WIN32 master crashes

[Michael Rickmann]

I just gave Ekiga master a try with yesterday's Ptlib and Opal. Result
is Ekiga does not start up any longer.
1) I get a crash from opal/src/sip/handlers.cxx:1656
m_byAOR.erase(handler->m_byAOR); - backtrace and d4 log are attached.

Ok, then I went back to Ekiga d6ff5f8bfcacbb4db 2009-09-04 plus "Cope
with opal's recent changes 2009-09-20" with the new Opal and Ptlib ().

2) Then I got an assertion error from
ptlib/src/ptlib/common/safecoll.cxx:81 when starting Ekiga without
debugger. When running under gdb I get the error described in 1).

This indicates that 1) is caused by a heap problem as Windows uses
special heap checking when a program is started from a debugger.
Michael


_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Michael Rickmann]

Just replying to myself: I think I have localized the culprit, Opal
commits 23554 (Fixed locating correct authorisation credentials for
INVITE when AOR user is different from authorisation ID. Big code tidy
up of SIP handler index maps) and 23559 (Fixed possible double deletion
if SIP handler indexes). For Win32 Ekiga it does not work. With Opal
23545 and Ptlib 23543 - I had to a bit further back to find a matching
pair, about 6 days ago - the crash does not happen.
Has anybody tried an up to date Opal trunk under Linux ?
Michael


Michael Rickmann schrieb:

> I just gave Ekiga master a try with yesterday's Ptlib and Opal. Result
> is Ekiga does not start up any longer.
> 1) I get a crash from opal/src/sip/handlers.cxx:1656
> m_byAOR.erase(handler->m_byAOR); - backtrace and d4 log are attached.
>
> Ok, then I went back to Ekiga d6ff5f8bfcacbb4db 2009-09-04 plus "Cope
> with opal's recent changes 2009-09-20" with the new Opal and Ptlib ().
>
> 2) Then I got an assertion error from
> ptlib/src/ptlib/common/safecoll.cxx:81 when starting Ekiga without
> debugger. When running under gdb I get the error described in 1).
>
> This indicates that 1) is caused by a heap problem as Windows uses
> special heap checking when a program is started from a debugger.
> Michael
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Ekiga-devel-list mailing list
> Ekiga-devel-list@...
> http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Damien Sandras]

Le dimanche 04 octobre 2009 à 12:15 +0200, Michael Rickmann a écrit :
> Just replying to myself: I think I have localized the culprit, Opal
> commits 23554 (Fixed locating correct authorisation credentials for
> INVITE when AOR user is different from authorisation ID. Big code tidy
> up of SIP handler index maps) and 23559 (Fixed possible double deletion
> if SIP handler indexes). For Win32 Ekiga it does not work. With Opal
> 23545 and Ptlib 23543 - I had to a bit further back to find a matching
> pair, about 6 days ago - the crash does not happen.
> Has anybody tried an up to date Opal trunk under Linux ?

I'm using an older trunk, but I will try.

Not sure I will be able to try today though, and on Wednesday I'm
leaving for 3 days.

--
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras@...
                       

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Damien Sandras]

Le jeudi 01 octobre 2009 à 10:33 +0200, Michael Rickmann a écrit :

> I just gave Ekiga master a try with yesterday's Ptlib and Opal. Result
> is Ekiga does not start up any longer.
> 1) I get a crash from opal/src/sip/handlers.cxx:1656
> m_byAOR.erase(handler->m_byAOR); - backtrace and d4 log are attached.
>
> Ok, then I went back to Ekiga d6ff5f8bfcacbb4db 2009-09-04 plus "Cope
> with opal's recent changes 2009-09-20" with the new Opal and Ptlib ().
>
> 2) Then I got an assertion error from
> ptlib/src/ptlib/common/safecoll.cxx:81 when starting Ekiga without
> debugger. When running under gdb I get the error described in 1).
>
> This indicates that 1) is caused by a heap problem as Windows uses
> special heap checking when a program is started from a debugger.
> Michael
> _______________________________________________
> Ekiga-devel-list mailing list
> Ekiga-devel-list@...
> http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

