Hi there,

great news, information-wise:

i just felt adventurous and downgraded xf86-video-intel back to 2.14.0
and there the issue didn't occur, at all.

So at least at my site, there must have happened something between
2.14.0 and 2.14.903, which is connected to the bug. I don't feel like
bisecting right now, but i'll think about it.

Is anyone, who is also hit by the bug able to confirm this?

Cheers,

Jonas

Denis Auroux <aur...@math.mit.edu> writes:

> Hi Jonas,
>
> Very interesting, thank you very much for the insightful message!
>
> Based on the packages updated, this seems to be an X server related
> issue indeed. I'd say it's either the server itself (there might have
> been changes in the order in which events/requests are processed,
> causing some deadlock in the processing either of pointer motion
> events or of display refresh requests), or the intel driver (because
> the bug only hits when rotated!). The wacom driver would have been
> another candidate normally, but since the bug has been reported by
> others to happen even when drawing with a mouse, it's not too likely.
>
> It's going to be very hard to debug though, in part because it seems
> that trying to produce debugging output will by itself cause the bug
> to disappear...
>
> Question: does the bug still appear if you run "xournal --sync"
> instead of the usual? (This makes X server calls synchronous, so in
> case the issue is a conflict between event processing and rendering
> requests, this might help...). If --sync fixes the bug, can you
> further assess whether adding the line
>
> gdk_display_sync(gdk_display_get_default());
>
> at the place where you inserted the printf() would be sufficient
> (perhaps with a lower performance cost than --sync, since it'd
> synchronize things only in the motion event processing code) ?
>
> Sorry I'm not testing myself, but given the obsolescence of my current
> distribution, I'd probably have to download an entire distribution
> image to do any testing (and might not even be hit by the bug if it's
> dependent on hardware or caused by patches to the X server...).
>
>  Denis
>
>
>
> On 05/09/2011 04:36 PM, Jonas wrote:
>> Hey,
>>
>> i've been hit by the same bug.
>>
>> I noted mostly the update of xf86-input-wacom from 0.10.8 to 0.10.9, but
>> there might be another candidate.
>>
>> I then also commented #define ENABLE_XINPUT_BUGFIX to no effect.
>>
>> But, when I then uncommented #define INPUT_DEBUG, the bug
>> disappeared.
>>
>> By playing around I narrowed it down to: As soon as there is something
>> which outputs a newline to stdout in on_canvas_motion_notify_event in
>> xo-callbacks.c, there are no further drawing delays.
>>
>> To be explicit: if I just apply the following patch, the bug does not
>> occur (although the side-effect of all those newlines is bothersome)
>>
>> diff -r1.29 xo-callbacks.c
>> 2699a2700,2701
>>>    printf("\n");
>>>
>>
>> I have no idea, why this happens; i'm just reciting what i found out.
>> But I hope, it helps you kill it for real. ;D
>>
>> All the best and thanks for xournal,
>>
>> Jonas
>>
>>
>>
>>
>>
>> [2011-04-13 15:25] upgraded audacity (1.3.12-3 ->  1.3.13-1)
>> [2011-04-13 15:25] upgraded bluedevil (1.0.3-1 ->  1.0.4-1)
>> [2011-04-13 15:25] upgraded crda (1.1.1-1 ->  1.1.1-3)
>> [2011-04-13 15:25] upgraded eventlog (0.2.12-1 ->  0.2.12-2)
>> [2011-04-13 15:25] upgraded filesystem (2010.12-1 ->  2011.04-1)
>> [2011-04-13 15:25] upgraded fixesproto (4.1.2-1 ->  5.0-1)
>> [2011-04-13 15:25] upgraded gstreamer0.10-good (0.10.27-2 ->  0.10.28-1)
>> [2011-04-13 15:25] upgraded gstreamer0.10-good-plugins (0.10.27-2 ->  
>> 0.10.28-1)
>> [2011-04-13 15:25] upgraded xz (5.0.1-1 ->  5.0.2-1)
>> [2011-04-13 15:25] upgraded imagemagick (6.6.9.1-1 ->  6.6.9.4-1)
>> [2011-04-13 15:25] upgraded libdrm (2.4.23-2 ->  2.4.25-1)
>> [2011-04-13 15:25] upgraded libgl (7.10.1-1 ->  7.10.2-2)
>> [2011-04-13 15:25] upgraded intel-dri (7.10.1-1 ->  7.10.2-2)
>> [2011-04-13 15:25] upgraded lftp (4.2.1-1 ->  4.2.2-1)
>> [2011-04-13 15:25] upgraded libtasn1 (2.8-1 ->  2.9-1)
>> [2011-04-13 15:25] upgraded libtiff (3.9.4-1 ->  3.9.5-1)
>> [2011-04-13 15:25] upgraded libxfixes (4.0.5-1 ->  5.0-1)
>> [2011-04-13 15:25] upgraded mesa (7.10.1-1 ->  7.10.2-2)
>> [2011-04-13 15:25] upgraded mlocate (0.23.1-2 ->  0.24-1)
>> [2011-04-13 15:25] upgraded namcap (2.8.2-1 ->  3.0.6-1)
>> [2011-04-13 15:25] upgraded ncurses (5.8-1 ->  5.9-1)
>> [2011-04-13 15:25] upgraded ngrep (1.45-5 ->  1.45-6)
>> [2011-04-13 15:25] upgraded sudo (1.8.0-4 ->  1.8.1-1)
>> [2011-04-13 15:25] upgraded vlc (1.1.8-2 ->  1.1.9-1)
>> [2011-04-13 15:25] upgraded xf86-input-evdev (2.6.0-2 ->  2.6.0-3)
>> [2011-04-13 15:25] upgraded xf86-input-synaptics (1.4.0-1 ->  1.4.0-2)
>> [2011-04-13 15:25] upgraded xf86-input-wacom (0.10.8-2 ->  0.10.9-1)
>> [2011-04-13 15:25] upgraded xf86-video-intel (2.14.0-3 ->  2.14.903-1)
>> [2011-04-13 15:25] upgraded xkeyboard-config (2.1-1 ->  2.2.1-1)
>> [2011-04-13 15:26] upgraded xorg-server-common (1.9.4.901-1 ->  1.10.0.902-1)
>> [2011-04-13 15:26] upgraded xorg-server (1.9.4.901-1 ->  1.10.0.902-1)
>> [2011-04-13 15:26] upgraded xsel (1.2.0-2 ->  1.2.0-3)
>> [2011-04-13 15:26] upgraded xvidcore (1.3.0-1 ->  1.3.1-1)
>>
>>>
>>> On 05/07/2011 10:58 AM, dmg wrote:
>>>> This might or might not be related.
>>>>
>>>> I installed Ubuntu 11.04 on an X61T and xournal in some cases the
>>>> strokes are not displayed
>>>> when created (but they are saved). Only when the page is refreshed.
>>>>
>>>> I didn't really look much into it, but it was odd, given that I use
>>>> in
>>>> Ubuntu 10.10.
>>>>
>>>> --dmg

------------------------------------------------------------------------------
Achieve unprecedented app performance and reliability
What every C/C++ and Fortran developer should know.
Learn how Intel has extended the reach of its next-generation tools
to help boost performance applications - inlcuding clusters.
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
Xournal-devel mailing list
Xournal-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xournal-devel

Reply via email to