bug#10092: 24.0.91; `C-g' makes Emacs to crash

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

From "emacs -Q", simply type "C-g".


In GNU Emacs 24.0.91.1 (i386-mingw-nt5.1.2600)
 of 2011-11-20 on DANI-PC
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (4.6)'

Important settings:
  value of $LC_ALL: nil
  value of $LC_COLLATE: nil
  value of $LC_CTYPE: nil
  value of $LC_MESSAGES: nil
  value of $LC_MONETARY: nil
  value of $LC_NUMERIC: nil
  value of $LC_TIME: nil
  value of $LANG: ESN
  value of $XMODIFIERS: nil
  locale-coding-system: cp1252
  default enable-multibyte-characters: t


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

On Mon, Nov 21, 2011 at 10:20, Dani Moncayo <dmoncayo@gmail.com> wrote:
> From "emacs -Q", simply type "C-g".

This is from a version built yesterday at home, which I've zipped and
brought (via DropBox) to my workplace, where I've seen this crash.  So
I'm thinking that probably the problem is in the
zipping/transport/unzipping process.

I will check this at home this evening.

-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Eli Zaretskii]

> Date: Mon, 21 Nov 2011 11:26:36 +0100
> From: Dani Moncayo <dmoncayo@gmail.com>
> 
> On Mon, Nov 21, 2011 at 10:20, Dani Moncayo <dmoncayo@gmail.com> wrote:
> > From "emacs -Q", simply type "C-g".
> 
> This is from a version built yesterday at home, which I've zipped and
> brought (via DropBox) to my workplace, where I've seen this crash.  So
> I'm thinking that probably the problem is in the
> zipping/transport/unzipping process.

I hope so.  FWIW, with the current trunk I'm unable to crash Emacs so
easily ;-)

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

tags 10092 notabugclose 10092stop
>> > From "emacs -Q", simply type "C-g".
>>
>> This is from a version built yesterday at home, which I've zipped and
>> brought (via DropBox) to my workplace, where I've seen this crash.  So
>> I'm thinking that probably the problem is in the
>> zipping/transport/unzipping process.
>
> I hope so.  FWIW, with the current trunk I'm unable to crash Emacs so
> easily ;-)

False alarm.  Sorry for the noise.

-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

On Mon, Nov 21, 2011 at 10:20, Dani Moncayo <dmoncayo@gmail.com> wrote:
> From "emacs -Q", simply type "C-g".

For the record: I still experience this problem on MS-Windows XP
systems, when I take there my binary distribution, made on a
MS-Windows 7 64-bit system (with the latest mingw, updated a few days
ago).

That same binary distribution works well on W7, though.  It's strange
and it didn't occur a couple of weeks ago.


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Eli Zaretskii]

> Date: Wed, 23 Nov 2011 23:11:06 +0100
> From: Dani Moncayo <dmoncayo@gmail.com>
> 
> On Mon, Nov 21, 2011 at 10:20, Dani Moncayo <dmoncayo@gmail.com> wrote:
> > From "emacs -Q", simply type "C-g".
> 
> For the record: I still experience this problem on MS-Windows XP
> systems, when I take there my binary distribution, made on a
> MS-Windows 7 64-bit system (with the latest mingw, updated a few days
> ago).

Please run Emacs under GDB and post the backtrace.

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

>> > From "emacs -Q", simply type "C-g".
>>
>> For the record: I still experience this problem on MS-Windows XP
>> systems, when I take there my binary distribution, made on a
>> MS-Windows 7 64-bit system (with the latest mingw, updated a few days
>> ago).
>
> Please run Emacs under GDB and post the backtrace.

I've never done such thing.  Is the procedure documented somewhere?


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

>>> For the record: I still experience this problem on MS-Windows XP
>>> systems, when I take there my binary distribution, made on a
>>> MS-Windows 7 64-bit system (with the latest mingw, updated a few days
>>> ago).
>>
>> Please run Emacs under GDB and post the backtrace.
>
> I've never done such thing.  Is the procedure documented somewhere?

Ok, I've found a description in "nt/INSTALL".

Here's the backtrace:
----------------------------------------------------
(gdb) c
Continuing.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 140.0x90]
0x77c3554a in msvcrt!_abnormal_termination ()
   from C:\WINDOWS\system32\msvcrt.dll
(gdb) bt
#0  0x77c3554a in msvcrt!_abnormal_termination ()
   from C:\WINDOWS\system32\msvcrt.dll
#1  0x77c39bc6 in strerror () from C:\WINDOWS\system32\msvcrt.dll
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb)
----------------------------------------------------


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Juanma Barranquero]

On Thu, Nov 24, 2011 at 20:39, Dani Moncayo <dmoncayo@gmail.com> wrote:

>> I've never done such thing.  Is the procedure documented somewhere?
>
> Ok, I've found a description in "nt/INSTALL".

