System freezes when clicking shutdown button

[Pedro I. Sanchez]

This bug report started as a comment in the thread associated with bug 38915 (https://bugs.launchpad.net/ubuntu/+bug/38915), but as suggested by Sitsofe Wheeler I'm opening a new bug report here.

Problem: 7 out of 10 times my PC freezes when I logout from my Ubuntu session. I get a black screen and my LCD says that there is no video input. Ctrl+Backspace and Ctrl+Alt+Del keyboard salutes don't work, and remote ssh sessions into my machine die; the only way out is a hard reset of my PC. As far as I can say nothing is logged in either the syslog or the X log files.

Environment:
  * Ubuntu 8.04 running on an ASUS P5E-VM mother board with the Intel G35 (3500) chipset and 2 GB of RAM.
  * My home directory is not local but NFS-mounted from another machine running Ubuntu server 8.04.
  * My xorg.conf file is the barebones one created by running "sudo dpkg-reconfigure -phigh xserver-xorg"; it
     has no mention of a specific driver or monitor details; I guess these days everything X-related is found
     dynamically as X takes over.
  * My PC was upgraded from Gutsy to Hardy. While on Gutsy, my video chipset was not well supported, I
     couldn't enable desktop effects for example, but I never experienced these freezes.

Things that I have done include:

1. Booted Hardy with the Feisty kernel. No changes.
2. Created a brand new user with a local home directory (not NFS-mounted) and tried login in and out. No changes.
3. Tried turning off desktop effects. No changes
4. Modified my xorg.conf file to force X to use the VESA driver ... success, no more freezes after logout.

Of course with this last option I get a sub-optimal desktop but this makes me conclude that the problem is X-related. My first call would be to blame the new Intel driver in xorg.

[Marcelo Barros (marcelobarros)]

Same problem with me (dell d505, intel 801 card).
I have tried to chance several options inside /etc/defaults/acpi-support, as discussed in
https://bugs.launchpad.net/ubuntu/+bug/38915/comments/61
No success. So, I decided to test if my system could hibernate from console, using /etc/acpi/hibernate.sh from command line. This gave me some freezes as well and a clue: acpi and kernel.
The only think that it is working at this moment: adding acpi=off inside menu.lst.
I can not hybernate anymore but system is working after logout.
So, problem seems to be related to kernel+acpi+video.
I can help in debugging if someone say me what to do (files, logs, etc).

[Jean-Baptiste Lallement (jibel)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. Unfortunately we can't fix it, because your description didn't include enough information.

Please include the information requested at https://wiki.ubuntu.com/DebuggingACPI as separate attachments.

[Tom Verdaat (tom-verdaat)]

Have about the same issue:

* Shutting down always works as expected
* Logging out often results in a black screen when restarting GDM.
    - The default Ubuntu sound is played just like when the login box is presented, only nothing is displayed.
    - The complete system freezes. nothing responds: monitor is in standby mode, I see no mouse pointer and the keyboard does nothing. Not even the light of the NumLock or CapsLock light when pushing those buttons.
    - Only a hard reset is possible. no ctrl+alt+del, ctrl+alt+backspace, ctrl+alt+f1, etc.
* Often it doesn't hang and I do get the login screen. Then logging in almost always results in the computer freezing just after that with just a yellow screen. No GNOME menu bars, windows or messages yet. Just like above, nothing works except a hard reset. This happens regardless of whether I'm logging in the same user again or a different one.

I have no idea how to provide additional debug information, because it freezes completely. This problem only applies to Hardy. Had Dapper, Edgy, Feisty and Gutsy on the same hardware before and it worked fine. Running amd64 and the nvdia propriatary video driver.

[Jean-Baptiste Lallement (jibel)]

Tom,

Please include the information requested at https://wiki.ubuntu.com/DebuggingACPI as separate attachments.

Can you try booting with the "acpi=off" kernel parameter and tell us if the error still occurs.

Thanks

[Pedro I. Sanchez (pirivan)]

Jean-Baptiste,

You are assuming that this bug is somehow related to ACPI but I believe it is not. As I said in my original message, changing my xorg.conf file to use the VESA driver eliminates the problem. I just tried again login in and out 10 times in a row and my PC never froze. I can't do the same when the Intel driver is loaded.

But I did the excercise of booting with "acpi=off" anyways. I'm sorry to say that it doesn't change anything for me. My PC still freezes upon logout.

