kde4 KDM does not restart X when killed with ctrl+alt+backspace

Hello Alan.

Actually, if I am not mistaken, the key combo itself is for restarting X, not killing it. That would also explain why it works on all of my machines. I would assume there is some flaw in your xorg.conf or (which is more likely) the graphics driver, preventing X from restarting properly.
I am moving this bug to the xorg package for further investigation.

Thank you for this report, and have a nice day :)

Harald Sitter wrote:
> Hello Alan.
>
> Actually, if I am not mistaken, the key combo itself is for restarting X, not killing it. That would also explain why it works on all of my machines. I would assume there is some flaw in your xorg.conf or (which is more likely) the graphics driver, preventing X from restarting properly.
> I am moving this bug to the xorg package for further investigation.
>

I think you are mistaken.

Try it without a login manager. Just run "X :1", either from a terminal
or a console. The 1 assumes you're only using display :0; increment it
if necessary. That should start X with no clients at all, and the
eye-watering default background. Press ctrl+alt+backspace and it will
go away - it will not restart.

Thanks for looking at this!
Alan

No, the key combo is for killing X. The display manager should handle the respawning.

Okies.

Then I can only imagine kdm crashes while trying to restart. Alan, please attach the file /var/log/kdm.log (or /var/log/kdm.log.1 if you do it from the restarted session).

Thanks in advance.

Here. Sorry for the delay. Also, I corrupted my 8.10 install :-). So this logfile is from a Jaunty pre-release which also exhibited the bug.

(Note: The version of X in Jaunty had "DontZap" set to "true" by default. If you want to use Ctrl+Alt+Backspace to try and recover from a bad X session, you have to edit xorg.conf.)

Hi.

Checking the log indicates that KDM properly tries to restart X, but X startup fails with error
error setting MTRR (base = 0xd0000000, size = 0x10000000, type = 1) Invalid argument (22)
 ddxSigGiveUp: Closing log

Bouncing the bug back to xorg again.

Have a nice day :)

Hmm. Somehow I managed to miss all the X errors :-).

I'm not sure about your logic though. The logfile seems to contain output for 3 different runs of the X server. If you're right, we are interested in the last two runs. But *all* of them end in this same X error.

The *difference* between the last two runs is the error from kdmgreet in the second run:

kdmgreet: Fatal IO error: client killed
error setting MTRR (base = 0xd0000000, size = 0x10000000, type = 1) Invalid argument (22)
 ddxSigGiveUp: Closing log

Surely that kdmgreet error is what you would expect to happen just after ctrl+alt+backspace. So it's not restarting X?

The other alternative is that kdmgreet is silent in response to ctrl+alt+backspace, but when it tried to restart X, something *other* than the MTRR error went wrong, which X did not log an error for, and kdmgreet then died.

Also, as far as I know it's not a fatal error to fail to setup an MTRR. It just slows things down :-). So I think the MTRR error is unlikely to be related to a crash.

Finally, X crashes are normally a bit spectacular - you at least see the screen flicker or something. I didn't see anything like that, it *looks* as if X just dies and is not restarted.

If it is an X crash, you need to collect a full backtrace - see http://wiki.ubuntu.com/X/Backtracing for directions.

Probably ought to update the title and description of this bug report too to remove mention of CAB since that seems irrelevant to the issue.

We're closing this bug since it is has been some time with no response from the original reporter. However, if the issue still exists please feel free to reopen with the requested information. Also, if you could, please test against the latest development version of Ubuntu, since this confirms the bug is one we may be able to pass upstream for help.