Guidance on how to collect debugging information for compositing issue in gnome-shell

[Alex Villacís Lasso]

El 07/08/13 09:10, Alex Villací­s Lasso escribió:
> El 03/08/13 21:38, Alex Villacís Lasso escribió:
>> El 28/07/13 19:56, Alex Villacís Lasso escribió:
>>> El 28/07/13 13:08, walter harms escribió:
>>>> 1. I have no clue but maybe some hints:
>>>>
>>>> * Do you get the failure also when you do nothing, just stating X ?
>>>> * Can you trigger that behavier at will or more at random ?
>>>> * Did you upgrade ? or is that the first installation ?
>>>>
>>> Currently I have no way to trigger the failure at will. This Fedora 
>>> 18 installation is clean, not an upgrade from a previous version. 
>>> However, it has been regularly updated to the most recent Fedora 18 
>>> packages.
>>> _______________________________________________
>>> xorg at lists.x.org: X.Org support
>>> Archives: http://lists.freedesktop.org/archives/xorg
>>> Info: http://lists.x.org/mailman/listinfo/xorg
>>> Your subscription address: a_villacis at palosanto.com
>> I might be entirely wrong, but the following behavior looks like a 
>> resource leak of some sort. It seems that Xv is the most efficient 
>> way to trigger the bug, but it *will* require quite a few tries. The 
>> bug still persists with xorg-x11-drv-intel-2.21.12-1.fc18.x86_64 . 
>> With the G33 chipset, you can try:
>>
>> 1) starting standard gnome desktop with gnome-shell and compositing
>> 2) find a program that invokes Xv, such as totem, mplayer, etc.
>> 3) find any video source playable by the chosen program
>> 4) run the chosen program with the chosen video for 10 seconds, then 
>> quit the program
>> 5) repeat steps 2 through 4 on the same gnome-shell session about 30 
>> times, or until the next application window/menu window appears as a 
>> black rectangle instead of valid data
>>
>> I do not know whether this can be triggered in anything other than 
>> the G33 chipset.
>>
>> The fact that the problem is solvable by restarting Xorg suggests 
>> that the leak is in userspace, not kernel space.
>> _______________________________________________
>> xorg at lists.x.org: X.Org support
>> Archives: http://lists.freedesktop.org/archives/xorg
>> Info: http://lists.x.org/mailman/listinfo/xorg
>> Your subscription address: a_villacis at palosanto.com
> Still happens with xorg-x11-drv-intel-2.21.12-1.fc18.x86_64 under 
> 3.11.0-rc4 .
> _______________________________________________
> xorg at lists.x.org: X.Org support
> Archives: http://lists.freedesktop.org/archives/xorg
> Info: http://lists.x.org/mailman/listinfo/xorg
> Your subscription address: a_villacis at palosanto.com
Still happens with 3.10.7-100.fc18.x86_64. I get this at the end of the 
Xorg.0.log if I attempt to use "r" to restart Gnome Shell using the 
Execute Command textbox invoked with Alt-F2. I also get the "Oh No!" 
black screen, and then I get sent back to the login screen.


[    38.192] (II) intel(0): Modeline "1280x1024"x0.0  108.00  1280 1328 
1440 1688  1024 1025 1028 1066 +hsync +vsync (64.0 kHz eP)
[    45.252] (II) intel(0): EDID vendor "LNX", prod id 0
[    45.252] (II) intel(0): Using hsync ranges from config file
[    45.252] (II) intel(0): Using vrefresh ranges from config file
[    45.252] (II) intel(0): Printing DDC gathered Modelines:
[    45.252] (II) intel(0): Modeline "1280x1024"x0.0  108.00  1280 1328 
1440 1688  1024 1025 1028 1066 +hsync +vsync (64.0 kHz eP)
[ 34459.661] (EE) intel(0): Failed to submit batch buffer, expect 
rendering corruption: Resource deadlock avoided.
  (II) evdev: ImPS/2 Generic Wheel Mouse: Close
[ 34521.333] (II) UnloadModule: "evdev"
[ 34521.352] (II) evdev: AT Translated Set 2 keyboard: Close
[ 34521.352] (II) UnloadModule: "evdev"
[ 34521.352] (II) evdev: Power Button: Close
[ 34521.352] (II) UnloadModule: "evdev"
[ 34521.352] (II) evdev: Power Button: Close
[ 34521.352] (II) UnloadModule: "evdev"
[ 34521.856] Server terminated successfully (0). Closing log file.