[iGM45] Connecting external monitor causes system to lock up.

[Ben James]

Binary package hint: xserver-xorg-video-intel

--------------------------------------------------------------------------------------

Summary:

When attaching an external monitor via VGA port my system locks up - there is no response to Ctrl+Alt+Backspace or Ctrl+Alt+F1, etc, there is also no response to other key presses and mouse (touchpad) input.

--------------------------------------------------------------------------------------

Hardware:

Type: Laptop
Manufacturer: Sony
Model: Vaio VGN-NS10J/S
Graphics adapter: Mobile Intel® GM45 Express Chipset

--------------------------------------------------------------------------------------

Software:

OS: Ubuntu 8.10 (Intrepid Ibex) AMD64
unane -a: Linux newton 2.6.27-9-generic #1 SMP Thu Nov 20 22:15:32 UTC 2008 x86_64 GNU/Linux

--------------------------------------------------------------------------------------

[lspci]
00:00.0 Host bridge [0600]: Intel Corporation Mobile 4 Series Chipset Memory Controller Hub [8086:2a40] (rev 07)
     Subsystem: Sony Corporation Device [104d:9045]
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 4 Series Chipset Integrated Graphics Controller [8086:2a42] (rev 07)
     Subsystem: Sony Corporation Device [104d:9045]

[Ben James (something-for-the-pain)]

The external "monitor" that was used was a Sony KDL32V2000 LCD TV; 1366x768 resolution.

[Geir Ove Myhr (gomyhr)]

Thank you for reporting this bug and helping making ubuntu better. Could you please attach the following:
- the output of `lspci -vvnn`
- /var/log/Xorg.0.log from when the problem occurs (if you can't get the file because the computer is locked up, it will be moved to Xorg.0.log.old next time X starts).
- /etc/X11/xorg.conf if you have changed this from the default

[Ben James (something-for-the-pain)]

As noted before, the laptop freezes completely when plugged into an external monitor, therefore the attached file is the Xorg.0.log.old version.

[Ben James (something-for-the-pain)]

More (possibly related) buggy behaviour...

I tried booting with the external display plugged in and found that the display is correctly detected (albeit at a low resolution, maybe 800x600), however once Ubuntu finishes booting and the desktop is displayed I get a critical battery level warning and the laptop subsequently shuts itself down (note that the laptop is running on a full battery at this point with no mains cable).

Ever since booting with the external display connected the laptop gives a critical battery warning and shuts itself down as soon as the main power cable is removed (see screenshot).

Additionally, some of the panel widgets lost their settings (for example the speed-step core frequency monitor changed from text to icon view and the system monitor also lost its colour settings).

[Bryce Harrington (bryce)]

Hi something-for-the-pain,

Please attach the output of `lspci -vvnn` too.

[This is an automated message. If this script has reached you erroneously, please accept our apologies; any reply to this message will be sufficient to prevent it from doing further automated processing.]

[Ben James (something-for-the-pain)]

I've repeated this with root privileges, just in case the extra info is useful.

[Geir Ove Myhr (gomyhr)]

This smells a bit like a bug in the kernel, an X-driver bug shouldn't be able to freeze the system completely. There are a few tests you can do to find out this, and we would be interested in the result of all of them.
- When the system freezes, does CapsLock still turn on and off the keyboard LED? (if it does, it is probably not a kernel bug)
- If you have another computer and network available: Could you install openssh-server and see if it is possible to log in after the freeze? (if you can't it's likely a kernel bug)
- Try and do a Ctrl+Alt+F1 before you plug in the external monitor. Does the system freeze and dump debugging information to the screen? If it does, it is a kernel bug, and that text on the screen is key information for fixing it. Since the system is frozen, you will either have to take a photo or write it down manually.
- Next time you start the computer after a freeze, is there any information in /var/log/kern.log or /var/log/kern.log.0 from the time the computer froze?

[Ben James (something-for-the-pain)]

After a lock-up caps lock does nothing (no response from LED).

Connecting the monitor while in console mode shows no output - no response from keyboard, no stack dump, no panic.

I've attached the kernel logs.

[Ben James (something-for-the-pain)]

One thing to note is that I've installed the following automatic update:

xserver-xorg-video-intel (2:2.4.1-1ubuntu10.3).

Even with this update, the problem persists.

[Geir Ove Myhr (gomyhr)]

Did you check the time when you plugged in the monitor? We would only be interested in those lines in /var/log/kern.log that appear at the time you plug in your monitor (if any). If you type `date` on the command line just before you plug it in, you can see next time you boot if there are any lines from the following seconds.

About the driver update: The updates to intrepid only fixes specific problems. The 10.3 was supposed to fix the problems detailed in bug 316672, and everything else is the same.

