Xglamo stability and questions

[Dodji Seketeli]

On Mon, 31 Mar 2008 23:28:35 +0200,
Holger Freyther <zecke at openmoko.org> wrote :

> Hey Dodji,
> 
> I have pretty big stability issues with the Xglamo, Enlightenment and
> Qtopia combo I'm using. I can make something easily misbehave with
> normal input and I don't know what to blame.
> 
> Currently I see three cursors, illume is not painting itself, newly
> apps are not shown, the screen is not redrawn correctly. And I do not
> know what to blame. This can be either the window manager misbehaving
> or Xglamo.

Hmmh, that is really weird and I am not sure which path to follow.

Which window manager are you using ?
 
> 
> e/illume:
> (gdb) bt
> #0  0x4050f250 in poll () from /lib/libc.so.6
> #1  0x4032d8a0 in _XWaitForReadable () from /usr/lib/libX11.so.6
> #2  0x4032dc4c in _XRead () from /usr/lib/libX11.so.6
> #3  0x4032e634 in _XReply () from /usr/lib/libX11.so.6
> #4  0x4032332c in XQueryPointer () from /usr/lib/libX11.so.6
> #5  0x4004a7dc in from /usr/lib/libecore_x.so.0
> #6  0x00046eb4 in _e_pointer_cb_idle_timer_pre ()
> #7  0x4041fa00 in _ecore_timer_call () from /usr/lib/libecore.so.0
> #8  0x4041f9e4 in _ecore_timer_call () from /usr/lib/libecore.so.0
> #9  0x4041c8bc in _ecore_main_loop_iterate_internal () 
> from /usr/lib/libecore.so.0
> #10 0x4041ccd4 in ecore_main_loop_begin () from /usr/lib/libecore.so.0
> #11 0x000291f4 in main ()
> 
> 
> it looks like e is stuck there. Sadly gdb is a bit broken with EABI
> on ARM, so I don't know if my breakpoint in ecore_x_pointer_xy_get is
> not triggered. If gdb/arm/EABI is working allright then e is stuck in
> the XQueryPointer method. Hehe, just finished the sentence and then
> it stopped there.
> 
> 
> 
> 
> Xglamo:
> 	Should I update the Xglamo version and then try again? Is
> there something to take care of when updating?
> 
> 	With a gdb attached it looks to behave allright. This is a
> backtrace and it goes back to Dispatch and then is doing work again.
> 
> 0x4028328c in gettimeofday () from /lib/libc.so.6
> (gdb) bt
> #0  0x4028328c in gettimeofday () from /lib/libc.so.6
> #1  0x00014fd4 in GLAMOEngineWait ()
> #2  0x000156d0 in GLAMOWaitMarker ()
> #3  0x00043504 in kaaWaitSync ()
> #4  0x00015118 in GLAMODoneCopy ()
> #5  0x00044aec in kaaCopyNtoN ()
> #6  0x00056140 in fbCopyRegion ()
> #7  0x000564b0 in fbDoCopy ()
> #8  0x00043d0c in kaaCopyArea ()
> #9  0x000cc840 in damageCopyArea ()
> #10 0x0007a66c in miDCRestoreUnderCursor ()
> #11 0x000856d0 in miSpriteRemoveCursor ()
> #12 0x000862c8 in miSpriteSetCursor ()
> #13 0x0007fc90 in miPointerUpdate ()
> #14 0x0004a060 in ProcessInputEvents ()
> #15 0x0002e988 in Dispatch ()
> #16 0x00018d44 in main ()
> #17 0x4021ca00 in __libc_start_main () from /lib/libc.so.6
> #18 0x00012388 in _start ()
> 

Hmmh do you mean that when E is stuck in XQueryPointer, Xglamo is stuck
in a a gettimeofday ? or is busy looping over gettimeofday ? If yes,
then that is a problem ... What is happening in this backtrace of
Xglamo is that X is updating the "image" of the cursor. For doing so,
it is copying data over using the glamo bit blitting. If you are stuck
in there, then there is a problem we need to investigate.


Just in case, which window manager are you using ?

Cheers,

Dodji.

-- 
OpenedHand Ltd.
Unit R, Homesdale Business Center / 216 - 218 Homesdale Road /
Bromley / BR1 2QZ / United Kingdom.  Tel,fax: +44 (0) 208 819 6559

Expert Open Source For Consumer Devices - http://www.openedhand.com