[gm45] X freeze on X200s with gm45 on Karmic

[Luka Renko]

Binary package hint: xorg

I am opening new bug, as the bugs I though I had (bug 424055, bug 396048) were closed as Fixed.

I have X200s with gm45 graphics with up-to-date Kubuntu Karmic. I am currently running X drivers from x-updates PPA (2.9.0 release), but this problem is reproducible with both stock X drivers from Karmic as well as xorg-edgers PPA. I do not use kwin compositing.

Laptop is used with only internal display (home, travel), as well in dual-head config at work (X200s Docking Base -> DisplayPort -> DVI -> HP L2045W). I can say that hangs are occurring in both configs, potentially a bit more often with dual-head config, but this may be subjective due to more time of using it at work. I would say that on average, I experience 1 hang per day.

When X freeze/hang occurs, I can ping the system, but I was not able to login via ssh. Just today, I had ssh session opened from the morning, just to catch it, but it did not help, as ssh session was stuck, so I was still not able to collect GPU dump as proposed in bug 388467 :-(

I did not see any obvious errors in the logs (kernel oops, Xorg.0.log errors or similar), so I am really not sure what I can do to help debug this problem.

Just for the reference (but probably not related to this problem):
- sometimes I get non-KMS boot as described in bug 392017 - not sure if hang occurs with non-KMS too, as I always reboot to get KMS boot
- X crash on using some applications - bug 415357
- random X crash with assert - bug 418194 (I did not reproduce this recently)
- random X crash with backtrace - bug 416421 (I did not reproduce this recently)
- wrong resolution in dual-head case - bug 286001

ProblemType: Bug
Architecture: amd64
Date: Fri Oct 2 13:12:05 2009
DistroRelease: Ubuntu 9.10
MachineType: LENOVO 74705HG
Package: xserver-xorg 1:7.4+3ubuntu5
ProcCmdLine: BOOT_IMAGE=/vmlinuz-2.6.31-11-generic root=/dev/mapper/plain-root ro single
ProcEnviron:
 LANGUAGE=
 PATH=(custom, user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-11.36-generic
RelatedPackageVersions:
 xserver-xorg 1:7.4+3ubuntu5
 libgl1-mesa-glx 7.6.1~git20090930.49fbdd18-0ubuntu1~xup~1
 libdrm2 2.4.13-1ubuntu1
 xserver-xorg-video-intel 2:2.9.0-1ubuntu1~xup~1
 xserver-xorg-video-ati 1:6.12.99+git20090929.7968e1fb-0ubuntu1~xup~1
SourcePackage: xorg
Uname: Linux 2.6.31-11-generic x86_64
XorgConf: Error: [Errno 2] No such file or directory: '/etc/X11/xorg.conf'
dmi.bios.date: 12/19/2008
dmi.bios.vendor: LENOVO
dmi.bios.version: 6DET38WW (2.02 )
dmi.board.name: 74705HG
dmi.board.vendor: LENOVO
dmi.board.version: Not Available
dmi.chassis.asset.tag: No Asset Information
dmi.chassis.type: 10
dmi.chassis.vendor: LENOVO
dmi.chassis.version: Not Available
dmi.modalias: dmi:bvnLENOVO:bvr6DET38WW(2.02):bd12/19/2008:svnLENOVO:pn74705HG:pvrThinkPadX200s:rvnLENOVO:rn74705HG:rvrNotAvailable:cvnLENOVO:ct10:cvrNotAvailable:
dmi.product.name: 74705HG
dmi.product.version: ThinkPad X200s
dmi.sys.vendor: LENOVO
fglrx: Not loaded
system:
 distro: Ubuntu
 architecture: x86_64kernel: 2.6.31-11-generic

[lspci]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller [8086:2a42] (rev 07)
     Subsystem: Lenovo Device [17aa:20e4]

[Geir Ove Myhr (gomyhr)]

This looks like a difficult one since you are not able to ssh in (you are connected to the network via cable and not wireless, right?) Would it be an option to make a script that takes a batchbuffer dump and then run it from cron every 15 minutes? You would just have to make sure it is not overwritten when you reboot before you copy the files.

[Luka Renko (lure)]

Yes, I have tried to ssh with both wlan0 as well as eth0. Even when I had opened ssh session for whole day, I could not use it, as it was blocked on something (could not type in commands). At the same time, system was still responding to keypress (NumLock key/LED).

OK, I will setup a cron-job that will log every 15 minutes the dumps. What do I need to collect:
- intel_reg_dumper
- intel_gpu_dump (BTW: does not return anything on my system)

Anything else?

[Geir Ove Myhr (gomyhr)]

There is a full list at https://wiki.ubuntu.com/X/Troubleshooting/Freeze . /sys/kernel/debug/dri/0/i915* seems to be important. Strange that intel_gpu_dump doesn't return anything. I assume you are running it as root? You could try with xorg-edgers (https://launchpad.net/~xorg-edgers/+archive/ppa) since it in addition to bleeding edge drivers has a newer version of intel_gpu_dump.

[Luka Renko (lure)]

OK, I have installed just intel-gpu-tools from ppa (not sure I want all the rest yet) and it works now.
Have checked wiki page to include all necessary files in my cron script.

Do I need to run gdm/kdm with INTEL_DEBUG=batch ?

[Luka Renko (lure)]

I have got hang, but no logs were written. :-(
The hang happened at 17:10, while the last log collected was at 17:00, even though I have waited until 17:19 in order to give enough time to write logs. I will add another "sync" command into script, just in case (even though at least files should exist, even if empty).

It seems that this hang is more than just GPU hang. :-(
I was still able to press NumLock and get LED response, but nothing else was possible, but only hard-power-down (holding power button for 10 seconds).

[bugbot (bugbot)]

This bug appears to involve a freeze of Xorg, probably due to a GPU
lockup in the kernel drm code, in a version of Ubuntu before
natty (or we couldn't determine the version from the report).

The X team has been investigating GPU lockups in Intel's video driver
during the natty development cycle. Many of these issues are now either
resolved or at least well understood. There are also better tools for
capturing and diagnosing these problems, but these tools are either
unavailable or incomplete in older versions of ubuntu, so debugging this
issue on pre-natty versions is unlikely to be productive. We're closing
out the bug report for this reason.

However, you may not yet be ready to upgrade to natty. It turned out
that a large number of these freezes were solved by updates to the
kernel. Thus, updating your kernel may be a workaround for you.
Updated kernel packages are available a variety of places, such as:

  http://kernel.ubuntu.com/~kernel-ppa/mainline/

If you find a newer kernel fixes it, and think the fix should be
backported to earlier Ubuntu versions, please file a request against the
linux kernel in Ubuntu, indicating the broken and the fixed kernel
versions.

If it does not fix it, and you do not wish to upgrade to natty, please
follow up through technical support channels (e.g. http://askubuntu.com
is a good place for free support for older Ubuntu versions).

On the other hand, if you find after upgrading to natty and updating to
latest that you still experience GPU lockups, please reproduce the
problem, ssh into the frozen machine, and collect the file
/sys/kernel/debug/dri/0/i915_error_state and the output of 'dmesg >
dmesg.txt', and attach both to a new bug report.