[Jean-Baptiste Lallement (jibel)]

Thanks. I don't assume anything. It's quite hard to guess the causes of a crash when you're not behind the PC. So we need to eliminate the most common source for this kind of issue. That's why I've asked you to turn off ACPI.

For the sake of completeness can you attach the following files to your report:
    * uname -a > uname-a.log
    * dmesg > dmesg.log
    * sudo lspci -vvnn > lspci-vvnn.log
    * /var/log/Xorg.0.log
    * /var/log/Xorg.0.log.old after a crash
    * /etc/X11/xorg.conf

After a crash see if there is any backtrace in /var/log/kern.log.0 (or /var/log/kern.log). Please attach this file if anything was captured.

Thanks in advance.

[Jean-Baptiste Lallement (jibel)]

One more question: are you connected via VGA or HDMI ?

[Pedro I. Sanchez (pirivan)]

I'm connected via VGA only. Support for the HDMI port still needs work by the X team, but that's another story.

Attached you will find the files you requested. Note the following:

~$ cat kern.log |grep -c segfault
10

GDM seems to be crashing ... but only when using the Intel driver.

I installed the XDM window manager to test the logout process. It didn't crash the first time I logged out. However for some weird reason I cannot login again (I get the login window again after attempting to login for the second time). Tried with other user accounts and the same happened, I was able to login/logout once ... but not a single crash, which is good.

Let me know if you want me to run any other tests.

Thanks.

[Marcelo Barros (marcelobarros)]

I have reported that acpi=off solved the problem after trying logging in/out 10 times, all of them with success. However, today a did more tests and even with acpi=off the system frozen, like reported by Pedro Sanchez.

Additional information:
- it is not possible to use ssh to log in from other machine, even ping does not respond. I had a similar problem in other machine but it was possible to log in. In that case, compiz was using 100% of CPU and I found a bug related to it at launchpad. In the current issue, I removed compiz to avoid more problems.
- I tried all options cited in "DebuggingACPI", like acpi=ht, noapic, pnpacpi=off, pci=noirq, nolapic ... freezing as well.

I am attaching all files requested by Jean-Baptiste Lallement, in the hope they help us to find the solution. Any other debug action, please let me know.

BTW:
What I need to do for using vesa driver ?
dpkg-reconfigure does not give me this option, it finishes after keyboard configuration.

[Pedro I. Sanchez (pirivan)]

Marcelo,

Attached you will find the xorg config file that I use to force the VESA driver to load.

1. For the BusID value in the "Device" section use the command "lspci" and look for the ID associated with the "VGA compatible controller", or something similar.

2. In the "Monitor" section update the frequency values to match your monitor, or for testing purposes I believe you can just comment them out.

It would be interesting to see if the freeze dissapears for you as well using the VESA driver.

[Jean-Baptiste Lallement (jibel)]

These bugs are likely due to the intel driver.

We have exactly the same case as Marcello's bug upstream. You can track and make comments there : https://bugs.freedesktop.org/show_bug.cgi?id=15168 .

Pedro, your platform is a bit different, so it's very hard to conclude this is the same bug even if the intel driver is probably the culprit too.

Does switching to the console (Ctrl-Alt-F1) then back to X (Alt-F7) reproduces the bug ?

Pedro: I'll open a new report upstream tomorrow for your particular case as we have none with a G35.

Marcello: can you file a new bug report (you have a different hardware and it's already identified upstream) and post a link to your new bug report here.

Thanks

[Marcelo Barros (marcelobarros)]

I added my comments to the following bug in freedesktop site:

https://bugs.freedesktop.org/show_bug.cgi?id=15168

[Pedro I. Sanchez (pirivan)]

Jean-Baptiste,

No, switching between console and X doesn't reproduce the bug. I can do that without problems.

Thank you for filing the bug upstream. Please let me know the bug number so that I can follow up on it.

Thanks.

[Pedro I. Sanchez (pirivan)]

Jean Baptiste,

Note that a duplicate of this bug, bug 228242, is also about the same mother board that I have.

[Jean-Baptiste Lallement (jibel)]

This is a regression in intel driver v2.2.1 and fixed in 2.3.0

[Pedro I. Sanchez (pirivan)]

Will the fix be part of the 8.04.1 point release?

