[i965] X freezes after screen saving or logout/login on Dell Hybrid Studio

[Fred]

System: Ubuntu Intrepid Ibex (beta) 8.10 up to date as of 11 Oct 2008.

Linuxroot@fdetienn-desktop:~# uname -a
 fdetienn-desktop 2.6.27-7-generic #1 SMP Fri Oct 10 03:55:24 UTC 2008 i686 GNU/Linux

root@fdetienn-desktop:~# lsb_release -rd
Description: Ubuntu intrepid (development branch)
Release: 8.10

This is going to be a messy report as I see things but do not know in which order they happen or even if they are truly related. They are correlated though.

I am running Ubuntu 8.10 on a Dell Hybrid Studio. 4GB RAM and X3100 Intel accelerators.

When leaving X running for some time (undetermined yet) with an OpenGL screensaver (Hypertorus in this case), the system display finally hangs and I can not return to session (not even a password prompt). The screen remains black and backlit.

SSH still works and I see Hypertorus spinning at 100% CPU.

If I kill Hypertorus and gnome-screensaver, the screen flashes once but nothing better happens. I then see Xorg at 100% CPU.

If I kill X, gdm restarts it and I get a user prompt. After typing my username/password, X wipes the screen and I see an orange background but then X goes back spinning at 100% CPU. This takes me to the previous point. There is never a way to get a working X display.

The VT consoles are also broken; they remain with corrupted leftovers of the X display (I see part of the orange background and the mouse pointer + some colorful areas).

The only outstanding message is:

[42627.033483] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
[42772.634036] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
[42810.157626] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining

from kernel.log (also visible in dmesg).

I will see if the problem happens without OpenGL screensavers.

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile PM965/GM965/GL960 Memory Controller Hub [8086:2a00] (rev 03)
     Subsystem: Dell Device [1028:0279]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile GM965/GL960 Integrated Graphics Controller [8086:2a02] (rev 03)
     Subsystem: Dell Device [1028:0279]

[Fred (f-detienne)]

I just made further tests and I can add a few things:

 - opengl is not guilty - had a GL screensaver (hypertorus) running for hours
 - the MTRR error does NOT seem to be the root cause of the problem; I systematically get an MTRR at boot

It seems ACPI is the issue:
   - causes the above symptoms (X hang at 100% CPU; not recoverable) after restore from screen power saving
   - ACPI sometimes causes hang at boots
   - hang at resume after suspend

[Bryce Harrington (bryce)]

Hi f-detienne,

Please attach the output of `lspci -vvnn`, and attach your /var/log/Xorg.0.log file from after reproducing this issue. If you've made any customizations to your /etc/X11/xorg.conf please attach that as well.

[Fred (f-detienne)]

Hi,

I attached the logs; Xorg.conf is default (untouched).

The problem has now evolved and the same situation happens even without screen blanking.

I just need to let the computer run for a few hours (seemingly random time) for the problem to happen.

The visible symptoms are:
 - no windows are visible; no gnome menus, dockbar -- just an empty background (orange)
 - mouse pointer still moves
 - keyboard dead (ctl-alt-backspace does not work)

SSH'ing to the computer shows Xorg at 100% CPU. This is the moment I took the logs above.

Killing Xorg takes me back to the gdm greeter but Xorg immediately goes back to 100% CPU when I log in again.

thanks,

   Fred

[asp_foo (ashleypoet)]

Confirmed on a Dell Optiplex GX280 with a standard hardware configuration, so this isn't unique to the Hybrid Studio.

[Victor Trac (victor-trac)]

This doesn't have anything to do with OpenGL, I don't think. On a brand new, pretty much virgin 8.10 install on my Dell D630 laptop with an Intel GM965 video chipset, I had this happen when returning from the blank screen saver. It doesn't happen everytime. All I see is the background and the mouse with no menus or windows. The mouse responds normally, but the only recourse is to power off the computer and reboot.

