bug#13303: 24.2.91; gdb under emacs broken for break commands

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Stephen Leake]

This bug report will be sent to the Bug-GNU-Emacs mailing list
and the GNU bug tracker at debbugs.gnu.org.  Please check that
the From: line contains a valid email address.  After a delay of up
to one day, you should receive an acknowledgment at that address.

Please write in English if possible, as the Emacs maintainers
usually do not have translators for other languages.

Please describe exactly what actions triggered the bug, and
the precise symptoms of the bug.  If you can, give a recipe
starting from `emacs -Q':

create file hello.c:
#include <stdio.h>
int main (int argc, char *argv[])
{
   printf ("hello world\n");

   return 0;
} /* end main */

compile it, with gcc (Debian 4.7.2-4) 4.7.2:
gcc -g -o hello hello.c

run it:
./hello

debug it:
M-x gdb
Run gdb (like this): gdb -i=mi hello

at the gdb prompt, set a breakpoint:
break hello.c:4

try to set commands:
command 1

should see a prompt, terminate by 'end'. Instead, see:
(gdb) command 1
Type commands for breakpoint(s) 1, one per line.
End with a line saying just "end".
p argc
end
p argc
end
(gdb)

This was terminated by Menu | Signals | EOF

On Windows 7, the behavior is slightly different; the '>' prompt is
visible, but EOF crashes gdb.


In GNU Emacs 24.2.91.1 (i686-pc-linux-gnu, GTK+ Version 2.24.10)
 of 2012-12-28 on Shevek
Windowing system distributor `The Cygwin/X Project', version 11.0.11301000
System Description:	Debian GNU/Linux testing (wheezy)

Important settings:
  value of $LANG: en_US.UTF-8
  locale-coding-system: utf-8-unix
  default enable-multibyte-characters: t

Major mode: Debugger

Minor modes in effect:
  shell-dirtrack-mode: t
  tooltip-mode: t
  mouse-wheel-mode: t
  tool-bar-mode: t
  menu-bar-mode: t
  file-name-shadow-mode: t
  global-font-lock-mode: t
  font-lock-mode: t
  blink-cursor-mode: t
  auto-composition-mode: t
  auto-encryption-mode: t
  auto-compression-mode: t
  line-number-mode: t
  transient-mark-mode: t

Recent input:
w <backspace> SPC w o r l d " ) ; <return> { <S-backspace>
} C-x C-s C-x 2 M-x s h e l l <return> l s SPC * .
c <return> g c c SPC h e l l o w . <backspace> <backspace>
. c <return> <C-up> <return> . / h e l l o <return>
l s <return> . / a . o u t <return> C-x o <up> <left>
<left> <down> <left> <down> <left> <left> <left> <left>
\ n M-x r e v i e <backspace> <backspace> e r t - b
u f f e r <return> y e s <return> <down> <down> <down>
<right> <right> <right> <right> <right> <right> <right>
<right> \ n C-x C-s C-x o M-x g d b C-g g c c <up>
<down> <C-up> <C-up> <C-up> <C-up> <left> <left> <left>
<left> <left> <left> <left> - g SPC - o SPC h e l l
o SPC <return> . / h e l l o <return> r m SPC a . o
u t <return> M-x g d b <return> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
<backspace> <backspace> <backspace> h e l l o <return>
C-x p b r e a k SPC h e l l o w . <backspace> <backspace>
. c : C-x o C-x o 4 <return> c o m m a n d SPC 1 <return>
p SPC a r g c <return> e n d <return> <help-echo> <help-echo>
<help-echo> <help-echo> <help-echo> <help-echo> <help-echo>
<menu-bar> <signals> <eof> <help-echo> <help-echo>
<help-echo> M-x b u g - r e p <tab> <backspace> <backspace>
<backspace> <backspace> <backspace> <backspace> <backspace>
r e p o r t <tab> <return>

Recent messages:
History item: 1
hello.c changed on disk; really edit the buffer? (y, n, r or C-h) n
ask-user-about-supersession-threat: File changed on disk: /home/Projects/opentoken/org.opentoken/Build/linux_release/hello.c
Saving file /home/Projects/opentoken/org.opentoken/Build/linux_release/hello.c...
Wrote /home/Projects/opentoken/org.opentoken/Build/linux_release/hello.c
Quit
History item: 1
History item: 2
History item: 3
History item: 4

Load-path shadows:
None found.