[Michael Wright (bodyjarrocks)]

Hi all,

Just saw this thread. I have the same problem I think. I can reproduce it via startx. I tried this:

1. Boot up without starting gdm or usplash.
2. startx
3. kill X
4. startx
5. kill X
6. startx
 -> Kernel LOCKS (freezes) HARD.

I can't ssh into the box.

I have a ASUS P5E-VM HDMI, just like the OP.

I'm on Hardy Heron. I have some addition complexities like:
- My box is diskless.
- It boots from USB key and gets rootfs from another Hardy box (Mythbuntu-diskless)

I'm pretty sure this is the same bug.

Is there an easy way to try Intel version 2.3.0?

Mike.

[Michael Wright (bodyjarrocks)]

>
> This is a regression in intel driver v2.2.1 and fixed in 2.3.0
>

Jean Baptiste,

Just for interest, how did you find out about this? Is there a particular patch in Xorg that fixes this? I'm looking at:
http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/log/?h=xf86-video-intel-2.3-branch.

Thanks,
Mike.

[Michael Wright (bodyjarrocks)]

Hi Jean Baptiste,

Here is an update. I just pulled the latest (3 hours ago) xf86-video-intel-2.3-branch branch from GIT. I only pulled the 2D Xorg driver. I compiled it and tried in on my box. It still locks the machine hard.

1. startx
2. Ctrl-Alt-Backspace.
3. startx
 -> Locks hard.

So 2.3 does not fix the issue. Do you think I should I also try pulling the DRM (kernel) and Mesa 3d GL parts and build them as well?

Mike.

[4lorne (4lorne)]

The logout freeze issue went away for me after I updated to gdm 2.20.6-0ubuntu1 this morning. It potentially was due one of the other updates, but gdm seems most likely.

Though, I'm guessing this only fixed a symptom of a problem.

[Jean-Baptiste Lallement (jibel)]

Hi Mike,

The comment from the developer is here : https://bugs.freedesktop.org/show_bug.cgi?id=15923#c1
I'm here to help triaging bug reports. I'm not a Xorg developer and I'm sorry not being able to help you any further with what's going wrong with this version of the driver.

You should report directly to the upstream report ( https://bugs.freedesktop.org/show_bug.cgi?id=14591 ) and post here any new information if there is still an issue.

Regards,

[Michael Wright (bodyjarrocks)]

Thanks Jean-Baptiste,

I'd like to post an update. I can confirm that driver 2.3.0 + latest DRM does fix the problem. 2.3.0 alone still locks hard. You need the latest DRM (I used HEAD on Friday 16th May). That is the kernel part. Makes sense since it is pretty hard to lock things hard from user space.

I need to raise another bug now since the modesetting code seems to be borked since I can't get 1360x768@60Hz on my Pioneer PDP427XD plasma :-(

To anyone else who hits this problem, just follow the instructions on: http://intellinuxgraphics.org/install.html

Mike.

[Jean-Baptiste Lallement (jibel)]

Thanks for your feedback Mike.

Can you file a new bug report for the other bug if you haven't already done so.

Thanks again and we appreciate your help.

[Michael Wright (bodyjarrocks)]

No need for another bug report. Option "NoDDC" "true" fixed that problem. I think my pioneer must have been reporting its EDID data incorrectly, and the driver rejected the modes.

Mike

[Job Noorman (jobnoorman)]

I have the same problem here but not with an Intel card. When using fglrx for my ATI car, my machine hangs on logout. This problem disappears when using vesa.
Please let me know if I can give more info.

Job

[Hannes Koller (hannes-koller)]

Enabling hardy-proposed and installing the latest updates fixed the Logout-Hang issue on my system (ATI card, fglrx driver).

[Jean-Baptiste Lallement (jibel)]

Note that it's helpful to the X team if bug reports target a specific bug against a specific set of hardware. Even though you may be experiencing the same symptom reported here, it will often require fixes in different drivers based on the hardware you are using. Because of this, if you have hardware which differs from the original bug reporter I'd encourage you to please open a new bug report. We can easily mark bugs as duplicates later on if necessary. Please, when reporting X bugs, also attach your X server configuration file (/etc/X11/xorg.conf) and X server log file (/var/log/Xorg.0.log) to the bug report as individual uncompressed file attachments.

I apologize for any inconvenience this may cause but appreciate your cooperation.