Try cd'ing to the src/ directory and doing

  gdb -ex run --args ..\bin\emacs.exe {rest of your arguments to call emacs}

preferably with a non-optimized build.

    Juanma

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

>>> I've never done such thing.  Is the procedure documented somewhere?
>>
>> Ok, I've found a description in "nt/INSTALL".
>
> Try cd'ing to the src/ directory and doing
>
>  gdb -ex run --args ..\bin\emacs.exe {rest of your arguments to call emacs}

This gives the same backtrace showed before.

> preferably with a non-optimized build.

I've just made an non-optimized build, and I'm unable to reproduce the
crash with it.

:(


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Juanma Barranquero]

On Thu, Nov 24, 2011 at 22:26, Dani Moncayo <dmoncayo@gmail.com> wrote:

> This gives the same backtrace showed before.

Yes, of course. It's just easier to run from a .bat script. I suggest
doing that and making sure you're in src/.

> I've just made an non-optimized build, and I'm unable to reproduce the
> crash with it.

Which compiler, and release, are you using?

    Juanma

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

> Which compiler, and release, are you using?

* gcc version 4.6.1
* mingw-get version 0.4-alpha-1


-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Juanma Barranquero]

> * gcc version 4.6.1
> * mingw-get version 0.4-alpha-1

When building with optimizations, are you passing the
-fno-omit-frame-pointer flag to GCC?

There was a discussion in emacs-devel about one month ago. Eli said:

   This thread:

    http://lists.gnu.org/archive/html/qemu-devel/2011-10/msg01697.html

   suggests that it probably is.  IIUC, there's an issue with setjmp in
   the MS runtime not working when -fomit-frame-pointer is used, and the
   MinGW GCC 4.6.1 turns on -fomit-frame-pointer by default when
   optimizing.  Since Emacs uses setjmp/longjmp _a_lot_, and in
   particular C-g throws to the top level via this mechanism, it seems
   like we have a smoking gun.

   Perhaps rebuild Emacs with -fno-omit-frame-pointer, and see if that
   helps.


    Juanma

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Eli Zaretskii]

> Date: Fri, 25 Nov 2011 00:32:54 +0100
> From: Dani Moncayo <dmoncayo@gmail.com>
> Cc: Eli Zaretskii <eliz@gnu.org>, 10092@debbugs.gnu.org
> 
> > Which compiler, and release, are you using?
> 
> * gcc version 4.6.1
> * mingw-get version 0.4-alpha-1

From etc/PROBLEMS:

  ** Emacs compiled with GCC 4.6.1 crashes on MS-Windows when C-g is pressed

  This is known to happen when Emacs is compiled with MinGW GCC 4.6.1
  with the -O2 option (which is the default in the Windows build).  The
  reason is a bug in MinGW GCC 4.6.1; to work around, either add the
  `-fno-omit-frame-pointer' switch to GCC or compile without
  optimizations (`--no-opt' switch to the configure.bat script).

If the build with -fno-omit-frame-pointer does not crash (please try
and report), we can close this bug.

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Dani Moncayo]

> From etc/PROBLEMS:
>
>  ** Emacs compiled with GCC 4.6.1 crashes on MS-Windows when C-g is pressed
>
>  This is known to happen when Emacs is compiled with MinGW GCC 4.6.1
>  with the -O2 option (which is the default in the Windows build).  The
>  reason is a bug in MinGW GCC 4.6.1;

I hope someone has reported the problem to the MinGW maintainers.

> to work around, either add the
>  `-fno-omit-frame-pointer' switch to GCC or compile without
>  optimizations (`--no-opt' switch to the configure.bat script).
>
> If the build with -fno-omit-frame-pointer does not crash (please try
> and report),

I've just tried that, and yes, passing that flag to GCC does the trick.

> we can close this bug.

This bug was already closed.

Thank you.

-- 
Dani Moncayo

bug#10092: 24.0.91; `C-g' makes Emacs to crash

[Eli Zaretskii]

> Date: Fri, 25 Nov 2011 17:03:08 +0100
> From: Dani Moncayo <dmoncayo@gmail.com>
> Cc: lekktu@gmail.com, 10092@debbugs.gnu.org
> 
> > From etc/PROBLEMS:
> >
> >  ** Emacs compiled with GCC 4.6.1 crashes on MS-Windows when C-g is pressed
> >
> >  This is known to happen when Emacs is compiled with MinGW GCC 4.6.1
> >  with the -O2 option (which is the default in the Windows build).  The
> >  reason is a bug in MinGW GCC 4.6.1;
> 
> I hope someone has reported the problem to the MinGW maintainers.

The URL that Juanma cited was a discussion of this bug by MinGW GCC
maintainers.  That's how I knew about the problem in the first place.