Comment 73 for bug 702090

Bryce Harrington (bryce) wrote :

Hello Travis and everyone reporting dupes!

A possible fix for this bug has been created and is available for testing. Please follow these steps to test:

A. Verify you still have the bug. Run this command:

  $ dmesg | egrep "drm.*ERROR"

This should display something like this:

  [ 10.625985] [drm:i915_report_and_clear_eir] *ERROR* EIR stuck: 0x00000010, masking

This indicates you've had a GPU lockup. "masking" means that the kernel was able to reset the GPU and keep going. So you probably never noticed any problems (except the apport error reporter dialog).

If you don't see anything, try rebooting a few times and repeating the command. If you never see the error message after several boots, you can stop testing here. No need to provide feedback.

B. Update your system to absolute latest

 $ sudo apt-get update
 $ sudo apt-get upgrade
 $ sudo apt-get dist-upgrade

If you encounter any problems with this step, DO NOT comment on this bug (else you will spam a lot of people on dupes). Resolve the issues before proceeding.

C. Please install the following PPA with the proposed fix:

$ sudo apt-add-repository ppa:apw/purple
$ sudo apt-get update
$ sudo apt-get upgrade

This should pull in new gdm, linux, plymouth, and udev and you should see "~vesafb1*" in their version strings. It also pulls in an update to module-init-tools.

D. Next, reboot your machine. Verify you have the correct kernel version booted:

$ dmesg | grep 'Linux version'

This should indicate you're running "(Ubuntu 2.6.38-6.34vesafb1*"

E. Verify the kernel has not encountered the GPU hang:

  $ dmesg | egrep "drm.*ERROR"

Ideally, no output should be printed.

Please repeat steps D and E several times, just in case your GPU hang is intermittent.

Finally, post a comment here to this bug. If you do not ever see output in step E, simply say so and indicate the number of boot attempts you tried. If you do see output in step E, please provide the error messages.