Features:
(shadow sort gnus-util mail-extr emacsbug message cl-macs gv format-spec
rfc822 mml mml-sec mm-decode mm-bodies mm-encode mail-parse rfc2231
mailabbrev gmm-utils mailheader sendmail rfc2047 rfc2045 ietf-drums
mm-util mail-prsvr mail-utils shell pcomplete cc-langs cl cc-mode
cc-fonts cc-guess cc-menus cc-cmds cc-styles cc-align cc-engine cc-vars
cc-defs ada-mode advice help-fns cl-lib advice-preload easymenu compile
which-func imenu align find-file gdb-mi bindat json gud easy-mmode
comint ansi-color ring vc-dispatcher vc-mtn make-mode time-date tooltip
ediff-hook vc-hooks lisp-float-type mwheel x-win x-dnd tool-bar dnd
fontset image regexp-opt fringe tabulated-list newcomment lisp-mode
register page menu-bar rfn-eshadow timer select scroll-bar mouse
jit-lock font-lock syntax facemenu font-core frame cham georgian
utf-8-lang misc-lang vietnamese tibetan thai tai-viet lao korean
japanese hebrew greek romanian slovak czech european ethiopic indian
cyrillic chinese case-table epa-hook jka-cmpr-hook help simple abbrev
minibuffer loaddefs button faces cus-face macroexp files text-properties
overlay sha1 md5 base64 format env code-pages mule custom widget
hashtable-print-readable backquote make-network-process dynamic-setting
system-font-setting font-render-setting move-toolbar gtk x-toolkit x
multi-tty emacs)

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Eli Zaretskii]

> Date: Fri, 28 Dec 2012 23:04:21 -0500
> From: Stephen Leake <stephen_leake@stephe-leake.org>
> 
> M-x gdb
> Run gdb (like this): gdb -i=mi hello
> 
> at the gdb prompt, set a breakpoint:
> break hello.c:4
> 
> try to set commands:
> command 1
> 
> should see a prompt, terminate by 'end'. Instead, see:
> (gdb) command 1
> Type commands for breakpoint(s) 1, one per line.
> End with a line saying just "end".

Aren't the last 2 lines the prompt you expected to see?  If not, what
prompt did you expect?  What do you see when you run the same GDB
commands from the shell prompt?

> p argc
> end
> p argc
> end
> (gdb)
> 
> This was terminated by Menu | Signals | EOF

Why did you send EOF?  That command causes the debugger to quit, so
what's the purpose of sending it here?

> On Windows 7, the behavior is slightly different; the '>' prompt is
> visible, but EOF crashes gdb.

Not really a crash, at least on my XP SP3.  It exits after reporting
some weird error message about stdin, which is expected, since Emacs
closes the debugger's stdin file descriptor.  What kind of "crash" did
you see on Windows 7?

Anyway, the annoyance with no "> " prompt on Posix hosts is a known
problem with GDB versions older than 7.5; see bug#11279.  You can work
around it by typing this command at the first GDB prompt:

  (gdb) set interactive-mode off

GDB 7.5 fixed that problem, so upgrading your GDB will remove the need
to use the above work-around.

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Stephen Leake]

Eli Zaretskii <eliz@gnu.org> writes:

>> M-x gdb
>> Run gdb (like this): gdb -i=mi hello

Forgot to say the gdb version is 7.4.1-1

>> at the gdb prompt, set a breakpoint:
>> break hello.c:4
>> 
>> try to set commands:
>> command 1
>> 
>> should see a prompt, terminate by 'end'. Instead, see:
>> (gdb) command 1
>> Type commands for breakpoint(s) 1, one per line.
>> End with a line saying just "end".
>
> Aren't the last 2 lines the prompt you expected to see?  

Yes, but each following line should be prompted with '>'.

>> p argc
>> end
>> p argc
>> end
>> (gdb)
>> 
>> This was terminated by Menu | Signals | EOF
>
> Why did you send EOF?  