[Fred (f-detienne)]

I confirm Victor's comments. I tried with a blank screensaver and the problem remains. If I disable the screensaver, the system remains stable much longer (there seem to be _other_ problems with X).

So bottom line:

recreate 1:
--------------
 - configure a screen saver (blank)
 - let the computer idle for a few hours
 - when returning from the screensaver, Xorg sits at 100% CPU and yields a moving cursor over empty background. No windows, no menu etc. Keyboard not responsive either (no ctl-alt-backspace, no ctl-alt-F1...)
 - killing X takes you to "recreate 2"

recreate 2:
--------------
 - log in under GDM (enter a gnome session)
 - log out
 - log in again => X starts spinning at 100% and yields a moving cursor over empty background. No windows, no menu etc. Keyboard not responsive either (no ctl-alt-backspace, no ctl-alt-F1...)

In either case, the only option is to reboot the computer.

I have the _impression_ that using a GL screensaver (such as hypetorus) accelerates the recreate 1 procedure. In this scenario, hypertorus would sit at 100% CPU. Killing it would yield an Xorg at 100% CPU, and then continue from step 3 of recreate 1.

[Matthew Sage (flamemaster)]

Confirmed on an Intel DG31PR mobo + Q6600 cpu running Intrepid amd64. I haven't tried Fred's "recreate 1" steps, but by trying "recreate 2" the end result is the same. I'm also attaching my logs in case they become useful.

# uname -a
Linux navi 2.6.27-9-generic #1 SMP Thu Nov 20 22:15:32 UTC 2008 x86_64 GNU/Linux.

[Matthew Sage (flamemaster)]

Here's Xorg.0.log. I'm using the default Xorg.conf

Also, just like Fred, the mtrr error is also showing when I try to log back in:

Dec 7 13:23:11 navi kernel: [ 732.102546] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
Dec 7 13:23:11 navi kernel: [ 732.389942] set status page addr 0x00033000
Dec 7 13:23:23 navi kernel: [ 744.420590] ecryptfs_parse_options: eCryptfs: unrecognized option 'rw'
Dec 7 13:23:23 navi kernel: [ 744.421269] ecryptfs_parse_options: eCryptfs: unrecognized option 'user=rafael'

[Sammy Spets (sammys)]

I have the same issue on Intrepid with GM965/GL960 (X3100) hardware and blank screen saver. I manage to get this bug happening when I lock the screen.

Compiz appears to have problems when it switches desktops. That causes compiz.real process to spin at 100% CPU. Although the symptoms of this bug is X running consuming 100% CPU I think we need to eliminate compiz as the cause. There is a desktop switch when the screensaver is switched out. I've had the same symptoms occur when using rdesktop in full screen mode and I go to switch back to windowed mode.

Another behaviour of compiz consuming 100% is ALT+SYSREQ (PRTSCR key) successfully switches off raw keyboard mode in X. This allows a switch to TTY1 using CTRL+ALT+F1. This is not possible when X is using 100%.

Since i've not found anyone else making progress with this bug i'll try to move things forward a little. Could you all report the following information:
 * Does ALT+PRTSCR enable tty switching for you (might need to use ALT+SYSREQ+R)?
 * Does the bug occur with compiz disabled/uninstalled?

References for the compiz problems (just some of them):
https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/176589
https://bugs.launchpad.net/ubuntu/+source/compiz/+bug/186058
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/285390

[Sammy Spets (sammys)]

Just a follow-on to post my log.

[Weiqi Gao (weiqigao)]

After uninstalling compiz, the bug stopped occuring and my GNOME session behaves normally now.

[strangedata (rcteigao)]

I'm also having the same problem on my lenovo desktop. If I leave the screensaver running for a few hours (2 or 3), everything works fine, but when I leave it overnight, the next day when I try to unlock it, it takes my password and only displays an empty background (i.e. I can see the background image, but no window is available; the mouse still moves, but the keyboard is unresponsive).

My graphic card is an Intel 82Q35 Express Integrated Graphics Controller, please see the attached file for the output of `lspci -vvnn`.