It would also be very useful if you could confirm that this problem also exists in Jaunty, which you can do by using a LiveCD. There is a recent alpha (alpha-3) at http://cdimage.ubuntu.com/releases/9.04/ . The intel-driver in Jaunty was updated yesterday (from 2.5.1 to 2.6.1), so maybe the latest daily live at http://cdimage.ubuntu.com/daily-live is a better option (unlike the alpha, it hasn't been tested for basic functionality, though).

[Ben James (something-for-the-pain)]

Attached is an edited kernel log; I checked the time before plugging in the external monitor (Jan 25 09:37:03) - everything before that time has been removed, so hopefully the log makes more sense now.

I'll give the LiveCD a go and see if the latest alpha, however it might take me a while to download the image, burn the CD, etc.

More info come...

[Ben James (something-for-the-pain)]

...I've just tried the LiveCD from the second link you posted (jaunty-desktop-amd64.iso - daily image); it suffers from the same symptoms as Intrepid Ibex.

I've noticed that the status for this bug is "Incomplete" - is there any additional info I can provide?

[Geir Ove Myhr (gomyhr)]

Okay, so since there are no messages around 09:37:03 and the first messages at 09:38:27 looks like a typical boot, plugging in the monitor doesn't show anything in kern.log. This means that I'm still not sure if this is a kernel bug or not. Do you have the possibility to perform the last test I proposed above (installing openssh-server and check if it's possible to log in - maybe also if it answers on ping)?

About the status: The "Incomplete" status is used to keep track of where we're awaiting more information (if I have understood it correctly). Since I always had another question for you, I didn't set it to confirmed yet.

[Ben James (something-for-the-pain)]

Unfortunately I don't currently have another PC that can SSH into the laptop, however I've tried pinging it and it does not respond once the VGA cable is attached - it seems more and more like a kernel bug to me.

[Geir Ove Myhr (gomyhr)]

I agree, it's looking more and more like a kernel bug. I have added the package linux to this bug report. For kernel bugs there is a whole new list of required information (see https://wiki.ubuntu.com/KernelTeamBugPolicies). At least you will need to upload the output of `uname -a`, `cat /proc/version/signature`, and `dmesg`. These should be uploaded as separate files. If you want, you can do it by sending the files as attachements to <bugnr>@bugs.launchpad.net, as this will only generate one comment (the web interface is limited in this respect, see bug 82652).

About your last test, when you pinged the computer and it stopped replying after you put the VGA cable in, was your computer on wireless or wired ethernet? I think it may be easier to bring down the wireless, since it is only started after you log in, but I'm not sure. Also I'm impressed that you could ping your computer without having another computer available - how did you do that?

[Geir Ove Myhr (gomyhr)]

I just came over bug 213191 which has a similar issue (though this was on switching to external, not on connecting). I haven't read it in detail yet (and I have no more time today), but in that case it looks like it is still a driver bug, even though CapsLock and network does not respond.

[Ben James (something-for-the-pain)]

> Also I'm impressed that you could ping your computer
> without having another computer available - how did
> you do that?

A friend with a Windows laptop was using the network - he didn't want to install an SSH client (work laptop), but could ping from the command line as that functionality is built-in.

The laptop was on a wireless connection - I'll have another try using a wired connection when I can get my hands on another machine (on which I can install an SSH client).

I'll also fill in the kernel bug info when I get some time tomorrow.

Thanks for the help so far.

[Bryce Harrington (bryce)]

This sort of sounds like another GM45 lockup error that was recently fixed:
https://bugs.freedesktop.org/show_bug.cgi?id=17292

Please test with latest updates to Jaunty and see if the bug still exists.

[Geir Ove Myhr (gomyhr)]

Is 17292 the right freedesktop bug? The freeze is mentioned in comments following #203, but there seems to be no freedesktop bug filed for this issue yet.

And he already tested with daily-live as of Jan 24, which contains xserver-xorg-video-intel 2:2.6.1-1ubuntu1. I can't see any updates to the driver after that. Are there updates to any other packages after that?

Btw, comment #204 to the freedesktop bug indicates that this happens also if X is using the VESA driver and even if X isn't running. Ben, maybe you can verify this?

[Ben James (something-for-the-pain)]

I've just tried the previous remaining suggestion (plugging in the VGA cable and attempting to PING over ethernet) to no avail - all packets are lost, so there is a definite lock-up here.

I'll try and replicate the freeze without X running...

[Ben James (something-for-the-pain)]

I still get the same freeze with X not running, here are some details:

Ctrl+Alt+F1
sudo killall gdm

----- Plug in VGA cable ------

Cursor continues to flash, but no input is registered - no response from Ctrl+Alt+F1/F2/F3/Backspace, power button (unless held down for 5 seconds), etc.

How do I try with the VESA driver?

Should I still complete a kernel bug report, or is this not definitely a kernel issue?

