Bugzilla – Bug 301101
intel: Xserver crashing on MacBook on shutdown
Last modified: 2007-10-21 19:27:11 UTC
Created attachment 157998 [details] yellow screen For the last ten days or so my system is crashing on every third logout or so. The screen goes yellow with a really mangled display as it can be seen in attached screenshots. Then the screen stays yellow for 2-3 seconds and the machine reboots. When disabling /etc/init.d/boot.videobios this seems to hapen on every switch to tty1.
Created attachment 157999 [details] my xorg.conf Here is the xorg.conf. # rpm -q kernel-default xorg-x11-driver-video kernel-default-2.6.22.2-5 xorg-x11-driver-video-7.2-152
Created attachment 158000 [details] hwinfo --gfxcard This is my graphicscard.
Please attach the /var/log/Xorg.0.log and /var/log/Xorg.0.log.old after it happened.
Created attachment 158007 [details] Xorg.0.log.old It is not contain "I830WaitLpRing crash" as I expected.
Created attachment 158009 [details] Xorg.0.log here it is.
Right. There's nothing obviously wrong in the logfiles. Joachim, do we have such a Macbook with 945GM chipset available?
Timo Hoenig had one he did this years Linuxtag presentation about suspending on a MacBook.
Timo, you still have it?
Yes, you're welcome to grab it for further investigation.
Thanks.
I grabbed the Macbook and finally managed to boot from CD. Unfortunately the keyboard does not work. Therefore I could not select "Installation" to install a system on it. Obviously there is no Linux system on it yet. This machine is simply unusable and therefore useless.
Joachim, do we have a different 945GM based laptop for testing available?
(In reply to comment #12 from Stefan Dirsch) > This machine is simply unusable and therefore useless. Is this a hardware defect? I never had any problems with the keyboard on that machine. Which CD did you try? I could see whether it boots here. I installed mine with alpha3 and have since then been following factory.
Tried "10.3 ALpha7 x86_64 DVD" and "SLED10 CD". Doesn't matter if I do a soft boot from MacOS or hard boot. Keyboard doesn't work. Tried about 10 times. Sorry, but these Macbooks are just crap.
I just downloaded openSUSE-10.3-Beta1-KDE-i386.iso and the installation started with a perfectly working keyboard. I have always been using 32bit openSUSE on this machine.
I used the "32bit SLED10 CD". I even don't know if this is a 32 or 64bit machine. I don't think it's possible to figure this out with MacOS X.
Ok; I downloaded the 64 bit beta1 too and got the following: "This is the 32-bit computer. You cannot use 64-bit software on it." After some googling I found out: The MacBook of early 2006 is 32 bit with Core Duo and the new one should be capable of 64 bit too. BUT BootCamp does not support 64 bit yet. Stefan: To install openSUSE on it you would need BootCamp and refit.
> Stefan: To install openSUSE on it you would need BootCamp and refit. I have no idea what you're talking of. :-(
(In reply to comment #19 from Stefan Dirsch) > > Stefan: To install openSUSE on it you would need BootCamp and refit. > I have no idea what you're talking of. :-( You might want to have a look at: http://en.opensuse.org/Installation_on_MacBook
It seems to be better at the moment. The last days (maybe 10 reboots) my MacBook has not crashed will report back. So maybe one of ... 2007-08-14 - sndirsch@suse.de - updated to intel bugfix driver release 2.1.1 * Fix typo in intel.man * Allow 965 composite acceleration to A8 destinations. * i965: fix memcpy of the sf_kernel when a mask is needed * i965: increase composite vertex buffer size and alignment to be safe * Fix some physical address handling for >4GB addresses. * Quirk away the nonexistent TV connector on the Panasonic CF-Y4. * Add the file mode for bios_dumper output so it doesn't have 000 permissions. * Bug #11593: Remove dead struct vch_bdb_20 which was angering the sun compiler. * Legacy backlight changes: - add support for 965GM - make sure legacy enabled systems don't reduce the range of backlight values we can present to the user * Eliminate bogus (and harmful) blanking adjustment for load detect. * Ensure pipe/output active before doing load detection. * Remove hard-coded CRT blanking frobbing for load detection. * Clean up tv mode name allocation and copy. * Add quirk support * Add another Lenovo TV output quirk * Fix device id info for 945GME, 965GME * Update Lenovo TV quirk info ... helped me.
Ok. Then let's close it as fixed for now. Don't hesitate to reopen if it occurs again. Thanks.
I am sorry but I have to reopen this one. It crashed and rebooted the last three days. When I wrote this bugreport it happened nearly all the time. Now it seems to just happen if the system was under "some" load before. ;)
Ok.
Timo, can you reproduce this issue on your Macbook with openSUSE 10.3 Beta2/Beta3?
I finally managed to install openSUSE 10.3 Beta3 (x86_64) on such a notebook. I also get a yellow screen for a short time similar to your one when logging out, but the machine doesn't crash. Tried about 10 times. VT switching works as well. console looks fine.
I tried this with your xorg.conf.
Do you use Suspend-To-Disk or Suspend-To-Ram before this occurs?
Issue is still not reproducable after doing Suspend-To-Ram. Suspend-To-Disk seems to freeze the machine.
(In reply to comment #28 from Stefan Dirsch) > Do you use Suspend-To-Disk or Suspend-To-Ram before this occurs? It did occur with suspend-to-ram too, but this stopped to work reasonable with beta1. It still does suspend and resume for me with factory but takes 3-4 minutes to resume ... ( bug #305162 ) The crash just happened yesterday to me again. When I want to shutdown the system and it switches back to the first console it hangs and shows the yellow screen. The screen is there for (~~5 seconds) after this the system reboots. The bootloader (refit) starts and I select the penguin, but then the machine hangs (grub does not start). I have to press the power switch to turn the machine on again.
So it's not related to STR and even rebooting does not work you at all. I'm setting this to LATER, since it doesn't make sense to investigate this any longer, as long as I can't reproduce it at all.
For the record: Suspendig "works" if one adds "install acpi_cpufreq /bin/true" to /etc/modprobe.d/suspend. The Xserver crashes on shutdown and suspend. I think just switching from X to console is enough to crash the system. For me this seems similar to bug #327064. I happens nearly always after compiling a kernel.
Ok. So this looks like a kernel issue to me.
Sorry I think I was not clear. Suspend (the resume part of it) is totally broken with acpi_cpufreq loaded (see bug #305162). Without acpi_cpufreq loaded a complete suspend cycle (i.e. suspend and resume) have a chance to be successful. This was just meant to keep this report up to date. The X-Server still crashes on suspend or shutdown (at the moment switching to console) after compiling a kernel with a chance of probably 75%. So I am sorry but this report probably still belongs to Xorg. :-( I will try using the intel-drivers git repository in two weeks after I have been on vacation an see whether it helps or this bug can be resolved upstream.
later again.
This is fixed for me with current factory. I have no idea which update fixed it.
Probably the Xserver update. The intel driver didn't change.