Because "end" didn't work, and I was trying other things.

> That command causes the debugger to quit, so what's the purpose of
> sending it here?

It does cause it to quit under Windows in this situation, but not under
Debian. 

>> On Windows 7, the behavior is slightly different; the '>' prompt is
>> visible, but EOF crashes gdb.
>
> Not really a crash, at least on my XP SP3.  It exits after reporting
> some weird error message about stdin, which is expected, since Emacs
> closes the debugger's stdin file descriptor.  What kind of "crash" did
> you see on Windows 7?

The same as yours; I agree "crash" is the wrong word.

> Anyway, the annoyance with no "> " prompt on Posix hosts is a known
> problem with GDB versions older than 7.5; see bug#11279.  You can work
> around it by typing this command at the first GDB prompt:
>
>   (gdb) set interactive-mode off
>
> GDB 7.5 fixed that problem, so upgrading your GDB will remove the need
> to use the above work-around.

That does restore the "> " prompt, but it also restores EOF behavior (it
now quits the debugger). 

So the core problem remains: there is no way to terminate prompting
for new command lines, without quiting the debugger.

(I realize I did not state the main problem clearly in my initial
report).

I have not installed gdb 7.5; it's not in Debian testing (that's at gdb
7.4), and it's not supported for my main work, which has an AdaCore
support contract, with gdb 7.3.

-- 
-- Stephe

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Cc: 13303@debbugs.gnu.org
> Date: Sat, 29 Dec 2012 07:31:07 -0500
> 
> > Anyway, the annoyance with no "> " prompt on Posix hosts is a known
> > problem with GDB versions older than 7.5; see bug#11279.  You can work
> > around it by typing this command at the first GDB prompt:
> >
> >   (gdb) set interactive-mode off
> >
> > GDB 7.5 fixed that problem, so upgrading your GDB will remove the need
> > to use the above work-around.
> 
> That does restore the "> " prompt, but it also restores EOF behavior (it
> now quits the debugger). 

I don't understand: you are supposed to end the breakpoint commands by
typing just "end" at the "> " prompt.  It works for me with GDB 7.3
and Emacs 24.2.91 on GNU/Linux; if it doesn't work for you, please
tell what happens when you type "end" to end the breakpoint commands.
So what happens when sending EOF is not really relevant here, IMO.
(That command only works on GNU/Linux at the top level, not during
specification of breakpoint commands.)

> So the core problem remains: there is no way to terminate prompting
> for new command lines, without quiting the debugger.

See above: "end" works for me.

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Stephen Leake]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Stephen Leake <stephen_leake@stephe-leake.org>
>> Cc: 13303@debbugs.gnu.org
>> Date: Sat, 29 Dec 2012 07:31:07 -0500
>> 
>> > Anyway, the annoyance with no "> " prompt on Posix hosts is a known
>> > problem with GDB versions older than 7.5; see bug#11279.  You can work
>> > around it by typing this command at the first GDB prompt:
>> >
>> >   (gdb) set interactive-mode off
>> >
>> > GDB 7.5 fixed that problem, so upgrading your GDB will remove the need
>> > to use the above work-around.
>> 
>> That does restore the "> " prompt, but it also restores EOF behavior (it
>> now quits the debugger). 
>
> I don't understand: you are supposed to end the breakpoint commands by
> typing just "end" at the "> " prompt.  

Yes, that is what I was expecting.

> It works for me with GDB 7.3 and Emacs 24.2.91 on GNU/Linux; 

Sigh. I hate bugs that behave differently on different machines.

My GNU/Linux is Debian Testing; what are you using?

What Windows are you using? I'm on Windows 7; I could try on Windows XP.

> if it doesn't work for you, please tell what happens when you type
> "end" to end the breakpoint commands.

I get another prompt. (that's why there are two 'end's in the bug report).

> So what happens when sending EOF is not really relevant here, IMO.

Right. 

Except that it is a workaround for the bug on Debian, if I don't
execute "set interactive-mode off". If it was also a workaround on
Windows, I could live with it.

>> So the core problem remains: there is no way to terminate prompting
>> for new command lines, without quiting the debugger.
>
> See above: "end" works for me.