[Geir Ove Myhr (gomyhr)]

From the comments in the freedesktop bug (#205) it seems that it is really a hardware bug in the Sony Vaio computers. It may be possible to work around it, but I'm not sure in which package. It could be ACPI, which may also explain the battery behaviour.

To use the VESA driver, add the line
Driver "vesa"
to the Device section of your xorg.conf and restart X (e.g. by logging in and out).
It's nice if you can test it, but this doesn't seem to be X-related anymore.

Also, could you verify the driver version on the LiveCD you used? I see that you use AMD64 and I only checked that version 2.6.1 of the driver was included in the i386 version of 24. jan LiveCD. I saw Timo mention in an unrelated comment that it the build had failed on amd64 the first time, so it may only have been included later (it is now).

I don't know if it will help exclude other things, but maybe you can start you computer in recovery mode (ESC on startup gives you a boot menu if you don't already see it). After booting into recovery mode, you can select a root shell. The computer will now run with only a minumum of services enabled. If the computer still freezes when you plug in the VGA it's definitely something low-level.

It wouldn't hurt if you uploaded the files for the kernel debugging. ACPI bugs are also filed against linux. There is some additional information on https://wiki.ubuntu.com/DebuggingACPI with some tests you can run to troubleshoot.

[Ben James (something-for-the-pain)]

I've tried with the Vesa driver and I get exactly the same behaviour - this is almost certainly a kernel bug.

Note: I've opened a new bug with a link back to this one. The new bug is filed against the kernel, but has, so far, had no response.

[Geir Ove Myhr (gomyhr)]

It wasn't really necessary to file a new bug (bug 325477) for the
kernel. A bug can be filed for many packages, and I had already filed
this bug for the kernel (at the top of the bug page you can see that
this bug affects "linux (Ubuntu)" and "xserver-xorg-video-intel
(Ubuntu)". Sometimes it can be useful to close the old bug and file
new one if there is a lot of comments that turn out to be irrelevant,
but in this case you've referred to this bug for information anyway.

I'm setting the other bug to a duplicate of this one, and attaching
the kernel relevant files you uploaded to this one.

[Geir Ove Myhr (gomyhr)]

Closing this bug for xserver-xorg-video-intel, since this the computer hangs even if the Xserver is not running.

[Ben James (something-for-the-pain)]

I can confirm that this behaviour is still seen on the latest LiveCD release - Jaunty 9.04.

[Leann Ogasawara (leannogasawara)]

Hi Guys,

Would you be willing to test the latest mainline kernel build, 2.6.30-rc8 as of this posting.

https://wiki.ubuntu.com/KernelMainlineBuilds

We'll want to open an upstream bug report to make sure the upstream developers are aware if this does exist upstream. Otherwise we'll want to figure out what patch may have fixed this upstream and will need to be applied to the actively developed Ubuntu kernel. Please let us know your results. Thanks.

[Ben James (something-for-the-pain)]

Just tried the new kernel (2.6.30-rc8 amd64) and the symptoms are exactly the same. It would seem that this bug is probably a mainline issue.

[Leann Ogasawara (leannogasawara)]

Ben, thanks for testing and the feedback. Would you be willing to also open an upstream bug report at bugzilla.kernel.org since you've been able to reproduce the issue on an upstream kernel? If you could post the link to the upstream report here for others to also track that would be great. I'll also open an upstream bug watch which we can update with the upstream bug # later. Refer to https://wiki.ubuntu.com/KernelTeam/KernelTeamBugPolicies#Reporting Bugs Upstream for more information. Thanks.

[Ben James (something-for-the-pain)]

After a quick search of BugZilla it seems that this is a known issue (bug 12904) and there is a work-around - add the following to the boot command line:

acpi_osi="!Windows 2006"

I've added a link to this page on the BugZilla bug. Thanks one and all for the support on this so far - hopefully there will be a proper fix in time for 9.10 from upstream :-)

[Andy Whitcroft (apw)]

This has been fixed by the commit below which was included in v2.6.31-rc1. This formed the base of the 2.6.31-1.13 ubuntu kernel. Therefore I am marking this Fix Released for karmic.

  commit 35a7c64fbc77bab4ca8ae477e8ab278ccd679ce2
  Author: Zhang Rui <email address hidden>
  Date: Mon Jun 22 11:31:17 2009 +0800

    ACPI: DMI to disable Vista compatibility on some Sony laptops

    Linux claims Vista compatibility to the BIOS for a number of
    reasons, but this brings hard lockup on some Sony laptops.

    Disable Vista compatibility via DMI for these laptops unless
    we can figure out what Vista is doing for this platform.
    http://bugzilla.kernel.org/show_bug.cgi?id=12904

    Signed-off-by: Zhang Rui <email address hidden>
    Signed-off-by: Len Brown <email address hidden>

[Emil Pavlov (emil-p-pavlov)]

This bug persists in Natty. I haven´t been able to test the proposed workarounds.