On Sat, 18 Feb 2012 23:31:28 +0100 Martin Jansa <martin.ja...@gmail.com> said:
> On Wed, Feb 15, 2012 at 07:39:24PM +0100, Martin Jansa wrote:
> > On Tue, Feb 14, 2012 at 04:15:28PM +0100, hannes.janet...@gmail.com wrote:
> > > On Tue, Feb 14, 2012 at 12:47 PM, Carsten Haitzler <ras...@rasterman.com>
> > > wrote:
> > > > On Mon, 6 Feb 2012 22:20:04 +0100 "hannes.janet...@gmail.com"
> > > > <hannes.janet...@googlemail.com> said:
> > > >
> > > > ok - i'm stumped. it's doing soem bizarre list logic where it keeps
> > > > taking things off the rect list and putting them back again, taking
> > > > them off, putting them back, and finding out why is hard due it it
> > > > taking a long time to trigger and then trying to figure out the logic
> > > > hole then by stepping is painful. i'm going to have to leave this bug
> > > > and get onto other things. :(
> > > >
> > >
> > > sorry for having made this much trouble. had this bug lying around for
> > > a while and thought to at least mention it. i think i've seen cases
> > > where 'rects' list has loops if that helps... anyway no hurry for me
> > > with it.
> >
> > I'm not sure if it's caused by this issue or not, but one simple script
> > we had started to cause OOM killer going after enlightenment process.
> >
> > After I have stripped all parts of that script which are not needed to
> > provoke OOM killer it's pretty simple:
> >
> > SHR root@gjama ~ $ cat oom.py
> > #!/usr/bin/env python
> >
> > import elementary
> >
> > print "Start"
> >
> > elementary.init()
> >
> > win = elementary.Window('oom',0);
> > bg = elementary.Background(win)
> > bg.show()
> > win.title_set('OOM')
> > win.show()
> > win.resize_object_add(bg)
> > win.fullscreen_set(True)
> >
> > elementary.run()
> > elementary.shutdown()
> > exit()
> >
> > First I've seen this with r67883, and maybe 100 revisions back it was still
> > working.
>
> I can still reproduce it with r68118 and the key point is
> elm_win_fullscreen_set(win, EINA_TRUE);
> without fullscreen it works (well does nothing as expected).
>
> Debuging testcase doesn't show clearly which instruction caused it,
> because enlightenment process (not the testcase) is causing OOM. And
> when trying to debug enlightenment process I wasn't able to get
> reasonable bt.. trying to bisect now as this is relatively new.
>
> Here is simple testcase without python:
>
> #include <eina_types.h>
> #include <Evas.h>
> #include <Ecore_X.h>
> #include <Edje.h>
> #include <elm_general.h>
> #include <elm_object.h>
> #include <elm_win.h>
>
> int
> elm_main(int argc, char *argv[])
> {
> Evas_Object *win;
>
> win = elm_win_add(NULL, "win-example", ELM_WIN_BASIC);
> elm_win_title_set(win, "Elm_Win Example");
> elm_win_fullscreen_set(win, EINA_TRUE);
> evas_object_show(win);
>
> elm_run();
>
> return 0;
> }
> ELM_MAIN();
i dont think this is an issue - some window goes fullscreen. fine - i have that
all the time on e and it doesnt go eating memory - WHY is your oom killer
wanting to kill e? and if it ACTUALLY is using lots of memory - what is it
using at that time it's killed? (instead of killing it, find out where all its
allocations have gone). i dont see anything odd here :/
--
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler) ras...@rasterman.com
------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/
_______________________________________________
enlightenment-devel mailing list
enlightenment-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