[Marcelo Barros (marcelobarros)]

Following the comment:

https://bugs.freedesktop.org/show_bug.cgi?id=15168#c31

I added ForceEnablePipeA to xorg.conf, like below

Section "Device"
 ... <other options here> ...
 Option "ForceEnablePipeA" "true"
EndSection

I tried 10x to reproduce the problem without success (login/logout). I was possible to hibernate and suspend as well. This option seems to solve the problem. I am using a newer gdm version as well (2.20.6-0ubuntu1) but problem was happening with it too.

Just to keep the bug tracker up to date. I will test more.

Other reference:
https://bugs.launchpad.net/debian/+source/xserver-xorg-video-intel/+bug/138256

[Pedro I. Sanchez (pirivan)]

Unfortunately the ForceEnablePipeA option doesn't work for me. My system still freezes, and as a bonus I get a chopped screen (15% of my screen at the bottom in black) and poor fonts.

~$ lspci -vvnn | grep -A1 "VGA compat"
00:02.0 VGA compatible controller [0300]: Intel Corporation 82G35 Express Integrated Graphics Controller [8086:2982] (rev 03) (prog-if 00 [VGA controller])
 Subsystem: ASUSTeK Computer Inc. Unknown device [1043:8276]

I will try updating from hardy-proposed to see if that fixes the problem.

[Pedro I. Sanchez (pirivan)]

:-( hardy-proposed doesn't work.

This bug is listed as "fixed committed." Can anybody please confirm whether the official fix will be included in 8.04.1 or not?

Thanks.

[Daniel R. (danielr-es)]

Hello,

I experimented this bug with Ubuntu 8.04 Hardy, too. Now I'm using Debian Testing (Lenny) and also have the same - or quite similar - bug (see: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=482819 ).

Just for completeness: I have performed several more tests, and in some cases now I also suffer the freeze when switching to VT (CTRL+ALT+F1), specially if I do it soon (less than 1 min ?) after entering my gnome session.

I guess it may be in fact a DRI related problem, however I do NOT bet on it. Please tell if adding the following to your (original auto-configured xorg.conf) exacerbates the problem:

Section "DRI"
 Mode 0666
EndSection

(It seemed in my case, but I'm not sure).

On the other hand, test if this limits the problem (once again, to your original xorg.conf):

Section "Module"
        Disable "glx"
        Disable "dri"
EndSection

(It seems it limits it in my case? - not quite sure either)

In addition, I have to say that in my Debian Lenny, package libgl-mesa-dri was not installed by default. After installing it I removed some warning messages in my Xorg.0.log, but with little variation in behaviour (I dare to say that crashes when switching to console started after that??).

Just another remote possibility: my laptop also has a WIFI ipw2200 interface. I think I have noticed higher X crash probability when Network Manager applet cannot connect to my wireless network ?.

Ah, I also have the console vesa framebuffer enabled, by adding the vga=788 parameter to my kernel boot.

Anyway, as this is all quite random, I am not sure at all of anything.

Regards,

[Jean-Baptiste Lallement (jibel)]

Pedro, the maintainer as been informed and he'll take a look. I've no more info at the moment.

[Bryce Harrington (bryce)]

Marcelo's issue is a dupe of bug #235643, and is unrelated to the other commenters in this bug report. I should be able to get that fixed up tomorrow or friday.

Other people's issues sound like unrelated bugs that just share a similar symptom, and I'd concur with Jean-Baptiste Lallement's assessment in comment #27. Jean-Baptiste, I would suggest you adjust the title of this bug to more precisely match the original reporter's issue, to prevent further incorrect 'me-too'ing.

[Bryce Harrington (bryce)]

Returning to Triaged since original reporter says the pipe-A solution didn't fix it. (Comment #29).

Pedro, to answer your question, if setting PipeA didn't fix the issue for you, then no, unfortunately there won't be a fix in 8.04.1 for your card.

[Daniel R. (danielr-es)]

Hint:

Setting the "ForceEnablePipeA" option apparently solves the problem for me. Previous experiments did not have any positive effect.

Note i'm currently using the xserver-xorg-video-intel version 2:2.3.1-1 in Debian Lenny ( +/- Ubuntu's "Unstable" brother ).

Anyway, it seems even 2.3.1 version of upstream driver is a bit raw and not quite appropriate for production, yet.