I get another crash :
Program received signal SIGSEGV, Segmentation fault.
0xb66f47c5 in std::string::assign(std::string const&) ()
from /usr/lib/libstdc++.so.6
(gdb) bt
#0  0xb66f47c5 in std::string::assign(std::string const&) ()
from /usr/lib/libstdc++.so.6
#1  0xb549edbf in std::string::operator= (this=0xbfffe8b4,
_search_filter=...) at /usr/include/c++/4.3/bits/basic_string.h:502
#2  Evolution::Book::set_search_filter (this=0xbfffe8b4,
_search_filter=...) at ../../plugins/evolution/evolution-book.cpp:290
#3  0x0824705d in on_updated (data=0x8560080)
at ../lib/engine/gui/gtk-frontend/book-view-gtk.cpp:240
#4  0x08193ed6 in boost::signal0<void, boost::last_value<void>, int,
std::less<int>, boost::function0<void> >::operator()() ()
#5  0xb549f481 in Evolution::Book::on_view_contacts_added
(this=0x8529bac, econtacts=0x0)
    at ../../plugins/evolution/evolution-book.cpp:80
#6  0xb549f514 in on_view_contacts_added_c (contacts=0x8688510,
data=0x8529b00) at ../../plugins/evolution/evolution-book.cpp:51
#7  0xb783ca4c in g_cclosure_marshal_VOID__POINTER ()
from /usr/lib/libgobject-2.0.so.0
#8  0xb782fe43 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#9  0xb7843e5f in ?? () from /usr/lib/libgobject-2.0.so.0
#10 0xb78452a9 in g_signal_emit_valist ()
from /usr/lib/libgobject-2.0.so.0
#11 0xb7845706 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#12 0xb54638f1 in ?? () from /usr/lib/libebook-1.2.so.9
#13 0xb783ca4c in g_cclosure_marshal_VOID__POINTER ()
from /usr/lib/libgobject-2.0.so.0
#14 0xb782fe43 in g_closure_invoke () from /usr/lib/libgobject-2.0.so.0
#15 0xb7843e5f in ?? () from /usr/lib/libgobject-2.0.so.0
#16 0xb78452a9 in g_signal_emit_valist ()
from /usr/lib/libgobject-2.0.so.0
#17 0xb7845706 in g_signal_emit () from /usr/lib/libgobject-2.0.so.0
#18 0xb546290e in ?? () from /usr/lib/libebook-1.2.so.9
#19 0xb77a70b1 in ?? () from /lib/libglib-2.0.so.0
#20 0xb77a8e98 in g_main_context_dispatch () from /lib/libglib-2.0.so.0
#21 0xb77ac623 in ?? () from /lib/libglib-2.0.so.0
#22 0xb77acaea in g_main_loop_run () from /lib/libglib-2.0.so.0
#23 0xb7ce7eb9 in IA__gtk_main () at /build/buildd-gtk
+2.0_2.18.1-1-i386-d2TqLS/gtk+2.0-2.18.1/gtk/gtkmain.c:1205
#24 0x0813da09 in main (argc=0, argv=0x0) at gui/main.cpp:4554


--
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras@...
                       

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Damien Sandras]

Le dimanche 04 octobre 2009 à 12:15 +0200, Michael Rickmann a écrit :
> Just replying to myself: I think I have localized the culprit, Opal
> commits 23554 (Fixed locating correct authorisation credentials for
> INVITE when AOR user is different from authorisation ID. Big code tidy
> up of SIP handler index maps) and 23559 (Fixed possible double deletion
> if SIP handler indexes). For Win32 Ekiga it does not work. With Opal
> 23545 and Ptlib 23543 - I had to a bit further back to find a matching
> pair, about 6 days ago - the crash does not happen.
> Has anybody tried an up to date Opal trunk under Linux ?

I have forwarded to Robert as I get another crash.

