Bug 93577 – valgrind detects illegal read access during Gtk::Main destruction

After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.

Bug 93577 - valgrind detects illegal read access during Gtk::Main destruction


Summary:	valgrind detects illegal read access during Gtk::Main destruction


Status:	RESOLVED NOTABUG

Product:	gtkmm
Classification:	Bindings
Component:	general
Version:	2.0
Hardware:	Other other

Importance:	Normal normal
Target Milestone:	---
Assigned To:	gtkmm-forge
QA Contact:	gtkmm-forge

URL:
Whiteboard:

Depends on:
Blocks:

Reported:	2002-09-18 15:33 UTC by Tassos Bassoukos
Modified:	2004-12-22 21:47 UTC

See Also:
GNOME target:	---
GNOME version:	---

Description Tassos Bassoukos 2002-09-18 15:33:28 UTC

program:

#include <gtkmm.h>

int main(int argc, char ** argv){
        Gtk::Main kit(argc,argv);
        Gtk::Window* win=new Gtk::Window();
        kit.run(*win);
        return 0;
}

Compiles correctly, runs correctly (pops up window, exits
when window is closed)

When running under valgrind, an invalid read is detected.
GDB backtrace:

+ Trace 27458

#0 vg_do_syscall3
at vg_mylibc.c line 91
#1 ??
#2 SigC::Control_::destroy
from /usr/lib/libsigc-1.1.so.5
#3 SigC::ObjectBase::~ObjectBase
from /usr/lib/libsigc-1.1.so.5
#4 Gtk::Main::~Main
from /usr/lib/libgtkmm-1.3.so.16
#5 main
at other_test.cc line 7


This happens (obviously) after pressing the close button.
However, the program exits normally.

Tassos

Comment 1 Murray Cumming 2002-09-18 15:39:08 UTC

Confirmed on RH7.2 with gcc 2.96.

Comment 2 Murray Cumming 2002-09-19 09:01:29 UTC

Here's some fuller information, using --num-callers=20. This shows
what freed the memory that is being incorrectly read again later.

==25837== Invalid read of size 4
==25837==    at 0x404F7E85: SigC::Control_::destroy(void) (object.cc:72)
==25837==    by 0x404F8020: SigC::ObjectBase::~ObjectBase(void)
(object.cc:117)
==25837==    by 0x4044BD2C: Gtk::Main::~Main(void) (main.cc:360)
==25837==    by 0x8048CAC: main (main.cc:7)
==25837==    by 0x40938507: __libc_start_main
(../sysdeps/generic/libc-start.c:129)
==25837==    by 0x8048B81: Gtk::Main::Main(int, int &, char **&, bool)
(in
/gnome/gnome-2-0/cvs/c++/gtkmm-1.3/tests/delete_cpp_child/.libs/lt-test)
==25837==    by <bogus frame pointer> ???
==25837==    by <bogus frame pointer> ???
==25837==    Address 0x41F8BD3C is 36 bytes inside a block of size 44
free'd
==25837==    at 0x40040729: __builtin_delete (vg_clientfuncs.c:186)
==25837==    by 0x404F7C6E:
SigC::ObjectSlotNode::~ObjectSlotNode(void) (object_slot.cc:31)
==25837==    by 0x404F7077: SigC::Node::clear(void) const (node.cc:47)
==25837==    by 0x404F7290: SigC::SlotNode::notify(bool) (slot.cc:78)
==25837==    by 0x404F7C0F: SigC::ObjectSlotNode::notify(bool)
(object_slot.cc:24)
==25837==    by 0x404F7E85: SigC::Control_::destroy(void) (object.cc:72)
==25837==    by 0x404F8020: SigC::ObjectBase::~ObjectBase(void)
(object.cc:117)
==25837==    by 0x4044BD2C: Gtk::Main::~Main(void) (main.cc:360)
==25837==    by 0x8048CAC: main (main.cc:7)
==25837==    by 0x40938507: __libc_start_main
(../sysdeps/generic/libc-start.c:129)
==25837==    by 0x8048B81: Gtk::Main::Main(int, int &, char **&, bool)
(in
/gnome/gnome-2-0/cvs/c++/gtkmm-1.3/tests/delete_cpp_child/.libs/lt-test)
==25837==    by <bogus frame pointer> ???
==25837==    by <bogus frame pointer> ???

Comment 3 Murray Cumming 2002-09-19 09:12:33 UTC

This doesn't happen if you delete the Gtk::Window before the end of
main(). And you should do that. So I don't think this is a bug.

Comment 4 Murray Cumming 2002-09-19 17:01:13 UTC

Do you agree?

Comment 5 Tassos Bassoukos 2002-09-19 17:29:55 UTC

<ponders>... Ok.