Re: [Om-synth] Double free or corruption crashes

[Benjamin Flaming]

On Friday 14 July 2006 12:18 pm, Dave Robillard wrote:
> On Fri, 2006-07-14 at 01:12 -0500, Benjamin Flaming wrote:
> > On Thursday 13 July 2006 10:52 pm, Dave Robillard wrote:
> > > On Thu, 2006-07-13 at 12:40 -0500, Benjamin Flaming wrote:
> > > > *** glibc detected *** om_gtk: double free or corruption (out):
> > > > 0xb70f91c8
> > >
> > > IIRC this is caused by an old version of liblo (though the "how" part
> > > is beyond me).
> > >
> I remember this happening to someone, but I don't remember why.  Pretty
> sure it was some kind of liblo version thing.  Maybe try 0.22?  That's
> what I happen to have on my laptop at the moment...  That
> lo_send_message_from function that was in the backtrace is new as of
> 0.23

Thanks for the help.  Building and running against liblo 0.22 seems to move 
things a bit further at least.  The Om engine starts as before.  The 
interface then outputs the following:

------------------------------------------------------------
[GladeFactory] Loading widgets from /home/benji/32bit/share/om/om_gtk.glade
[Configuration] Unable to open settings file /home/benji/.omgtkrc
lash_open_socket: could not connect to host 'localhost', service 'lash'
lash_comm_connect_to_server: could not create server connection
lash_init: could not connect to server 'localhost' - disabling LASH
lash_init: LASH_START_SERVER unset, not attempting to start server 
automatically
[LashController] Failed to connect to LASH.  Session management will not 
function.
[OSCController] Started OSC listener on port 19493
------------------------------------------------------------

At this point, it pops up a "Connecting to Engine" window with the 
text "Connecting to engine at ...".  After that, the interface repeats the 
following text indefinitely, once per second:

------------------------------------------------------------
[OSCController] Attaching to engine at osc.udp://bdesk:16180/
------------------------------------------------------------

If I press the "Launch Engine" button, I get the following output (each time 
the button is pressed):

------------------------------------------------------------
Launching engine..Xlib: unexpected async reply (sequence 0x509)!
------------------------------------------------------------

The "attaching to engine" messages continue uninterrupted by the Xlib 
message.  After receiving one of the Xlib messages, however, the 
window "sticks" when I try to close the interface (either via the "Cancel" 
button or via Ctrl-C).  Control is given back to bash, but the window stays 
open (without repainting itself) until I "killall om_gtk".  This doesn't 
happen if I don't press the "Launch Engine" button.

I'm still going to try linking against an SVN build of liblo (and try both 
liblo versions with the last official Om release), but let me know if this 
bug report triggers any other ideas for avenues of exploration.

|)
|)enji