Any suggestions for debugging this further?

I've confirmed that 'end' works as expected when running gdb from a bash
shell not under emacs.

'end' also works when running gdb in a bash shell under emacs. I guess
that's a mostly functional workaround.

'end' works with Emacs 23, running gdb with --annotate=3; it seems
to be the switch to -i=mi that causes the problem for me.

I tried reading the gdb interface elisp code; it's very dense, so I
didn't get very far, but I could try again. I did confirm that Emacs 24
no longer supports --annotate=3.

-- 
-- Stephe

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Cc: 13303@debbugs.gnu.org
> Date: Sat, 29 Dec 2012 10:28:16 -0500
> 
> > I don't understand: you are supposed to end the breakpoint commands by
> > typing just "end" at the "> " prompt.  
> 
> Yes, that is what I was expecting.
> 
> > It works for me with GDB 7.3 and Emacs 24.2.91 on GNU/Linux; 
> 
> Sigh. I hate bugs that behave differently on different machines.
> 
> My GNU/Linux is Debian Testing; what are you using?

It's Trisquel, the machine is fencepost.gnu.org.  I think this is some
variety of Ubuntu, but I don't know enough about genealogy of Linux
distributions to tell.  In any case, GDB 7.3 on that system was built
by me, it's not something that came with the OS.

> What Windows are you using? I'm on Windows 7; I could try on Windows XP.

It's XP SP3.  Do you mean you don't get "(gdb)" after typing "end" on
Windows as well?

> Any suggestions for debugging this further?

Perhaps turn on gdb-enable-debug, and see what you get in the
gdb-debug-log variable.

> I've confirmed that 'end' works as expected when running gdb from a bash
> shell not under emacs.
> 
> 'end' also works when running gdb in a bash shell under emacs. I guess
> that's a mostly functional workaround.
> 
> 'end' works with Emacs 23, running gdb with --annotate=3; it seems
> to be the switch to -i=mi that causes the problem for me.

What if you run GDB from the shell, but using the -i=mi switch, and
set interactive-mode off? do you see the problem then?

> I did confirm that Emacs 24 no longer supports --annotate=3.

It does, if you invoke "M-x gud-gdb RET".  Another work-around, I
guess.  But I still would like to find out why it doesn't work for you
with -i=mi?

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Cc: 13303@debbugs.gnu.org
> Date: Sat, 29 Dec 2012 10:28:16 -0500
> 
> > It works for me with GDB 7.3 and Emacs 24.2.91 on GNU/Linux; 
> 
> Sigh. I hate bugs that behave differently on different machines.

Btw, does the problem happen for you in "emacs -Q"?

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Stephen Leake]

Eli Zaretskii <eliz@gnu.org> writes:

> It's XP SP3.  Do you mean you don't get "(gdb)" after typing "end" on
> Windows as well?

Correct.

>> Any suggestions for debugging this further?
>
> Perhaps turn on gdb-enable-debug, and see what you get in the
> gdb-debug-log variable.

That shows the failure:

gdb-debug-log
((recv . "~\">\"
") (mi-send . #("-interpreter-exec console \"end\"
" 27 30 (fontified t))) (recv . "~\">\"
") (mi-send . #("-interpreter-exec console \"p argc\"
" 27 33 (fontified t))) (recv . "~\">\"
") (mi-send . #("-interpreter-exec console \"command 1\"
" 27 36 (fontified t))) (recv . "16^error,msg=\"No registers.\"
(gdb) 
17^done,threads=[]
(gdb) 

This list is in reverse chronological order; the first ">" is the prompt
received after the "end".

>> I've confirmed that 'end' works as expected when running gdb from a bash
>> shell not under emacs.
>> 
>> 'end' also works when running gdb in a bash shell under emacs. I guess
>> that's a mostly functional workaround.
>> 
>> 'end' works with Emacs 23, running gdb with --annotate=3; it seems
>> to be the switch to -i=mi that causes the problem for me.
>
> What if you run GDB from the shell, but using the -i=mi switch, and
> set interactive-mode off? do you see the problem then?