[Matthew Sage (flamemaster)]

>
> * Does ALT+PRTSCR enable tty switching for you (might need to use ALT+SYSREQ+R)?
No, it doesn't. I can't switch to tty, I can't even use Alt+SysReq+B.

> * Does the bug occur with compiz disabled/uninstalled?
Because of some speed issues, I have compiz disabled, but even then the
bug is happening.

My igpu is Intel Corporation 82G33/G31 Express Integrated Graphics.

[Bryce Harrington (bryce)]

[This is an automatic notice.]

We'd like to forward your bug upstream, however upstream requires
that you first test it against their newer driver code.

To save you the effort of building the driver from source, we've built
packages for the driver and its new dependencies.

So you have a couple options:

 1. Download and test .debs for intrepid, from:
     https://edge.launchpad.net/~intel-gfx-testing/+archive

 -or-

 2. Download and test the Jaunty alpha-2 (or newer) Live CD,
     (which includes a beta of the new xserver 1.6 as well).
     See http://cdimage.ubuntu.com/releases/9.04/ for ISOs

Thanks ahead of time! You can simply reply to this email to report your
findings.

P.S., if you wish to forward your bug upstream yourself, please follow
these directions to do so:
  http://intellinuxgraphics.org/how_to_report_bug.html

[Fred (f-detienne)]

Hi Bryce,

I tried the new driver on Intrepid (option 1 in your proposals) and I am
still seeing the problem.

It seems this bug can be sent upstream.

thanks for your help!

 fred

On Sat, 2008-12-20 at 02:26 +0000, Bryce Harrington wrote:
> [This is an automatic notice.]
>
> We'd like to forward your bug upstream, however upstream requires
> that you first test it against their newer driver code.
>
> To save you the effort of building the driver from source, we've built
> packages for the driver and its new dependencies.
>
> So you have a couple options:
>
> 1. Download and test .debs for intrepid, from:
> https://edge.launchpad.net/~intel-gfx-testing/+archive
>
> -or-
>
> 2. Download and test the Jaunty alpha-2 (or newer) Live CD,
> (which includes a beta of the new xserver 1.6 as well).
> See http://cdimage.ubuntu.com/releases/9.04/ for ISOs
>
>
> Thanks ahead of time! You can simply reply to this email to report your
> findings.
>
>
> P.S., if you wish to forward your bug upstream yourself, please follow
> these directions to do so:
> http://intellinuxgraphics.org/how_to_report_bug.html
>
>
> ** Changed in: xserver-xorg-video-intel (Ubuntu)
> Status: Confirmed => Incomplete
>

[Bryce Harrington (bryce)]

Thanks for testing; this should go upstream. Setting to Triaged for now.

[Tom Sillence (tom-sillence)]

I think this is a regression from Hardy, provided it's the same bug that I am experiencing: I have an intel 82G33/G31 on a lenovo desktop. X hangs after several hours (unable to even move mouse pointer, change VT or do ctrl-alt-bksp etc). It's not related to the screensaver, but haven't tried without compiz yet.

When X has hung, there is no process that's using 100% cpu and a kill -9 of X itself is the only thing that will unlock things.

[Victor Trac (victor-trac)]

Tom: No, I'm fairly certain you're experiencing a different issue. I only experienced it when coming out of sleep mode, not just after several hours of use. The mouse moves, but nothing else works.

Bryce: FWIW, I used the new driver and haven't yet been able to reproduce the problem.

[Fred (f-detienne)]

On Tue, 2009-01-06 at 16:53 +0000, Victor Trac wrote:
[...]
> Bryce: FWIW, I used the new driver and haven't yet been able to
> reproduce the problem.

I use the new driver and the problem is still there... just log out and
log in again or let go to screen saver (power saving mode) then return
to work will cause it.

 fred

[Jeff Powell (jeffp-bangtherockstogether)]

I, too, am seeing what may or may not be this problem. I've seen it on two different systems running 8.10.

