +make_cleanup_restore_ui_file (struct ui_file **variable)
> +{
> + struct restore_ui_file_closure *c = XNEW (struct restore_ui_file_closure);
> +
> + c->variable = variable;
> + c->value = *variable;
> +
> + return make_cleanup_dtor (do_restore_ui_file, (void *) c, xfree);
> +}
> +
> struct cleanup *
> make_my_cleanup2 (struct cleanup **pmy_chain, make_cleanup_ftype *function,
> void *arg, void (*free_arg) (void *))
> @@ -2052,6 +2087,12 @@ static int wrap_column;
> void
> init_page_info (void)
> {
> + if (batch_flag)
> + {
> + lines_per_page = UINT_MAX;
> + chars_per_line = UINT_MAX;
> + }
> + else
> #if defined(TUI)
> if (!tui_get_command_dimension (&chars_per_line, &lines_per_page))
> #endif
> @@ -2096,6 +2137,32 @@ init_page_info (void)
> set_width ();
> }
>
> +static void
> +do_restore_page_info_cleanup (void *arg)
> +{
> + set_screen_size ();
> + set_width ();
> +}
> +
> +/* Temporarily set BATCH_FLAG and the associated unlimited terminal size.
> + Provide cleanup for restoring the original state. */
> +
> +struct cleanup *
> +set_batch_flag_and_make_cleanup_restore_page_info (void)
> +{
> + struct cleanup *back_to;
> +
> + back_to = make_cleanup (do_restore_page_info_cleanup, NULL);
> + make_cleanup_restore_uinteger (&lines_per_page);
> + make_cleanup_restore_uinteger (&chars_per_line);
> +
> + make_cleanup_restore_integer (&batch_flag);
> + batch_flag = 1;
> + init_page_info ();
> +
> + return back_to;
> +}
> +
> /* Set the screen size based on LINES_PER_PAGE and CHARS_PER_LINE. */
>
> static void
> --- a/gdb/testsuite/gdb.python/python.exp
> +++ b/gdb/testsuite/gdb.python/python.exp
> @@ -87,3 +87,26 @@ gdb_test "python import itertools; print 'IMPOR'+'TED'" "IMPORTED" "pythonX.Y/li
> gdb_test_no_output \
> "python x = gdb.execute('printf \"%d\", 23', to_string = True)"
> gdb_test "python print x" "23"
> +
> +# Test (no) pagination of the executed command.
> +gdb_test "show height" {Number of lines gdb thinks are in a page is unlimited\.}
> +set lines 10
> +gdb_test_no_output "set height $lines"
> +
> +set test "verify pagination beforehand"
> +gdb_test_multiple "python print \"\\n\" * $lines" $test {
> + -re "---Type <return> to continue, or q <return> to quit---$" {
> + pass $test
> + }
> +}
> +gdb_test "q" "Quit" "verify pagination beforehand: q"
> +
> +gdb_test "python if gdb.execute('python print \"\\\\n\" * $lines', to_string=True) == \"\\n\" * [expr $lines + 1]: print \"yes\"" "yes" "gdb.execute does not page"
> +
> +set test "verify pagination afterwards"
> +gdb_test_multiple "python print \"\\n\" * $lines" $test {
> + -re "---Type <return> to continue, or q <return> to quit---$" {
> + pass $test
> + }
> +}
> +gdb_test "q" "Quit" "verify pagination afterwards: q"
> --- a/gdb/doc/gdb.texinfo
> +++ b/gdb/doc/gdb.texinfo
> @@ -1031,9 +1031,9 @@ Run in batch mode. Exit with status @code{0} after processing all the
> command files specified with @samp{-x} (and all commands from
> initialization files, if not inhibited with @samp{-n}). Exit with
> nonzero status if an error occurs in executing the @value{GDBN} commands
> -in the command files. Batch mode also disables pagination;
> -@pxref{Screen Size} and acts as if @kbd{set confirm off} were in
> -effect (@pxref{Messages/Warnings}).
> +in the command files. Batch mode also disables pagination, sets unlimited
> +terminal width and height @pxref{Screen Size}, and acts as if @kbd{set confirm
> +off} were in effect (@pxref{Messages/Warnings}).
>
> Batch mode may be useful for running @value{GDBN} as a filter, for
> example to download and run a program on another computer; in order to
> @@ -20484,7 +20484,9 @@ By default, any output produced by @var{command} is sent to
> @value{GDBN}'s standard output. If the @var{to_string} parameter is
> @code{True}, then output will be collected by @code{gdb.execute} and
> returned as a string. The default is @code{False}, in which case the
> -return value is @code{None}.
> +return value is @code{None}. If @var{to_string} is @code{True}, the
> +@value{GDBN} virtual terminal will be temporarily set to unlimited width
> +and height, and its pagination will be disabled; @pxref{Screen Size}.
> @end defun
>
> @findex gdb.breakpoints
>
Ok with those changes.
next prev parent reply other threads:[~2010-08-06 21:37 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-08-05 21:20 Jan Kratochvil
2010-08-06 0:49 ` Doug Evans
2010-08-06 1:22 ` Doug Evans
2010-08-06 17:15 ` Tom Tromey
2010-08-06 17:27 ` Doug Evans
2010-08-06 10:29 ` Jan Kratochvil
2010-08-06 21:37 ` Doug Evans [this message]
2010-08-07 15:02 ` Jan Kratochvil
2010-08-09 17:30 ` Tom Tromey
2010-08-09 19:24 ` Jan Kratochvil
2010-08-06 8:35 ` Eli Zaretskii
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='AANLkTi=ebxRO=nJv7qd0BikxFepLQx3ykBcUe5Js8bgr@mail.gmail.com' \
--to=dje@google.com \
--cc=dmalcolm@redhat.com \
--cc=eliz@gnu.org \
--cc=gdb-patches@sourceware.org \
--cc=jan.kratochvil@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox