Hello!

Henry Gebhardt has written on Sunday, 25 August, at 19:51:
>On Wed, Aug 21, 2013 at 12:07:30PM +0200, Jörg-Volker Peetz wrote:
>> Andrej N. Gritsenko wrote, on 08/19/2013 20:39:
>> > Wolfgang Scheicher has written on Monday, 19 August, at 19:17:

>> >> Trying to provide additional and useful info to the problem,
>> >> which is probably the same as
>> >> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=720109 

>> I submitted this report.
>> I'm using lxpanel together with openbox (without lxsession).

>> Meanwhile I tried also to deliver some additional information by starting
>> lxpanel with ltrace. The last few lines before terminating the frozen panel 
>> with
>> Ctrl-C are:

>> ...
>> gdk_threads_enter(0x7fde6c195640, 3, 0x7fde6c195660, 0x1e457a0, 0 
>> <unfinished ...>

>Seems like gdk_threads_enter() is called twice. Does Debian patch
>gtk_main() to call gdk_threads_enter()?

    No, it doesn't. I'm afraid it may be something else, I've stepped into
something similar when I wrote new lxshortcut. You might have some init
callback which does gdk_threads_enter() and that callback blocks gtk_main
from entering main loop because it never returns. I don't know how it is
done in lxpanel but may be gdk_threads_enter() was missed in lxpanel's
main() on purpose? I have left lxshortcut without that call on purpose.
And I don't think lxpanel ever would need GDK threading support, I don't
see many windows updated simultaneously in lxpanel so I would suggest to
remove gdk_threads_init() and all that jazz. :) I think that would make
it even faster because no locking and context switching will be involved.
But as you wish, of course. (I mean only gdk_threads_init() removal, any
gdk_threads_enter/gdk_threads_leave will just do nothing then)

>Also, on the bug you say:
>> This was on another system with a single CPU. My main system has two cores.

>Does this mean it occurs on single-core machines only?

    I think on multi-core CPU some init function may be fast enough to
slip critical section already. What again confirms the theory above may
be right.

>Thanks,
>Henry

    Cheers!
    Andriy.

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
Lxde-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/lxde-list

Reply via email to