The first was an ancient p3-800MHz that is now retired.

The current is a Dell Optiplex 330. I'll attach output pf lspci --vvnn for those who know more than me about these things.

In my case, I see two related failures:

The screen saver need not have been running for hours. Sometimes just a few minutes will do the trick, but it can be hours. I've seen it with a blank screen and with other screen savers selected. The initial thing is that the system hangs on trying to exit from the screen saver. Where the two failure modes differ is that on occasion (with the old system) I was able to use Ctrl-Alt-F# to switch to a tty. However, having just seen this problem on the new Dell, that was not possible.

Sadly I had no access to the system from any other place at the time, so I don't know what was going on. As I could get nothing to work, I forced a hard reboot. I've setup now with a second system that I can ssh in with to see what I can learn when I next see it.

This is a significant problem. I hope someone can resolve it.

[Jeff Powell (jeffp-bangtherockstogether)]

Sorry. Forgot to mention one thing. The old system that showed this bug was running 32 bit 8.10. The new system is running 64 bit 8.10 (amd64 architecture). Given the above discussion it probably doesn't matter, but I should have said it in my original comments.

[Jeff Powell (jeffp-bangtherockstogether)]

This just hit me again, but this time I was able to recover.

Logging in remotely I found nothing obviously wrong. System was almost entirely idle.

So I went back to the system keyboard and tried again. Initially I could not get Ctrl-Alt-F# to give me a tty, but eventually I got one after trying several times. Looking around there I still saw nothing abnormal.

Finally I hit Ctrl-Alt-F2 and was back at the screen saver password entry dialog. I entered the password and everything worked fine.

I'm not convinced this is the same bug that is being discussed above because I don't see X (or any other processes) chewing up CPU. However, the overall failure mode - sporadic inability to exit the screen saver properly - sure seems related.

If there are things I can provide to help debug this please let me know.

[Bryce Harrington (bryce)]

Forwarding this bug from a Ubuntu reporter:
https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/282081

[Problem]
X hangs on return from any screensaver, with Xorg at 100% and a moving cursor over an empty background. Killing and restarting X results in same condition.

[Steps to Reproduce]
recreate 1:
--------------
 - configure a screen saver (blank, 3D, OpenGL, whatever...)
 - let the computer idle for a few hours
 - when returning from the screensaver, Xorg sits at 100% CPU and yields a moving cursor over empty background. No windows, no menu etc. Keyboard not responsive either (no ctl-alt-backspace, no ctl-alt-F1...)
 - killing X takes you to "recreate 2"

recreate 2:
--------------
 - log in under GDM (enter a gnome session)
 - log out
 - log in again => X starts spinning at 100% and yields a moving cursor over empty background. No windows, no menu etc. Keyboard not responsive either (no ctl-alt-backspace, no ctl-alt-F1...)

In either case, the only option is to reboot the computer.

I have the _impression_ that using a GL screensaver (such as hypetorus) accelerates the recreate 1 procedure. In this scenario, hypertorus would sit at 100% CPU. Killing it would yield an Xorg at 100% CPU, and then continue from step 3 of recreate 1.

[Original Report]
System: Ubuntu Intrepid Ibex (beta) 8.10 up to date as of 11 Oct 2008.

Linuxroot@fdetienn-desktop:~# uname -a
 fdetienn-desktop 2.6.27-7-generic #1 SMP Fri Oct 10 03:55:24 UTC 2008 i686 GNU/Linux

root@fdetienn-desktop:~# lsb_release -rd
Description: Ubuntu intrepid (development branch)
Release: 8.10

This is going to be a messy report as I see things but do not know in which order they happen or even if they are truly related. They are correlated though.

I am running Ubuntu 8.10 on a Dell Hybrid Studio. 4GB RAM and X3100 Intel accelerators.

When leaving X running for some time (undetermined yet) with an OpenGL screensaver (Hypertorus in this case), the system display finally hangs and I can not return to session (not even a password prompt). The screen remains black and backlit.