Please consider packaging the last upstream git version of this driver and including it in ubuntu security updates, or as backport. I guess this may be affecting a big amount of Ubuntu users.

[Bryce Harrington (bryce)]

Daniel, thanks for confirming the ForceEnablePipeA option helped in your case.

Putting the latest git version of the driver into ubuntu-updates is not possible, however specific fixes can be backported. In the case of people running into the "pipe A problems" we are in fact quirking the driver for each specific chipset encountering it, and this is definitely a possibility we can do for you. Please refer to bug #138256 for background, and then file a new bug according to the directions in that bug's description.

[Jerson (jerson)]

Why you guys assume is intel graphic related only?
I have exactly same problem with 2 computers
with nvidia card and ati
All the updates by today June 8 2008 and the problems is still there.
Ubuntu Hardy 64 bit

One click on the log out applet and the system freezes everything.

[Jerson (jerson)]

If you have keytouch installed, try to remove it. That solved my problem with 2 computers (nvidia and ATI)
Ubuntu Hardy 64 bit

http://ubuntuforums.org/showthread.php?t=800013

If not, try the alternative from here.

http://ubuntuforums.org/showthread.php?p=4858236#post4858236

[f4hy (bmfahy+bug)]

I have had the same issue for months and do not have keytouch installed.

The problem is not due to Intel drivers as I am experiencing it using nvidia.

A work around seems to be to turn compiz off by running 'metacity --replace &'

The interesting thing is that I do not get this problem every time. It seems to be about 80%. Even when it does hang it will finally respond after about 15 minutes and show the logout dialog. In the mean time killing the session with ctrl+alt+backspace appears to work.

[Nick2000 (monpetitbeurre)]

Thank you but I am not using compiz and when this happens, no key
combination works at all and it does stay frozen a long time (I left
it on and it was still frozen in the morning)

Unfortunately, I have another variant now. I do not know if it's
related but the keyboard becomes erratic in a gnome session, but works
at the login dialog and in a new session...

I guess that I need to learn how to troubleshoot/trace what is going on.

On Mon, Jul 28, 2008 at 10:35 PM, f4hy <email address hidden> wrote:
> I have had the same issue for months and do not have keytouch installed.
>
> The problem is not due to Intel drivers as I am experiencing it using
> nvidia.
>
> A work around seems to be to turn compiz off by running 'metacity
> --replace &'
>
> The interesting thing is that I do not get this problem every time. It
> seems to be about 80%. Even when it does hang it will finally respond
> after about 15 minutes and show the logout dialog. In the mean time
> killing the session with ctrl+alt+backspace appears to work.
>
>
> ** Attachment added: "xorg.conf"
> http://launchpadlibrarian.net/16374197/xorg.conf
>
> --
> System freezes when clicking shutdown button
> https://bugs.launchpad.net/bugs/229043
> You received this bug notification because you are a direct subscriber
> of a duplicate bug.
>

[Bryce Harrington (bryce)]

According to the upstream bug report, there was a register being accessed on all G35 systems, however it's not supported by all G35 chip variants (such as 0x2982 (rev 01)), so on those that don't support it, it can cause a crash.

So it looks like we need this patch:
http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/diff/?id=02a3b80ee8341968dbce5b7cb871b187db5ff805

(I'll also need to doublecheck that a better patch didn't get added later.)

[f4hy (bmfahy+bug)]

Since I have this problem but I am not using intel drivers should I file a separate bug report?

[Bryce Harrington (bryce)]

f4hy, yes filing a separate bug report would make sense. You could reference this bug perhaps. See http://wiki.ubuntu.com/X/Reporting for tips on making a good bug report that can be easily troubleshot.

[Pedro I. Sanchez (pirivan)]

My machine still freezes often but my hopes for the LTS status of 8.04 to give me a fix are still alive.

I hope that https://bugs.launchpad.net/ubuntu/+source/fglrx-installer/+bug/38915/comments/82 doesn't mean that this bug is not going to be fixed.

Any updates on the fix for the Intel driver? Will it be part of Intrepid?

[Bryce Harrington (bryce)]

Yes, this patch is included in the version we have in Intrepid.

Probably ought to be backported for Hardy as well.

[Rolf Leggewie (r0lf)]

Hardy has seen the end of its life and is no longer receiving any updates. Marking the Hardy task for this ticket as "Won't Fix".