Bug 202254

Created attachment 35779 [details]
The screen before the error

Created attachment 35780 [details]
The screen after the error

also, i apologize for the poor image quality, i had to use my cell phone camera for this.

The error happens earlier and more often when i have an flash-based website open in firefox or konqueror.
Sometimes, i can use ctrl+alt+backspace to restart the x screen, but most times that does not work and i have to resort to alt+print+(s,u,b) to reboot the entire system. And there were a few times when even that did not work :(

Err. this is probably a graphics drivers bug. Let's reassing (unrelated to Plasma) and ask KWin guys.
Are you using compositing/effects ? What is your drivers version? Is it updated?
Regards

assumption 1: your external monitor has a different resolution than the NB dev?
assumption 2: you're using compositing
assumption 3: you're running kwin, not compiz

-> press alt+shift+f12 (to suspend compositing) -> switch your display -> press alt+shift+f12 again (to resume compositing)

works?

@Dario Andres:
   compositing / kwin effects: yes
   driver version: 180.44-0ubuntu1 (which is the driver version recommended by envyng)
   updated: uhm is assume you mean if its the newest version available? it is, at least via envy

@Thomas Lübking:
   assumption 1: nope, both run at 1900x1200
   assumption 2: yes
   assumption 3: yes
   solution: sory, once the screen is garbled up like that, it won't respond to anything (mostly). On rare occasions i have been able to salvage the session by switching to a console, killing plasma and kwin, switching back and restarting them. But in most cases all the keyboard shortcuts that are caught by the X Server do not work anymore, this goes for alt+shift+f12 as well.

Thanks for your quick responses :)

i meant to suspend /before/ you mess up your system (by switching displays) and resume /after/ (in case it didn't mess - due compositing being disabled)

180.44 is however pretty outdated - i know it's not great regarding a clean package system, but could you please try to install the latest (stable+beta) drivers from here
http://www.nvidia.com/object/unix.html
and here (beta)
http://www.nvidia.com/Download/Find.aspx?lang=en-us

note: it appears nVidia has pulled back the /current/ ia32 beta driver for whatever issue (at least the link is dead)
a) that makes it a bad long-term solution ;-)
b) direct link:
http://us.download.nvidia.com/XFree86/Linux-x86/190.18/NVIDIA-Linux-x86-190.18-pkg1.run

Hi,

  i can try that, but it will probably take a while. The point is, that the screen is not messed up like that "immediately" sometimes it takes just a few minutes, the next time it may be 3-4 hours before something happens. I even had entire sessions on the external monitor where nothing happened at all.

@Dominik: did you try even with other desktop enviroments (and window managers, of course) ?

so usually you can operate on the ext. monitor for a while?
(i.e. it's not /directly/ related to the switch. in case, please confirm so the but get get marked "upstream")

in this case i'd blame nVidia (or you for using outdated drivers ;-)

to trace this issue you should look into /valr/log/Xorg.0.log (esp. for EE)
can you ssh the broken machine for "in-crash" inspection?

note the def. approach is alt+sysreq+r/e/i/s/u/b (not just s/u/b) and can leave use with a running system just w/o X11 (also you should wait a few seconds between the sysreq's)

Hi,

  i upgraded to the 185 driver (i had to use the amd64 version for some reason) and the error has not appeared since then. I might be celebrating too soon, but i think i'm in the clear :)
  thanks a lot for all your suggestions :)

Dominik