SSH still works and I see Hypertorus spinning at 100% CPU.

If I kill Hypertorus and gnome-screensaver, the screen flashes once but nothing better happens. I then see Xorg at 100% CPU.

If I kill X, gdm restarts it and I get a user prompt. After typing my username/password, X wipes the screen and I see an orange background but then X goes back spinning at 100% CPU. This takes me to the previous point. There is never a way to get a working X display.

The VT consoles are also broken; they remain with corrupted leftovers of the X display (I see part of the orange background and the mouse pointer + some colorful areas).

The only outstanding message is:

[42627.033483] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
[42772.634036] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining
[42810.157626] mtrr: type mismatch for d0000000,10000000 old: write-back new: write-combining

from kernel.log (also visible in dmesg).

 - opengl is not guilty - had a GL screensaver (hypertorus) running for hours
 - the MTRR error does NOT seem to be th...

[Bryce Harrington (bryce)]

Created an attachment (id=21739)
Xorg.0.log

[Bryce Harrington (bryce)]

Okay, we're starting to get multiple confirmations with varying symptoms and varying hardware, which is kind of muddying things up a bit. It's quite likely that people are seeing unrelated bugs that just happen to all get triggered during screensaver/idle situations. The bugs may be related, but with this kind of bug it's often fairly hardware specific.

So I'm going to focus this bug on the original report of seeing the issue on Dell Hybrid Studio. If you're seeing a similar bug on different hardware, please file a new bug. Attach your Xorg.0.log[.old] from after reproducing the hang, the output of lspci -vvnn, and your xorg.conf if you've customized it. Detailed steps to reproduce the issue are *very* important for this kind of bug - see Fred's writeup in comment #8 as the perfect way to write it.

[Bryce Harrington (bryce)]

I've forwarded the bug upstream at https://bugs.freedesktop.org/show_bug.cgi?id=19434 - Fred, please subscribe to that bug in case upstream needs more information or wishes you to test something.

Also, it's very possible the crash is not due to a bug in the -intel driver as we're assuming, but rather the xserver, mesa, or something else. I would strongly recommend testing against Jaunty, which has the more up to date X bits and pieces. You can get an ISO from http://cdimages.ubuntu.com/releases/jaunty/.

[Sammy Spets (sammys)]

Thank you Bryce for giving this bug some attention. My experience with system level coding tells me the problem smacks of a race condition when switching desktops. I've turned off compiz and have run the system for 11 days non-stop with screen locking and full-screen windows all over the place.

Compiz must be exacerbating the race condition by slowing things down with the animations. Same thing with heavy loads like OpenGL (especially with the low frame-rate witnessed on hardy).

[Matthew Sage (flamemaster)]

Nevermind, just got to read the other messages.

[Jeff Powell (jeffp-bangtherockstogether)]

As requested I have opened a new bug for my problem. Bug 315219

[Jesse Barnes (jbarnes-virtuousgeek)]

This bug sounds famliar.... any way you could get a profile with oprofile or sysprof that might help point us in the right direction? If it's ACPI related, a BIOS update might help; are there any available for your system?

[Bryce Harrington (bryce)]

(Information has been requested of original reporter)

[Bryce Harrington (bryce)]

Hi Fred,

Upstream is requesting some additional information:

 "This bug sounds familiar.... any way you could get a profile with oprofile or
 sysprof that might help point us in the right direction? If it's ACPI related,
 a BIOS update might help; are there any available for your system?"

For some further information about how to use oprofile, see file:///usr/share/doc/oprofile/oprofile.html, or `man oprofile`, or hit up google for online resources. For more info on sysprof, see
http://www.daimi.au.dk/~sandmann/sysprof/. Hope this helps.

[Jesse Barnes (jbarnes-virtuousgeek)]

I'm also curious to know if this happens with newer drivers as well.

[Jesse Barnes (jbarnes-virtuousgeek)]

Timeout, closing.

[Bryce Harrington (bryce)]

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.