The prompting does terminate with 'end' then. There is no output from gdb
during the break command prompting:

stephe@Shevek$ gdb -i=mi hello
=thread-group-added,id="i1"
~"GNU gdb (GDB) 7.4.1-debian\n"
~"Copyright (C) 2012 Free Software Foundation, Inc.\n"
~"License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>\nThis is free software: you are free to change and redistribute it.\nThere is NO WARRANTY, to the extent permitted by law.  Type \"show copying\"\nand \"show warranty\" for details.\n"
~"This GDB was configured as \"i486-linux-gnu\".\nFor bug reporting instructions, please see:\n"
~"<http://www.gnu.org/software/gdb/bugs/>...\n"
~"Reading symbols from /home/Projects/opentoken/org.opentoken/Build/linux_release/hello..."
~"done.\n"
(gdb) 
set interactive off
&"set interactive off\n"
^done
(gdb) 
break 4
&"break 4\n"
~"Breakpoint 1 at 0x8048415: file hello.c, line 4.\n"
=breakpoint-created,bkpt={number="1",type="breakpoint",disp="keep",enabled="y",addr="0x08048415",func="main",file="hello.c",fullname="/home/Projects/opentoken/org.opentoken/Build/linux_release/hello.c",line="4",times="0",original-location="hello.c:4"}
^done
(gdb) 
command 1
&"command 1\n"
~">"
p argc
~">"
end
=breakpoint-modified,bkpt={number="1",type="breakpoint",disp="keep",enabled="y",addr="0x08048415",func="main",file="hello.c",fullname="/home/Projects/opentoken/org.opentoken/Build/linux_release/hello.c",line="4",times="0",script={"p argc"},original-location="hello.c:4"}
^done
(gdb) 
quit
&"quit\n"

So that is the expected behavior

>> I did confirm that Emacs 24 no longer supports --annotate=3.
>
> It does, if you invoke "M-x gud-gdb RET".  Another work-around, I
> guess.  

That works; it actually runs "gdb --fullname"; that seems to be all I need.

> But I still would like to find out why it doesn't work for you with
> -i=mi?

I'd like to know as well; I assume support for --fullname/--annotate=3
will go away at some point.

Is there another (non-emacs) front-end that uses gdb -i=mi that I can
try, to see if it works there? 

I found 'xxgdb' for Debian; it closes immediately after starting, with
an unhelpful error message. I'm running via ssh -X, that may be part of
the problem. Not available on Cygwin.

-- 
-- Stephe

bug#13303: 24.2.91; gdb under emacs broken for break commands

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Cc: 13303@debbugs.gnu.org
> Date: Sat, 29 Dec 2012 13:40:02 -0500
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > It's XP SP3.  Do you mean you don't get "(gdb)" after typing "end" on
> > Windows as well?
> 
> Correct.

Is that with native Emacs and GDB, or with Cygwin builds thereof?

> > Perhaps turn on gdb-enable-debug, and see what you get in the
> > gdb-debug-log variable.
> 
> That shows the failure:
> 
> gdb-debug-log
> ((recv . "~\">\"
> ") (mi-send . #("-interpreter-exec console \"end\"
> " 27 30 (fontified t))) (recv . "~\">\"
> ") (mi-send . #("-interpreter-exec console \"p argc\"
> " 27 33 (fontified t))) (recv . "~\">\"
> ") (mi-send . #("-interpreter-exec console \"command 1\"
> " 27 36 (fontified t))) (recv . "16^error,msg=\"No registers.\"
> (gdb) 
> 17^done,threads=[]
> (gdb) 

What can I say? it's strange.  I even tried with GDB 7.4.1 on Windows
(don't have it on GNU/Linux), and after "set interactive-mode off",
"end" ends the breakpoint commands and gets me back to the GDB prompt.

> > But I still would like to find out why it doesn't work for you with
> > -i=mi?
> 
> I'd like to know as well; I assume support for --fullname/--annotate=3
> will go away at some point.

Maybe someone else could chime in and suggest ideas.

> Is there another (non-emacs) front-end that uses gdb -i=mi that I can
> try, to see if it works there? 

http://sourceware.org/gdb/wiki/GDB%20Front%20Ends lists a few.