--
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras@...
                       

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Eugen Dedu]

Michael Rickmann wrote:
> Just replying to myself: I think I have localized the culprit, Opal
> commits 23554 (Fixed locating correct authorisation credentials for
> INVITE when AOR user is different from authorisation ID. Big code tidy
> up of SIP handler index maps) and 23559 (Fixed possible double deletion
> if SIP handler indexes). For Win32 Ekiga it does not work. With Opal
> 23545 and Ptlib 23543 - I had to a bit further back to find a matching
> pair, about 6 days ago - the crash does not happen.
> Has anybody tried an up to date Opal trunk under Linux ?

Hi,

I have not yet resumed and tested debian snapshots, I am stuck with an
EchoCanceler link error in simpleopal (and very probably ekiga too).  As
the time passes, it seems better that I add a workaround for this error,
I will probably get a try this evening.

Eugen
_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Damien Sandras]

Le dimanche 04 octobre 2009 à 14:07 +0200, Damien Sandras a écrit :
> Le jeudi 01 octobre 2009 à 10:33 +0200, Michael Rickmann a écrit :
> > I just gave Ekiga master a try with yesterday's Ptlib and Opal. Result
> > is Ekiga does not start up any longer.
> > 1) I get a crash from opal/src/sip/handlers.cxx:1656
> > m_byAOR.erase(handler->m_byAOR); - backtrace and d4 log are attached.
> >
> > Ok, then I went back to Ekiga d6ff5f8bfcacbb4db 2009-09-04 plus "Cope
> > with opal's recent changes 2009-09-20" with the new Opal and Ptlib ().
> >

[...]

> I get another crash :
> Program received signal SIGSEGV, Segmentation fault.
> 0xb66f47c5 in std::string::assign(std::string const&) ()
> from /usr/lib/libstdc++.so.6
> (gdb) bt
> #0  0xb66f47c5 in std::string::assign(std::string const&) ()
> from /usr/lib/libstdc++.so.6
> #1  0xb549edbf in std::string::operator= (this=0xbfffe8b4,
> _search_filter=...) at /usr/include/c++/4.3/bits/basic_string.h:502
> #2  Evolution::Book::set_search_filter (this=0xbfffe8b4,
> _search_filter=...) at ../../plugins/evolution/evolution-book.cpp:290
> #3  0x0824705d in on_updated (data=0x8560080)

It was a false alarm due to a bug in the compilation process.

Michael, I can not reproduce your crash on Linux. However, I have
forwarded the backtrace and information to Robert. I got no answer yet.

--
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras@...
                       

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list

[Damien Sandras]

Le lundi 05 octobre 2009 à 19:04 +0200, Damien Sandras a écrit :

> Le dimanche 04 octobre 2009 à 14:07 +0200, Damien Sandras a écrit :
> > Le jeudi 01 octobre 2009 à 10:33 +0200, Michael Rickmann a écrit :
> > > I just gave Ekiga master a try with yesterday's Ptlib and Opal. Result
> > > is Ekiga does not start up any longer.
> > > 1) I get a crash from opal/src/sip/handlers.cxx:1656
> > > m_byAOR.erase(handler->m_byAOR); - backtrace and d4 log are attached.
> > >
> > > Ok, then I went back to Ekiga d6ff5f8bfcacbb4db 2009-09-04 plus "Cope
> > > with opal's recent changes 2009-09-20" with the new Opal and Ptlib ().
> > >
>
> [...]
>

Another problem is taht when we are in a call, we never reach the
"connected" state anymore.

Trunk is in good shape ;-))

--
 _     Damien Sandras
(o-      
//\    Ekiga Softphone : http://www.ekiga.org/
v_/_   Be IP           : http://www.beip.be/
       FOSDEM          : http://www.fosdem.org/
       SIP Phone       : sip:dsandras@...
                       

_______________________________________________
Ekiga-devel-list mailing list
Ekiga-devel-list@...
http://mail.gnome.org/mailman/listinfo/ekiga-devel-list