virtualbox causes system to lockup

Thank you for using Ubuntu and reporting a bug. It is odd that the new apparmor userspace would break virtualbox because the update only changed an abstraction to have an updated path for thunderbird and virtualbox is not confined by default in Ubuntu. Can you perform the following command which will attach various files to this bug to aid in debugging this problem:
$ apport-collect -p apparmor 1012627

Note, please perform this command after reinstalling apparmor and after your system exhibits the problem.

Attached a set of Virtualbox logfiles from today. For most recent log it was running ok without apparmor.

Attached logs from a second machine experiencing the same issue.

@Jamie: I have noticed that there should not be a change related to virtualbox (have read the changelog) and it took me a while to discover apparmor being the reason.

I will reinstall apparmor and run "apport-collect -p apparmor 1012627" in the evening - now I am pretty under stress to get my work done.

I had another lockup/freeze now after apparmor already uninstalled for a while. Several attempts to start the virtual machine resulted in lockup/freeze. I found that disabling USB support in the virtual machine solved the issue (we will see if it solved the problem for granted now).

I already thought of probably being a kernel problem, but my last kernel update is already a longer while ago (march 31) and it worked everything well without hassle since today morning. I know it seems strange. I will continue with the apparmor test as recommended in the evening - as already promised.

apport information

apport information

apport information

apport information

apport information

apport information

Maybe the problem is not apparmor itself but the following:
"ureadahead will be reprofiled on next reboot"
what is displayed for example after executing
"sudo apt-get purge apparmor"
Maybe that reprofiling is needed to make it work again...

It works again after purgin apparmor - I mentioned earlier some doubts that it may have to do (also) with usb support in the virtual machine: In my last test (when posting results of "apport-collect -p apparmor 1012627") usb support was (and still is) disabled.

It is completely reproducable that having apparmor installed causes the issue and removing it solves the problem.

Of course uninstalling apparmor is not the desired solution.

Note: The apport-collect result was collected after reboot as I couldn't either switch to a text console anymore after the freeze. Interestingly the machine still responded to a ping from another machine but an attempt to do an ssh session from remote to the machine caused the remote terminal to hang either (maybe just waiting for a timeout but I canceled the attempt after about 30 seconds).

Hope this helps to identify the issue. BTW: Thanks for the quick response to my bug report.

Thank you for your feedback. There is nothing in the logs that you submitted that would indicate this is an apparmor problem. You also mentioned that you encountered the problem when apparmor was disabled, so I am going to move the bug assignment to virtualbox instead. It could be an interaction between the kernel and the virtualbox kernel modules, but the intermittent failures make me think it might be a hardware problem. I suggest you run memtest and also examine /var/log/kern.log for any reported errors, and report back. Thanks.

Attached my kernel log.

Memtest finished without errors. So if really a hardware issue, it does not seem to be the memory.

Tested on a second machine with still Ubuntu 10.04, kernel 2.6.32-41-generic-pae #90-Ubuntu SMP Tue May 22 12:48:11 UTC 2012 i686 GNU/Linux and apparmor 2.5.1 - works - no issue with this configuration. Hardware is the same (both tested laptops have the same hardware specifications).

The second (working) machine has the following related packages installed:
apparmor
apparmor-utils
libapparmor-perl
libapparmor1

The Ubuntu 12.04 only has apparmor and dh-apparmor installed; and apparmor-utils, libapparmor-perl and libapparmor1 are missing. I wonder if those modules would create a difference. Tried installing apparmor + libapparmor1 + apparmor-utils (the latter asks a lot of perl stuff to be installed too).

I investigated dh-apparmor - that seems to be a dependency of alien and google-earth. Even when removing apparmor completely, the dh-apparmor remains and if I want to uninstall that also, it wants to uninstall alien and google-earth also.

Now the 12.04 machine has the following installed:
apparmor
apparmor-utils
dh-apparmor
libapparmor-perl
libapparmor1

First test after these changes: Virtual Machine booted ok - I was happy - too early. Second and third attempt after another reboot (to make sure) caused again the freeze. I will attach another kernel log where the freeze occurred at 1:58 and next boot at 2:03.

