Created attachment 15245 [details]
Xorg log when failing with driver 2.2.1
On a Toshiba Libretto U105 (82852/855GM), using an external screen (the LCD is bust), driver 2.2.1 locks up (with a text cursor and parasite dots) at startup. The console is unusable, but if you ssh into the machine, you can kill X and have your console back. However, the device is left in a bad state, and even driver 2.1.1 crashes similarly if tried afterwards. On reboot, driver 2.1.1 happily works (after other woes, I've got into the habit of backing up my known working display drivers).
xorg.conf and Xorg.log.0 attached.
I can provide on request Xorg.log for working 2.1.1 driver or failing 2.1.1 driver (after 2.2.1 messed up with the device state), or any other information.
Machine is running Debian i386 testing. Relevant package is xserver-xorg-video-intel 2:2.2.1-1.
After a quick search through the existing bugs, I find that this may or may not be related to bug 13376.
Created attachment 15246 [details]
NB: Linux kernel 126.96.36.199, using intelfb. Nothing significant in dmesg.
I'm available to give more information on request, although I may be slow as this is not the highest priority to me since I have a working fallback and this isn't my main machine.
please refer to http://www.intellinuxgraphics.org/how_to_report_bug.html.
Would you please get rid of intelfb kernel module and retest?
Francois-Rene, the intelfb driver doesn't work well in many situations, can you reproduce your problem without it? If you need the framebuffer interface for some reason, you can also try vesafb, but let's try to narrow down the problem w/o any fb drivers loaded first.
Unless you can tell me a way to disable intelfb at the kernel commandline (I didn't find such a way), I will have to recompile a new custom kernel for that machine (a generic kernel won't do because I need special crypto and wifi drivers).
I will do that eventually and recontact you with the requested info, but that may take time.
Thanks for your support. In the meantime, I'll stick to driver 2.1.1.
Yeah, I think you'll have to rebuild the kernel, I don't know of an option to disable things at boot time. Just get us the info when you get a chance, and we can continue debugging then. Thanks.
Francois-Rene Rideau, are you able to give us an update of your testing now? thanks.
ping the reporter
Now works again.
I failed to install an upgraded kernel (because the loop-aes support is a pain to configure) and so couldn't try without the framebuffer support. However, after many driver upgrades and many reboots, it now seems to work with intel_drv 2.3.1.
Thanks to all of you, and particularly to whoever fixed that bug.