Here is the kern.log again. I investigated it a little - the last entries are:
Jun 14 01:57:37 wildi-torres kernel: [ 120.001052] [UFW BLOCK] IN=wlan0 OUT= MAC=00:22:fb:52:22:68:00:14:7f:ee:ac:b5:08:00 SRC=173.194.39.84 DST=10.0.0.9 LEN=52 TOS=0x00 PREC=0x00 TTL=57 ID=61755 PROTO=TCP SPT=443 DPT=58058 WINDOW=360 RES=0x00 ACK FIN URGP=0
Jun 14 01:57:38 wildi-torres kernel: [ 121.591505] [UFW BLOCK] IN=wlan0 OUT= MAC=00:22:fb:52:22:68:00:14:7f:ee:ac:b5:08:00 SRC=10.0.0.138 DST=10.0.0.9 LEN=340 TOS=0x00 PREC=0x00 TTL=64 ID=7501 PROTO=UDP SPT=1900 DPT=53691 LEN=320
Jun 14 01:57:38 wildi-torres kernel: [ 121.595531] [UFW BLOCK] IN=wlan0 OUT= MAC=00:22:fb:52:22:68:00:14:7f:ee:ac:b5:08:00 SRC=10.0.0.138 DST=10.0.0.9 LEN=340 TOS=0x00 PREC=0x00 TTL=64 ID=7502 PROTO=UDP SPT=1900 DPT=38780 LEN=320
Jun 14 01:58:31 wildi-torres kernel: [ 174.564539] device vboxnet0 entered promiscuous mode
Jun 14 01:58:41 wildi-torres kernel: [ 184.458380] [UFW BLOCK] IN=wlan0 OUT= MAC=00:22:fb:52:22:68:00:14:7f:ee:ac:b5:08:00 SRC=212.41.224.91 DST=10.0.0.9 LEN=265 TOS=0x00 PREC=0x00 TTL=55 ID=20812 DF PROTO=TCP SPT=993 DPT=40309 WINDOW=124 RES=0x00 ACK PSH URGP=0
Jun 14 01:58:42 wildi-torres kernel: [ 185.200030] vboxnet0: no IPv6 routers present

I compared settings to other 10.04 virtualbox installation and network adapter settings are the same. Anyway I am trying now with changed network settings and firewall disabled...

I tried on another machine with apparmor enabled where it worked until the updates were applied (including kernel update from 3.2.0-23-generic-pae to 23 to 253.2.0-25-generic-pae.

Then it began to happen - first once it worked, once not after changing several virtualbox settings (including USB support, disabling hostonlyadapters etc). But then happened again.

Finally I found the following ticket on virtualbox site:
https://www.virtualbox.org/ticket/10528

Now it works on both machines, so the reason is/was in the kernel and not apparmor or whatever other package.

The virtualbox guys have fixed that in SVN already. We will see, when is released. Until then, workaround to fix it is:
1. open a terminal window
2. type "sudo gedit /etc/default/grub" (without the quotes) and press enter
3. Comment out the line GRUB_CMDLINE_LINUX_DEFAULT="quiet splash" by adding a "#" character in front.
4. Add new line below: GRUB_CMDLINE_LINUX_DEFAULT="quiet splash nmi_watchdog=0 nowatchdog"
5. Save and exit
6. type "sudo update-grub" (without the quotes) and press enter
7. reboot.

Status changed to 'Confirmed' because the bug affects multiple users.

This bug was fixed in the package virtualbox - 4.1.18-dfsg-1

---------------
virtualbox (4.1.18-dfsg-1) unstable; urgency=low

  * New upstream release.
    - Fixes host freezes with 64-bit guests on 32-bit hosts. (LP: #1012627)
    - Fixes odd permissions of files in shared folders. (LP: #1013368)
  * Drop 37-fix-unregister-netdevice.patch, fixed upstream.
  * Compress orig tarball and binary packages with xz.

 -- Felix Geyer <email address hidden> Thu, 21 Jun 2012 12:31:17 +0200